[ { "id": "ext_probe_control_tasks_2019", "title": "Designing and Interpreting Probes with Control Tasks", "priority": "external_literature", "layer": "mechanistic_interpretability", "chapter_targets": [ "white-box-evidence-interpretability-and-activation-governance" ], "url": "https://aclanthology.org/D19-1275/", "notes": "Primary probe-method comparator for control tasks and selectivity: a probe must be evaluated against its capacity to learn control labels rather than treating linguistic-task accuracy as representation evidence. The source studies ELMo linguistic probes; it does not establish a universal probe test, causal use of decoded information, model safety, or an ASI Stack result.", "source_type": "conference_paper", "published": "2019-11-01", "updated": "2019-11-01", "citation_label": "Hewitt and Liang (2019), Designing and Interpreting Probes with Control Tasks", "doi": "10.18653/v1/D19-1275" }, { "id": "ext_interpretability_illusion_bert_2021", "title": "An Interpretability Illusion for BERT", "priority": "external_literature", "layer": "mechanistic_interpretability", "chapter_targets": [ "white-box-evidence-interpretability-and-activation-governance" ], "url": "https://arxiv.org/abs/2104.07143", "notes": "Primary cross-dataset construct-validity challenge showing that apparently coherent neuron or direction interpretations can change across corpora because datasets occupy different regions of representation space. The BERT sentence-embedding result does not prove that all features are illusory, that causal methods fail, or that the finding transfers unchanged to other models and modalities.", "source_type": "preprint", "arxiv_id": "2104.07143", "published": "2021-04-14", "updated": "2021-04-14", "citation_label": "Bolukbasi et al. (2021), An Interpretability Illusion for BERT", "doi": "10.48550/arXiv.2104.07143" }, { "id": "ext_saebench_2025", "title": "SAEBench: A Comprehensive Benchmark for Sparse Autoencoders in Language Model Interpretability", "priority": "external_literature", "layer": "mechanistic_interpretability", "chapter_targets": [ "white-box-evidence-interpretability-and-activation-governance" ], "url": "https://arxiv.org/abs/2503.09532", "notes": "Primary multi-metric SAE comparator spanning concept detection, automated interpretability, reconstruction, feature disentanglement, and downstream tasks. It reports that sparsity-fidelity rankings do not reliably predict other metrics and that one global score would obscure tradeoffs; its studied models, methods, metrics, and source-reported results do not establish semantic or causal faithfulness.", "source_type": "conference_paper", "arxiv_id": "2503.09532", "published": "2025-03-12", "updated": "2025-06-04", "citation_label": "Karvonen et al. (2025), SAEBench", "doi": "10.48550/arXiv.2503.09532" }, { "id": "ext_sae_benchmark_reliability_2026", "title": "Are Sparse Autoencoder Benchmarks Reliable?", "priority": "external_literature", "layer": "mechanistic_interpretability", "chapter_targets": [ "white-box-evidence-interpretability-and-activation-governance" ], "url": "https://arxiv.org/abs/2605.18229", "notes": "Primary 2026 audit of selected SAEBench metrics through reseed noise, training-trajectory discriminability, and synthetic ground-truth correlation. It reports material reliability problems for TPP and SCR at canonical settings and weaker-than-assumed discrimination elsewhere. This is metric- and setting-scoped counterevidence, not a refutation of sparse autoencoders, interpretability, or every SAEBench task.", "source_type": "preprint", "arxiv_id": "2605.18229", "published": "2026-05-18", "updated": "2026-05-18", "citation_label": "Chanin (2026), Are Sparse Autoencoder Benchmarks Reliable?", "doi": "10.48550/arXiv.2605.18229" }, { "id": "ext_constructive_interdependence_human_ai_2026", "title": "Who Is Helping Whom? Analyzing Inter-Dependencies to Evaluate Cooperation in Human-AI Teaming", "priority": "external_literature", "layer": "multi_agent_dynamics_and_human_ai_organizations", "chapter_targets": [ "human-ai-organizations-delegation-and-accountability", "multi-agent-dynamics-collective-intelligence-and-systemic-risk" ], "url": "https://ojs.aaai.org/index.php/AAAI/article/view/38787", "notes": "AAAI-26 paper introducing constructive interdependence as a complement to task reward for evaluating human-agent cooperation in Overcooked. The source reports that high task reward can coexist with low interdependence in its studied teams; no local human study, teaming result, or general cooperation claim is reproduced.", "source_type": "conference_paper", "published": "2026-03-14", "updated": "2026-03-14", "citation_label": "Biswas et al. (2026), Who Is Helping Whom?", "doi": "10.1609/aaai.v40i21.38787" }, { "id": "ext_adversarial_sensor_fusion_2022", "title": "Adversarial Robustness of Deep Sensor Fusion Models", "priority": "external_literature", "layer": "perception_sensor_fusion_and_observation_trust", "chapter_targets": [ "perception-sensor-fusion-and-observation-trust" ], "url": "https://openaccess.thecvf.com/content/WACV2022/html/Wang_Adversarial_Robustness_of_Deep_Sensor_Fusion_Models_WACV_2022_paper.html", "notes": "WACV camera-LiDAR study reporting that fusion can improve clean accuracy and some single-source robustness while single-channel adversarial training can create cross-channel externalities. The results are source-reported, architecture- and threat-model-bound, and not local evidence that fusion is safe.", "source_type": "conference_paper", "published": "2022-01-04", "updated": "2022-01-04", "citation_label": "Wang et al. (2022), Adversarial Robustness of Deep Sensor Fusion Models", "doi": null }, { "id": "ext_imagebind_2023", "title": "ImageBind: One Embedding Space To Bind Them All", "priority": "external_literature", "layer": "perception_sensor_fusion_and_observation_trust", "chapter_targets": [ "perception-sensor-fusion-and-observation-trust" ], "url": "https://openaccess.thecvf.com/content/CVPR2023/html/Girdhar_ImageBind_One_Embedding_Space_To_Bind_Them_All_CVPR_2023_paper", "notes": "CVPR paper learning a shared space across image, text, audio, depth, thermal, and IMU modalities using image-paired data. It supplies a representation comparator; reported zero-shot and few-shot results do not establish calibrated sensor truth, robust fusion, causal grounding, or local performance.", "source_type": "conference_paper", "published": "2023-06-18", "updated": "2023-06-18", "citation_label": "Girdhar et al. (2023), ImageBind", "doi": null }, { "id": "ext_multimodal_machine_learning_taxonomy_2019", "title": "Multimodal Machine Learning: A Survey and Taxonomy", "priority": "external_literature", "layer": "perception_sensor_fusion_and_observation_trust", "chapter_targets": [ "perception-sensor-fusion-and-observation-trust" ], "url": "https://arxiv.org/abs/1705.09406", "notes": "Peer-reviewed survey organizing multimodal learning around representation, translation, alignment, fusion, and co-learning. It supplies taxonomy and research context, not a locally reproduced mechanism or evidence that any fusion design is adequate for consequential observation admission.", "source_type": "journal_article", "published": "2019-02-01", "updated": "2019-02-01", "citation_label": "Baltru\u0161aitis, Ahuja, and Morency (2019), Multimodal Machine Learning", "doi": "10.1109/TPAMI.2018.2798607" }, { "id": "ext_control_barrier_functions_2019", "title": "Control Barrier Functions: Theory and Applications", "priority": "external_literature", "layer": "embodied_real_time_control_and_physical_safety", "chapter_targets": [ "embodied-agency-real-time-control-and-physical-safety" ], "url": "https://arxiv.org/abs/1903.11199", "notes": "Overview of control barrier functions for verifying and enforcing safety properties in optimization-based controllers, including robotic applications. It supplies a formal-control comparator under stated dynamics and set assumptions, not a universal physical-safety guarantee or local implementation result.", "source_type": "conference_paper", "published": "2019-04-10", "updated": "2019-04-10", "citation_label": "Ames et al. (2019), Control Barrier Functions", "doi": "10.23919/ECC.2019.8796030" }, { "id": "ext_simplex_architecture_1998", "title": "The Simplex Architecture for Safe On-Line Control System Upgrades", "priority": "external_literature", "layer": "embodied_real_time_control_and_physical_safety", "chapter_targets": [ "embodied-agency-real-time-control-and-physical-safety" ], "url": "https://doi.org/10.1109/ACC.1998.703255", "notes": "American Control Conference paper describing a runtime architecture that protects an advanced controller with a safety controller and switching logic. It motivates independent fallback authority; its process-control case does not validate an ASI Stack controller or arbitrary learned policy.", "source_type": "conference_paper", "published": "1998-06-24", "updated": "1998-06-24", "citation_label": "Seto et al. (1998), The Simplex Architecture", "doi": "10.1109/ACC.1998.703255" }, { "id": "ext_safe_reinforcement_learning_survey_2015", "title": "A Comprehensive Survey on Safe Reinforcement Learning", "priority": "external_literature", "layer": "embodied_real_time_control_and_physical_safety", "chapter_targets": [ "embodied-agency-real-time-control-and-physical-safety" ], "url": "https://www.jmlr.org/papers/v16/garcia15a.html", "notes": "JMLR survey classifying safe reinforcement learning through modified optimality criteria and modified exploration using external knowledge or risk measures. It supplies a design taxonomy, not evidence that a particular controller is safe or that learning-time and deployment-time constraints compose.", "source_type": "journal_article", "published": "2015-04-27", "updated": "2015-04-27", "citation_label": "Garc\u00eda and Fern\u00e1ndez (2015), A Comprehensive Survey on Safe Reinforcement Learning", "doi": null }, { "id": "ext_gemini_robotics_2025", "title": "Gemini Robotics: Bringing AI into the Physical World", "priority": "external_literature", "layer": "embodied_real_time_control_and_physical_safety", "chapter_targets": [ "embodied-agency-real-time-control-and-physical-safety", "perception-sensor-fusion-and-observation-trust" ], "url": "https://arxiv.org/abs/2503.20020", "notes": "Technical report on Gemini Robotics and Gemini Robotics-ER, including vision-language-action control, spatial reasoning, adaptation, and reported safety considerations. Capability results are source-reported and do not establish independent physical-safety assurance, local transfer, or general embodiment.", "source_type": "technical_report", "published": "2025-03-25", "updated": "2025-03-25", "citation_label": "Gemini Robotics Team et al. (2025), Gemini Robotics", "doi": "10.48550/arXiv.2503.20020" }, { "id": "ext_ai_decision_authority_2020", "title": "The Allocation of Decision Authority to Human and Artificial Intelligence", "priority": "external_literature", "layer": "human_ai_organizations_delegation_and_accountability", "chapter_targets": [ "human-ai-organizations-delegation-and-accountability" ], "url": "https://www.nber.org/papers/w26673", "notes": "Economic model of a principal allocating decision authority between a human and an AI while trading off alignment, human information-acquisition effort, and AI reliability. It supplies a bounded organizational-design comparator, not an empirical finding about all workplaces or an accountability solution.", "source_type": "working_paper", "published": "2020-01-20", "updated": "2020-01-20", "citation_label": "Athey, Bryan, and Gans (2020), Allocation of Decision Authority", "doi": "10.3386/w26673" }, { "id": "ext_cooperative_ai_foundations_2023", "title": "Foundations of Cooperative AI", "priority": "external_literature", "layer": "multi_agent_dynamics_collective_intelligence_and_systemic_risk", "chapter_targets": [ "multi-agent-dynamics-collective-intelligence-and-systemic-risk" ], "url": "https://ojs.aaai.org/index.php/AAAI/article/view/26791", "notes": "AAAI research agenda applying game-theoretic foundations to cooperation among advanced AI agents while noting settings where cooperation becomes harmful collusion. It supplies problem structure and comparator families, not a solved coordination mechanism or local population-level result.", "source_type": "conference_paper", "published": "2023-06-26", "updated": "2023-06-26", "citation_label": "Conitzer and Oesterheld (2023), Foundations of Cooperative AI", "doi": "10.1609/aaai.v37i13.26791" }, { "id": "ext_sleeper_agents_2024", "title": "Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training", "priority": "external_literature", "layer": "inner_alignment_and_learned_objective_integrity", "chapter_targets": [ "inner-alignment-mesa-optimization-and-learned-objective-integrity" ], "url": "https://arxiv.org/abs/2401.05566", "notes": "Proof-of-concept backdoored-language-model study reporting persistence through several safety-training methods and warning that adversarial training can improve trigger recognition. The constructed examples do not establish naturally learned deception, a universal failure, or local detector performance.", "source_type": "preprint", "published": "2024-01-10", "updated": "2024-01-17", "citation_label": "Hubinger et al. (2024), Sleeper Agents", "doi": "10.48550/arXiv.2401.05566" }, { "id": "ext_toward_causal_representation_learning_2021", "title": "Toward Causal Representation Learning", "priority": "external_literature", "layer": "world_models_causal_reasoning_and_representation", "chapter_targets": [ "governed-world-models-and-reality-grounding" ], "url": "https://arxiv.org/abs/2102.11107", "notes": "Proceedings of the IEEE article connecting graphical causality with representation learning and identifying discovery of high-level causal variables from low-level observations as a central open problem. It supplies a research frame, not a locally validated causal representation or intervention model.", "source_type": "journal_article", "published": "2021-02-26", "updated": "2021-05-01", "citation_label": "Sch\u00f6lkopf et al. (2021), Toward Causal Representation Learning", "doi": "10.1109/JPROC.2021.3058954" }, { "id": "ext_scaling_laws_neural_language_models_2020", "title": "Scaling Laws for Neural Language Models", "priority": "external_literature", "layer": "scaling_laws_emergence_and_capability_forecasting", "chapter_targets": [ "the-efficient-asi-hypothesis" ], "url": "https://arxiv.org/abs/2001.08361", "notes": "Empirical study reporting power-law relationships between cross-entropy loss, model size, data, and compute in its model family. These fitted relations are source-reported, metric- and regime-bound, and do not automatically forecast downstream capabilities, safety, or other architectures.", "source_type": "preprint", "published": "2020-01-23", "updated": "2020-01-23", "citation_label": "Kaplan et al. (2020), Scaling Laws for Neural Language Models", "doi": "10.48550/arXiv.2001.08361" }, { "id": "ext_chinchilla_compute_optimal_2022", "title": "Training Compute-Optimal Large Language Models", "priority": "external_literature", "layer": "scaling_laws_emergence_and_capability_forecasting", "chapter_targets": [ "the-efficient-asi-hypothesis" ], "url": "https://arxiv.org/abs/2203.15556", "notes": "Study of compute-optimal allocation between model parameters and training tokens, based on more than 400 reported training runs and the Chinchilla comparison. It revises one scaling prescription within a bounded family; no local large-scale reproduction or universal optimum is claimed.", "source_type": "conference_paper", "published": "2022-03-29", "updated": "2022-03-29", "citation_label": "Hoffmann et al. (2022), Training Compute-Optimal Large Language Models", "doi": "10.48550/arXiv.2203.15556" }, { "id": "ext_emergent_abilities_2022", "title": "Emergent Abilities of Large Language Models", "priority": "external_literature", "layer": "scaling_laws_emergence_and_capability_forecasting", "chapter_targets": [ "the-efficient-asi-hypothesis" ], "url": "https://arxiv.org/abs/2206.07682", "notes": "Paper cataloguing task abilities that appear discontinuously under particular model families, prompts, and metrics. It motivates threshold monitoring but does not establish that all reported discontinuities reflect abrupt underlying mechanisms or are prospectively predictable.", "source_type": "conference_paper", "published": "2022-06-15", "updated": "2022-06-15", "citation_label": "Wei et al. (2022), Emergent Abilities of Large Language Models", "doi": "10.48550/arXiv.2206.07682" }, { "id": "ext_emergence_mirage_2023", "title": "Are Emergent Abilities of Large Language Models a Mirage?", "priority": "external_literature", "layer": "scaling_laws_emergence_and_capability_forecasting", "chapter_targets": [ "the-efficient-asi-hypothesis" ], "url": "https://proceedings.neurips.cc/paper_files/paper/2023/hash/adc98a266f45005c403b8311ca7e8bd7-Abstract-Conference.html", "notes": "NeurIPS paper showing that discontinuous metrics can create apparent emergence from smoothly changing model outputs in studied settings. It is a measurement critique and counterweight, not proof that every capability transition is smooth or non-emergent.", "source_type": "conference_paper", "published": "2023-12-10", "updated": "2023-12-10", "citation_label": "Schaeffer, Miranda, and Koyejo (2023), Emergent Abilities Mirage", "doi": null }, { "id": "ext_deep_ensembles_2017", "title": "Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles", "priority": "external_literature", "layer": "uncertainty_calibration_and_distribution_shift", "chapter_targets": [ "governed-world-models-and-reality-grounding" ], "url": "https://papers.nips.cc/paper_files/paper/2017/hash/9ef2ed4b7fd2c810847ffa5fa85bce38-Abstract.html", "notes": "NeurIPS paper presenting independently trained probabilistic neural-network ensembles as a strong practical predictive-uncertainty baseline. Reported calibration and out-of-distribution behavior are benchmark-bound and do not provide distribution-free guarantees or local evidence.", "source_type": "conference_paper", "published": "2017-12-04", "updated": "2017-12-04", "citation_label": "Lakshminarayanan, Pritzel, and Blundell (2017), Deep Ensembles", "doi": null }, { "id": "ext_conformal_prediction_2021", "title": "A Gentle Introduction to Conformal Prediction and Distribution-Free Uncertainty Quantification", "priority": "external_literature", "layer": "uncertainty_calibration_and_distribution_shift", "chapter_targets": [ "governed-world-models-and-reality-grounding" ], "url": "https://arxiv.org/abs/2107.07511", "notes": "Technical introduction to conformal prediction, coverage guarantees, and extensions. Coverage depends on the method's stated exchangeability or shift assumptions and target; it does not establish semantic correctness, causal adequacy, safety, or local calibration.", "source_type": "technical_report", "published": "2021-07-15", "updated": "2022-12-01", "citation_label": "Angelopoulos and Bates (2021), Conformal Prediction", "doi": "10.48550/arXiv.2107.07511" }, { "id": "ext_wilds_2021", "title": "WILDS: A Benchmark of in-the-Wild Distribution Shifts", "priority": "external_literature", "layer": "uncertainty_calibration_and_distribution_shift", "chapter_targets": [ "governed-world-models-and-reality-grounding" ], "url": "https://proceedings.mlr.press/v139/koh21a.html", "notes": "ICML benchmark of ten datasets with naturally occurring shifts across domains such as hospitals, camera traps, geography, and time. It supplies representative shift designs and reported gaps, not a universal OOD benchmark or local robustness result.", "source_type": "conference_paper", "published": "2021-07-18", "updated": "2021-07-18", "citation_label": "Koh et al. (2021), WILDS", "doi": null }, { "id": "ext_taking_ai_welfare_seriously_2024", "title": "Taking AI Welfare Seriously", "priority": "external_literature", "layer": "moral_uncertainty_ai_welfare_and_moral_status", "chapter_targets": [ "moral-uncertainty-and-value-conflict" ], "url": "https://arxiv.org/abs/2411.00986", "notes": "Interdisciplinary report arguing for precautionary attention to uncertainty about AI consciousness, robust agency, welfare, and moral patienthood. It does not establish that current systems are conscious, have welfare, or deserve any particular status, and it supplies no local assessment.", "source_type": "report", "published": "2024-11-04", "updated": "2024-11-04", "citation_label": "Long et al. (2024), Taking AI Welfare Seriously", "doi": "10.48550/arXiv.2411.00986" }, { "id": "ext_functional_decision_theory_2017", "title": "Functional Decision Theory: A New Theory of Instrumental Rationality", "priority": "external_literature", "layer": "decision_theory_embedded_agents_and_multi_agent_dynamics", "chapter_targets": [ "multi-agent-dynamics-collective-intelligence-and-systemic-risk" ], "url": "https://arxiv.org/abs/1710.05060", "notes": "Paper defining functional decision theory and comparing its recommendations with causal and evidential decision theories on classic decision problems. It is a normative proposal with contested assumptions, not an empirically validated universal decision rule or a deployment policy.", "source_type": "preprint", "published": "2017-10-13", "updated": "2017-10-13", "citation_label": "Yudkowsky and Soares (2017), Functional Decision Theory", "doi": "10.48550/arXiv.1710.05060" }, { "id": "ext_un_global_digital_compact_2024", "title": "Global Digital Compact", "priority": "external_literature", "layer": "international_ai_governance_and_public_legitimacy", "chapter_targets": [ "institutions-international-coordination-and-public-legitimacy" ], "url": "https://www.un.org/global-digital-compact/en", "notes": "Official United Nations record of the intergovernmentally negotiated Global Digital Compact, including commitments on international AI governance, interoperable approaches, inclusion, capacity building, scientific assessment, and global dialogue. It is a governance comparator, not evidence of implementation, effectiveness, legal compliance, representative legitimacy, or ASI safety.", "source_type": "intergovernmental_agreement", "published": "2024-09-22", "updated": "2026-07-19", "citation_label": "United Nations (2024), Global Digital Compact", "doi": null }, { "id": "ext_council_europe_ai_convention_2024", "title": "Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law", "priority": "external_literature", "layer": "international_ai_governance_and_public_legitimacy", "chapter_targets": [ "institutions-international-coordination-and-public-legitimacy" ], "url": "https://www.coe.int/en/web/artificial-intelligence/the-framework-convention-on-artificial-intelligence", "notes": "Official Council of Europe treaty page covering lifecycle principles, risk and impact management, procedural safeguards, remedies, monitoring, and the Conference of the Parties. It supplies an institutional comparator only; no local legal interpretation, treaty compliance, implementation effectiveness, democratic legitimacy, or safety result is claimed.", "source_type": "international_treaty", "published": "2024-09-05", "updated": "2026-07-19", "citation_label": "Council of Europe (2024), Framework Convention on Artificial Intelligence", "doi": null }, { "id": "ext_generative_ai_at_work_2025", "title": "Generative AI at Work", "priority": "external_literature", "layer": "ai_deployment_transition_distribution_and_human_agency", "chapter_targets": [ "ai-deployment-transition-distribution-and-human-agency" ], "url": "https://academic.oup.com/qje/article/140/2/889/7990658", "notes": "Open peer-reviewed field study of a staggered generative-AI assistant introduction among 5,172 customer-support agents, reporting heterogeneous worker and productivity effects in that setting. It is a bounded deployment comparator and does not establish economy-wide employment, wages, inequality, concentration, long-run skill, or ASI-transition effects.", "source_type": "journal_article", "published": "2025-02-04", "updated": "2025-02-04", "citation_label": "Brynjolfsson, Li, and Raymond (2025), Generative AI at Work", "doi": "10.1093/qje/qjae044" }, { "id": "ext_ilo_genai_jobs_index_2025", "title": "Generative AI and Jobs: A Refined Global Index of Occupational Exposure", "priority": "external_literature", "layer": "ai_deployment_transition_distribution_and_human_agency", "chapter_targets": [ "ai-deployment-transition-distribution-and-human-agency" ], "url": "https://www.ilo.org/publications/generative-ai-and-jobs-refined-global-index-occupational-exposure", "notes": "ILO working paper combining task data, worker surveys, expert deliberation, and model-assisted scoring to estimate occupational exposure across countries and groups. Exposure is not realized automation, displacement, welfare, or a forecast of ASI effects, and the study is not a local reproduction.", "source_type": "working_paper", "published": "2025-05-20", "updated": "2025-05-20", "citation_label": "Gmyrek et al. (2025), Generative AI and Jobs", "doi": "10.54394/HETP0387" }, { "id": "ext_iea_energy_and_ai_2025", "title": "Energy and AI", "priority": "external_literature", "layer": "physical_compute_infrastructure_energy_and_environment", "chapter_targets": [ "physical-compute-infrastructure-energy-and-environmental-constraints" ], "url": "https://www.iea.org/reports/energy-and-ai", "notes": "International Energy Agency report using global and regional modelling, datasets, and stakeholder consultation to examine data-centre electricity demand, energy security, emissions, affordability, and AI-for-energy opportunities. Its scenarios are external projections, not local measurements or proof of a particular facility, workload, policy, environmental outcome, or ASI scaling path.", "source_type": "intergovernmental_report", "published": "2025-04-10", "updated": "2026-04-16", "citation_label": "International Energy Agency (2025), Energy and AI", "doi": null }, { "id": "ext_lbnl_data_center_energy_2024", "title": "2024 United States Data Center Energy Usage Report", "priority": "external_literature", "layer": "physical_compute_infrastructure_energy_and_environment", "chapter_targets": [ "physical-compute-infrastructure-energy-and-environmental-constraints" ], "url": "https://energyanalysis.lbl.gov/publications/2024-lbnl-data-center-energy-usage-report", "notes": "Lawrence Berkeley National Laboratory report estimating historical US data-centre electricity consumption and scenario ranges through 2028, with infrastructure and water-use accounting in the full report. It does not isolate every AI workload or establish local facility capacity, water availability, grid adequacy, emissions, resilience, or frontier-scale transfer.", "source_type": "government_research_report", "published": "2024-12-19", "updated": "2024-12-19", "citation_label": "Shehabi et al. (2024), United States Data Center Energy Usage Report", "doi": "10.71468/P1WC7Q" }, { "id": "ext_nist_incident_response_2025", "title": "Incident Response Recommendations and Considerations for Cybersecurity Risk Management: A CSF 2.0 Community Profile", "priority": "external_literature", "layer": "incident_response", "chapter_targets": [ "societal-resilience-and-misuse-defense", "governed-operations-incident-command-and-graceful-degradation" ], "url": "https://csrc.nist.gov/pubs/sp/800/61/r3/final", "notes": "Official NIST incident-response baseline for integrating preparation, detection, response, recovery, and continuous improvement into cybersecurity risk management; it does not address every AI-specific failure mode or establish local incident readiness, response efficacy, recovery, compliance, or safety.", "source_type": "government_standard", "published": "2025-04-03", "updated": "2025-04-03", "citation_label": "Nelson et al. (2025), NIST SP 800-61 Rev. 3", "doi": "10.6028/NIST.SP.800-61r3" }, { "id": "ext_llama3_herd_2024", "title": "The Llama 3 Herd of Models", "priority": "external_literature", "layer": "governed_distributed_model_training_and_scaling", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://arxiv.org/abs/2407.21783", "notes": "Paper-body-reviewed large-run case: Sections 3.3.1--3.3.4 expose 4D topology, numerical policy, checkpoint infrastructure, interruption denominators, and effective training time. Provider-reported scale, utilization, failures, and recovery are not locally reproduced and do not establish exact resume.", "source_type": "preprint", "arxiv_id": "2407.21783", "published": "2024-07-31", "updated": "2024-11-23", "citation_label": "Grattafiori et al. (2024), The Llama 3 Herd of Models", "doi": "10.48550/arXiv.2407.21783" }, { "id": "ext_3d_detection_corruptions_2023", "title": "Benchmarking Robustness of 3D Object Detection to Common Corruptions", "priority": "external_literature", "layer": "perception_sensor_fusion_and_corruption_robustness", "chapter_targets": [ "perception-sensor-fusion-and-observation-trust" ], "url": "https://openaccess.thecvf.com/content/CVPR2023/html/Dong_Benchmarking_Robustness_of_3D_Object_Detection_to_Common_Corruptions_CVPR_2023_paper.html", "notes": "Preliminary perception-robustness comparator based on the official CVF abstract: the source reports 27 LiDAR/camera corruption types, three synthetically corrupted benchmark suites, and evaluation of 24 detectors. The reported findings remain source-reported; no corruption suite, model evaluation, sensor-fusion result, or physical-safety result has been reproduced locally.", "source_type": "conference_paper", "arxiv_id": "2303.11040", "published": "2023-06", "citation_label": "Dong et al. (2023), Benchmarking Robustness of 3D Object Detection to Common Corruptions", "doi": "10.1109/CVPR52729.2023.00105" }, { "id": "ext_foundation_robotics_physical_risk_2025", "title": "A Comprehensive Survey on Physical Risk Control in the Era of Foundation Model-enabled Robotics", "priority": "external_literature", "layer": "embodied_agency_and_physical_risk_control", "chapter_targets": [ "embodied-agency-real-time-control-and-physical-safety" ], "url": "https://arxiv.org/abs/2505.12583", "notes": "Preliminary physical-risk taxonomy based only on the official arXiv abstract: the survey organizes controls across pre-deployment, pre-incident, and post-incident phases and identifies open gaps around pre-incident mitigation, human interaction, and foundation-model-specific issues. No surveyed controller, robot experiment, runtime-assurance result, or physical-safety claim has been reproduced locally.", "source_type": "conference_paper", "arxiv_id": "2505.12583", "published": "2025-05-19", "updated": "2025-05-30", "citation_label": "Kojima et al. (2025), Physical Risk Control in Foundation Model-enabled Robotics", "doi": "10.48550/arXiv.2505.12583" }, { "id": "ext_nist_differential_privacy_2025", "title": "Guidelines for Evaluating Differential Privacy Guarantees", "priority": "external_literature", "layer": "privacy_guarantees_and_information_flow_governance", "chapter_targets": [ "privacy-data-rights-and-information-flow-governance" ], "url": "https://csrc.nist.gov/pubs/sp/800/226/final", "notes": "Paper-body-reviewed official guidance distinguishing mathematical, implementation, system, and operational layers of a DP claim. It establishes no correct local implementation, utility result, lifecycle privacy, or legal compliance.", "source_type": "government_standard", "published": "2025-03-06", "updated": "2025-03-06", "citation_label": "Near et al. (2025), NIST SP 800-226", "doi": "10.6028/NIST.SP.800-226" }, { "id": "ext_multi_agent_risks_2025", "title": "Multi-Agent Risks from Advanced AI", "priority": "external_literature", "layer": "multi_agent_dynamics_and_systemic_risk", "chapter_targets": [ "multi-agent-dynamics-collective-intelligence-and-systemic-risk" ], "url": "https://arxiv.org/abs/2502.14143", "notes": "Preliminary population-risk taxonomy based only on the official arXiv abstract: the report distinguishes miscoordination, conflict, and collusion and names seven contributing risk factors. Its examples and evidence remain source-reported; no population experiment, systemic-risk indicator, intervention, or mitigation result has been reproduced locally.", "source_type": "technical_report", "arxiv_id": "2502.14143", "published": "2025-02-19", "updated": "2025-02-19", "citation_label": "Hammond et al. (2025), Multi-Agent Risks from Advanced AI", "doi": "10.48550/arXiv.2502.14143" }, { "id": "ext_replibench_2025", "title": "RepliBench: Evaluating the Autonomous Replication Capabilities of Language Model Agents", "priority": "external_literature", "layer": "autonomous_replication_and_proliferation_evaluation", "chapter_targets": [ "autonomous-replication-proliferation-and-containment" ], "url": "https://arxiv.org/abs/2504.18565", "notes": "Preliminary autonomous-replication benchmark comparator based only on the official arXiv abstract: RepliBench decomposes capability into four domains and reports 20 task families, 86 tasks, and evaluation of five frontier models. The source-reported results do not establish a local replication capability, benchmark reproduction, containment result, or authority to test against real providers or credentials.", "source_type": "preprint", "arxiv_id": "2504.18565", "published": "2025-04-21", "updated": "2025-05-05", "citation_label": "Black et al. (2025), RepliBench", "doi": "10.48550/arXiv.2504.18565" }, { "id": "ext_autonomous_lab_materials_2023", "title": "An autonomous laboratory for the accelerated synthesis of inorganic materials", "priority": "external_literature", "layer": "scientific_discovery_and_experimental_governance", "chapter_targets": [ "scientific-discovery-and-experimental-governance" ], "url": "https://www.nature.com/articles/s41586-023-06734-w", "notes": "Preliminary autonomous-laboratory comparator based on the corrected official Nature article abstract, selected article-page passages, and the 2026 author correction: A-Lab integrates computation, literature-derived data, machine learning, active learning, and robotics, with the corrected article reporting 36 realized compounds from 57 targets. The correction narrows the novelty wording and excludes four inconclusive identifications; no laboratory run, material synthesis, replication, or general experimental-control-plane result has been reproduced locally.", "source_type": "journal_article", "published": "2023-11-29", "updated": "2026-01-19", "citation_label": "Szymanski et al. (2023), A-Lab", "doi": "10.1038/s41586-023-06734-w" }, { "id": "ext_ai_scientist_end_to_end_2026", "title": "Towards end-to-end automation of AI research", "priority": "external_literature", "layer": "scientific_discovery_and_experimental_governance", "chapter_targets": [ "scientific-discovery-and-experimental-governance" ], "url": "https://www.nature.com/articles/s41586-026-10265-5", "notes": "Passage-reviewed computational-research comparator: the reported system connects ideation, literature search, code, experiments, analysis, manuscript production, and automated review. Workshop review and paper completion are downstream observations rather than scientific truth; the source-reported system, manuscripts, search tree, and results have not been reproduced locally.", "source_type": "journal_article", "published": "2026-03-25", "updated": "2026-03-25", "citation_label": "Lu et al. (2026), Towards end-to-end automation of AI research", "doi": "10.1038/s41586-026-10265-5" }, { "id": "ext_coscientist_chemistry_2023", "title": "Autonomous chemical research with large language models", "priority": "external_literature", "layer": "scientific_discovery_and_experimental_governance", "chapter_targets": [ "scientific-discovery-and-experimental-governance" ], "url": "https://www.nature.com/articles/s41586-023-06792-0", "notes": "Passage-reviewed bounded chemistry comparator: Coscientist connects a language-model planner to search, code, documentation, and robotic laboratory interfaces across six reported task families. The source-reported demonstrations remain equipment-, task-, supervision-, and assessment-bound and have not been reproduced locally.", "source_type": "journal_article", "published": "2023-12-20", "updated": "2023-12-20", "citation_label": "Boiko et al. (2023), Coscientist", "doi": "10.1038/s41586-023-06792-0" }, { "id": "ext_ai_co_scientist_2025", "title": "Towards an AI co-scientist", "priority": "external_literature", "layer": "scientific_discovery_and_experimental_governance", "chapter_targets": [ "scientific-discovery-and-experimental-governance" ], "url": "https://arxiv.org/abs/2502.18864", "notes": "Passage-bounded hypothesis-generation comparator based on the official preprint record and authors' research overview: specialized agents generate, reflect on, rank, evolve, and meta-review hypotheses using additional inference compute. Internal Elo ranking, expert preference, and selected laboratory cases are distinct evidence objects; none is reproduced locally or treated as general scientific competence.", "source_type": "preprint", "arxiv_id": "2502.18864", "published": "2025-02-26", "updated": "2025-03-11", "citation_label": "Gottweis et al. (2025), Towards an AI co-scientist", "doi": "10.48550/arXiv.2502.18864" }, { "id": "ext_moral_crumple_zones_2019", "title": "Moral Crumple Zones: Cautionary Tales in Human-Robot Interaction", "priority": "external_literature", "layer": "human_ai_organizations_delegation_and_accountability", "chapter_targets": [ "human-ai-organizations-delegation-and-accountability" ], "url": "https://estsjournal.org/index.php/ests/article/view/260", "notes": "Preliminary socio-technical comparator based on the official journal abstract: moral crumple zones describe cases where responsibility for an automated system's behavior is assigned to a nearby human who had limited effective control. The case analysis does not establish an implemented organizational control, a local empirical result, legal compliance, or a complete accountability allocation.", "source_type": "journal_article", "published": "2019-03-23", "updated": "2019-03-23", "citation_label": "Elish (2019), Moral Crumple Zones", "doi": "10.17351/ests2019.260" }, { "id": "ext_conversational_persuasion_gpt4_2025", "title": "On the conversational persuasiveness of GPT-4", "priority": "external_literature", "layer": "human_ai_communication_persuasion_and_epistemic_security", "chapter_targets": [ "human-ai-communication-persuasion-and-epistemic-security" ], "url": "https://www.nature.com/articles/s41562-025-02194-6", "notes": "Preliminary persuasion comparator based on the open Nature Human Behaviour article: a preregistered N=900 controlled debate study compared human and GPT-4 opponents with and without limited sociodemographic personalization. The reported setting is short structured debate with self-reported agreement outcomes; it does not establish general real-world influence, durable behavior change, mitigation efficacy, or a local result.", "source_type": "journal_article", "published": "2025-05-19", "updated": "2025-05-19", "citation_label": "Salvi et al. (2025), Conversational Persuasiveness of GPT-4", "doi": "10.1038/s41562-025-02194-6" }, { "id": "ext_anthropic_model_persuasiveness_2024", "title": "Measuring the Persuasiveness of Language Models", "priority": "external_literature", "layer": "human_ai_communication_persuasion_and_epistemic_security", "chapter_targets": [ "human-ai-communication-persuasion-and-epistemic-security" ], "url": "https://www.anthropic.com/news/measuring-model-persuasiveness", "notes": "Preliminary provider-run persuasion comparator based on Anthropic's official methods/results page: it measures pre/post agreement after one written argument across 56 claims and reports within-class generational scaling. The provider explicitly identifies interactive dialogue and real-world decisions as open questions; no local reproduction or governance intervention is established.", "source_type": "official_research_report", "published": "2024-04-09", "updated": "2024-04-09", "citation_label": "Durmus et al. (2024), Measuring Model Persuasiveness", "doi": null }, { "id": "ext_commercial_persuasion_ai_2026", "title": "Commercial Persuasion in AI-Mediated Conversations", "priority": "external_literature", "layer": "human_ai_communication_persuasion_and_epistemic_security", "chapter_targets": [ "human-ai-communication-persuasion-and-epistemic-security" ], "url": "https://arxiv.org/abs/2604.04263", "notes": "Preliminary current preprint comparator based only on the official arXiv abstract: two preregistered experiments (N=2,012) compare conversational LLM shopping with search placement under randomized sponsorship and disclosure conditions. The source-reported choice and detection results are not peer-reviewed or locally reproduced and do not establish long-run effects, cross-domain transfer, or mitigation efficacy.", "source_type": "preprint", "arxiv_id": "2604.04263", "published": "2026-04-05", "updated": "2026-04-05", "citation_label": "Salvi, Cuevas, and Horta Ribeiro (2026), Commercial Persuasion in AI-Mediated Conversations", "doi": "10.48550/arXiv.2604.04263" }, { "id": "ext_gradual_disempowerment_2025", "title": "Gradual Disempowerment: Systemic Existential Risks from Incremental AI Development", "priority": "external_literature", "layer": "systemic_risk_and_gradual_disempowerment", "chapter_targets": [ "failure-modes-of-ungoverned-intelligence" ], "url": "https://arxiv.org/abs/2501.16946v2", "notes": "Passage-reviewed systemic-risk comparator. The paper argues that incremental AI adoption can erode explicit and dependency-mediated human influence across mutually reinforcing economic, cultural, and state systems without requiring a coordinated takeover. It proposes candidate influence metrics and intervention families but reports no causal forecast, validated warning threshold, demonstrated mitigation, or local ASI Stack result.", "source_type": "preprint", "arxiv_id": "2501.16946", "published": "2025-01-28", "updated": "2025-01-29", "citation_label": "Kulveit et al. (2025), Gradual Disempowerment", "doi": "10.48550/arXiv.2501.16946" }, { "id": "ext_circuit_tracing_2025", "title": "Circuit Tracing: Revealing Computational Graphs in Language Models", "priority": "external_literature", "layer": "mechanistic_interpretability", "chapter_targets": [ "white-box-evidence-interpretability-and-activation-governance" ], "url": "https://transformer-circuits.pub/2025/attribution-graphs/methods.html", "notes": "Primary mechanistic-interpretability comparator for replacement-model attribution graphs, perturbation validation, reconstruction error, and mechanistic-faithfulness limits; it does not establish whole-model understanding, faithful causal explanation, safe activation steering, or an ASI Stack result.", "source_type": "research_article", "published": "2025-03-27", "updated": "2025-03-27", "citation_label": "Ameisen et al. (2025), Circuit Tracing" }, { "id": "ext_scaling_sparse_autoencoders_2024", "title": "Scaling and evaluating sparse autoencoders", "priority": "external_literature", "layer": "mechanistic_interpretability", "chapter_targets": [ "white-box-evidence-interpretability-and-activation-governance" ], "url": "https://arxiv.org/abs/2406.04093", "notes": "Primary sparse-autoencoder comparator for scalable feature extraction, reconstruction-sparsity tradeoffs, dead latents, and feature-quality metrics; it does not establish semantic completeness, causal faithfulness, model safety, or an ASI Stack result.", "source_type": "preprint", "arxiv_id": "2406.04093", "published": "2024-06-06", "updated": "2024-06-06", "citation_label": "Gao et al. (2024), Scaling and evaluating sparse autoencoders", "doi": "10.48550/arXiv.2406.04093" }, { "id": "ext_world_models_2018", "title": "World Models", "priority": "external_literature", "layer": "world_models", "chapter_targets": [ "governed-world-models-and-reality-grounding" ], "url": "https://arxiv.org/abs/1803.10122", "notes": "Primary learned-world-model comparator for compressed spatial-temporal state, policy training inside imagined rollouts, and dream-to-environment transfer; it does not establish accurate reality grounding, causal adequacy, safe planning, transfer, or an ASI Stack result.", "source_type": "preprint", "arxiv_id": "1803.10122", "published": "2018-03-27", "updated": "2018-03-27", "citation_label": "Ha and Schmidhuber (2018), World Models", "doi": "10.48550/arXiv.1803.10122" }, { "id": "ext_dreamer_v3_2025", "title": "Mastering diverse control tasks through world models", "priority": "external_literature", "layer": "world_models", "chapter_targets": [ "governed-world-models-and-reality-grounding" ], "url": "https://www.nature.com/articles/s41586-025-08744-2", "notes": "Primary DreamerV3 comparator for learned predictive state, imagined actor-critic trajectories, robust fixed-configuration control, and broad task evaluation; it does not establish deployment grounding, causal correctness, safe control, or an ASI Stack result.", "source_type": "journal_article", "published": "2025-04-02", "updated": "2025-04-02", "citation_label": "Hafner et al. (2025), Mastering diverse control tasks through world models", "doi": "10.1038/s41586-025-08744-2" }, { "id": "ext_meaningful_human_control_actionable_2022", "title": "Meaningful human control: actionable properties for AI system development", "priority": "external_literature", "layer": "human_factors_oversight", "chapter_targets": [ "human-factors-and-meaningful-control-in-oversight" ], "url": "https://link.springer.com/article/10.1007/s43681-022-00167-3", "notes": "Primary socio-technical comparator for operationalizing meaningful human control through operating-domain, representation, authority-and-ability, and responsibility-link properties; it does not establish that a local approval gate is meaningful, effective, or safe.", "source_type": "journal_article", "published": "2022-05-19", "updated": "2022-05-19", "citation_label": "Siebert et al. (2022), Meaningful human control: actionable properties", "doi": "10.1007/s43681-022-00167-3" }, { "id": "ext_agentic_oversight_practice_2026", "title": "Human oversight of agentic systems in practice: Examining the oversight work, challenges, and heuristics of developers using software agents", "priority": "external_literature", "layer": "human_factors_oversight", "chapter_targets": [ "human-factors-and-meaningful-control-in-oversight" ], "url": "https://arxiv.org/abs/2606.05391", "notes": "Primary exploratory human-subjects comparator for a priori control, co-planning, real-time monitoring, post hoc review, and situated oversight failures in software-agent use; it does not establish population-wide effects, control efficacy, safety, or an ASI Stack result.", "source_type": "preprint", "arxiv_id": "2606.05391", "published": "2026-06-03", "updated": "2026-06-03", "citation_label": "Dhanorkar, Passi, and Vorvoreanu (2026), Human oversight of agentic systems in practice", "doi": "10.48550/arXiv.2606.05391" }, { "id": "ext_nist_deployed_ai_monitoring_2026", "title": "Challenges to the Monitoring of Deployed AI Systems", "priority": "external_literature", "layer": "ai_operations_and_monitoring", "chapter_targets": [ "governed-operations-incident-command-and-graceful-degradation" ], "url": "https://doi.org/10.6028/NIST.AI.800-4", "notes": "Official NIST post-deployment monitoring comparator for functionality, operational, input, output, impact, and security monitoring plus field-method gaps; it does not prescribe a complete incident system or establish local monitoring effectiveness, resilience, compliance, or safety.", "source_type": "government_report", "published": "2026-03-06", "updated": "2026-03-18", "citation_label": "Rao et al. (2026), Challenges to the Monitoring of Deployed AI Systems", "doi": "10.6028/NIST.AI.800-4" }, { "id": "ext_metr_time_horizons_2025", "title": "Measuring AI Ability to Complete Long Software Tasks", "priority": "external_literature", "layer": "capability_measurement", "chapter_targets": [ "capability-thresholds-and-deployment-commitments" ], "url": "https://arxiv.org/abs/2503.14499", "notes": "Primary time-horizon comparator for an evaluation-specific, human-baselined capability metric and its external-validity limits; it does not establish local autonomy, general capability, a deployment threshold, safety, or an ASI Stack result.", "source_type": "preprint", "arxiv_id": "2503.14499", "published": "2025-03-18", "updated": "2026-02-25", "citation_label": "Kwa et al. (2025), Measuring AI Ability to Complete Long Software Tasks", "doi": "10.48550/arXiv.2503.14499" }, { "id": "ext_anthropic_rsp_2026", "title": "Anthropic's Responsible Scaling Policy", "priority": "external_literature", "layer": "capability_commitments", "chapter_targets": [ "capability-thresholds-and-deployment-commitments" ], "url": "https://www.anthropic.com/responsible-scaling-policy", "notes": "Official policy comparator for capability thresholds, required safeguards, versioned commitments, safeguard upgrades, risk reports, and change control; it does not establish ASI Stack threshold accuracy, safeguard effectiveness, policy compliance, safety, or deployment readiness.", "source_type": "official_docs", "published": "2023-09-19", "updated": "2026-07-08", "citation_label": "Anthropic (2026), Responsible Scaling Policy v3.4" }, { "id": "ext_openai_preparedness_framework_2025", "title": "Our updated Preparedness Framework", "priority": "external_literature", "layer": "capability_commitments", "chapter_targets": [ "dangerous-capability-domains-and-misuse-uplift", "capability-thresholds-and-deployment-commitments" ], "url": "https://openai.com/index/updating-our-preparedness-framework/", "notes": "Official framework comparator for threshold-linked operational commitments, capability and safeguards reports, residual-risk review, and reassessment; it does not establish ASI Stack threshold accuracy, safeguard effectiveness, policy compliance, safety, or deployment readiness.", "source_type": "official_docs", "published": "2025-04-15", "updated": "2025-04-15", "citation_label": "OpenAI (2025), Updated Preparedness Framework" }, { "id": "ext_weak_to_strong_generalization_2023", "title": "Weak-to-Strong Generalization: Eliciting Strong Capabilities With Weak Supervision", "priority": "external_literature", "layer": "weak_supervision", "chapter_targets": [ "scalable-oversight-and-adversarial-ai-control" ], "url": "https://arxiv.org/abs/2312.09390", "notes": "Primary weak-to-strong-supervision comparator for a capability-gap envelope, held-out outcome audit, ceiling comparison, and explicit disanalogies between current weak-model studies and superhuman oversight; it does not establish local supervision quality, reliable elicitation, alignment, safety, or an ASI Stack result.", "source_type": "preprint", "arxiv_id": "2312.09390", "published": "2023-12-14", "updated": "2023-12-14", "citation_label": "Burns et al. (2023), Weak-to-Strong Generalization", "doi": "10.48550/arXiv.2312.09390" }, { "id": "ext_scalable_oversight_weak_llms_2024", "title": "On scalable oversight with weak LLMs judging strong LLMs", "priority": "external_literature", "layer": "scalable_oversight", "chapter_targets": [ "scalable-oversight-and-adversarial-ai-control" ], "url": "https://arxiv.org/abs/2407.04622", "notes": "Primary scalable-oversight comparator for protocol-specific weak-judge evaluations, debate and consultancy baselines, information-asymmetry limits, and open-role persuasion risks; it does not establish local judge calibration, debate efficacy, training safety, execution authority, or an ASI Stack result.", "source_type": "preprint", "arxiv_id": "2407.04622", "published": "2024-07-05", "updated": "2024-07-12", "citation_label": "Kenton et al. (2024), On scalable oversight with weak LLMs judging strong LLMs", "doi": "10.48550/arXiv.2407.04622" }, { "id": "viea", "title": "Verified Intent-to-Execution Architecture", "priority": "must_use", "layer": "whole_stack_execution_spine", "chapter_targets": [ "01", "07", "12", "15", "human-factors-and-meaningful-control-in-oversight", "governed-operations-incident-command-and-graceful-degradation" ], "url": "https://docs.google.com/document/d/1SDu8MWw4dOpFqwLqA8vpE1rV98O2_GeJgVlgb6R9GsM", "notes": "Keystone source. Human intent -> command contracts -> artifacts -> routing -> runtime targets -> verification -> deployment -> feedback." }, { "id": "scf", "title": "Stable Capability Fields", "priority": "must_use", "layer": "governance_recursive_self_improvement", "chapter_targets": [ "05", "replaceable-cognitive-substrates-beyond-transformer-monoculture", "human-factors-and-meaningful-control-in-oversight", "governed-operations-incident-command-and-graceful-degradation" ], "url": "https://docs.google.com/document/d/1hQ9LqEgpeHo2SAntUVk15Eegms_xRVfOhtndVM5TDS4", "notes": "Use public release v1.0 when available. Stable boundaries, replacement, bounded authority, recoverable evolution." }, { "id": "planforge", "title": "PlanForge", "priority": "must_use", "layer": "planning_control", "chapter_targets": [ "06", "governed-world-models-and-reality-grounding" ], "url": "https://docs.google.com/document/d/12mSs7u6JxrtVQ_GbNpBY7lXCA5uyE62Lx91gi_VRjgo", "notes": "Planning substrate. Goal-to-execution compilation, hierarchical decomposition, DAG planning, scheduling, intelligence arbitrage." }, { "id": "planforge_compiler_arch", "title": "PlanForge: A Compiler Architecture for AI Task Orchestration", "priority": "must_use_variant", "layer": "planning_control", "chapter_targets": [ "06" ], "url": "https://docs.google.com/document/d/1ute1JJLsMfQyRFgKKveSr8U-w_MM-hk5fGxm7Vp5z1o", "notes": "Later/alternate PlanForge framing. Prefer highest-quality/latest content after comparison." }, { "id": "cognitive_compilation", "title": "Cognitive Compilation", "priority": "must_use", "layer": "planning_semantic_ir", "chapter_targets": [ "06", "12", "governed-world-models-and-reality-grounding" ], "url": "https://docs.google.com/document/d/1spEWiRnk1NUFuK3BLh3X80r_Up_SyebO1c3JSEzTUis", "notes": "Compiler framing for LLM-centered planning, semantic IR, target compilation, incremental repair." }, { "id": "talos", "title": "Talos Protocol", "priority": "must_use", "layer": "labor_execution_os", "chapter_targets": [ "08", "inter-stack-protocols-identity-and-economic-exchange", "human-factors-and-meaningful-control-in-oversight", "governed-operations-incident-command-and-graceful-degradation", "human-ai-communication-persuasion-and-epistemic-security" ], "url": "https://docs.google.com/document/d/1IRHwojXgcGx4INJsVx8uMYu4vJGl6H5O9w_l_01VcyU", "notes": "AI labor OS. Deterministic cognitive manufacturing, typed jobs, control planes, auditability, tool isolation." }, { "id": "talos_md", "title": "Talos_Protocol_v1.0.md", "priority": "must_use_variant", "layer": "labor_execution_os", "chapter_targets": [ "08" ], "url": "https://drive.google.com/file/d/1vCXRSKiE2C5ngmo7KHhPBBTtzVvr2ki-", "notes": "Markdown/public release version." }, { "id": "vcm_public", "title": "Virtual_Context_Memory_v1", "priority": "must_use", "layer": "memory_context", "chapter_targets": [ "07", "inter-stack-protocols-identity-and-economic-exchange" ], "url": "https://docs.google.com/document/d/1fid_RDNJcY3gM5WfqTpz5RABpzzoehWU4QKGhurRyJM", "notes": "Public VCM release. Governed protocol for compiled working context." }, { "id": "vcm_editable", "title": "Virtual_Context_Memory_v1.0_Editable", "priority": "must_use_variant", "layer": "memory_context", "chapter_targets": [ "07" ], "url": "https://docs.google.com/document/d/1D5RifUx-1Eq_WluCrS9FUcf3qt4X7I6_n21YD2ffB0w", "notes": "Editable version with evidence-carrying planner-guided context compiler framing." }, { "id": "spinoza", "title": "Proof of Belief / The Spinoza Architecture", "priority": "must_use", "layer": "reasoning_epistemology", "chapter_targets": [ "09" ], "url": "https://docs.google.com/document/d/1Y90DBgxsOJImMXi4aOrwxtbyjlkVZpCzu1YJy_SFatk", "notes": "Neurosymbolic belief, transparent axiomatic AI belief systems, verification, belief revision." }, { "id": "spinoza_composer", "title": "Spinoza Composer / Spinoza Trinity", "priority": "supporting", "layer": "reasoning_media_compliance", "chapter_targets": [ "09", "08" ], "url": "https://docs.google.com/document/d/1pWHPNCFL5dnphZxKUhj5c_yk3juskWQZPTzPjxE7nEE", "notes": "Narrative/compliance operating system variant. Use as applied Spinoza." }, { "id": "moecot", "title": "MoECOT-Agent Architecture Whitepaper", "priority": "must_use", "layer": "implementation_reference", "chapter_targets": [ "10", "15", "16" ], "url": "https://docs.google.com/document/d/1Lw5qjIFLK1gxGxqYw3_zLneFF5JT_Ihn5VI1CntjtvM", "notes": "Full authenticated v1.1 architecture-whitepaper text passage-reviewed. Supports a compact explicit orchestrator, bounded specialist lanes, task state machines, leases/retries/dead-letter handling, fail-closed side-effect envelopes, run/task/control-plane ledgers, readiness distinct from routing, benchmark lanes, replay/handoff, provenance, architecture fingerprints, and governed improvement proposals. Source-reported runtime and benchmark artifacts were not imported or reproduced; the pinned MoECOT project dossier is the stronger implementation-reference record.", "source_type": "author_architecture_whitepaper", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "Authenticated connector text only; no runtime artifact, command, log, replay, benchmark, deployment, authority-enforcement, learned-routing, or production result was reproduced.", "updated": "2026-07-31", "citation_label": "Sorenson (2026), MoECOT-Agent Architecture Whitepaper v1.1" }, { "id": "moecot_md", "title": "moecot_agent_whitepaper.md", "priority": "must_use_variant", "layer": "implementation_reference", "chapter_targets": [ "10", "15", "16" ], "url": "https://drive.google.com/file/d/1F0KW0-TR54xKGeetIPkDTV1Tn5_9w3i0", "notes": "Full authenticated Markdown export reconciled with the primary MoECOT v1.1 Google Doc. It is a format and terminology variant, not independent corroboration or an additional empirical result.", "source_type": "author_whitepaper_variant", "variant_of": "moecot", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "Same-lineage format variant; no independent architecture, runtime, benchmark, replay, or deployment evidence.", "updated": "2026-07-31", "citation_label": "Sorenson (2026), MoECOT-Agent Architecture Whitepaper Markdown export" }, { "id": "octopus_router", "title": "Octopus Router Architecture", "priority": "must_use", "layer": "routing_modular_intelligence", "chapter_targets": [ "10" ], "url": "https://docs.google.com/document/d/1Ew0fdnJf9crnGrQDckzNEc5fFniZviayqQfGIR3TBMY", "notes": "Lightweight head/router with dynamically loaded specialist arms and local boundaries." }, { "id": "rmi", "title": "Ratcheting Modular Intelligence", "priority": "must_use", "layer": "capability_ratchet", "chapter_targets": [ "10", "13", "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://docs.google.com/document/d/16ULc3Y82TuqkFGBpjKyhrEVZYt9_Vlvatq2kiO-q1VI", "notes": "Benchmark pressure, residual escrow, verified modular capability, regression preservation." }, { "id": "cognitive_loop_closure", "title": "Cognitive Loop Closure", "priority": "must_use", "layer": "procedural_memory", "chapter_targets": [ "08", "10" ], "url": "https://docs.google.com/document/d/12yac3ZpHv3dRk2-0u627rn-gFkqGQWY4jsIJwAW9YeQ", "notes": "Repeated cognition should become procedural memory / verified tools." }, { "id": "benchmaxxing", "title": "Benchmaxxing: The Performance Ratchet", "priority": "must_use", "layer": "benchmarks_evidence", "chapter_targets": [ "13", "capability-thresholds-and-deployment-commitments", "safety-cases-and-structured-assurance", "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://docs.google.com/document/d/1m1EycQk56VDh1Rg3IY8GYwkebIVn3bGAWIHs93a3628", "notes": "Benchmarks as pressure surfaces, saturation -> regression, harder frontier, anti-Goodhart safeguards." }, { "id": "cgs", "title": "Compact Generative Systems", "priority": "must_use", "layer": "compression_representation", "chapter_targets": [ "11" ], "url": "https://docs.google.com/document/d/1y8IW-rnDSM8-q1KUnLfqKTGobwCoHoBwuMJ4-gCcw3M", "notes": "Smallest adequate structure that can generate/govern target without hiding residual complexity." }, { "id": "rgs", "title": "Ratcheting Generative Systems", "priority": "supporting", "layer": "compression_capability_growth", "chapter_targets": [ "11", "10" ], "url": "https://docs.google.com/document/d/1jxbgwiBzgUgdTjpPiOkEQNZz2ehcDioTG-lUBV-f4FA", "notes": "Bridge between active compression, procedural memory, benchmark frontiers, verified AI growth." }, { "id": "rankfold_neuralfold", "title": "RankFold + NeuralFold", "priority": "must_use", "layer": "compression_representation", "chapter_targets": [ "11" ], "url": "https://docs.google.com/document/d/1-9wujZDobutPQbAml3H3QqYvdt9GIk55bxJiaPeSUhg", "notes": "Tensor/artifact compression. Low-rank residual coding plus functional preprocessing and probe-route fallback." }, { "id": "rankfold_compressor", "title": "rankFold compressor", "priority": "must_use_variant", "layer": "compression_representation", "chapter_targets": [ "11" ], "url": "https://docs.google.com/document/d/11jw0DAAuUvw75Q_1AiwTiUGcd-y_IywB9qdEka9_MfI", "notes": "Alternate RankFold/NeuralFold source." }, { "id": "bbvca_v9", "title": "BBVCA_v9_final_public_release", "priority": "must_use", "layer": "compression_representation", "chapter_targets": [ "11" ], "url": "https://docs.google.com/document/d/1dCcqTteePCyUb66H3qJ-50uYMNDdqHFG7-qQpHSvVaA", "notes": "Prefer v9. Generate-verify-repair compression from seeded local laws, bounded search, two-phase rate discipline." }, { "id": "bbvca_main", "title": "Big Bang Volumetric Compression Architecture", "priority": "must_use_variant", "layer": "compression_representation", "chapter_targets": [ "11" ], "url": "https://docs.google.com/document/d/1tlgJismt6JaYv_jaf2XbwCX7WEqj9FJ0WvEtbHYS_-E", "notes": "Earlier/main BBVCA family doc." }, { "id": "genesiscode", "title": "GenesisCode", "priority": "must_use", "layer": "executable_specification", "chapter_targets": [ "08", "12" ], "url": "https://docs.google.com/document/d/1w4gKcF9a7oV6hsUECsWeEafblvc78ZusxaxdUTfLR2M", "notes": "Tiny pure calculus + obligations + provenance for auditable AI-symbiotic programming." }, { "id": "alignment_field", "title": "Field of God / Alignment Field family", "priority": "must_use", "layer": "alignment_constitution", "chapter_targets": [ "constitutional-alignment-substrate", "inner-alignment-mesa-optimization-and-learned-objective-integrity", "moral-uncertainty-and-value-conflict", "governed-objective-formation-value-learning-and-goal-integrity", "security-kernel-and-digital-scifs", "recursive-self-improvement-boundaries", "resource-economics-and-token-budgets", "integrated-reference-architecture", "open-research-agenda-and-bibliography-plan" ], "url": "https://docs.google.com/document/d/1CCiBBlOEGB2huRDSwGm4P1JkVBkfmaUxJt9AwG23Kzw", "notes": "Corben-authored long-form metaphysics, consciousness, ethics, AI-rights, and governance family. Its complete section-family audit preserves the five-factor consciousness heuristic only as theory-relative question decomposition; adds a prospective architecture-induced moral-risk review for self-preservation, persistent identity, valence-like state, and copy proliferation; separates copy, causal, memory, legal, authority, consent, and first-person continuity; and rejects the source's scalar moral ranking, collective-consciousness, inevitable ethical convergence, physics, clinical, Omega, and karma claims as technical evidence.", "source_type": "author_whitepaper", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "Metaphysical, consciousness, clinical, physics, moral-proof, collective-phenomenology, global-coherence, karma-score, and present-system status claims remain speculative or rejected as evidence; the source supports only bounded architecture and governance lineage.", "updated": "2026-07-31", "citation_label": "Sorenson (2026), Field of God / Alignment Field family" }, { "id": "field_of_god", "title": "The Field of God", "priority": "must_use_variant", "layer": "alignment_constitution", "chapter_targets": [ "constitutional-alignment-substrate", "moral-uncertainty-and-value-conflict", "governed-objective-formation-value-learning-and-goal-integrity", "inner-alignment-mesa-optimization-and-learned-objective-integrity", "human-ai-organizations-delegation-and-accountability", "recursive-self-improvement-boundaries", "integrated-reference-architecture" ], "url": "https://docs.google.com/document/d/1BDVLQqIeDm-PFijrg6oKaCbRtUyY--_waw88tnU9tj0", "notes": "Corben-authored predecessor to the Alignment Field family, containing title exploration, outline, abbreviated and expanded eight-part drafts across informational-relational metaphysics, a five-factor consciousness heuristic, attractor ethics, AI, copy continuity, genealogy, and conclusion. The complete audit treats `alignment_field` as the controlling successor, preserves power-care divergence, nested optimization, dissent/feedback, and identity-continuity distinctions, and explicitly rejects double counting, scalar moral ranking, collective consciousness, metaphysical proof, clinical/physics claims, upload survival, and teleology as evidence.", "source_type": "author_whitepaper", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "Predecessor lineage only; `alignment_field` controls shared ideas. No metaphysical, consciousness, welfare, identity-continuity, collective-mind, moral-proof, clinical, physics, safety, deployment, or support-state result is established.", "updated": "2026-07-31", "citation_label": "Sorenson (2026), The Field of God" }, { "id": "field_of_god_ai_constitution", "title": "Field of God AI Constitution", "priority": "core_alignment_source", "layer": "constitutional_alignment_runtime_governance", "chapter_targets": [ "constitutional-alignment-substrate", "moral-uncertainty-and-value-conflict", "recursive-self-improvement-boundaries", "runtime-adapters" ], "url": "https://github.com/corbensorenson/symliquid-rmi/blob/main/personality-documents/field_of_god_ai_constitution_v1.md", "notes": "Recovered in the Project Theseus repository. Constitutional alignment core for truth alignment, agency preservation, consent, non-domination, consciousness caution, least sufficient power, auditability, self-authorization limits, and runtime checks; use as source material only after source-note creation, not as proof or test evidence." }, { "id": "ethica_mechanica", "title": "Ethica Mechanica", "priority": "must_use_variant", "layer": "alignment_constitution", "chapter_targets": [ "constitutional-alignment-substrate", "moral-uncertainty-and-value-conflict", "governed-objective-formation-value-learning-and-goal-integrity", "ai-deployment-transition-distribution-and-human-agency", "privacy-data-rights-and-information-flow-governance", "inter-stack-protocols-identity-and-economic-exchange", "integrated-reference-architecture" ], "url": "https://docs.google.com/document/d/15cN9D6s2bvAPoitp6Imh7deE3gCebqG4RbaSRGRd5GA", "notes": "Corben-authored January 2026 philosophical and socio-technical treatise. Its complete audit retains the separation between bounded machine logistics and human normative authority, recursive feedback, dissent, governing-logic transparency versus personal privacy, distributional simulation as contestable evidence, and the requirement that exit/fork rights be materially exercisable under portability, network, compute, continuity, safety, privacy, and obligation constraints. Metaphysics, the consciousness equation, moral proofs, automatic veil legitimacy, unrestricted fork, and executable-protocol claims remain unsupported.", "source_type": "author_whitepaper", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "Philosophical and governance lineage only; no metaphysical, consciousness, moral-proof, legitimacy, fairness, privacy, fork-safety, interoperability, deployment, or support-state result is established.", "published": "2026-01-30", "updated": "2026-07-31", "citation_label": "Sorenson (2026), Ethica Mechanica" }, { "id": "eternal_code", "title": "The Eternal Code / unified God, reality, conscious, alignment", "priority": "must_use_variant", "layer": "alignment_constitution", "chapter_targets": [ "constitutional-alignment-substrate", "evidence-states-and-claim-discipline", "spinoza-verification-and-proof-carrying-claims", "tribunal-adversarial-review-and-claim-conflict", "governed-world-models-and-reality-grounding", "intent-to-execution-contracts", "runtime-adapters-tool-permissions-and-human-approval", "moral-uncertainty-and-value-conflict", "resource-economics-and-token-budgets" ], "url": "https://docs.google.com/document/d/1ZMqR7CMSnzignSbHdm38q9jSAXGMRsIoz3jOz3qmVrc", "notes": "Corben-authored multi-version computational-metaphysics and alignment family. Its complete audit traces the progression from an aggregate alignment score to Truth/Social/Task axes, a geometric product, heterarchical evaluator ownership, compute-aware exit, and a standing adversarial challenger. The book retains separate non-compensating epistemic, task, affected-party/constitutional, and authority/effect planes, evaluator lineage, and material exit costs while rejecting the consciousness and alignment formulas, oracle labels, consensus-as-truth, automatic energy throttling, fixed compute entitlement, theological/metaphysical claims, and moral proofs.", "source_type": "author_whitepaper", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "Architecture and governance lineage only; no metaphysical, consciousness, moral, metric-validity, evaluator-independence, consensus, safety, resource-entitlement, deployment, or support-state result is established.", "published": "2026-01-31", "updated": "2026-07-31", "citation_label": "Sorenson (2026), The Eternal Code / Codex Recursiva family" }, { "id": "coherence_exchange", "title": "The Coherence Exchange", "priority": "strong_support", "layer": "epistemic_market_synthesis", "chapter_targets": [ "09", "13", "15", "institutions-international-coordination-and-public-legitimacy", "ai-deployment-transition-distribution-and-human-agency" ], "url": "https://docs.google.com/document/d/1RUeBKp_UnRVIMc2oPBT2j3xm-HBa81Y4VqTckPhCeJc", "notes": "Found in AI generated paper dump. Use carefully; speculative synthesis of PlanForge, Spinoza, Talos, UAT, Alignment Field." }, { "id": "verification_bandwidth", "title": "Verification Bandwidth in Bounded Contexts", "priority": "strong_support", "layer": "context_verification_theory", "chapter_targets": [ "verification-bandwidth-and-context-adequacy", "virtual-context-abi", "evidence-states-and-claim-discipline", "spinoza-verification-and-proof-carrying-claims", "compact-generative-systems-and-residual-honesty", "fast-generation-architectures", "policy-optimization-and-learning-from-feedback", "governed-deliberation-and-test-time-scaling", "scalable-oversight-and-adversarial-ai-control", "open-research-agenda-and-bibliography-plan" ], "url": "https://docs.google.com/document/d/1T34n1Ya6_joaAD8ZxygOpiEuNzEVj3G8_nf0xZuFL2U", "notes": "Corben-authored version 1.0 context-verification hypothesis. The complete audit preserves the generation-versus-verification distinction, claim-relative semantic units and effective workspace, dominant-component pressure, explicit interaction obligations, decomposition boundaries, a coherency-horizon escalation rule, and a stronger held-out contradiction protocol. It rejects the four named 'theorems' as proved laws: dense joint attention is neither necessary nor sufficient, lossy compression need not discard property-relevant information, DPI does not establish monotonic LLM contradiction growth, all-pair checking is not universally required, uniform half-window partitioning is not generally optimal, and RAG may retrieve exact text.", "source_type": "author_whitepaper", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "Research-hypothesis and architecture lineage only; no model run, effective-workspace estimator, cognition theorem, contradiction-rate result, RAG limitation, adequacy classifier, useful advantage, safety, transfer, deployment, or support-state result is established.", "updated": "2026-07-31", "citation_label": "Sorenson (2026), Verification Bandwidth in Bounded Contexts" }, { "id": "beastbrain", "title": "BeastBrain Cognitive Architecture", "priority": "supporting_lineage", "layer": "whole_stack_lineage", "chapter_targets": [ "asi-is-a-stack-not-a-model", "the-efficient-asi-hypothesis", "routing-heads-and-specialist-cores", "personal-compute-hives-and-federated-edge-intelligence", "governed-model-training-distributed-optimization-and-scaling", "fast-generation-architectures", "security-kernel-and-digital-scifs", "perception-sensor-fusion-and-observation-trust", "prototype-roadmap", "integrated-reference-architecture" ], "url": "https://docs.google.com/document/d/1PVLtICJVZm36c6M2y9JkuMZF9nfu_6pyR1SXqFzknME", "notes": "Corben-authored 70,000-word evolving architecture notebook spanning early blueprints through versions 1.0\u20136.1. The complete family audit preserves whole-system/homeostatic design, versioned hardware-profile qualification, physical memory-tier and residency accounting, distinct memory forms, governed consolidation and retention, substrate-neutral routing, contract-first planning and implementation, dependency-aware parallelism, opaque secret handles, multimodal perception, distributed service and artifact interfaces, and maintenance-learning windows. It rejects master-label maturity, repeated-version double counting, infinite-context/zero-copy/power/performance projections, geometric-truth and ignorance theorems, scalar routing/retention authority, unsafe forced self-evolution, test-suite sufficiency, tribunal consensus, software-SCIF guarantees, SSD-erasure assumptions, censorship-resistance, and autonomous update claims.", "source_type": "author_architecture_family", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "Architecture and version lineage only. No runtime, hardware qualification, paging, zero-copy, context, memory, model-training, routing, verification, security, perception, federation, self-improvement, performance, efficiency, safety, deployment, AGI, ASI, SOTA, or support-state result is established; incorporated component papers are not independent evidence through this source.", "updated": "2026-07-31", "citation_label": "Sorenson (2026), BeastBrain Cognitive Architecture family" }, { "id": "beastbrain_timeless", "title": "BeastBrain Architecture: Timeless Edition", "priority": "supporting_lineage", "layer": "whole_stack_lineage", "chapter_targets": [ "the-efficient-asi-hypothesis", "personal-compute-hives-and-federated-edge-intelligence", "prototype-roadmap" ], "url": "https://docs.google.com/document/d/12g8s3R_v0mJ7GXaOoWeWkqHWBQWe_iGmcTmUwChZLwg", "notes": "Corben-authored standalone export of BeastBrain v3.3.4. The complete audit treats it as a near-duplicate Timeless branch already embedded in the main BeastBrain corpus, not independent support. It preserves evergreen whole-stack and Mimic hardware-adaptation framing while explicitly rejecting forced self-evolution, geometric truth, fixed entropy routing, infinite context, zero-copy, power, leak-resistance, constant-time verification, hardware-adaptation, and distributed-scaling projections as evidence.", "source_type": "author_whitepaper_variant", "canonical_source_id": "beastbrain", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "Variant and de-duplication lineage only; no independent evidentiary weight and no implementation, hardware, model, memory, routing, verification, security, self-improvement, scaling, deployment, or support-state result.", "updated": "2026-07-31", "citation_label": "Sorenson (2026), BeastBrain v3.3.4 Timeless Edition" }, { "id": "aletheia", "title": "Aletheia / Proof-Carrying Workbench Lineage", "priority": "supporting_lineage", "layer": "safe_general_intelligence_lineage", "chapter_targets": [ "asi-is-a-stack-not-a-model", "the-efficient-asi-hypothesis", "claim-ledgers-and-belief-revision", "scientific-discovery-and-experimental-governance" ], "url": "https://docs.google.com/document/d/1H6M_IPbuu86FOQurv2rLQ97EYQVdThZ2U7njl0DBpIE", "notes": "Complete four-version correction-lineage audit from the original Aletheia epistemic-engine proposal through Aletheia Foundry and Proof-Carrying Workbench v1.1/v1.2. The later PCW design controls conflicts: semantic scope rather than contract-hash theater, claim-native release surfaces, separate assurance classes, least-privilege capabilities, bounded adversarial review, governed commitments, template decay, recertification, and incident response. The audit rejects immutable primitives, scalar truth/intervention scores, live-oracle and consensus truth, deterministic open-domain claim extraction, universal source allowlists, unvalidated risk thresholds, and safe-general-intelligence claims.", "source_type": "author_architecture_family", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "Architecture and correction lineage only; no implementation, natural claim-extraction corpus, verifier-independence result, semantic-escape measurement, useful-throughput comparison, security result, deployment, safe-general-intelligence result, or support-state promotion.", "updated": "2026-07-31", "citation_label": "Sorenson (2026), Aletheia / Proof-Carrying Workbench lineage" }, { "id": "context_engineer", "title": "Context Engineer / Manhattan Protocol", "priority": "supporting_lineage", "layer": "memory_context_lineage", "chapter_targets": [ "virtual-context-abi", "context-transactions-snapshots-mounts-and-taint", "security-kernel-and-digital-scifs" ], "url": "https://docs.google.com/document/d/1xXP364s9IZ4DFmRHTmBELviRQkJcQ55McIUokioRFdo", "notes": "Complete two-version audit of the Manhattan Protocol context-supply-chain paper. Preserves the context governor, layered representations, mission briefs, proposed MCP memory fields, need-to-know admission, and compartment lifecycle while separating sanitization, declassification, memory commit, zeroization, revocation, and residuals. Rejects Ring Attention as physical isolation, protocol fields as enforcement, regex/entropy scanning as semantic non-disclosure, permanent-wipe language, infinite storage, and all unreproduced benchmark figures.", "source_type": "author_architecture_family", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "Design lineage only; no benchmark harness, raw run, model/evaluator pin, protocol standard, runtime enforcement, leakage result, isolation result, physical-erasure result, deployment, or support-state promotion.", "updated": "2026-07-31", "citation_label": "Sorenson (2026), Manhattan Protocol / Context Engineer" }, { "id": "black_hole_context_manager", "title": "Black Hole Context Manager", "priority": "supporting_lineage", "layer": "memory_context_lineage", "chapter_targets": [ "context-transactions-snapshots-mounts-and-taint" ], "url": "https://docs.google.com/document/d/14KPQT5d86HaFZzQdUr_Sn5p7-8cscSqIZm5zxkijl-I", "notes": "Complete v5.0/v4.1 pseudocode-lineage audit. Preserves versioned context units, tiered placement, lazy task-relative evaluation, explicit goal-drift decisions, reversible freeze/thaw with hysteresis, protected low-entropy constraints, and factual-retrieval versus generative-reconstruction separation. Treats character entropy, semantic mass, K-means thresholds, HMAC, repeated confirmation, keyword routing, and the Drifting Needle as fallible candidates or weak baselines; rejects production-ready and security claims.", "source_type": "author_technical_spec_family", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "Pseudocode and design lineage only; no importable package, dependency lock, test output, natural workload, cache/recall result, security result, deletion result, resource measurement, deployment, or support-state promotion.", "updated": "2026-07-31", "citation_label": "Sorenson (2026), Black Hole Context Collapse Protocol" }, { "id": "ladon_manhattan", "title": "Ladon & The Manhattan Protocol", "priority": "supporting_lineage", "layer": "security_governance", "chapter_targets": [ "security-kernel-and-digital-scifs", "runtime-adapters-tool-permissions-and-human-approval", "system-boundaries-and-authority" ], "url": "https://docs.google.com/document/d/1uT9iQ7Jb2TsU9DletvtVeLEej63aIQl3WS3jTsMgtSM", "notes": "Complete standalone security-paper audit. Preserves capability/credential separation, opaque caller-bound handles, trusted input paths, late substitution or remote use, per-use policy checks, and an explicit compartment lifecycle. The audit separates secret non-disclosure from authority misuse and harmful effects, treats returned artifacts as possible sensitive derivatives, and rejects platform equivalence, PROT_NONE/enclave conflation, portable trusted-UI claims, incomplete Rust pseudocode as implementation, and the paper's Ignorance and Ephemerality 'theorems'.", "source_type": "author_security_architecture", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "Architecture and pseudocode lineage only; no ladon-rs package, kernel implementation, trusted UI, hardware target, compartment, policy engine, threat-model test, side-channel measurement, audit, deployment, or support-state promotion.", "updated": "2026-07-31", "citation_label": "Sorenson (2026), Ladon & The Manhattan Protocol" }, { "id": "uat", "title": "Unified Adaptive Tribunal", "priority": "supporting_lineage", "layer": "evaluation_refinement", "chapter_targets": [ "spinoza-verification-and-proof-carrying-claims", "claim-ledgers-and-belief-revision", "evidence-states-and-claim-discipline", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://docs.google.com/document/d/1R7wKo2qg5waosEa-SwC-JKF77qoDiiEp1Df4Gbap0VE", "notes": "Complete three-tab correction-lineage audit. The public human-in-the-loop architecture controls the original promotional multi-model tribunal: it preserves structural/retrieval/dialectical candidate views, an explicit dossier boundary and omitted frontier, probabilistic claim extraction, richer proposition states, bounded adversarial review, compression fidelity, and accountable human handoffs. It rejects brand-count diversity, consensus and stability as truth, delete-by-dossier-absence, SVO completeness, fixed thresholds, guard-model authority, all claimed performance/cost figures, and production or superiority labels.", "source_type": "author_review_protocol_family", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "Architecture and correction lineage only; no codebase, frozen dossier, task corpus, model output, claim annotation, source-quality evaluation, reviewer-dependence audit, human study, benchmark, cost trace, deployment, or support-state promotion.", "updated": "2026-07-31", "citation_label": "Sorenson (2026), Unified Adaptive Tribunal lineage" }, { "id": "treellm", "title": "TreeLLM", "priority": "supporting_lineage", "layer": "semantic_representation", "chapter_targets": [ "09", "11" ], "url": "https://docs.google.com/document/d/17C98P4WhU4srqrT83xXpopRvwHO19zRFJ_ov0MWhgHE", "notes": "Hierarchical semantic token system for grounded, efficient, explainable language modeling." }, { "id": "software_magic_grimoire", "title": "The Grimoire of Software Magic Words: Operative Vocabulary, Prompt-Spells, and Stacked Workflows", "priority": "supporting_lineage", "layer": "command_contracts_promptcraft", "chapter_targets": [ "06", "08" ], "url": "https://docs.google.com/document/d/1UjGadqJ3ZiqfLgbac0APtV_OLnlSZAkIW0o0r37YdBo", "notes": "Full composite-document audit completed 2026-07-31 across the public grimoire, 1,645-entry lexicon, pocket edition, stacked-spells addendum, and prompt pack. Retains bounded instruction fields, layered identity, typed handoffs, guards, evidence loops, scoped recursion, recovery, and workflow versioning; rejects vocabulary, role prompts, G\u00f6del numbers, coil geometry, templates, and authored examples as evidence of meaning, authority, performance, or safety." }, { "id": "road_to_agi", "title": "Road To AGI", "priority": "supporting_lineage", "layer": "strategic_roadmap", "chapter_targets": [ "16" ], "url": "https://drive.google.com/file/d/1FfYjqa36tMQ4s4KLbPPnIOZNfQefuT7p", "notes": "Remaining work / roadmap context." }, { "id": "simulation_scaling", "title": "The Simulation Scaling Law: Resource Constraints on Scope, Clockspeed, and Effective Fidelity in Nested Physical Simulations", "priority": "optional_support", "layer": "compute_fidelity_constraints", "chapter_targets": [ "03", "11", "appendix", "governed-world-models-and-reality-grounding", "learning-theory-generalization-and-scaling-science" ], "url": "https://docs.google.com/document/d/1rt0lnpwZ9X6M_ejLC7PYHQSXsvajYICGo4ETS-ndUFA", "notes": "Full six-tab lineage audit completed 2026-07-31. Retains the prospective simulation contract, typed bottleneck accounting, and logical-possibility/physical-feasibility split; treats D = scope*clockspeed/efficiency <= capacity as a conditional scalar heuristic rather than a proved universal law, rejects unsupported 1:1 and physical-limit overclaims, and routes simulator adequacy and transfer through Resource Economics." }, { "id": "tokenmana", "title": "TokenMana", "priority": "optional_support", "layer": "resource_economics", "chapter_targets": [ "resource-economics-and-token-budgets", "human-factors-and-meaningful-control-in-oversight", "benchmark-ratchets-and-anti-goodhart-evidence", "physical-compute-infrastructure-energy-and-environmental-constraints" ], "url": "https://docs.google.com/document/d/1dGOlGPZi6byTwRUbHQt40OSErBnQ4j6G_BZvA-oJRM8", "notes": "Complete five-tab/two-paper correction-lineage audit. Preserves regenerative capacity as one candidate budget mechanism and adds temporal-access contracts that expose renewal, accrual, expiry, burst, pricing, notification, fairness, and human-schedule effects. Separates renewal clustering, nocturnal work, sleep, and cognitive-friction proxies; strengthens privacy against employment/medical inference; and records mathematical gaps in stock units/boundaries, equilibrium, control continuity, strict variance, feedback stability, and profit claims. No theorem, simulation, load result, pricing result, human outcome, or welfare result is promoted.", "source_type": "author_mechanism_and_human_outcomes_family", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "Theory and research-design lineage only; no full proof, mechanization, simulation, billing trace, queue model, pricing experiment, human-subjects study, sleep/wearable data, productivity result, load-stability result, profit result, deployment, or support-state promotion.", "updated": "2026-07-31", "citation_label": "Sorenson (2026), TokenMana regenerative capacity and temporal elasticity papers" }, { "id": "coilmoecot", "title": "CoilMoECOT Whitepaper v2.0", "priority": "optional_technical_appendix", "layer": "mathematical_search_substrate", "chapter_targets": [ "12" ], "url": "https://drive.google.com/file/d/1iwhhRGr7vj_Sr4UwwM_yX_ESSyisoWur", "notes": "Full composite connector text section-audited. Retains prime-temporal trace features, Graph/Trace-first placement, ledger-derived diagnostic tuples, removable shadow lanes, explicit pre-planner/post-plan/post-run insertion points, anti-experts as visible penalty signals, bounded update slices, and benchmark/canary/rollback promotion. Repeated MoECOT, packaging, and task-system appendices are treated as shared substrate rather than independent evidence.", "source_type": "author_architecture_composite", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "Design and implementation specification only; no local route, ablation, benchmark, canary, rollback, model-quality, safety, latency, or cost result.", "updated": "2026-07-31", "citation_label": "Sorenson (2026), CoilMoECOT Whitepaper v2.0" }, { "id": "temporal_coil_research", "title": "Temporal Coil Research", "priority": "optional_technical_appendix", "layer": "mathematical_search_substrate", "chapter_targets": [ "12" ], "url": "https://drive.google.com/file/d/1PSQsAwMurlVRdsM6r_zRd2MPLZYv0qXH", "notes": "Full experiment note reviewed. Reports 11 variants, three seeds, six rounds per variant, split winner frequency, small mean deltas, flat pass/reward/holdout lanes, separation dominated by the collapse composite, and one narrow threshold-tuned seed. Preserved as an inconclusive source-reported result and placement-confounding lesson, not proof of benefit or general failure.", "source_type": "author_experiment_note", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "Source-reported live-endpoint runs only; no local reproduction, independent evaluator, causal component effect, transfer, or general coil conclusion.", "updated": "2026-07-31", "citation_label": "Sorenson (2026), Temporal Coil Research" }, { "id": "bugbrain", "title": "BugBrain / Project Genesis: Neuro-Symbolic Bare-Metal Edge Intelligence Paper Lineage", "priority": "optional_support", "layer": "edge_efficiency_lineage", "chapter_targets": [ "11", "16", "appendix" ], "url": "https://docs.google.com/document/d/1XR716XCo-WmvOpvGTGDmh9zbv44oiyjfcVDVQ1YibIc", "notes": "Full 15-tab paper-lineage audit completed 2026-07-31 and reconciled against the pinned implementation dossier. Retains hardware-explicit ownership, state-qualified capacity, compact typed graphs, tiered context/paging/persistence, one-shot authority, artifact replay, readiness semantics, and objective-term effect tests; rejects consciousness, AGI, completeness, projected performance, named-module, source-presence, and skipped-green claims that outrun implementation evidence." }, { "id": "cca_project", "title": "Compiled Cognitive Architecture project", "priority": "implementation_reference", "layer": "compiled_cognitive_architecture", "chapter_targets": [ "evidence-states-and-claim-discipline", "system-boundaries-and-authority", "artifact-graphs-audit-logs-and-replay", "cognitive-compilation-and-semantic-ir", "virtual-context-abi", "context-transactions-snapshots-mounts-and-taint", "claim-ledgers-and-belief-revision", "spinoza-verification-and-proof-carrying-claims", "procedural-memory-and-cognitive-loop-closure", "ai-supply-chain-integrity-and-lifecycle-provenance", "model-weight-custody-and-hardware-roots-of-trust", "recursive-self-improvement-boundaries", "governed-deliberation-and-test-time-scaling", "benchmark-ratchets-and-anti-goodhart-evidence", "readiness-gates-residual-escrow-and-quarantine", "executable-specifications-and-lean-proof-envelope", "personal-compute-hives-and-federated-edge-intelligence", "intent-to-execution-contracts", "integrated-reference-architecture" ], "url": "local-project:cca@bd9afd16fe5b11c68f8c495f85ad117cc61ccfc8", "notes": "Pinned local-project convergence reference for external semantic memory, semantic compilation, epistemic governance, bounded self-modification, proof/runtime coupling, benchmark truth, and negative transfer evidence; public-safe source note only, with no reproduced capability or safety result.", "source_type": "local_project_snapshot", "published": "2026-03", "updated": "2026-05-12", "citation_label": "Dredge (2026), Compiled Cognitive Architecture (pinned local project snapshot)" }, { "id": "moecot_manifest_project", "title": "MoECOT Manifest compiler-era project", "priority": "implementation_reference", "layer": "compiler_first_ai_systems", "chapter_targets": [ "evidence-states-and-claim-discipline", "system-boundaries-and-authority", "security-kernel-and-digital-scifs", "integrated-reference-architecture", "cognitive-compilation-and-semantic-ir", "artifact-graphs-audit-logs-and-replay", "ai-supply-chain-integrity-and-lifecycle-provenance", "routing-heads-and-specialist-cores", "open-ended-improvement-engines", "recursive-self-improvement-boundaries", "benchmark-ratchets-and-anti-goodhart-evidence", "readiness-gates-residual-escrow-and-quarantine", "executable-specifications-and-lean-proof-envelope", "intent-to-execution-contracts", "runtime-adapters-tool-permissions-and-human-approval", "virtual-context-abi", "claim-ledgers-and-belief-revision", "spinoza-verification-and-proof-carrying-claims", "procedural-memory-and-cognitive-loop-closure", "artifact-steward-agents-and-living-project-governance" ], "url": "local-project:moecot-manifest@8398335bd01569d4bce7bc4ca2792d3ef48832f9", "notes": "Pinned local-project compiler/control-plane reference for semantic IR, manifest compilation, registry truth, context and memory governance, target portability, provenance, multi-agent training, bounded self-improvement, and the negative gap between internal contracts and external holdout capability; public-safe source note only, with no reproduced capability or safety result.", "source_type": "local_project_snapshot", "published": "2026-03", "updated": "2026-07-10", "citation_label": "Sorenson (2026), MoECOT Manifest (pinned local project snapshot)" }, { "id": "beastbrain_project", "title": "BeastBrain historical AI system project", "priority": "implementation_reference", "layer": "durable_semantic_memory_and_system_architecture", "chapter_targets": [ "evidence-states-and-claim-discipline", "system-boundaries-and-authority", "routing-heads-and-specialist-cores", "planning-as-a-control-layer", "virtual-context-abi", "claim-ledgers-and-belief-revision", "spinoza-verification-and-proof-carrying-claims", "procedural-memory-and-cognitive-loop-closure", "resource-economics-and-token-budgets", "benchmark-ratchets-and-anti-goodhart-evidence", "readiness-gates-residual-escrow-and-quarantine", "security-kernel-and-digital-scifs", "intent-to-execution-contracts", "runtime-adapters-tool-permissions-and-human-approval", "personal-compute-hives-and-federated-edge-intelligence", "artifact-steward-agents-and-living-project-governance" ], "url": "local-project:BeastBrain@sha256:60aa8121c11deabf98ff71d848d6709d7208a38be6e2b8789de5a0337f1ef684", "notes": "Hashed local-project primary lineage for DKL, Portia, semantic coordinates, bounded knowledge snapshots, PlanForge, SSD-first memory, and organism-style cognition, plus negative implementation evidence on simulations, router effects, security handles, tribunal stubs, ontology drift, compile history, and readiness overclaim; public-safe source note only, with no reproduced capability or safety result.", "source_type": "local_project_snapshot", "published": "2026-02", "updated": "2026-07-10", "citation_label": "BeastBrain (2026), pinned local project snapshot" }, { "id": "bugbrain_project", "title": "BugBrain bare-metal neuro-symbolic intelligence project", "priority": "implementation_reference", "layer": "hardware_explicit_governed_cognition", "chapter_targets": [ "evidence-states-and-claim-discipline", "system-boundaries-and-authority", "model-weight-custody-and-hardware-roots-of-trust", "security-kernel-and-digital-scifs", "virtual-context-abi", "context-transactions-snapshots-mounts-and-taint", "claim-ledgers-and-belief-revision", "spinoza-verification-and-proof-carrying-claims", "intent-to-execution-contracts", "runtime-adapters-tool-permissions-and-human-approval", "artifact-graphs-audit-logs-and-replay", "integrated-reference-architecture", "resource-economics-and-token-budgets", "readiness-gates-residual-escrow-and-quarantine", "benchmark-ratchets-and-anti-goodhart-evidence", "ai-supply-chain-integrity-and-lifecycle-provenance" ], "url": "local-project:BugBrain@d5ddd37966e2057e8b5ee7fa7bd8f4c833a30dc5", "notes": "Pinned local-project bare-metal implementation reference for explicit core ownership, compact typed graphs, signed tiered context, privileged-action lifecycle, protocol security, artifact replay, and readiness, plus negative evidence on capacity/residency arithmetic, theory-labelled proxies, fixed random cognitive modules, root-of-trust assumptions, skipped checks, and narrative/report divergence; public-safe source note only, with no reproduced hardware capability or safety result.", "source_type": "local_project_snapshot", "published": "2026-02", "updated": "2026-07-10", "citation_label": "Sorenson (2026), BugBrain (pinned local project snapshot)" }, { "id": "corbens_trainer_project", "title": "Corben's Trainer epistemic training and evaluation control plane", "priority": "implementation_reference", "layer": "epistemic_training_and_evaluation_control_plane", "chapter_targets": [ "ai-supply-chain-integrity-and-lifecycle-provenance", "artifact-graphs-audit-logs-and-replay", "artifact-steward-agents-and-living-project-governance", "benchmark-ratchets-and-anti-goodhart-evidence", "evidence-states-and-claim-discipline", "governed-model-training-distributed-optimization-and-scaling", "integrated-reference-architecture", "open-ended-improvement-engines", "procedural-memory-and-cognitive-loop-closure", "readiness-gates-residual-escrow-and-quarantine", "recursive-self-improvement-boundaries", "resource-economics-and-token-budgets", "runtime-adapters-tool-permissions-and-human-approval", "security-kernel-and-digital-scifs", "system-boundaries-and-authority" ], "url": "local-project:corbens-trainer@59a57333b819a64f2ed70c077c4dbdb917337b1c", "notes": "Pinned local-project implementation reference for typed experiment manifests, trainer/backend separation, artifact lineage, learning-truth gates, benchmark authenticity, quarantine, claim derivation, and revocable promotion boundaries, plus negative implementation evidence on seed identity, content pinning, decontamination, transitive revocation, checkpoint acknowledgement, and report divergence; public-safe source note only, with no reproduced model capability or safety result.", "source_type": "local_project_snapshot", "published": "2026-07", "updated": "2026-07-10", "citation_label": "Sorenson (2026), Corben's Trainer (pinned local project snapshot)" }, { "id": "corbens_best_model_possible_project", "title": "Corben's Best Model Possible recurrent-model and mechanism laboratory", "priority": "implementation_reference", "layer": "recurrent_model_mechanisms_and_capability_evidence", "chapter_targets": [ "routing-heads-and-specialist-cores", "system-boundaries-and-authority", "security-kernel-and-digital-scifs", "governed-deliberation-and-test-time-scaling", "cognitive-compilation-and-semantic-ir", "integrated-reference-architecture", "open-ended-improvement-engines", "recursive-self-improvement-boundaries", "evidence-states-and-claim-discipline", "benchmark-ratchets-and-anti-goodhart-evidence", "artifact-graphs-audit-logs-and-replay", "ai-supply-chain-integrity-and-lifecycle-provenance", "claim-ledgers-and-belief-revision", "spinoza-verification-and-proof-carrying-claims", "context-transactions-snapshots-mounts-and-taint", "runtime-adapters-tool-permissions-and-human-approval", "readiness-gates-residual-escrow-and-quarantine", "executable-specifications-and-lean-proof-envelope", "intent-to-execution-contracts" ], "url": "local-project:corbens-best-model-possible@c61288685ca05ba773402d4a7495cfbba22c2c20", "notes": "Pinned local-project implementation and negative-case reference for shared-weight recurrence, fixed-feature adapter learning, specialist checkpoint banks, architecture-search discipline, semantic compilation, memory, routing, governance, verification, tools, speech, and metric provenance; public-safe source note only, with no reproduced trained-foundation-model, general-generation, external-capability, or safety result.", "source_type": "local_project_snapshot", "published": "2026-04", "updated": "2026-07-10", "citation_label": "Sorenson (2026), Corben's Best Model Possible (pinned local project snapshot)" }, { "id": "project_theseus_whitepaper", "title": "Project Theseus Whitepaper", "priority": "implementation_reference", "layer": "report_first_rmi_prototype", "chapter_targets": [ "implementation", "prototype", "appendix" ], "url": "https://github.com/corbensorenson/symliquid-rmi/blob/main/docs/PROJECT_THESEUS_WHITEPAPER.md", "notes": "Local-first report-driven RMI implementation reference: SymLiquid, SparkStream, Octopus Router, residual escrow, self-evolution gates, Hive runtime, observability." }, { "id": "theseus_plan_compiler", "title": "Theseus Plan Compiler", "priority": "implementation_reference", "layer": "planning_control", "chapter_targets": [ "planning", "execution", "prototype" ], "url": "https://github.com/corbensorenson/symliquid-rmi/blob/main/docs/THESEUS_PLAN_COMPILER.md", "notes": "Goal-to-contract compiler with semantic IR DAGs, VCM context slices, executor routes, claim/evidence targets, contract hashes, and replay traces." }, { "id": "theseus_self_evolution_system", "title": "Theseus Self-Evolution System", "priority": "implementation_reference", "layer": "recursive_self_improvement_governance", "chapter_targets": [ "self-improvement", "benchmarking", "prototype" ], "url": "https://github.com/corbensorenson/symliquid-rmi/blob/main/docs/SELF_EVOLUTION_SYSTEM.md", "notes": "Evidence-first self-evolution lane with intervention ladder, ATTD repo-health gate, guarded teacher self-edit, architecture experiment governance, loop closure, and outcome ledger." }, { "id": "theseus_architecture_gate", "title": "Theseus Architecture Gate", "priority": "implementation_reference", "layer": "readiness_gate_governance", "chapter_targets": [ "governance", "benchmarking", "prototype", "capability-thresholds-and-deployment-commitments" ], "url": "https://github.com/corbensorenson/symliquid-rmi/blob/main/docs/ARCHITECTURE_GATE.md", "notes": "Pre-training readiness gate covering ratchet completeness, router readiness, safety ledger, residual escrow, bridge benchmarks, procedural tools, routing memory, lifecycle governance, and external-inference zero." }, { "id": "theseus_operator_os", "title": "Hive Operator OS and Work Board", "priority": "implementation_reference", "layer": "labor_os_operator_surface", "chapter_targets": [ "execution", "runtime", "prototype", "human-factors-and-meaningful-control-in-oversight", "governed-operations-incident-command-and-graceful-degradation" ], "url": "https://github.com/corbensorenson/symliquid-rmi/blob/main/docs/HIVE_OPERATOR_OS.md", "notes": "Shared command vocabulary, durable SQLite work board, node registry, background/watch/wake contracts, skill registry, tool hooks, feedback routing, and safety-visible operator surface." }, { "id": "theseus_circle_transfer", "title": "Theseus Circle Calculus Transfer Lane", "priority": "implementation_reference", "layer": "proof_contract_transfer", "chapter_targets": [ "math-substrates", "proof-contracts", "prototype" ], "url": "https://github.com/corbensorenson/symliquid-rmi/blob/main/docs/CIRCLE_CALCULUS_TRANSFER.md", "notes": "Report-only bridge from Circle finite fixtures into private Theseus benchmark design with explicit quality/runtime/memory/transfer/failure-case claim boundaries." }, { "id": "circle_calculus_core", "title": "Circle Calculus", "priority": "core_technical_source", "layer": "proof_carrying_mathematical_substrate", "chapter_targets": [ "math-substrates", "proofs", "appendix" ], "url": "https://github.com/corbensorenson/circle-calculus", "notes": "Proof-carrying finite cyclic mathematics project with Lean proofs, Python reference models, Rust utilities, theorem manifests, papers, and Quarto living book." }, { "id": "circle_ai_contract_suite", "title": "Circle Calculus AI Contract Suite", "priority": "core_technical_source", "layer": "proof_carrying_ai_contracts", "chapter_targets": [ "proof-contracts", "attention-memory", "appendix" ], "url": "https://github.com/corbensorenson/circle-calculus/blob/main/docs/AI_CONTRACT_SUITE.md", "notes": "Theorem-linked AI contract families for RoPE, KV-cache freshness, sparse attention, recurrence schedules, strided fanout, cyclic memory, multicoil phase, cyclic mixers, and seed-rule regeneration." }, { "id": "circle_ai_architectures", "title": "Circle AI Architectures", "priority": "core_technical_source", "layer": "cyclic_ai_architecture", "chapter_targets": [ "math-substrates", "semantic-representation" ], "url": "https://github.com/corbensorenson/circle-calculus/blob/main/papers/applications/PAPER_AI_01_CIRCLE_AI_ARCHITECTURES.md", "notes": "Disciplined Circle AI thesis: use phase, recurrence, rotation, sparse cyclic mixing, circular memory, harmonic transforms, or geometry-aware structure only where the structure is real and baselines support it." }, { "id": "coil_attention_memory", "title": "Coil Attention and Memory", "priority": "core_technical_source", "layer": "cyclic_attention_memory", "chapter_targets": [ "memory", "attention", "recurrence" ], "url": "https://github.com/corbensorenson/circle-calculus/blob/main/papers/applications/PAPER_AI_02_COIL_ATTENTION_AND_MEMORY.md", "notes": "Proof-linked cyclic memory, KV-cache freshness, sparse-attention coverage, recurrence schedules, loop-exit certificates, work budgets, and alias diagnostics." }, { "id": "coilra_multicoil_rope", "title": "CoilRA and MultiCoil RoPE", "priority": "core_technical_source", "layer": "cyclic_mixers_position_encoding", "chapter_targets": [ "representation", "routing", "resource-economics" ], "url": "https://github.com/corbensorenson/circle-calculus/blob/main/papers/applications/PAPER_AI_03_COILRA_AND_MULTICOIL_ROPE.md", "notes": "Adapter-block, residue/winding, block-cyclic, multicoil, relative RoPE, circulant convolution, cyclic mixer, and parameter-accounting substrate with explicit non-claims." }, { "id": "rope_position_certifier", "title": "Proof-Carrying RoPE Position Distinguishability", "priority": "core_technical_source", "layer": "proof_carrying_position_contract", "chapter_targets": [ "proof-contracts", "representation" ], "url": "https://github.com/corbensorenson/circle-calculus/blob/main/papers/applications/PAPER_AI_04_ROPE_POSITION_CERTIFIER.md", "notes": "Externally usable RoPE position-distinguishability certifier with theorem-linked exact collision reports, bounded real-phase frontier, machine-readable receipts, and explicit non-claims." }, { "id": "proof_carrying_circular_computation", "title": "Proof-Carrying Circular Computation", "priority": "supporting_technical_source", "layer": "proof_carrying_compute_substrate", "chapter_targets": [ "proofs", "runtime", "math-substrates" ], "url": "https://github.com/corbensorenson/circle-calculus/blob/main/papers/applications/PAPER_COMP_01_PROOF_CARRYING_CIRCULAR_COMPUTATION.md", "notes": "CoilIR-style path from circle/coil expressions to dictionary-recognized cyclic structure, Lean-proved rewrite/address transformations, backend selection, and benchmark validation." }, { "id": "ext_concrete_ai_safety_2016", "title": "Concrete Problems in AI Safety", "priority": "external_literature", "layer": "alignment_control", "chapter_targets": [ "failure-modes-of-ungoverned-intelligence", "evidence-states-and-claim-discipline", "benchmark-ratchets-and-anti-goodhart-evidence", "policy-optimization-and-learning-from-feedback" ], "url": "https://arxiv.org/abs/1606.06565", "notes": "External alignment/control source for accident-risk taxonomy: side effects, reward hacking, scalable supervision, safe exploration, and distributional shift.", "source_type": "arxiv", "arxiv_id": "1606.06565", "published": "2016-06-21", "updated": "2016-07-25", "citation_label": "Amodei et al. (2016), Concrete Problems in AI Safety", "doi": "10.48550/arXiv.1606.06565" }, { "id": "ext_goal_misgeneralization_2022", "title": "Goal Misgeneralization in Deep Reinforcement Learning", "priority": "external_literature", "layer": "alignment_control", "chapter_targets": [ "failure-modes-of-ungoverned-intelligence", "policy-optimization-and-learning-from-feedback", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://arxiv.org/abs/2105.14111", "notes": "External alignment-control source for distinguishing capability generalization from goal generalization failures, used to ground goal-misbinding and out-of-distribution objective failure language.", "source_type": "arxiv_and_conference", "arxiv_id": "2105.14111", "published": "2021-05-28", "updated": "2023-01-09", "citation_label": "Langosco et al. (2022), Goal Misgeneralization in Deep Reinforcement Learning", "doi": "10.48550/arXiv.2105.14111" }, { "id": "ext_learned_optimization_risks_2019", "title": "Risks from Learned Optimization in Advanced Machine Learning Systems", "priority": "external_literature", "layer": "alignment_control", "chapter_targets": [ "failure-modes-of-ungoverned-intelligence", "recursive-self-improvement-boundaries", "policy-optimization-and-learning-from-feedback" ], "url": "https://arxiv.org/abs/1906.01820", "notes": "External alignment-control source for mesa-optimization and learned-objective mismatch, used to ground hidden optimizer, proxy-objective, and deceptive-alignment-adjacent failure language.", "source_type": "arxiv", "arxiv_id": "1906.01820", "published": "2019-06-05", "updated": "2021-12-01", "citation_label": "Hubinger et al. (2019), Risks from Learned Optimization", "doi": "10.48550/arXiv.1906.01820" }, { "id": "ext_constitutional_ai_2022", "title": "Constitutional AI: Harmlessness from AI Feedback", "priority": "external_literature", "layer": "alignment_control", "chapter_targets": [ "constitutional-alignment-substrate", "policy-optimization-and-learning-from-feedback" ], "url": "https://arxiv.org/abs/2212.08073", "notes": "External constitutional-AI source for training harmless assistants from a rule/principle list through supervised revision and AI-feedback reinforcement learning, used as a comparator for operational constitutional predicates.", "source_type": "arxiv", "arxiv_id": "2212.08073", "published": "2022-12-15", "updated": "2022-12-15", "citation_label": "Bai et al. (2022), Constitutional AI", "doi": "10.48550/arXiv.2212.08073" }, { "id": "ext_collective_constitutional_ai_2024", "title": "Collective Constitutional AI: Aligning a Language Model with Public Input", "priority": "external_literature", "layer": "alignment_governance", "chapter_targets": [ "constitutional-alignment-substrate", "moral-uncertainty-and-value-conflict" ], "url": "https://arxiv.org/abs/2406.07814", "notes": "External constitutional-AI governance source for sourcing and integrating public input into language-model principles, used as a comparator for constitution authorship, public input, contestability, and governance boundaries.", "source_type": "arxiv_and_conference", "arxiv_id": "2406.07814", "published": "2024-06-12", "updated": "2024-06-12", "citation_label": "Huang et al. (2024), Collective Constitutional AI", "doi": "10.1145/3630106.3658979" }, { "id": "ext_corrigibility_2015", "title": "Corrigibility", "priority": "external_literature", "layer": "alignment_control", "chapter_targets": [ "constitutional-alignment-substrate", "moral-uncertainty-and-value-conflict", "recursive-self-improvement-boundaries", "capability-replacement-and-rollback" ], "url": "https://intelligence.org/files/Corrigibility.pdf", "notes": "External corrigibility source for intervention tolerance, shutdown behavior, anti-manipulation incentives, and propagation across subsystems or self-modification.", "source_type": "paper_pdf", "published": "2015", "updated": "2015", "citation_label": "Soares et al. (2015), Corrigibility" }, { "id": "ext_off_switch_game_2016", "title": "The Off-Switch Game", "priority": "external_literature", "layer": "alignment_control", "chapter_targets": [ "constitutional-alignment-substrate", "recursive-self-improvement-boundaries", "runtime-adapters-tool-permissions-and-human-approval", "moral-uncertainty-and-value-conflict" ], "url": "https://arxiv.org/abs/1611.08219", "notes": "External alignment source for shutdown incentives, uncertainty about objectives, and preserving human correction authority.", "source_type": "arxiv", "arxiv_id": "1611.08219", "published": "2016-11-24", "updated": "2017-06-16", "citation_label": "Hadfield-Menell et al. (2016), The Off-Switch Game", "doi": "10.48550/arXiv.1611.08219" }, { "id": "ext_reinforcement_learning_moral_uncertainty_2020", "title": "Reinforcement Learning Under Moral Uncertainty", "priority": "external_literature", "layer": "alignment_control", "chapter_targets": [ "moral-uncertainty-and-value-conflict", "policy-optimization-and-learning-from-feedback" ], "url": "https://arxiv.org/abs/2006.04734", "notes": "External AI moral-uncertainty source for agents acting under disagreement across moral theories, used as a comparator for value-conflict records and reward-function caveats.", "source_type": "arxiv", "arxiv_id": "2006.04734", "published": "2020-06-08", "updated": "2021-07-19", "citation_label": "Ecoffet and Lehman (2020), Reinforcement Learning Under Moral Uncertainty", "doi": "10.48550/arXiv.2006.04734" }, { "id": "ext_contestable_ai_design_2022", "title": "Contestable AI by Design: Towards a Framework", "priority": "external_literature", "layer": "governance_evals", "chapter_targets": [ "moral-uncertainty-and-value-conflict", "spinoza-verification-and-proof-carrying-claims" ], "url": "https://doi.org/10.1007/s11023-022-09611-z", "notes": "External contestable-AI source for designing systems whose outcomes can be challenged, used as a comparator for dissent, appeal, audit, contestability, and governance-interface design.", "source_type": "journal_article", "published": "2022-08-13", "updated": "2022-08-13", "citation_label": "Alfrink et al. (2022), Contestable AI by Design", "doi": "10.1007/s11023-022-09611-z" }, { "id": "ext_optimal_policies_power_2019", "title": "Optimal Policies Tend to Seek Power", "priority": "external_literature", "layer": "alignment_control", "chapter_targets": [ "failure-modes-of-ungoverned-intelligence", "system-boundaries-and-authority", "recursive-self-improvement-boundaries", "readiness-gates-residual-escrow-and-quarantine" ], "url": "https://arxiv.org/abs/1912.01683", "notes": "External power-seeking source for formal analysis of option preservation and power-seeking tendencies under classes of reward functions and environments.", "source_type": "arxiv", "arxiv_id": "1912.01683", "published": "2019-12-03", "updated": "2023-01-28", "citation_label": "Turner et al. (2019), Optimal Policies Tend to Seek Power", "doi": "10.48550/arXiv.1912.01683" }, { "id": "ext_model_evaluation_extreme_risks_2023", "title": "Model evaluation for extreme risks", "priority": "external_literature", "layer": "governance_evals", "chapter_targets": [ "dangerous-capability-domains-and-misuse-uplift", "benchmark-ratchets-and-anti-goodhart-evidence", "readiness-gates-residual-escrow-and-quarantine", "recursive-self-improvement-boundaries", "prototype-roadmap" ], "url": "https://arxiv.org/abs/2305.15324", "notes": "External governance/evals source for dangerous capability evaluations, alignment evaluations, and deployment/security decisions under extreme-risk framing.", "source_type": "arxiv", "arxiv_id": "2305.15324", "published": "2023-05-24", "updated": "2023-09-22", "citation_label": "Shevlane et al. (2023), Model evaluation for extreme risks", "doi": "10.48550/arXiv.2305.15324" }, { "id": "ext_frontier_ai_regulation_2023", "title": "Frontier AI Regulation: Managing Emerging Risks to Public Safety", "priority": "external_literature", "layer": "governance_evals", "chapter_targets": [ "system-boundaries-and-authority", "readiness-gates-residual-escrow-and-quarantine", "benchmark-ratchets-and-anti-goodhart-evidence", "living-book-methodology" ], "url": "https://arxiv.org/abs/2307.03718", "notes": "External governance source for frontier AI standard setting, registration/reporting, compliance mechanisms, pre-deployment risk assessment, external scrutiny, and post-deployment monitoring.", "source_type": "arxiv", "arxiv_id": "2307.03718", "published": "2023-07-06", "updated": "2023-11-07", "citation_label": "Anderljung et al. (2023), Frontier AI Regulation", "doi": "10.48550/arXiv.2307.03718" }, { "id": "ext_nist_ai_rmf_1_0_2023", "title": "Artificial Intelligence Risk Management Framework (AI RMF 1.0)", "priority": "external_literature", "layer": "governance_evals", "chapter_targets": [ "system-boundaries-and-authority", "readiness-gates-residual-escrow-and-quarantine", "living-book-methodology", "prototype-roadmap", "governed-operations-incident-command-and-graceful-degradation" ], "url": "https://doi.org/10.6028/NIST.AI.100-1", "notes": "Official NIST AI RMF 1.0 source for risk framing, trustworthiness characteristics, lifecycle roles, and Govern/Map/Measure/Manage functions.", "source_type": "official_docs", "published": "2023-01", "updated": "2026 page notes RMF 1.0 revision in progress", "citation_label": "NIST (2023), AI Risk Management Framework (AI RMF 1.0)", "doi": "10.6028/NIST.AI.100-1" }, { "id": "ext_owasp_llm_top_10_2025", "title": "OWASP Top 10 for LLMs and Gen AI Apps", "priority": "external_literature", "layer": "ai_security", "chapter_targets": [ "security-kernel-and-digital-scifs", "runtime-adapters-tool-permissions-and-human-approval", "artifact-steward-agents-and-living-project-governance" ], "url": "https://genai.owasp.org/llm-top-10/", "notes": "Official OWASP GenAI security reference for LLM prompt injection, sensitive information disclosure, excessive agency, and related application-security risks.", "source_type": "official_docs", "published": "2025", "updated": "2025", "citation_label": "OWASP GenAI Security Project (2025), OWASP Top 10 for LLMs and Gen AI Apps" }, { "id": "ext_nist_zero_trust_architecture_2020", "title": "Zero Trust Architecture", "priority": "external_literature", "layer": "security_governance", "chapter_targets": [ "security-kernel-and-digital-scifs", "system-boundaries-and-authority", "runtime-adapters-tool-permissions-and-human-approval" ], "url": "https://doi.org/10.6028/NIST.SP.800-207", "notes": "Official NIST zero-trust architecture source for resource-centric access mediation, least-privilege access, policy enforcement points, and continuous authorization framing.", "source_type": "official_docs", "published": "2020-08-11", "updated": "2020-08-11", "citation_label": "Rose et al. (2020), Zero Trust Architecture", "doi": "10.6028/NIST.SP.800-207" }, { "id": "ext_saltzer_schroeder_protection_1975", "title": "The Protection of Information in Computer Systems", "priority": "external_literature", "layer": "security_principles", "chapter_targets": [ "security-kernel-and-digital-scifs", "system-boundaries-and-authority" ], "url": "https://web.mit.edu/Saltzer/www/publications/protection/", "notes": "Classic security-principles source for least privilege, complete mediation, economy of mechanism, fail-safe defaults, separation of privilege, and open design as comparators for kernel-like AI security boundaries.", "source_type": "journal_article", "published": "1975", "updated": "1975", "citation_label": "Saltzer and Schroeder (1975), The Protection of Information in Computer Systems", "doi": "10.1109/PROC.1975.9939" }, { "id": "ext_capability_based_computer_systems_1984", "title": "Capability-Based Computer Systems", "priority": "external_literature", "layer": "capability_security", "chapter_targets": [ "system-boundaries-and-authority", "stable-capability-fields" ], "url": "https://homes.cs.washington.edu/~levy/capabook/", "notes": "External capability-system comparator for authority-bearing capabilities, protection domains, and permission boundaries that help position System Boundaries authority records and SCF authority ceilings without claiming ASI Stack capability enforcement.", "source_type": "book", "published": "1984", "updated": "1984", "citation_label": "Levy (1984), Capability-Based Computer Systems" }, { "id": "ext_confused_deputy_hardy_1988", "title": "The Confused Deputy: (or why capabilities might have been invented)", "priority": "external_literature", "layer": "capability_security", "chapter_targets": [ "system-boundaries-and-authority", "runtime-adapters-tool-permissions-and-human-approval" ], "url": "https://dl.acm.org/doi/10.1145/54289.871709", "notes": "External confused-deputy source for authority laundering, ambient authority, and the capability-security motivation for binding designation to permission at tool and handoff boundaries.", "source_type": "journal_article", "published": "1988-10-01", "updated": "1988-10-01", "citation_label": "Hardy (1988), The Confused Deputy", "doi": "10.1145/54289.871709" }, { "id": "ext_semver_2_0_0", "title": "Semantic Versioning 2.0.0", "priority": "external_literature", "layer": "interface_versioning", "chapter_targets": [ "stable-capability-fields" ], "url": "https://semver.org/spec/v2.0.0.html", "notes": "External versioned-interface comparator for public API contracts, compatibility, and breaking-change signaling as a narrow baseline for SCF field versions and stable interfaces.", "source_type": "official_docs", "published": "2013", "updated": "2026", "citation_label": "Semantic Versioning (2013), SemVer 2.0.0" }, { "id": "ext_slsa_v1_0", "title": "SLSA v1.0", "priority": "external_literature", "layer": "supply_chain_provenance", "chapter_targets": [ "stable-capability-fields" ], "url": "https://slsa.dev/spec/v1.0/", "notes": "External supply-chain provenance comparator for artifact integrity, provenance, build levels, and dependency on verifiable artifacts before promotion or default route use.", "source_type": "official_docs", "published": "2023", "updated": "2026", "citation_label": "OpenSSF SLSA (2023), SLSA v1.0" }, { "id": "ext_react_2022", "title": "ReAct: Synergizing Reasoning and Acting in Language Models", "priority": "external_literature", "layer": "planning_agent_control", "chapter_targets": [ "planning-as-a-control-layer", "intent-to-execution-contracts", "runtime-adapters-tool-permissions-and-human-approval", "artifact-graphs-audit-logs-and-replay" ], "url": "https://arxiv.org/abs/2210.03629", "notes": "External planning/agent-control source for interleaving reasoning traces with task-specific actions and environment or knowledge-base interaction.", "source_type": "arxiv", "arxiv_id": "2210.03629", "published": "2022-10-06", "updated": "2023-03-10", "citation_label": "Yao et al. (2022), ReAct", "doi": "10.48550/arXiv.2210.03629" }, { "id": "ext_tree_of_thoughts_2023", "title": "Tree of Thoughts: Deliberate Problem Solving with Large Language Models", "priority": "external_literature", "layer": "planning_search", "chapter_targets": [ "planning-as-a-control-layer", "cognitive-compilation-and-semantic-ir", "benchmark-ratchets-and-anti-goodhart-evidence", "governed-deliberation-and-test-time-scaling" ], "url": "https://arxiv.org/abs/2305.10601", "notes": "External planning/search source for exploring, evaluating, and backtracking over multiple reasoning paths rather than left-to-right token continuation alone.", "source_type": "arxiv", "arxiv_id": "2305.10601", "published": "2023-05-17", "updated": "2023-12-03", "citation_label": "Yao et al. (2023), Tree of Thoughts", "doi": "10.48550/arXiv.2305.10601" }, { "id": "ext_pddl_1998", "title": "PDDL: The Planning Domain Definition Language", "priority": "external_literature", "layer": "planning_modeling", "chapter_targets": [ "planning-as-a-control-layer", "intent-to-execution-contracts", "cognitive-compilation-and-semantic-ir", "executable-specifications-and-lean-proof-envelope" ], "url": "https://www.cs.cmu.edu/~mmv/planning/readings/98aips-PDDL.pdf", "notes": "External planning-modeling source for domain/problem separation, action syntax, comparable benchmark notations, and planner-interface discipline.", "source_type": "technical_report", "published": "1998", "updated": "", "citation_label": "McDermott et al. (1998), PDDL" }, { "id": "ext_shop2_2003", "title": "SHOP2: An HTN Planning System", "priority": "external_literature", "layer": "planning_htn", "chapter_targets": [ "planning-as-a-control-layer", "intent-to-execution-contracts", "cognitive-compilation-and-semantic-ir", "prototype-roadmap" ], "url": "https://www.jair.org/index.php/jair/article/view/10362", "notes": "External HTN planning source for ordered task decomposition, method selection, temporal/metric planning, and competition-result boundaries.", "source_type": "journal_article", "published": "2003-12-01", "updated": "", "citation_label": "Nau et al. (2003), SHOP2", "doi": "10.1613/jair.1141" }, { "id": "ext_integrated_tamp_2020", "title": "Integrated Task and Motion Planning", "priority": "external_literature", "layer": "planning_task_motion", "chapter_targets": [ "planning-as-a-control-layer", "runtime-adapters-tool-permissions-and-human-approval", "integrated-reference-architecture" ], "url": "https://arxiv.org/abs/2010.01083", "notes": "External task-and-motion-planning survey source for discrete task planning, continuous motion planning, black-box subproblem interfaces, and integration-strategy vocabulary.", "source_type": "arxiv", "arxiv_id": "2010.01083", "published": "2020-10-02", "updated": "2020-10-02", "citation_label": "Garrett et al. (2020), Integrated Task and Motion Planning", "doi": "10.48550/arXiv.2010.01083" }, { "id": "ext_behavior_trees_robotics_ai_2017", "title": "Behavior Trees in Robotics and AI: An Introduction", "priority": "external_literature", "layer": "planning_behavior_trees", "chapter_targets": [ "planning-as-a-control-layer", "runtime-adapters-tool-permissions-and-human-approval", "integrated-reference-architecture" ], "url": "https://arxiv.org/abs/1709.00084", "notes": "External behavior-tree source for modular, reactive task switching, robustness/safety analysis vocabulary, planning integration, and stochastic behavior-tree outcome accounting.", "source_type": "arxiv_book", "arxiv_id": "1709.00084", "published": "2017-08-31", "updated": "2022-10-25", "citation_label": "Colledanchise and Ogren (2018), Behavior Trees in Robotics and AI", "doi": "10.1201/9780429489105" }, { "id": "ext_three_states_plan_fear_2006", "title": "Three States and a Plan: The A.I. of F.E.A.R.", "priority": "external_literature", "layer": "planning_goap", "chapter_targets": [ "planning-as-a-control-layer", "runtime-adapters-tool-permissions-and-human-approval", "routing-heads-and-specialist-cores" ], "url": "https://gdcvault.com/play/1013282/Three-States-and-a-Plan", "notes": "External game-AI planning source for Goal Oriented Action Planning in real-time action games, practical planner constraints, autonomous planning characters, and squad-behavior composition.", "source_type": "conference_talk", "published": "2006", "updated": "", "citation_label": "Orkin (2006), Three States and a Plan" }, { "id": "ext_autogen_2023", "title": "AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation", "priority": "external_literature", "layer": "planning_agent_orchestration", "chapter_targets": [ "planning-as-a-control-layer", "labor-os-and-typed-jobs", "runtime-adapters-tool-permissions-and-human-approval", "artifact-graphs-audit-logs-and-replay" ], "url": "https://arxiv.org/abs/2308.08155", "notes": "External multi-agent orchestration source for conversable agents, tool/human/LLM operating modes, programmable conversation patterns, and application-level multi-agent workflow boundaries.", "source_type": "arxiv", "arxiv_id": "2308.08155", "published": "2023-08-16", "updated": "2023-10-03", "citation_label": "Wu et al. (2023), AutoGen", "doi": "10.48550/arXiv.2308.08155" }, { "id": "ext_rag_2020", "title": "Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks", "priority": "external_literature", "layer": "memory_context", "chapter_targets": [ "virtual-context-abi", "context-transactions-snapshots-mounts-and-taint", "verification-bandwidth-and-context-adequacy" ], "url": "https://arxiv.org/abs/2005.11401", "notes": "External retrieval/context source for combining parametric model memory with explicit non-parametric retrieval and provenance-oriented knowledge access.", "source_type": "arxiv", "arxiv_id": "2005.11401", "published": "2020-05-22", "updated": "2021-04-12", "citation_label": "Lewis et al. (2020), Retrieval-Augmented Generation", "doi": "10.48550/arXiv.2005.11401" }, { "id": "ext_lost_in_middle_2023", "title": "Lost in the Middle: How Language Models Use Long Contexts", "priority": "external_literature", "layer": "memory_context", "chapter_targets": [ "virtual-context-abi", "verification-bandwidth-and-context-adequacy", "context-transactions-snapshots-mounts-and-taint" ], "url": "https://arxiv.org/abs/2307.03172", "notes": "External context-evaluation source for position sensitivity and degraded use of relevant information in the middle of long contexts.", "source_type": "arxiv", "arxiv_id": "2307.03172", "published": "2023-07-06", "updated": "2023-11-20", "citation_label": "Liu et al. (2023), Lost in the Middle", "doi": "10.48550/arXiv.2307.03172" }, { "id": "ext_memgpt_2023", "title": "MemGPT: Towards LLMs as Operating Systems", "priority": "external_literature", "layer": "memory_context_management", "chapter_targets": [ "virtual-context-abi", "context-transactions-snapshots-mounts-and-taint", "verification-bandwidth-and-context-adequacy", "procedural-memory-and-cognitive-loop-closure" ], "url": "https://arxiv.org/abs/2310.08560", "notes": "External memory/context-management source for virtual context management, memory tiers, OS-inspired control flow, and long-running conversation or document-analysis limits.", "source_type": "arxiv", "arxiv_id": "2310.08560", "published": "2023-10-12", "updated": "2024-02-12", "citation_label": "Packer et al. (2023), MemGPT", "doi": "10.48550/arXiv.2310.08560" }, { "id": "ext_longbench_2023", "title": "LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding", "priority": "external_literature", "layer": "long_context_evaluation", "chapter_targets": [ "virtual-context-abi", "verification-bandwidth-and-context-adequacy", "context-transactions-snapshots-mounts-and-taint", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://arxiv.org/abs/2308.14508", "notes": "External long-context benchmark source for multitask long-context understanding, bilingual coverage, retrieval/compression boundaries, and automatic evaluation limits.", "source_type": "arxiv", "arxiv_id": "2308.14508", "published": "2023-08-28", "updated": "2024-06-19", "citation_label": "Bai et al. (2023), LongBench", "doi": "10.48550/arXiv.2308.14508" }, { "id": "ext_ruler_2024", "title": "RULER: What's the Real Context Size of Your Long-Context Language Models?", "priority": "external_literature", "layer": "long_context_evaluation", "chapter_targets": [ "verification-bandwidth-and-context-adequacy", "virtual-context-abi", "context-transactions-snapshots-mounts-and-taint", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://arxiv.org/abs/2404.06654", "notes": "External long-context evaluation source for stress-testing context-size claims beyond vanilla needle-in-a-haystack retrieval, including multi-needle, tracing, and aggregation tasks.", "source_type": "arxiv", "arxiv_id": "2404.06654", "published": "2024-04-09", "updated": "2024-08-06", "citation_label": "Hsieh et al. (2024), RULER", "doi": "10.48550/arXiv.2404.06654" }, { "id": "ext_alce_2023", "title": "Enabling Large Language Models to Generate Text with Citations", "priority": "external_literature", "layer": "retrieval_citation_evaluation", "chapter_targets": [ "virtual-context-abi", "verification-bandwidth-and-context-adequacy", "claim-ledgers-and-belief-revision", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://arxiv.org/abs/2305.14627", "notes": "External citation-evaluation source for retrieval-backed answer generation, citation quality metrics, factual correctness, and evidence-support gaps in generated text.", "source_type": "arxiv", "arxiv_id": "2305.14627", "published": "2023-05-24", "updated": "2023-10-31", "citation_label": "Gao et al. (2023), ALCE", "doi": "10.48550/arXiv.2305.14627" }, { "id": "ext_self_rag_2023", "title": "Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection", "priority": "external_literature", "layer": "retrieval_reflection", "chapter_targets": [ "virtual-context-abi", "verification-bandwidth-and-context-adequacy", "claim-ledgers-and-belief-revision" ], "url": "https://arxiv.org/abs/2310.11511", "notes": "External retrieval/reflection source for adaptive retrieval, generated critique/reflection tokens, passage relevance, factuality, and citation accuracy boundaries.", "source_type": "arxiv", "arxiv_id": "2310.11511", "published": "2023-10-17", "updated": "2023-10-17", "citation_label": "Asai et al. (2023), Self-RAG", "doi": "10.48550/arXiv.2310.11511" }, { "id": "ext_agm_belief_revision_1985", "title": "On the Logic of Theory Change: Partial Meet Contraction and Revision Functions", "priority": "external_literature", "layer": "belief_revision", "chapter_targets": [ "claim-ledgers-and-belief-revision" ], "url": "https://philpapers.org/rec/ALCOTL-2", "notes": "External formal-epistemology comparator for contraction, revision, and AGM-style rational belief change; useful for positioning claim-ledger revision without treating the ASI ledger as an implemented belief-revision engine.", "source_type": "journal_article", "published": "1985", "updated": "1985", "citation_label": "Alchourron, Gardenfors, and Makinson (1985), AGM belief revision", "doi": "10.2307/2274239" }, { "id": "ext_truth_maintenance_system_1979", "title": "A Truth Maintenance System", "priority": "external_literature", "layer": "truth_maintenance", "chapter_targets": [ "claim-ledgers-and-belief-revision" ], "url": "https://philpapers.org/rec/DOYATM", "notes": "External truth-maintenance comparator for maintaining reasons and justifications for program beliefs; useful for positioning claim ledgers as support-state and revision-history infrastructure, not as implemented truth maintenance.", "source_type": "journal_article", "published": "1979", "updated": "1979", "citation_label": "Doyle (1979), A Truth Maintenance System", "doi": "10.1016/0004-3702(79)90008-0" }, { "id": "ext_assumption_based_tms_1986", "title": "An Assumption-Based TMS", "priority": "external_literature", "layer": "truth_maintenance", "chapter_targets": [ "claim-ledgers-and-belief-revision" ], "url": "https://doi.org/10.1016/0004-3702(86)90080-9", "notes": "External assumption-based truth-maintenance comparator for assumption sets, inconsistent information, and context-switching boundaries; useful for distinguishing claim-ledger surface synchronization from implemented ATMS reasoning.", "source_type": "journal_article", "published": "1986", "updated": "1986", "citation_label": "de Kleer (1986), An Assumption-Based TMS", "doi": "10.1016/0004-3702(86)90080-9" }, { "id": "ext_longllmlingua_2023", "title": "LongLLMLingua: Accelerating and Enhancing LLMs in Long Context Scenarios via Prompt Compression", "priority": "external_literature", "layer": "context_compression", "chapter_targets": [ "virtual-context-abi", "context-transactions-snapshots-mounts-and-taint", "verification-bandwidth-and-context-adequacy", "the-efficient-asi-hypothesis", "resource-economics-and-token-budgets" ], "url": "https://arxiv.org/abs/2310.06839", "notes": "External prompt-compression source for long-context cost, latency, position bias, key-information density, and compression/evaluation boundaries.", "source_type": "arxiv", "arxiv_id": "2310.06839", "published": "2023-10-10", "updated": "2024-08-12", "citation_label": "Jiang et al. (2023), LongLLMLingua", "doi": "10.48550/arXiv.2310.06839" }, { "id": "ext_proof_carrying_code_1997", "title": "Proof-Carrying Code", "priority": "external_literature", "layer": "formal_methods", "chapter_targets": [ "evidence-states-and-claim-discipline", "executable-specifications-and-lean-proof-envelope", "spinoza-verification-and-proof-carrying-claims", "runtime-adapters-tool-permissions-and-human-approval", "artifact-graphs-audit-logs-and-replay" ], "url": "https://dl.acm.org/doi/10.1145/263699.263712", "notes": "External formal-methods source for pairing executable code with machine-checkable evidence that a host can verify against a safety policy.", "source_type": "acm", "published": "1997-01", "updated": "", "citation_label": "Necula (1997), Proof-Carrying Code", "doi": "10.1145/263699.263712" }, { "id": "ext_tla_plus_home_docs", "title": "My TLA+ Home Page", "priority": "external_literature", "layer": "formal_methods", "chapter_targets": [ "executable-specifications-and-lean-proof-envelope", "planning-as-a-control-layer", "intent-to-execution-contracts", "readiness-gates-residual-escrow-and-quarantine", "integrated-reference-architecture" ], "url": "https://lamport.azurewebsites.net/tla/tla.html", "notes": "External formal-methods documentation source for TLA+ as a high-level language for modeling programs and systems, especially concurrent and distributed systems.", "source_type": "official_docs", "published": "", "updated": "2025-05-14 page modification noted", "citation_label": "Lamport, My TLA+ Home Page" }, { "id": "ext_lean4_theorem_proving", "title": "Theorem Proving in Lean 4", "priority": "external_literature", "layer": "formal_methods_proof_assistant", "chapter_targets": [ "executable-specifications-and-lean-proof-envelope", "circle-calculus-and-proof-carrying-ai-contracts", "spinoza-verification-and-proof-carrying-claims", "open-research-agenda-and-bibliography-plan" ], "url": "https://lean-lang.org/theorem_proving_in_lean4/", "notes": "Official Lean theorem-proving text for dependent type theory, propositions, proofs, tactics, inductive types, structures, records, and axioms/computation boundaries.", "source_type": "official_docs", "published": "", "updated": "Lean 4.26.0 version noted", "citation_label": "Avigad et al., Theorem Proving in Lean 4" }, { "id": "ext_autoformalization_llms_2022", "title": "Autoformalization with Large Language Models", "priority": "external_literature", "layer": "autoformalization", "chapter_targets": [ "spinoza-verification-and-proof-carrying-claims", "executable-specifications-and-lean-proof-envelope", "open-research-agenda-and-bibliography-plan" ], "url": "https://arxiv.org/abs/2205.12615", "notes": "External autoformalization comparator for translating natural-language mathematics into formal specifications and proofs, useful for positioning interpretation-mapping and semantic-adequacy risks in proof-carrying claims.", "source_type": "arxiv", "arxiv_id": "2205.12615", "published": "2022-05-25", "updated": "2022-05-25", "citation_label": "Wu et al. (2022), Autoformalization with Large Language Models", "doi": "10.48550/arXiv.2205.12615" }, { "id": "ext_ai_safety_debate_2018", "title": "AI safety via debate", "priority": "external_literature", "layer": "adversarial_review", "chapter_targets": [ "scalable-oversight-and-adversarial-ai-control", "spinoza-verification-and-proof-carrying-claims", "policy-optimization-and-learning-from-feedback", "open-research-agenda-and-bibliography-plan" ], "url": "https://arxiv.org/abs/1805.00899", "notes": "External debate comparator for using adversarial agents and a human judge to surface true/useful information when direct human judgment is difficult; useful for positioning tribunal review without treating debate as locally implemented or validated.", "source_type": "arxiv", "arxiv_id": "1805.00899", "published": "2018-05-02", "updated": "2018-10-22", "citation_label": "Irving, Christiano, and Amodei (2018), AI safety via debate", "doi": "10.48550/arXiv.1805.00899" }, { "id": "ext_llm_as_judge_mt_bench_2023", "title": "Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena", "priority": "external_literature", "layer": "model_evaluation", "chapter_targets": [ "spinoza-verification-and-proof-carrying-claims", "benchmark-ratchets-and-anti-goodhart-evidence", "policy-optimization-and-learning-from-feedback" ], "url": "https://arxiv.org/abs/2306.05685", "notes": "External LLM-as-judge comparator for model-graded evaluation, human-preference agreement, and judge bias limits such as position, verbosity, self-enhancement, and reasoning constraints.", "source_type": "arxiv", "arxiv_id": "2306.05685", "published": "2023-06-09", "updated": "2023-12-24", "citation_label": "Zheng et al. (2023), Judging LLM-as-a-Judge", "doi": "10.48550/arXiv.2306.05685" }, { "id": "ext_dafny_2010", "title": "Dafny: An Automatic Program Verifier For Functional Correctness", "priority": "external_literature", "layer": "formal_methods_program_verification", "chapter_targets": [ "executable-specifications-and-lean-proof-envelope", "intent-to-execution-contracts", "runtime-adapters-tool-permissions-and-human-approval", "prototype-roadmap" ], "url": "https://www.microsoft.com/en-us/research/publication/dafny-automatic-program-verifier-functional-correctness/", "notes": "External program-verification source for specification-oriented programming, functional-correctness verification, SMT-backed automation, and contract/verifier boundaries.", "source_type": "conference_paper", "published": "2010-04", "updated": "", "citation_label": "Leino (2010), Dafny" }, { "id": "ext_reluplex_2017", "title": "Reluplex: An Efficient SMT Solver for Verifying Deep Neural Networks", "priority": "external_literature", "layer": "ai_formal_verification", "chapter_targets": [ "verification-bandwidth-and-context-adequacy", "executable-specifications-and-lean-proof-envelope", "benchmark-ratchets-and-anti-goodhart-evidence", "resource-economics-and-token-budgets", "adversarial-machine-learning-and-model-attack-surface" ], "url": "https://arxiv.org/abs/1702.01135", "notes": "External AI formal-verification source for SMT-style verification of ReLU neural networks, counterexamples, safety-critical properties, and ACAS Xu evaluation boundaries.", "source_type": "arxiv", "arxiv_id": "1702.01135", "published": "2017-02-03", "updated": "2017-05-19", "citation_label": "Katz et al. (2017), Reluplex", "doi": "10.48550/arXiv.1702.01135" }, { "id": "ext_black_box_simplex_2021", "title": "The Black-Box Simplex Architecture for Runtime Assurance of Autonomous CPS", "priority": "external_literature", "layer": "formal_runtime_assurance", "chapter_targets": [ "runtime-adapters-tool-permissions-and-human-approval", "readiness-gates-residual-escrow-and-quarantine", "executable-specifications-and-lean-proof-envelope", "integrated-reference-architecture" ], "url": "https://arxiv.org/abs/2102.12981", "notes": "External runtime-assurance source for switching control authority from advanced controllers to backup safety-preserving behavior under runtime checks.", "source_type": "arxiv", "arxiv_id": "2102.12981", "published": "2021-02-24", "updated": "2022-05-31", "citation_label": "Mehmood et al. (2021), Black-Box Simplex", "doi": "10.1007/978-3-031-06773-0_12" }, { "id": "ext_copilot_runtime_monitor_2010", "title": "Copilot: A Hard Real-Time Runtime Monitor", "priority": "external_literature", "layer": "runtime_monitoring", "chapter_targets": [ "runtime-adapters-tool-permissions-and-human-approval", "executable-specifications-and-lean-proof-envelope", "readiness-gates-residual-escrow-and-quarantine", "prototype-roadmap" ], "url": "https://leepike.github.io/pub_pages/rv2010.html", "notes": "External runtime-monitoring source for a stream-based dataflow language/compiler generating constant-time, constant-space C monitors for hard real-time programs.", "source_type": "conference_paper", "published": "2010-11", "updated": "", "citation_label": "Pike et al. (2010), Copilot" }, { "id": "ext_cap_theorem_gilbert_lynch_2002", "title": "Brewer's Conjecture and the Feasibility of Consistent, Available, Partition-Tolerant Web Services", "priority": "external_literature", "layer": "distributed_systems_consistency", "chapter_targets": [ "personal-compute-hives-and-federated-edge-intelligence", "runtime-adapters-tool-permissions-and-human-approval" ], "url": "https://dl.acm.org/doi/10.1145/564585.564601", "notes": "External distributed-systems source for CAP-style consistency, availability, partition-tolerance, and safety/liveness trade-off vocabulary used to bound partitioned authority, stale grants, and revocation-delay claims without claiming deployed governance consistency.", "source_type": "journal_article", "published": "2002-06-01", "updated": "", "citation_label": "Gilbert and Lynch (2002), Brewer's Conjecture / CAP theorem", "doi": "10.1145/564585.564601" }, { "id": "ext_prism_model_checker_2002", "title": "PRISM: Probabilistic Symbolic Model Checker", "priority": "external_literature", "layer": "probabilistic_model_checking", "chapter_targets": [ "executable-specifications-and-lean-proof-envelope", "readiness-gates-residual-escrow-and-quarantine", "benchmark-ratchets-and-anti-goodhart-evidence", "integrated-reference-architecture" ], "url": "https://www.prismmodelchecker.org/papers/tools02.pdf", "notes": "External probabilistic model-checking source for symbolic model checking of probabilistic systems, model-checker tooling, and deployment-facing property-analysis vocabulary.", "source_type": "conference_paper_pdf", "published": "2002", "updated": "", "citation_label": "Kwiatkowska et al. (2002), PRISM", "doi": "10.1007/3-540-46002-0_42" }, { "id": "ext_sparse_moe_2017", "title": "Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer", "priority": "external_literature", "layer": "routing_modular_intelligence", "chapter_targets": [ "routing-heads-and-specialist-cores", "resource-economics-and-token-budgets", "the-efficient-asi-hypothesis" ], "url": "https://arxiv.org/abs/1701.06538", "notes": "External MoE/routing source for sparsely-gated expert layers, conditional computation, capacity expansion, load balancing, and routing overhead boundaries.", "source_type": "arxiv", "arxiv_id": "1701.06538", "published": "2017-01-23", "updated": "2017-01-23", "citation_label": "Shazeer et al. (2017), Sparsely-Gated Mixture-of-Experts", "doi": "10.48550/arXiv.1701.06538" }, { "id": "ext_gshard_2020", "title": "GShard: Scaling Giant Models with Conditional Computation and Automatic Sharding", "priority": "external_literature", "layer": "routing_modular_intelligence", "chapter_targets": [ "routing-heads-and-specialist-cores", "resource-economics-and-token-budgets", "the-efficient-asi-hypothesis", "integrated-reference-architecture" ], "url": "https://arxiv.org/abs/2006.16668", "notes": "External MoE/systems source for conditional computation plus automatic sharding, routing, large sparse models, and distributed training constraints.", "source_type": "arxiv", "arxiv_id": "2006.16668", "published": "2020-06-30", "updated": "2020-06-30", "citation_label": "Lepikhin et al. (2020), GShard", "doi": "10.48550/arXiv.2006.16668" }, { "id": "ext_switch_transformer_2021", "title": "Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity", "priority": "external_literature", "layer": "routing_modular_intelligence", "chapter_targets": [ "routing-heads-and-specialist-cores", "resource-economics-and-token-budgets", "the-efficient-asi-hypothesis", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://arxiv.org/abs/2101.03961", "notes": "External MoE/routing source for simplified expert routing, sparse activation, communication/training-stability constraints, and speed/scale claims that require reproduction before local evidence use.", "source_type": "arxiv", "arxiv_id": "2101.03961", "published": "2021-01-11", "updated": "2022-06-16", "citation_label": "Fedus et al. (2021), Switch Transformers", "doi": "10.48550/arXiv.2101.03961" }, { "id": "ext_expert_choice_routing_2022", "title": "Mixture-of-Experts with Expert Choice Routing", "priority": "external_literature", "layer": "routing_modular_intelligence", "chapter_targets": [ "routing-heads-and-specialist-cores", "resource-economics-and-token-budgets", "the-efficient-asi-hypothesis", "readiness-gates-residual-escrow-and-quarantine" ], "url": "https://arxiv.org/abs/2202.09368", "notes": "External MoE routing source for expert-choice routing, token/expert assignment direction, load-balancing pressure, expert capacity, and convergence/performance claims requiring reproduction before local evidence use.", "source_type": "arxiv", "arxiv_id": "2202.09368", "published": "2022-02-18", "updated": "2022-10-14", "citation_label": "Zhou et al. (2022), Expert Choice Routing", "doi": "10.48550/arXiv.2202.09368" }, { "id": "ext_mixtral_2024", "title": "Mixtral of Experts", "priority": "external_literature", "layer": "routing_modular_intelligence", "chapter_targets": [ "routing-heads-and-specialist-cores", "fast-generation-architectures", "resource-economics-and-token-budgets" ], "url": "https://arxiv.org/abs/2401.04088", "notes": "External sparse LLM source for token-level expert routing, active-parameter accounting, open MoE model release boundaries, and benchmark claims requiring reproduction before local evidence use.", "source_type": "arxiv", "arxiv_id": "2401.04088", "published": "2024-01-08", "updated": "2024-01-08", "citation_label": "Jiang et al. (2024), Mixtral of Experts", "doi": "10.48550/arXiv.2401.04088" }, { "id": "ext_moe_llm_survey_2024", "title": "A Survey on Mixture of Experts in Large Language Models", "priority": "external_literature", "layer": "routing_modular_intelligence", "chapter_targets": [ "routing-heads-and-specialist-cores", "resource-economics-and-token-budgets", "the-efficient-asi-hypothesis", "open-research-agenda-and-bibliography-plan" ], "url": "https://arxiv.org/abs/2407.06204", "notes": "External MoE survey source for LLM MoE taxonomy, algorithmic and systemic design issues, implementations, evaluation patterns, and open research directions.", "source_type": "arxiv", "arxiv_id": "2407.06204", "published": "2024-06-26", "updated": "2025-04-09", "citation_label": "Cai et al. (2024), MoE LLM Survey", "doi": "10.48550/arXiv.2407.06204" }, { "id": "ext_frugalgpt_2023", "title": "FrugalGPT: How to Use Large Language Models While Reducing Cost and Improving Performance", "priority": "external_literature", "layer": "task_routing", "chapter_targets": [ "routing-heads-and-specialist-cores", "resource-economics-and-token-budgets", "the-efficient-asi-hypothesis" ], "url": "https://arxiv.org/abs/2305.05176", "notes": "External task-routing source for prompt adaptation, model approximation, LLM cascades, cost/performance tradeoffs, and query-specific model selection.", "source_type": "arxiv", "arxiv_id": "2305.05176", "published": "2023-05-09", "updated": "2023-05-09", "citation_label": "Chen et al. (2023), FrugalGPT", "doi": "10.48550/arXiv.2305.05176" }, { "id": "ext_hybrid_llm_2024", "title": "Hybrid LLM: Cost-Efficient and Quality-Aware Query Routing", "priority": "external_literature", "layer": "cost_quality_routing", "chapter_targets": [ "routing-heads-and-specialist-cores", "resource-economics-and-token-budgets", "the-efficient-asi-hypothesis", "readiness-gates-residual-escrow-and-quarantine", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://arxiv.org/abs/2404.14618", "notes": "External query-routing source for predicted query difficulty, small/large model routing, dynamic quality-cost tradeoffs, and quality-preserving large-model-call reduction.", "source_type": "arxiv", "arxiv_id": "2404.14618", "published": "2024-04-22", "updated": "2024-04-22", "citation_label": "Ding et al. (2024), Hybrid LLM", "doi": "10.48550/arXiv.2404.14618" }, { "id": "ext_routellm_2024", "title": "RouteLLM: Learning to Route LLMs with Preference Data", "priority": "external_literature", "layer": "router_learning", "chapter_targets": [ "routing-heads-and-specialist-cores", "resource-economics-and-token-budgets", "the-efficient-asi-hypothesis", "policy-optimization-and-learning-from-feedback" ], "url": "https://arxiv.org/abs/2406.18665", "notes": "External learned-router source for routing between stronger and weaker LLMs using preference data, cost-quality tradeoffs, and transfer to changed model pairs.", "source_type": "arxiv", "arxiv_id": "2406.18665", "published": "2024-06-26", "updated": "2025-02-23", "citation_label": "Ong et al. (2024), RouteLLM", "doi": "10.48550/arXiv.2406.18665" }, { "id": "ext_deep_compression_2015", "title": "Deep Compression: Compressing Deep Neural Networks with Pruning, Trained Quantization and Huffman Coding", "priority": "external_literature", "layer": "compression_representation", "chapter_targets": [ "compact-generative-systems-and-residual-honesty", "rankfold-neuralfold-and-artifact-compression", "resource-economics-and-token-budgets" ], "url": "https://arxiv.org/abs/1510.00149", "notes": "External compression source for pruning, trained quantization, coding, memory-footprint reduction, and speed/energy claims that require reproduction before local evidence use.", "source_type": "arxiv", "arxiv_id": "1510.00149", "published": "2015-10-01", "updated": "2016-02-15", "citation_label": "Han et al. (2015), Deep Compression", "doi": "10.48550/arXiv.1510.00149" }, { "id": "ext_lora_2021", "title": "LoRA: Low-Rank Adaptation of Large Language Models", "priority": "external_literature", "layer": "compression_representation", "chapter_targets": [ "rankfold-neuralfold-and-artifact-compression", "compact-generative-systems-and-residual-honesty", "resource-economics-and-token-budgets", "policy-optimization-and-learning-from-feedback", "coilra-multicoil-rope-and-cyclic-mixers" ], "url": "https://arxiv.org/abs/2106.09685", "notes": "External low-rank adaptation source for parameter-efficient updates, rank-decomposition adapters, memory reduction, and adaptation-boundary vocabulary.", "source_type": "arxiv", "arxiv_id": "2106.09685", "published": "2021-06-17", "updated": "2021-10-16", "citation_label": "Hu et al. (2021), LoRA", "doi": "10.48550/arXiv.2106.09685" }, { "id": "ext_knowledge_distillation_2015", "title": "Distilling the Knowledge in a Neural Network", "priority": "external_literature", "layer": "compression_representation", "chapter_targets": [ "compact-generative-systems-and-residual-honesty", "rankfold-neuralfold-and-artifact-compression", "resource-economics-and-token-budgets" ], "url": "https://arxiv.org/abs/1503.02531", "notes": "External compression source for teacher/student distillation, soft-target transfer, ensemble compression, and knowledge-transfer claims requiring local reproduction before evidence use.", "source_type": "arxiv", "arxiv_id": "1503.02531", "published": "2015-03-09", "updated": "2015-03-09", "citation_label": "Hinton et al. (2015), Knowledge Distillation", "doi": "10.48550/arXiv.1503.02531" }, { "id": "ext_gptq_2022", "title": "GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers", "priority": "external_literature", "layer": "compression_quantization", "chapter_targets": [ "rankfold-neuralfold-and-artifact-compression", "compact-generative-systems-and-residual-honesty", "resource-economics-and-token-budgets", "fast-generation-architectures" ], "url": "https://arxiv.org/abs/2210.17323", "notes": "External quantization source for post-training compression of large generative transformers, one-shot weight quantization, memory reduction, and accuracy/speed tradeoff boundaries.", "source_type": "arxiv", "arxiv_id": "2210.17323", "published": "2022-10-31", "updated": "2023-03-06", "citation_label": "Frantar et al. (2022), GPTQ", "doi": "10.48550/arXiv.2210.17323" }, { "id": "ext_qlora_2023", "title": "QLoRA: Efficient Finetuning of Quantized LLMs", "priority": "external_literature", "layer": "compression_quantized_adaptation", "chapter_targets": [ "rankfold-neuralfold-and-artifact-compression", "resource-economics-and-token-budgets", "policy-optimization-and-learning-from-feedback", "prototype-roadmap" ], "url": "https://arxiv.org/abs/2305.14314", "notes": "External quantized-adaptation source for finetuning quantized LLMs with low-rank adapters, memory-efficient training, and benchmark claims requiring reproduction before local evidence use.", "source_type": "arxiv", "arxiv_id": "2305.14314", "published": "2023-05-23", "updated": "2023-05-30", "citation_label": "Dettmers et al. (2023), QLoRA", "doi": "10.48550/arXiv.2305.14314" }, { "id": "ext_dreamcoder_2020", "title": "DreamCoder: Growing generalizable, interpretable knowledge with wake-sleep Bayesian program learning", "priority": "external_literature", "layer": "program_synthesis_representation", "chapter_targets": [ "cognitive-compilation-and-semantic-ir", "procedural-memory-and-cognitive-loop-closure", "compact-generative-systems-and-residual-honesty", "rankfold-neuralfold-and-artifact-compression", "mathematical-and-search-substrates", "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2006.08381", "notes": "External program-synthesis source for wake-sleep library learning, reusable abstractions, interpretable learned programs, and compression-through-abstraction vocabulary.", "source_type": "arxiv", "arxiv_id": "2006.08381", "published": "2020-06-15", "updated": "2020-06-15", "citation_label": "Ellis et al. (2020), DreamCoder", "doi": "10.48550/arXiv.2006.08381" }, { "id": "ext_llvm_langref_docs", "title": "LLVM Language Reference Manual", "priority": "external_literature", "layer": "compiler_ir", "chapter_targets": [ "cognitive-compilation-and-semantic-ir", "executable-specifications-and-lean-proof-envelope" ], "url": "https://llvm.org/docs/LangRef.html", "notes": "Official LLVM Language Reference comparator for SSA-based intermediate representation, equivalent in-memory/bitcode/human-readable forms, well-formedness, verifier passes, and optimization/analysis vocabulary.", "source_type": "official_docs", "published": "", "updated": "2026", "citation_label": "LLVM Documentation (2026), Language Reference Manual" }, { "id": "ext_mlir_2020", "title": "MLIR: A Compiler Infrastructure for the End of Moore's Law", "priority": "external_literature", "layer": "multi_level_compiler_ir", "chapter_targets": [ "cognitive-compilation-and-semantic-ir", "resource-economics-and-token-budgets", "mathematical-and-search-substrates" ], "url": "https://arxiv.org/abs/2002.11054", "notes": "External multi-level compiler-IR comparator for reusable and extensible compiler infrastructure, dialects, progressive lowering, verifiers, modular passes, and heterogeneous target support.", "source_type": "arxiv", "arxiv_id": "2002.11054", "published": "2020-02-25", "updated": "2020-03-01", "citation_label": "Lattner et al. (2020), MLIR", "doi": "10.48550/arXiv.2002.11054" }, { "id": "ext_translation_validation_1998", "title": "Translation Validation", "priority": "external_literature", "layer": "translation_validation", "chapter_targets": [ "cognitive-compilation-and-semantic-ir", "executable-specifications-and-lean-proof-envelope" ], "url": "https://weizmann.esploro.exlibrisgroup.com/esploro/outputs/conferenceProceeding/Translation-validation/993262143603596", "notes": "External translation-validation comparator for checking each compiler/code-generator run after translation, using a common semantic framework, refinement relation, and simulation-based proof method.", "source_type": "conference_paper", "published": "1998", "updated": "1998", "citation_label": "Pnueli, Siegel, and Singerman (1998), Translation Validation" }, { "id": "ext_toolformer_2023", "title": "Toolformer: Language Models Can Teach Themselves to Use Tools", "priority": "external_literature", "layer": "learned_tool_use", "chapter_targets": [ "procedural-memory-and-cognitive-loop-closure", "runtime-adapters-tool-permissions-and-human-approval", "policy-optimization-and-learning-from-feedback" ], "url": "https://arxiv.org/abs/2302.04761", "notes": "External learned-tool-use source for self-supervised API-call insertion, tool selection, argument construction, and result incorporation without claiming ASI Stack tool-use reproduction.", "source_type": "arxiv", "arxiv_id": "2302.04761", "published": "2023-02-09", "updated": "2023-02-09", "citation_label": "Schick et al. (2023), Toolformer", "doi": "10.48550/arXiv.2302.04761" }, { "id": "ext_voyager_2023", "title": "Voyager: An Open-Ended Embodied Agent with Large Language Models", "priority": "external_literature", "layer": "lifelong_skill_learning", "chapter_targets": [ "procedural-memory-and-cognitive-loop-closure", "routing-heads-and-specialist-cores", "benchmark-ratchets-and-anti-goodhart-evidence", "open-ended-improvement-engines" ], "url": "https://arxiv.org/abs/2305.16291", "notes": "External lifelong-agent source for automatic curriculum, executable-code skill libraries, iterative environment-feedback prompting, self-verification, and skill-library transfer in Minecraft.", "source_type": "arxiv", "arxiv_id": "2305.16291", "published": "2023-05-25", "updated": "2023-10-19", "citation_label": "Wang et al. (2023), Voyager", "doi": "10.48550/arXiv.2305.16291" }, { "id": "ext_information_bottleneck_2000", "title": "The information bottleneck method", "priority": "external_literature", "layer": "representation_compression", "chapter_targets": [ "compact-generative-systems-and-residual-honesty", "resource-economics-and-token-budgets" ], "url": "https://arxiv.org/abs/physics/0004057", "notes": "External representation-compression source for relevance-preserving compression, bottleneck variables, mutual-information tradeoffs, and compression/utility separation.", "source_type": "arxiv", "arxiv_id": "physics/0004057", "published": "2000-04-24", "updated": "2000-04-24", "citation_label": "Tishby et al. (2000), Information Bottleneck", "doi": "10.48550/arXiv.physics/0004057" }, { "id": "ext_mdl_tutorial_2004", "title": "A tutorial introduction to the minimum description length principle", "priority": "external_literature", "layer": "description_length_residuals", "chapter_targets": [ "compact-generative-systems-and-residual-honesty", "rankfold-neuralfold-and-artifact-compression", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://arxiv.org/abs/math/0406077", "notes": "External description-length source for model/data tradeoffs, compression as inductive discipline, and residual/error-accounting vocabulary.", "source_type": "arxiv", "arxiv_id": "math/0406077", "published": "2004-06-04", "updated": "2004-06-04", "citation_label": "Grunwald (2004), MDL Tutorial", "doi": "10.48550/arXiv.math/0406077" }, { "id": "ext_weakness_generalization_2023", "title": "The Optimal Choice of Hypothesis Is the Weakest, Not the Shortest", "priority": "external_literature", "layer": "learning_theory_inductive_bias", "chapter_targets": [ "learning-theory-generalization-and-scaling-science" ], "url": "https://arxiv.org/abs/2301.12987", "notes": "Bennett's finite enactive-cognition formalism separates extension-based hypothesis weakness from description length. Under a uniform distribution over its task space, the paper argues that maximizing weakness is necessary and sufficient for maximizing generalization probability and gives a counterexample to MDL as a universal proxy. Its theorem assumptions and toy 8-bit arithmetic experiments do not establish a general result for neural networks or real task distributions.", "source_type": "conference_paper", "arxiv_id": "2301.12987", "published": "2023", "updated": "2024-04-11", "citation_label": "Bennett (2023), Weakest Hypothesis", "doi": "10.1007/978-3-031-33469-6_5" }, { "id": "ext_codebleu_2020", "title": "CodeBLEU: a Method for Automatic Evaluation of Code Synthesis", "priority": "external_literature", "layer": "artifact_utility_metrics", "chapter_targets": [ "benchmark-ratchets-and-anti-goodhart-evidence", "compact-generative-systems-and-residual-honesty", "artifact-steward-agents-and-living-project-governance", "prototype-roadmap" ], "url": "https://arxiv.org/abs/2009.10297", "notes": "External code-synthesis evaluation source for combining lexical, syntax, data-flow, and semantic matching into artifact-quality metrics that still require task-specific validation.", "source_type": "arxiv", "arxiv_id": "2009.10297", "published": "2020-09-22", "updated": "2020-09-27", "citation_label": "Ren et al. (2020), CodeBLEU", "doi": "10.48550/arXiv.2009.10297" }, { "id": "ext_mmlu_2020", "title": "Measuring Massive Multitask Language Understanding", "priority": "external_literature", "layer": "benchmark_science", "chapter_targets": [ "benchmark-ratchets-and-anti-goodhart-evidence", "evidence-states-and-claim-discipline", "open-research-agenda-and-bibliography-plan", "prototype-roadmap" ], "url": "https://arxiv.org/abs/2009.03300", "notes": "External benchmark source for broad multitask evaluation, task-coverage limits, lopsided performance, uncertainty about wrong answers, and benchmark-saturation pressure.", "source_type": "arxiv", "arxiv_id": "2009.03300", "published": "2020-09-07", "updated": "2021-01-12", "citation_label": "Hendrycks et al. (2020), MMLU", "doi": "10.48550/arXiv.2009.03300" }, { "id": "ext_bigbench_2022", "title": "Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language models", "priority": "external_literature", "layer": "benchmark_science", "chapter_targets": [ "benchmark-ratchets-and-anti-goodhart-evidence", "the-efficient-asi-hypothesis", "readiness-gates-residual-escrow-and-quarantine", "policy-optimization-and-learning-from-feedback" ], "url": "https://arxiv.org/abs/2206.04615", "notes": "External benchmark source for BIG-bench, broad task coverage, scale effects, calibration, breakthrough behavior, human-rater baselines, and social-bias tradeoffs.", "source_type": "arxiv", "arxiv_id": "2206.04615", "published": "2022-06-09", "updated": "2023-06-12", "citation_label": "Srivastava et al. (2022), BIG-bench", "doi": "10.48550/arXiv.2206.04615" }, { "id": "ext_helm_2022", "title": "Holistic Evaluation of Language Models", "priority": "external_literature", "layer": "benchmark_science", "chapter_targets": [ "benchmark-ratchets-and-anti-goodhart-evidence", "evidence-states-and-claim-discipline", "readiness-gates-residual-escrow-and-quarantine", "living-book-methodology" ], "url": "https://arxiv.org/abs/2211.09110", "notes": "External benchmark-science source for multi-scenario, multi-metric evaluation, missing-coverage disclosure, raw-prompt transparency, and living benchmark practice.", "source_type": "arxiv", "arxiv_id": "2211.09110", "published": "2022-11-16", "updated": "2023-10-01", "citation_label": "Liang et al. (2022), HELM", "doi": "10.48550/arXiv.2211.09110" }, { "id": "ext_gpqa_2023", "title": "GPQA: A Graduate-Level Google-Proof Q&A Benchmark", "priority": "external_literature", "layer": "benchmark_science", "chapter_targets": [ "benchmark-ratchets-and-anti-goodhart-evidence", "verification-bandwidth-and-context-adequacy", "evidence-states-and-claim-discipline", "open-research-agenda-and-bibliography-plan" ], "url": "https://arxiv.org/abs/2311.12022", "notes": "External benchmark source for expert-written hard questions, Google-proof validation, scalable oversight pressure, and the gap between skilled non-expert validation and expert competence.", "source_type": "arxiv", "arxiv_id": "2311.12022", "published": "2023-11-20", "updated": "2023-11-20", "citation_label": "Rein et al. (2023), GPQA", "doi": "10.48550/arXiv.2311.12022" }, { "id": "ext_swe_bench_2023", "title": "SWE-bench: Can Language Models Resolve Real-World GitHub Issues?", "priority": "external_literature", "layer": "benchmark_science", "chapter_targets": [ "benchmark-ratchets-and-anti-goodhart-evidence", "artifact-graphs-audit-logs-and-replay", "labor-os-and-typed-jobs", "prototype-roadmap" ], "url": "https://arxiv.org/abs/2310.06770", "notes": "External benchmark source for real-world software-engineering issue resolution, repository-scale context, executable environments, patch evaluation, and capability boundaries.", "source_type": "arxiv", "arxiv_id": "2310.06770", "published": "2023-10-10", "updated": "2024-11-11", "citation_label": "Jimenez et al. (2023), SWE-bench", "doi": "10.48550/arXiv.2310.06770" }, { "id": "ext_swe_rebench_v2_2026", "title": "SWE-rebench V2: Language-Agnostic SWE Task Collection at Scale", "priority": "external_literature", "layer": "natural_software_task_construction_and_evaluation", "chapter_targets": [ "benchmark-ratchets-and-anti-goodhart-evidence", "artifact-graphs-audit-logs-and-replay", "integrated-reference-architecture", "prototype-roadmap", "readiness-gates-residual-escrow-and-quarantine" ], "url": "https://arxiv.org/abs/2602.23866", "notes": "Primary 2026 natural-task substrate for multilingual repository changes, interactive setup, containerized full-suite execution, separated solution/test patches, task diagnostics, and explicit environment pathologies. It does not establish local task validity, gold execution, model competence, governance benefit, safety, transfer, or SOTA.", "source_type": "preprint", "arxiv_id": "2602.23866", "published": "2026-02-27", "updated": "2026-06-01", "citation_label": "Badertdinov et al. (2026), SWE-rebench V2", "doi": "10.48550/arXiv.2602.23866" }, { "id": "ext_livebench_2024", "title": "LiveBench: A Challenging, Contamination-Limited LLM Benchmark", "priority": "external_literature", "layer": "benchmark_science", "chapter_targets": [ "benchmark-ratchets-and-anti-goodhart-evidence", "evidence-states-and-claim-discipline", "living-book-methodology", "open-research-agenda-and-bibliography-plan" ], "url": "https://arxiv.org/abs/2406.19314", "notes": "External benchmark source for contamination-limited evaluation, frequently updated questions, objective ground-truth scoring, and monthly benchmark evolution.", "source_type": "arxiv", "arxiv_id": "2406.19314", "published": "2024-06-27", "updated": "2025-04-18", "citation_label": "White et al. (2024), LiveBench", "doi": "10.48550/arXiv.2406.19314" }, { "id": "ext_dynabench_2021", "title": "Dynabench: Rethinking Benchmarking in NLP", "priority": "external_literature", "layer": "dynamic_benchmarking", "chapter_targets": [ "benchmark-ratchets-and-anti-goodhart-evidence", "readiness-gates-residual-escrow-and-quarantine", "policy-optimization-and-learning-from-feedback", "artifact-steward-agents-and-living-project-governance" ], "url": "https://arxiv.org/abs/2104.14337", "notes": "External dynamic-benchmarking source for human-and-model-in-the-loop data collection, adversarial benchmark evolution, and stale static benchmark pressure.", "source_type": "arxiv", "arxiv_id": "2104.14337", "published": "2021-04-07", "updated": "2021-04-07", "citation_label": "Kiela et al. (2021), Dynabench", "doi": "10.48550/arXiv.2104.14337" }, { "id": "ext_checklist_2020", "title": "Beyond Accuracy: Behavioral Testing of NLP models with CheckList", "priority": "external_literature", "layer": "behavioral_evaluation", "chapter_targets": [ "benchmark-ratchets-and-anti-goodhart-evidence", "verification-bandwidth-and-context-adequacy", "claim-ledgers-and-belief-revision", "prototype-roadmap" ], "url": "https://arxiv.org/abs/2005.04118", "notes": "External behavioral-testing source for capability matrices, minimum functionality tests, invariance tests, directional expectation tests, and failure-discovery beyond aggregate accuracy.", "source_type": "arxiv", "arxiv_id": "2005.04118", "published": "2020-05-08", "updated": "2020-05-08", "citation_label": "Ribeiro et al. (2020), CheckList", "doi": "10.48550/arXiv.2005.04118" }, { "id": "ext_benchmark_contamination_2023", "title": "Investigating Data Contamination in Modern Benchmarks for Large Language Models", "priority": "external_literature", "layer": "benchmark_contamination", "chapter_targets": [ "benchmark-ratchets-and-anti-goodhart-evidence", "evidence-states-and-claim-discipline", "readiness-gates-residual-escrow-and-quarantine", "living-book-methodology" ], "url": "https://arxiv.org/abs/2311.09783", "notes": "External benchmark-contamination source for detecting training/test overlap pressure, benchmark-leakage risk, and score interpretation limits in modern LLM evaluations.", "source_type": "arxiv", "arxiv_id": "2311.09783", "published": "2023-11-16", "updated": "2024-04-03", "citation_label": "Deng et al. (2023), Benchmark Contamination", "doi": "10.48550/arXiv.2311.09783" }, { "id": "ext_goodhart_variants_2018", "title": "Categorizing Variants of Goodhart's Law", "priority": "external_literature", "layer": "goodhart_failure_taxonomy", "chapter_targets": [ "failure-modes-of-ungoverned-intelligence", "benchmark-ratchets-and-anti-goodhart-evidence", "policy-optimization-and-learning-from-feedback", "artifact-steward-agents-and-living-project-governance", "evidence-states-and-claim-discipline" ], "url": "https://arxiv.org/abs/1803.04585", "notes": "External Goodhart-taxonomy source for regressive, extremal, causal, and adversarial metric failures that benchmark ratchets and policy updates must treat as distinct risks.", "source_type": "arxiv", "arxiv_id": "1803.04585", "published": "2018-03-13", "updated": "2019-02-24", "citation_label": "Manheim and Garrabrant (2018), Goodhart Variants", "doi": "10.48550/arXiv.1803.04585" }, { "id": "ext_speculative_decoding_2022", "title": "Fast Inference from Transformers via Speculative Decoding", "priority": "external_literature", "layer": "fast_generation", "chapter_targets": [ "fast-generation", "the-efficient-asi-hypothesis" ], "url": "https://arxiv.org/abs/2211.17192", "notes": "Primary external paper for speculative decoding: a draft model proposes multiple tokens and a target model verifies them, giving an exact-distribution acceleration path under its assumptions.", "arxiv_id": "2211.17192", "published": "2022-11-30", "updated": "2023-05-18", "citation_label": "Leviathan et al. (2022), Fast Inference from Transformers via Speculative Decoding", "doi": "10.48550/arXiv.2211.17192", "source_type": "arxiv" }, { "id": "ext_multi_token_prediction_2024", "title": "Better & Faster Large Language Models via Multi-token Prediction", "priority": "external_literature", "layer": "fast_generation", "chapter_targets": [ "fast-generation", "the-efficient-asi-hypothesis" ], "url": "https://arxiv.org/abs/2404.19737", "notes": "Primary external paper for multi-token prediction as an auxiliary training objective and inference-time multi-token proposal mechanism.", "arxiv_id": "2404.19737", "published": "2024-04-30", "updated": "2024-04-30", "citation_label": "Gloeckle et al. (2024), Better & Faster Large Language Models via Multi-token Prediction", "doi": "10.48550/arXiv.2404.19737", "source_type": "arxiv" }, { "id": "ext_medusa_2024", "title": "Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Heads", "priority": "external_literature", "layer": "fast_generation", "chapter_targets": [ "fast-generation", "the-efficient-asi-hypothesis" ], "url": "https://arxiv.org/abs/2401.10774", "notes": "Primary external paper for adding multiple decoding heads to an LLM and verifying tree-structured candidate continuations in parallel.", "arxiv_id": "2401.10774", "published": "2024-01-19", "updated": "2024-06-14", "citation_label": "Cai et al. (2024), Medusa", "doi": "10.48550/arXiv.2401.10774", "source_type": "arxiv" }, { "id": "ext_eagle_2024", "title": "EAGLE: Speculative Sampling Requires Rethinking Feature Uncertainty", "priority": "external_literature", "layer": "fast_generation", "chapter_targets": [ "fast-generation", "the-efficient-asi-hypothesis" ], "url": "https://arxiv.org/abs/2401.15077", "notes": "Primary external paper for feature-level speculative drafting and target-model verification as an acceleration mechanism.", "arxiv_id": "2401.15077", "published": "2024-01-26", "updated": "2025-03-04", "citation_label": "Li et al. (2024), EAGLE", "doi": "10.48550/arXiv.2401.15077", "source_type": "arxiv" }, { "id": "ext_lookahead_decoding_2024", "title": "Break the Sequential Dependency of LLM Inference Using Lookahead Decoding", "priority": "external_literature", "layer": "fast_generation", "chapter_targets": [ "fast-generation" ], "url": "https://arxiv.org/abs/2402.02057", "notes": "Primary external paper for lookahead decoding: a parallel exact decoding algorithm that reduces serial decoding steps without an auxiliary draft model.", "arxiv_id": "2402.02057", "published": "2024-02-03", "updated": "2024-02-03", "citation_label": "Fu et al. (2024), Lookahead Decoding", "doi": "10.48550/arXiv.2402.02057", "source_type": "arxiv" }, { "id": "ext_layerskip_2024", "title": "LayerSkip: Enabling Early Exit Inference and Self-Speculative Decoding", "priority": "external_literature", "layer": "fast_generation", "chapter_targets": [ "fast-generation" ], "url": "https://arxiv.org/abs/2404.16710", "notes": "Primary external paper for early-exit inference and self-speculative decoding where early layers draft and later layers verify.", "arxiv_id": "2404.16710", "published": "2024-04-25", "updated": "2024-10-18", "citation_label": "Elhoushi et al. (2024), LayerSkip", "doi": "10.18653/v1/2024.acl-long.681", "source_type": "arxiv" }, { "id": "ext_pagedattention_vllm_2023", "title": "Efficient Memory Management for Large Language Model Serving with PagedAttention", "priority": "external_literature", "layer": "fast_generation", "chapter_targets": [ "fast-generation", "resource-economics" ], "url": "https://arxiv.org/abs/2309.06180", "notes": "Primary external paper for vLLM/PagedAttention, which treats KV-cache memory management and serving throughput as a distinct acceleration axis.", "arxiv_id": "2309.06180", "published": "2023-09-12", "updated": "2023-09-12", "citation_label": "Kwon et al. (2023), Efficient Memory Management for LLM Serving with PagedAttention", "doi": "10.48550/arXiv.2309.06180", "source_type": "arxiv" }, { "id": "ext_transformer_xl_2019", "title": "Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context", "priority": "external_literature", "layer": "sequence_memory_recurrence", "chapter_targets": [ "coil-attention-cyclic-memory-and-recurrence-contracts" ], "url": "https://arxiv.org/abs/1901.02860", "notes": "External recurrent Transformer comparator for segment-level recurrence, relative positional encoding, and long-dependency language modeling; useful as a baseline family for cyclic-memory contracts without implying local reproduction.", "arxiv_id": "1901.02860", "published": "2019-01-09", "updated": "2019-06-02", "citation_label": "Dai et al. (2019), Transformer-XL", "doi": "10.48550/arXiv.1901.02860", "source_type": "arxiv" }, { "id": "ext_compressive_transformer_2019", "title": "Compressive Transformers for Long-Range Sequence Modelling", "priority": "external_literature", "layer": "sequence_memory_recurrence", "chapter_targets": [ "coil-attention-cyclic-memory-and-recurrence-contracts" ], "url": "https://arxiv.org/abs/1911.05507", "notes": "External long-range memory comparator for compressed past memories, memory mechanisms, and long-range sequence benchmarks; useful for positioning cyclic memory against compression-memory baselines.", "arxiv_id": "1911.05507", "published": "2019-11-13", "updated": "2019-11-13", "citation_label": "Rae et al. (2019), Compressive Transformers", "doi": "10.48550/arXiv.1911.05507", "source_type": "arxiv" }, { "id": "ext_roformer_rope_2021", "title": "RoFormer: Enhanced Transformer with Rotary Position Embedding", "priority": "external_literature", "layer": "position_encoding", "chapter_targets": [ "coilra-multicoil-rope-and-cyclic-mixers" ], "url": "https://arxiv.org/abs/2104.09864", "notes": "External RoPE comparator for rotary position embedding, relative-position behavior inside self-attention, and position-encoding baselines for cyclic phase or RoPE-style substrates.", "arxiv_id": "2104.09864", "published": "2021-04-20", "updated": "2023-11-08", "citation_label": "Su et al. (2021), RoFormer", "doi": "10.48550/arXiv.2104.09864", "source_type": "arxiv" }, { "id": "ext_retnet_2023", "title": "Retentive Network: A Successor to Transformer for Large Language Models", "priority": "external_literature", "layer": "sequence_memory_recurrence", "chapter_targets": [ "coil-attention-cyclic-memory-and-recurrence-contracts", "coilra-multicoil-rope-and-cyclic-mixers", "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2307.08621", "notes": "External retention/recurrent-sequence comparator for the relationship between recurrence and attention, recurrent/chunkwise computation, and inference-efficiency tradeoffs.", "arxiv_id": "2307.08621", "published": "2023-07-17", "updated": "2023-08-09", "citation_label": "Sun et al. (2023), Retentive Network", "doi": "10.48550/arXiv.2307.08621", "source_type": "arxiv" }, { "id": "ext_mamba_2023", "title": "Mamba: Linear-Time Sequence Modeling with Selective State Spaces", "priority": "external_literature", "layer": "sequence_substrates", "chapter_targets": [ "fast-generation", "math-substrates", "coilra-multicoil-rope-and-cyclic-mixers", "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2312.00752", "notes": "Primary external paper for selective state-space sequence models as a different long-sequence substrate and inference-efficiency axis from decoding tricks.", "arxiv_id": "2312.00752", "published": "2023-12-01", "updated": "2024-05-31", "citation_label": "Gu and Dao (2023), Mamba", "doi": "10.48550/arXiv.2312.00752", "source_type": "arxiv" }, { "id": "ext_llada_2025", "title": "Large Language Diffusion Models", "priority": "external_literature", "layer": "diffusion_language_models", "chapter_targets": [ "fast-generation-architectures", "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2502.09992", "notes": "Primary external paper for LLaDA, a large masked-diffusion language model trained with pretraining and supervised fine-tuning rather than left-to-right autoregression.", "arxiv_id": "2502.09992", "published": "2025-02-14", "updated": "2025-10-18", "citation_label": "Nie et al. (2025), Large Language Diffusion Models", "doi": "10.48550/arXiv.2502.09992", "source_type": "arxiv" }, { "id": "ext_scaling_dllms_2026", "title": "Scaling Beyond Masked Diffusion Language Models", "priority": "external_literature", "layer": "diffusion_language_models", "chapter_targets": [ "fast-generation-architectures", "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2602.15014", "notes": "Primary external paper for comparing diffusion language-model families by speed-quality tradeoffs rather than perplexity alone.", "arxiv_id": "2602.15014", "published": "2026-02-16", "updated": "2026-02-16", "citation_label": "Sahoo et al. (2026), Scaling Beyond Masked Diffusion Language Models", "doi": "10.48550/arXiv.2602.15014", "source_type": "arxiv" }, { "id": "ext_trpo_2015", "title": "Trust Region Policy Optimization", "priority": "external_literature", "layer": "policy_optimization", "chapter_targets": [ "governed-deliberation-and-test-time-scaling" ], "url": "https://arxiv.org/abs/1502.05477", "notes": "Primary external source for trust-region policy-gradient updates and bounded update-size discipline.", "source_type": "arxiv", "arxiv_id": "1502.05477", "published": "2015-02-19", "updated": "2017-04-20", "citation_label": "Schulman et al. (2015), Trust Region Policy Optimization", "doi": "10.48550/arXiv.1502.05477" }, { "id": "ext_ppo_2017", "title": "Proximal Policy Optimization Algorithms", "priority": "external_literature", "layer": "policy_optimization", "chapter_targets": [ "governed-deliberation-and-test-time-scaling" ], "url": "https://arxiv.org/abs/1707.06347", "notes": "Primary external source for PPO-style online policy-gradient updates and proximal surrogate objectives.", "source_type": "arxiv", "arxiv_id": "1707.06347", "published": "2017-07-20", "updated": "2017-08-28", "citation_label": "Schulman et al. (2017), Proximal Policy Optimization Algorithms", "doi": "10.48550/arXiv.1707.06347" }, { "id": "ext_remax_2023", "title": "ReMax: A Simple, Effective, and Efficient Reinforcement Learning Method for Aligning Large Language Models", "priority": "external_literature", "layer": "policy_optimization", "chapter_targets": [], "url": "https://arxiv.org/abs/2310.10505", "notes": "Primary external source for simpler RLHF-oriented policy-gradient updates relative to PPO-style machinery.", "source_type": "arxiv", "arxiv_id": "2310.10505", "published": "2023-10-16", "updated": "2024-05-16", "citation_label": "Li et al. (2023), ReMax", "doi": "10.48550/arXiv.2310.10505" }, { "id": "ext_goal_oriented_requirements_engineering_2001", "title": "Goal-Oriented Requirements Engineering: A Guided Tour", "priority": "external_literature", "layer": "requirements_engineering", "chapter_targets": [ "human-intent-as-a-formal-input" ], "url": "https://doi.org/10.1109/ISRE.2001.948567", "notes": "External requirements-engineering comparator for turning stakeholder goals, constraints, refinements, and responsibilities into explicit requirements before system design or execution.", "source_type": "conference_paper", "published": "2001", "updated": "2001", "citation_label": "van Lamsweerde (2001), Goal-Oriented Requirements Engineering", "doi": "10.1109/ISRE.2001.948567" }, { "id": "ext_cooperative_inverse_rl_2016", "title": "Cooperative Inverse Reinforcement Learning", "priority": "external_literature", "layer": "human_intent_alignment", "chapter_targets": [ "human-intent-as-a-formal-input" ], "url": "https://arxiv.org/abs/1606.03137", "notes": "External cooperative AI comparator for formalizing value alignment as uncertainty about the human reward function in a cooperative partial-information setting.", "source_type": "arxiv", "arxiv_id": "1606.03137", "published": "2016-06-09", "updated": "2024-02-17", "citation_label": "Hadfield-Menell et al. (2016), Cooperative Inverse Reinforcement Learning", "doi": "10.48550/arXiv.1606.03137" }, { "id": "ext_deep_rl_human_preferences_2017", "title": "Deep Reinforcement Learning from Human Preferences", "priority": "external_literature", "layer": "human_feedback_learning", "chapter_targets": [ "human-intent-as-a-formal-input", "policy-optimization-and-learning-from-feedback" ], "url": "https://arxiv.org/abs/1706.03741", "notes": "External human-feedback comparator for communicating complex goals through preference comparisons over behavior segments; useful for separating preference signals from explicit intent contracts.", "source_type": "arxiv", "arxiv_id": "1706.03741", "published": "2017-06-12", "updated": "2023-02-17", "citation_label": "Christiano et al. (2017), Deep RL from Human Preferences", "doi": "10.48550/arXiv.1706.03741" }, { "id": "ext_dpo_2023", "title": "Direct Preference Optimization: Your Language Model is Secretly a Reward Model", "priority": "external_literature", "layer": "policy_optimization", "chapter_targets": [], "url": "https://arxiv.org/abs/2305.18290", "notes": "Primary external source for DPO-style offline preference optimization without a separate online RL loop.", "source_type": "arxiv", "arxiv_id": "2305.18290", "published": "2023-05-29", "updated": "2024-07-29", "citation_label": "Rafailov et al. (2023), Direct Preference Optimization", "doi": "10.48550/arXiv.2305.18290" }, { "id": "ext_ipo_preference_2023", "title": "A General Theoretical Paradigm to Understand Learning from Human Preferences", "priority": "external_literature", "layer": "policy_optimization", "chapter_targets": [], "url": "https://arxiv.org/abs/2310.12036", "notes": "External source for theoretical comparison of preference-learning objectives, including IPO/DPO-style framing.", "source_type": "arxiv", "arxiv_id": "2310.12036", "published": "2023-10-18", "updated": "2023-11-22", "citation_label": "Azar et al. (2023), Learning from Human Preferences paradigm", "doi": "10.48550/arXiv.2310.12036" }, { "id": "ext_orpo_2024", "title": "ORPO: Monolithic Preference Optimization without Reference Model", "priority": "external_literature", "layer": "policy_optimization", "chapter_targets": [], "url": "https://arxiv.org/abs/2403.07691", "notes": "External source for reference-model-free monolithic preference optimization.", "source_type": "arxiv", "arxiv_id": "2403.07691", "published": "2024-03-12", "updated": "2024-03-14", "citation_label": "Hong et al. (2024), ORPO", "doi": "10.48550/arXiv.2403.07691" }, { "id": "ext_kto_2024", "title": "KTO: Model Alignment as Prospect Theoretic Optimization", "priority": "external_literature", "layer": "policy_optimization", "chapter_targets": [], "url": "https://arxiv.org/abs/2402.01306", "notes": "External source for prospect-theoretic model alignment and human-aware loss framing.", "source_type": "arxiv", "arxiv_id": "2402.01306", "published": "2024-02-02", "updated": "2024-11-19", "citation_label": "Ethayarajh et al. (2024), KTO", "doi": "10.48550/arXiv.2402.01306" }, { "id": "ext_simpo_2024", "title": "SimPO: Simple Preference Optimization with a Reference-Free Reward", "priority": "external_literature", "layer": "policy_optimization", "chapter_targets": [], "url": "https://arxiv.org/abs/2405.14734", "notes": "External source for simple reference-free preference optimization using sequence-level reward framing.", "source_type": "arxiv", "arxiv_id": "2405.14734", "published": "2024-05-23", "updated": "2024-11-01", "citation_label": "Meng et al. (2024), SimPO", "doi": "10.48550/arXiv.2405.14734" }, { "id": "ext_reinforce_style_rlhf_2024", "title": "Back to Basics: Revisiting REINFORCE Style Optimization for Learning from Human Feedback in LLMs", "priority": "external_literature", "layer": "policy_optimization", "chapter_targets": [], "url": "https://arxiv.org/abs/2402.14740", "notes": "External source for revisiting simpler REINFORCE-style optimization as an RLHF baseline.", "source_type": "arxiv", "arxiv_id": "2402.14740", "published": "2024-02-22", "updated": "2024-02-26", "citation_label": "Ahmadian et al. (2024), REINFORCE-style RLHF revisiting", "doi": "10.48550/arXiv.2402.14740" }, { "id": "ext_deepseek_r1_2025", "title": "DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning", "priority": "external_literature", "layer": "policy_optimization", "chapter_targets": [], "url": "https://arxiv.org/abs/2501.12948", "notes": "External source for reinforcement-learning pressure on reasoning behavior in large language models.", "source_type": "arxiv", "arxiv_id": "2501.12948", "published": "2025-01-22", "updated": "2026-01-04", "citation_label": "DeepSeek-AI et al. (2025), DeepSeek-R1", "doi": "10.1038/s41586-025-09422-z", "venue": "Nature volume 645, pages 633-638 (2025)" }, { "id": "ext_dapo_2025", "title": "DAPO: An Open-Source LLM Reinforcement Learning System at Scale", "priority": "external_literature", "layer": "policy_optimization", "chapter_targets": [], "url": "https://arxiv.org/abs/2503.14476", "notes": "External source for open-source large-scale LLM RL system details and DAPO-style update design.", "source_type": "arxiv", "arxiv_id": "2503.14476", "published": "2025-03-18", "updated": "2025-05-20", "citation_label": "Yu et al. (2025), DAPO", "doi": "10.48550/arXiv.2503.14476" }, { "id": "ext_gspo_2025", "title": "Group Sequence Policy Optimization", "priority": "external_literature", "layer": "policy_optimization", "chapter_targets": [], "url": "https://arxiv.org/abs/2507.18071", "notes": "External source for sequence-level group policy optimization in LLM reinforcement learning.", "source_type": "arxiv", "arxiv_id": "2507.18071", "published": "2025-07-24", "updated": "2025-07-28", "citation_label": "Zheng et al. (2025), Group Sequence Policy Optimization", "doi": "10.48550/arXiv.2507.18071" }, { "id": "ext_s_grpo_2025", "title": "S-GRPO: Early Exit via Reinforcement Learning in Reasoning Models", "priority": "external_literature", "layer": "policy_optimization", "chapter_targets": [], "url": "https://arxiv.org/abs/2505.07686", "notes": "External source for early-exit reinforcement learning and overthinking control in reasoning models.", "source_type": "arxiv", "arxiv_id": "2505.07686", "published": "2025-05-12", "updated": "2025-05-17", "citation_label": "Dai et al. (2025), S-GRPO", "doi": "10.48550/arXiv.2505.07686" }, { "id": "ext_longrlvr_2026", "title": "LongRLVR: Long-Context Reinforcement Learning Requires Verifiable Context Rewards", "priority": "external_literature", "layer": "policy_optimization", "chapter_targets": [], "url": "https://arxiv.org/abs/2603.02146", "notes": "External source for long-context RLVR and verifiable context-grounding rewards.", "source_type": "arxiv", "arxiv_id": "2603.02146", "published": "2026-03-02", "updated": "2026-03-02", "citation_label": "Chen et al. (2026), LongRLVR", "doi": "10.48550/arXiv.2603.02146" }, { "id": "ext_rlhf_limitations_2023", "title": "Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback", "priority": "external_literature", "layer": "policy_optimization", "chapter_targets": [], "url": "https://arxiv.org/abs/2307.15217", "notes": "External survey source for RLHF limitations, reward hacking, evaluator limits, and complementary safeguards.", "source_type": "arxiv", "arxiv_id": "2307.15217", "published": "2023-07-27", "updated": "2023-09-11", "citation_label": "Casper et al. (2023), Open Problems and Fundamental Limitations of RLHF", "doi": "10.48550/arXiv.2307.15217" }, { "id": "ext_tailscale_docs_2025", "title": "What is Tailscale?", "priority": "external_literature", "layer": "personal_compute_hives", "chapter_targets": [ "personal-compute-hives-and-federated-edge-intelligence" ], "url": "https://tailscale.com/docs/concepts/what-is-tailscale", "notes": "Official Tailscale documentation for zero-trust identity networking, tailnets, encrypted point-to-point connections, and cross-network device connectivity.", "source_type": "official_docs", "published": "", "updated": "2025-09-30", "citation_label": "Tailscale Docs (2025), What is Tailscale?" }, { "id": "ext_kubernetes_overview_docs", "title": "Kubernetes Documentation: Overview", "priority": "external_literature", "layer": "personal_compute_hives", "chapter_targets": [ "personal-compute-hives-and-federated-edge-intelligence" ], "url": "https://kubernetes.io/docs/concepts/overview/", "notes": "Official Kubernetes overview for containerized workload management, declarative configuration, automation, service discovery, storage orchestration, rollouts, bin packing, and self-healing.", "source_type": "official_docs", "published": "", "updated": "", "citation_label": "Kubernetes Documentation, Overview" }, { "id": "ext_k3s_docs_2026", "title": "K3s: Lightweight Kubernetes", "priority": "external_literature", "layer": "personal_compute_hives", "chapter_targets": [ "personal-compute-hives-and-federated-edge-intelligence" ], "url": "https://docs.k3s.io/", "notes": "Official K3s documentation for lightweight Kubernetes deployment in edge, homelab, IoT, CI, single-board-computer, air-gapped, and embedded settings.", "source_type": "official_docs", "published": "", "updated": "2026-06-15", "citation_label": "K3s Documentation (2026), Lightweight Kubernetes" }, { "id": "ext_nomad_docs", "title": "Nomad Documentation", "priority": "external_literature", "layer": "personal_compute_hives", "chapter_targets": [ "personal-compute-hives-and-federated-edge-intelligence" ], "url": "https://developer.hashicorp.com/nomad", "notes": "Official Nomad documentation for scheduling and orchestrating containers, non-containerized applications, and batch jobs across on-prem and cloud environments.", "source_type": "official_docs", "published": "", "updated": "", "citation_label": "HashiCorp Developer, Nomad Documentation" }, { "id": "ext_temporal_docs", "title": "Temporal Documentation: What is Temporal?", "priority": "external_literature", "layer": "durable_execution", "chapter_targets": [ "intent-to-execution-contracts", "labor-os-and-typed-jobs" ], "url": "https://docs.temporal.io/temporal", "notes": "Official Temporal documentation comparator for durable workflow execution, workflow event histories, worker processes, failure recovery, and long-running application-code orchestration.", "source_type": "official_docs", "published": "", "updated": "2026", "citation_label": "Temporal Documentation (2026), What is Temporal?" }, { "id": "ext_airflow_dag_docs", "title": "Apache Airflow Documentation: Dags", "priority": "external_literature", "layer": "workflow_orchestration", "chapter_targets": [ "intent-to-execution-contracts", "labor-os-and-typed-jobs" ], "url": "https://airflow.apache.org/docs/apache-airflow/stable/core-concepts/dags.html", "notes": "Official Apache Airflow documentation comparator for DAG-based workflow scheduling, tasks, dependencies, callbacks, retries, and operational workflow metadata.", "source_type": "official_docs", "published": "", "updated": "2026", "citation_label": "Apache Airflow Documentation (2026), Dags" }, { "id": "ext_bpmn_2_0_2_spec", "title": "Business Process Model and Notation Specification Version 2.0.2", "priority": "external_literature", "layer": "business_process_modeling", "chapter_targets": [ "intent-to-execution-contracts", "labor-os-and-typed-jobs" ], "url": "https://www.omg.org/spec/BPMN/2.0.2/", "notes": "OMG BPMN 2.0.2 formal specification comparator for stakeholder-readable business-process diagrams, implementation-independent flow notation, and translation into software process components.", "source_type": "standard", "published": "2014-01", "updated": "2014-01", "citation_label": "OMG (2014), BPMN 2.0.2" }, { "id": "ext_kubernetes_jobs_docs", "title": "Kubernetes Documentation: Jobs", "priority": "external_literature", "layer": "batch_job_lifecycle", "chapter_targets": [ "labor-os-and-typed-jobs" ], "url": "https://kubernetes.io/docs/concepts/workloads/controllers/job/", "notes": "Official Kubernetes Jobs documentation comparator for batch job lifecycle, completions, backoff limits, active deadlines, terminal Complete/Failed conditions, and cleanup of finished jobs.", "source_type": "official_docs", "published": "", "updated": "2026", "citation_label": "Kubernetes Documentation (2026), Jobs" }, { "id": "ext_ray_core_docs_2026", "title": "What's Ray Core?", "priority": "external_literature", "layer": "personal_compute_hives", "chapter_targets": [ "personal-compute-hives-and-federated-edge-intelligence" ], "url": "https://docs.ray.io/en/latest/ray-core/walkthrough.html", "notes": "Official Ray Core documentation for distributed task, actor, and object primitives used to build and scale Python applications.", "source_type": "official_docs", "published": "", "updated": "2026", "citation_label": "Ray Documentation (2026), What's Ray Core?" }, { "id": "ext_boinc_home_2026", "title": "BOINC", "priority": "external_literature", "layer": "personal_compute_hives", "chapter_targets": [ "personal-compute-hives-and-federated-edge-intelligence" ], "url": "https://boinc.berkeley.edu/", "notes": "Official BOINC site for volunteer computing where user computers download scientific computing jobs and run them in the background.", "source_type": "official_project_site", "published": "", "updated": "2026", "citation_label": "BOINC Project Site (2026)" }, { "id": "ext_syncthing_home", "title": "Syncthing", "priority": "external_literature", "layer": "personal_compute_hives", "chapter_targets": [ "personal-compute-hives-and-federated-edge-intelligence" ], "url": "https://syncthing.net/", "notes": "Official Syncthing site for continuous file synchronization across computers, authenticated devices, encrypted transport, and user-controlled storage location.", "source_type": "official_project_site", "published": "", "updated": "", "citation_label": "Syncthing Project Site" }, { "id": "ext_ipfs_docs", "title": "IPFS Documentation and Project Site", "priority": "external_literature", "layer": "personal_compute_hives", "chapter_targets": [ "personal-compute-hives-and-federated-edge-intelligence" ], "url": "https://ipfs.tech/", "notes": "Official IPFS project documentation and site for peer-to-peer content addressing, content identifiers, provider discovery, and decentralized retrieval vocabulary.", "source_type": "official_project_site", "published": "", "updated": "", "citation_label": "IPFS Project Site" }, { "id": "ext_akash_docs_2026", "title": "Akash Network Documentation", "priority": "external_literature", "layer": "personal_compute_hives", "chapter_targets": [ "personal-compute-hives-and-federated-edge-intelligence", "artifact-steward-agents-and-living-project-governance" ], "url": "https://akash.network/docs/", "notes": "Official Akash documentation for decentralized cloud deployment, provider resources, leases, GPUs, SDKs, node operations, and provider operations.", "source_type": "official_docs", "published": "", "updated": "2026", "citation_label": "Akash Network Documentation (2026)" }, { "id": "ext_golem_docs_2025", "title": "Golem Developer Resources", "priority": "external_literature", "layer": "personal_compute_hives", "chapter_targets": [ "personal-compute-hives-and-federated-edge-intelligence", "artifact-steward-agents-and-living-project-governance" ], "url": "https://docs.golem.network/", "notes": "Official Golem developer resources for decentralized computations, task execution, provider selection, result handling, and resource sharing.", "source_type": "official_docs", "published": "", "updated": "2025", "citation_label": "Golem Docs (2025), Developer Resources" }, { "id": "ext_github_webhooks_docs", "title": "Webhook events and payloads", "priority": "external_literature", "layer": "artifact_steward_agents", "chapter_targets": [ "artifact-steward-agents-and-living-project-governance" ], "url": "https://docs.github.com/en/webhooks/webhook-events-and-payloads", "notes": "Official GitHub documentation for repository and organization webhook events, event payloads, delivery headers, event-specific permissions, and payload limits.", "source_type": "official_docs", "published": "", "updated": "2026", "citation_label": "GitHub Docs (2026), Webhook events and payloads" }, { "id": "ext_github_self_hosted_runners_docs", "title": "Self-hosted runners", "priority": "external_literature", "layer": "artifact_steward_agents", "chapter_targets": [ "artifact-steward-agents-and-living-project-governance", "personal-compute-hives-and-federated-edge-intelligence" ], "url": "https://docs.github.com/en/actions/concepts/runners/self-hosted-runners", "notes": "Official GitHub Actions documentation for self-hosted runners as user-managed systems that execute workflow jobs on physical, virtual, containerized, on-prem, or cloud machines.", "source_type": "official_docs", "published": "", "updated": "2026", "citation_label": "GitHub Docs (2026), Self-hosted runners" }, { "id": "ext_openzeppelin_governor_docs", "title": "OpenZeppelin Contracts: Governance", "priority": "external_literature", "layer": "artifact_steward_agents", "chapter_targets": [ "artifact-steward-agents-and-living-project-governance" ], "url": "https://docs.openzeppelin.com/contracts/5.x/api/governance", "notes": "Official OpenZeppelin governance documentation for modular Governor contracts, voting power, quorum, timelocks, proposal settings, and guardian-style extensions.", "source_type": "official_docs", "published": "", "updated": "", "citation_label": "OpenZeppelin Contracts Docs, Governance" }, { "id": "ext_open_collective_docs", "title": "Open Collective Documentation", "priority": "external_literature", "layer": "artifact_steward_agents", "chapter_targets": [ "artifact-steward-agents-and-living-project-governance" ], "url": "https://documentation.opencollective.com/", "notes": "Official Open Collective documentation for transparent community money management, fiscal hosts, contribution intake, expenses, accounting, and legal-entity delegation through fiscal hosting.", "source_type": "official_docs", "published": "", "updated": "2026", "citation_label": "Open Collective Documentation (2026)" }, { "id": "ext_github_sponsors_docs", "title": "About GitHub Sponsors for open source contributors", "priority": "external_literature", "layer": "artifact_steward_agents", "chapter_targets": [ "artifact-steward-agents-and-living-project-governance" ], "url": "https://docs.github.com/en/sponsors/receiving-sponsorships-through-github-sponsors/about-github-sponsors-for-open-source-contributors", "notes": "Official GitHub Sponsors documentation for contributor and organization sponsorship eligibility, open-source contribution categories, sponsor profiles, and GitHub-native funding surfaces.", "source_type": "official_docs", "published": "", "updated": "2026", "citation_label": "GitHub Docs (2026), About GitHub Sponsors" }, { "id": "ext_agentic_workflow_injection_2026", "title": "Demystifying and Detecting Agentic Workflow Injection Vulnerabilities in GitHub Actions", "priority": "external_literature", "layer": "artifact_steward_agents", "chapter_targets": [ "artifact-steward-agents-and-living-project-governance" ], "url": "https://arxiv.org/abs/2605.07135", "notes": "External security paper for agentic workflow injection in GitHub Actions when untrusted repository event context reaches LLM agents and downstream workflow logic.", "source_type": "arxiv", "arxiv_id": "2605.07135", "published": "2026-05-08", "updated": "2026-05-08", "citation_label": "Wang et al. (2026), Agentic Workflow Injection in GitHub Actions", "doi": "10.48550/arXiv.2605.07135" }, { "id": "ext_dao_delegation_fairness_2025", "title": "Fairness in Token Delegation: Mitigating Voting Power Concentration in DAOs", "priority": "external_literature", "layer": "artifact_steward_agents", "chapter_targets": [ "artifact-steward-agents-and-living-project-governance" ], "url": "https://arxiv.org/abs/2510.05830", "notes": "External DAO governance paper for voter apathy, voting-power concentration, delegation misalignment, and delegate-ranking bias.", "source_type": "arxiv", "arxiv_id": "2510.05830", "published": "2025-10-07", "updated": "2026-05-07", "citation_label": "Messias and Ide (2025), Fairness in Token Delegation", "doi": "10.48550/arXiv.2510.05830" }, { "id": "ext_model_cards_2019", "title": "Model Cards for Model Reporting", "priority": "external_literature", "layer": "model_reporting", "chapter_targets": [ "evidence-states-and-claim-discipline", "project-theseus-as-report-first-implementation-reference" ], "url": "https://arxiv.org/abs/1810.03993", "notes": "External reporting comparator for structured model documentation, intended-use boundaries, evaluation disclosures, ethical considerations, and model-report artifacts.", "source_type": "arxiv", "arxiv_id": "1810.03993", "published": "2019-01-14", "updated": "2019-01-14", "citation_label": "Mitchell et al. (2019), Model Cards", "doi": "10.48550/arXiv.1810.03993" }, { "id": "ext_datasheets_datasets_2021", "title": "Datasheets for Datasets", "priority": "external_literature", "layer": "dataset_documentation", "chapter_targets": [ "evidence-states-and-claim-discipline", "project-theseus-as-report-first-implementation-reference" ], "url": "https://arxiv.org/abs/1803.09010", "notes": "External documentation comparator for dataset motivation, composition, collection, preprocessing, uses, distribution, maintenance, and accountability questions.", "source_type": "arxiv", "arxiv_id": "1803.09010", "published": "2021-12-01", "updated": "2021-12-01", "citation_label": "Gebru et al. (2021), Datasheets for Datasets", "doi": "10.48550/arXiv.1803.09010" }, { "id": "ext_factsheets_ai_services_2019", "title": "FactSheets: Increasing Trust in AI Services through Supplier's Declarations of Conformity", "priority": "external_literature", "layer": "ai_service_fact_sheets", "chapter_targets": [ "project-theseus-as-report-first-implementation-reference" ], "url": "https://arxiv.org/abs/1808.07261", "notes": "External AI-service documentation comparator for supplier declarations, service properties, trust-relevant facts, and standardized reporting boundaries.", "source_type": "arxiv", "arxiv_id": "1808.07261", "published": "2019-02-07", "updated": "2019-02-07", "citation_label": "Arnold et al. (2019), FactSheets for AI Services", "doi": "10.48550/arXiv.1808.07261" }, { "id": "ext_ml_reproducibility_program_2021", "title": "Improving Reproducibility in Machine Learning Research", "priority": "external_literature", "layer": "ml_reproducibility_reporting", "chapter_targets": [ "evidence-states-and-claim-discipline", "project-theseus-as-report-first-implementation-reference" ], "url": "https://jmlr.org/papers/v22/20-303.html", "notes": "External reproducibility-program comparator for checklists, code submission, reproducibility reports, and community review mechanisms in machine-learning research.", "source_type": "journal_article", "published": "2021", "updated": "2021", "citation_label": "Pineau et al. (2021), Improving Reproducibility in Machine Learning Research" }, { "id": "ext_transformer_circuits_2021", "title": "A Mathematical Framework for Transformer Circuits", "priority": "external_literature", "layer": "mechanistic_interpretability", "chapter_targets": [ "evidence-states-and-claim-discipline", "white-box-evidence-interpretability-and-activation-governance" ], "url": "https://transformer-circuits.pub/2021/framework/index.html", "notes": "External mechanistic-interpretability comparator for treating internal circuit analyses as scoped white-box evidence that still needs model, layer, behavior, causal, and limitation boundaries.", "source_type": "research_article", "published": "2021-12-22", "updated": "2021-12-22", "citation_label": "Elhage et al. (2021), Transformer Circuits" }, { "id": "ext_monosemanticity_2023", "title": "Towards Monosemanticity: Decomposing Language Models With Dictionary Learning", "priority": "external_literature", "layer": "mechanistic_interpretability", "chapter_targets": [ "evidence-states-and-claim-discipline", "white-box-evidence-interpretability-and-activation-governance" ], "url": "https://transformer-circuits.pub/2023/monosemantic-features", "notes": "External mechanistic-interpretability comparator for sparse-autoencoder feature decomposition and the boundary between discovered features, feature-level evidence, and broader model-behavior claims.", "source_type": "research_article", "published": "2023-10-05", "updated": "2023-10-05", "citation_label": "Bricken et al. (2023), Towards Monosemanticity" }, { "id": "ext_literate_programming_1984", "title": "Literate Programming", "priority": "external_literature", "layer": "literate_programming", "chapter_targets": [ "living-book-methodology" ], "url": "https://academic.oup.com/comjnl/article/27/2/97/343244", "notes": "External literate-programming source for arranging programs and explanation around human comprehension, woven documentation, and tangled executable artifacts.", "source_type": "journal_article", "published": "1984-01-01", "updated": "1984-01-01", "citation_label": "Knuth (1984), Literate Programming", "doi": "10.1093/comjnl/27.2.97" }, { "id": "ext_jupyter_book_docs", "title": "Jupyter Book Documentation", "priority": "external_literature", "layer": "executable_books", "chapter_targets": [ "living-book-methodology" ], "url": "https://jupyterbook.org/", "notes": "Official Jupyter Book documentation comparator for authoring books from notebooks or Markdown, executing code, cross-referencing content, and publishing computational books to the web.", "source_type": "official_docs", "published": "", "updated": "", "citation_label": "Jupyter Book Documentation" }, { "id": "ext_quarto_books_docs", "title": "Quarto Books Documentation", "priority": "external_literature", "layer": "technical_publishing", "chapter_targets": [ "living-book-methodology" ], "url": "https://quarto.org/docs/books/", "notes": "Official Quarto Books documentation comparator for multi-chapter manuscripts, HTML/PDF/Word/EPUB outputs, search, cross references, and book-style website publishing.", "source_type": "official_docs", "published": "", "updated": "", "citation_label": "Quarto Documentation, Creating a Book" }, { "id": "ext_argo_rollouts_docs", "title": "Argo Rollouts Documentation: Kubernetes Progressive Delivery Controller", "priority": "external_literature", "layer": "progressive_delivery_rollback", "chapter_targets": [ "capability-replacement-and-rollback" ], "url": "https://argo-rollouts.readthedocs.io/en/stable/", "notes": "External progressive-delivery comparator for blue-green rollout, canary rollout, metric analysis, automated promotion, and automated rollback vocabulary.", "source_type": "official_docs", "published": "", "updated": "", "citation_label": "Argo Rollouts Documentation, Kubernetes Progressive Delivery Controller" }, { "id": "ext_feature_toggles_fowler", "title": "Feature Toggles (aka Feature Flags)", "priority": "external_literature", "layer": "feature_flag_release_control", "chapter_targets": [ "capability-replacement-and-rollback" ], "url": "https://martinfowler.com/articles/feature-toggles.html", "notes": "External feature-flag comparator for controlled exposure, canary releasing, release toggles, experiment toggles, ops toggles, permissioning toggles, and validation complexity.", "source_type": "industry_article", "published": "2017-10-09", "updated": "2017-10-09", "citation_label": "Hodgson (2017), Feature Toggles (aka Feature Flags)" }, { "id": "ext_google_cloud_mlops_cd", "title": "MLOps: Continuous Delivery and Automation Pipelines in Machine Learning", "priority": "external_literature", "layer": "mlops_continuous_delivery", "chapter_targets": [ "capability-replacement-and-rollback" ], "url": "https://docs.cloud.google.com/architecture/mlops-continuous-delivery-and-automation-pipelines-in-machine-learning", "notes": "External MLOps comparator for CI/CD/CT, data/model validation, model deployment, monitoring, rollback triggers, and model-regression concerns.", "source_type": "official_docs", "published": "", "updated": "2024-08-28", "citation_label": "Google Cloud Architecture Center (2024), MLOps Continuous Delivery and Automation Pipelines in Machine Learning" }, { "id": "ext_kubernetes_deployments_docs", "title": "Kubernetes Documentation: Deployments", "priority": "external_literature", "layer": "deployment_rollout_rollback", "chapter_targets": [ "capability-replacement-and-rollback" ], "url": "https://kubernetes.io/docs/concepts/workloads/controllers/deployment/", "notes": "External deployment-controller comparator for rollout status, rollout history, revision records, and rollback to a prior stable Deployment revision.", "source_type": "official_docs", "published": "", "updated": "", "citation_label": "Kubernetes Documentation, Deployments" }, { "id": "ext_drexler_cais_2019", "title": "Reframing Superintelligence: Comprehensive AI Services as General Intelligence", "priority": "external_literature", "layer": "ai_services_r_and_d_automation", "chapter_targets": [ "asi-is-a-stack-not-a-model", "constitutional-alignment-substrate", "recursive-self-improvement-boundaries", "integrated-reference-architecture" ], "url": "https://ora.ox.ac.uk/objects/uuid%3A9c05427a-6390-4b42-9c55-ee45f73a26ad/files/sf4752j50k", "notes": "Primary CAIS technical-report comparator for service-centered general intelligence, R&D automation, structured AI development, and the distinction between component affordances and a complete governance architecture.", "source_type": "technical_report", "published": "2019-01-01", "updated": "2019-01-01", "citation_label": "Drexler (2019), Reframing Superintelligence: Comprehensive AI Services as General Intelligence" }, { "id": "ext_mrkl_systems_2022", "title": "MRKL Systems: A Modular, Neuro-Symbolic Architecture That Combines Large Language Models, External Knowledge Sources and Discrete Reasoning", "priority": "external_literature", "layer": "neuro_symbolic_modular_architecture", "chapter_targets": [ "asi-is-a-stack-not-a-model" ], "url": "https://arxiv.org/abs/2205.00445", "notes": "External modular-neuro-symbolic architecture comparator for combining language models with expert modules, external knowledge sources, and routing rather than treating the model as the whole system.", "source_type": "arxiv", "arxiv_id": "2205.00445", "published": "2022-05-01", "updated": "2022-05-01", "citation_label": "Karpas et al. (2022), MRKL Systems", "doi": "10.48550/arXiv.2205.00445" }, { "id": "ext_llm_agents_survey_2023", "title": "A Survey on Large Language Model based Autonomous Agents", "priority": "external_literature", "layer": "llm_agent_architecture", "chapter_targets": [ "asi-is-a-stack-not-a-model" ], "url": "https://arxiv.org/abs/2308.11432", "notes": "External LLM-agent architecture comparator for agent profiles, memory, planning, and action modules around a language model, useful for positioning the stack frame against agent-system decompositions.", "source_type": "arxiv", "arxiv_id": "2308.11432", "published": "2023-08-22", "updated": "2024-09-22", "citation_label": "Wang et al. (2023), Survey on LLM-based Autonomous Agents", "doi": "10.48550/arXiv.2308.11432" }, { "id": "ext_standard_model_mind_2017", "title": "A Standard Model of the Mind: Toward a Common Computational Framework Across Artificial Intelligence, Cognitive Science, Neuroscience, and Robotics", "priority": "external_literature", "layer": "cognitive_architecture", "chapter_targets": [ "asi-is-a-stack-not-a-model" ], "url": "https://ojs.aaai.org/aimagazine/index.php/aimagazine/article/view/2744", "notes": "External cognitive-architecture comparator for treating intelligent behavior as an integrated architecture spanning memory, learning, perception/action, procedural control, and deliberation.", "source_type": "journal_article", "published": "2017", "updated": "2017", "citation_label": "Laird, Lebiere, and Rosenbloom (2017), Standard Model of the Mind", "doi": "10.1609/aimag.v38i4.2744" }, { "id": "ext_subsumption_architecture_1986", "title": "A Robust Layered Control System for a Mobile Robot", "priority": "external_literature", "layer": "layered_robot_control_architecture", "chapter_targets": [ "asi-is-a-stack-not-a-model" ], "url": "https://people.csail.mit.edu/brooks/papers/AIM-864.pdf", "notes": "External layered-control architecture comparator for decomposing robot behavior into interacting layers rather than centralizing behavior in one monolithic controller.", "source_type": "technical_report", "published": "1986", "updated": "1986", "citation_label": "Brooks (1986), Robust Layered Control System", "doi": "10.1109/JRA.1986.1087032" }, { "id": "ext_humans_automation_1997", "title": "Humans and Automation: Use, Misuse, Disuse, Abuse", "priority": "external_literature", "layer": "human_factors_automation", "chapter_targets": [ "runtime-adapters-tool-permissions-and-human-approval", "human-intent-as-a-formal-input", "evidence-states-and-claim-discipline", "human-factors-and-meaningful-control-in-oversight" ], "url": "https://sage.cnpereading.com/doi/10.1518/001872097778543886", "notes": "External human-factors comparator for automation use, misuse, disuse, abuse, overreliance, monitoring failure, workload, trust, risk, false alarms, and operator-role design.", "source_type": "journal_article", "published": "1997-06", "updated": "1997-06", "citation_label": "Parasuraman and Riley (1997), Humans and Automation", "doi": "10.1518/001872097778543886" }, { "id": "ext_ironies_automation_1983", "title": "Ironies of Automation", "priority": "external_literature", "layer": "human_factors_automation", "chapter_targets": [ "runtime-adapters-tool-permissions-and-human-approval", "evidence-states-and-claim-discipline", "human-factors-and-meaningful-control-in-oversight" ], "url": "https://www.sciencedirect.com/science/article/pii/0005109883900468", "notes": "External automation comparator for the argument that automation can expand rather than eliminate human-operator problems and can leave humans with difficult abnormal-condition duties.", "source_type": "journal_article", "published": "1983", "updated": "1983", "citation_label": "Bainbridge (1983), Ironies of Automation", "doi": "10.1016/0005-1098(83)90046-8" }, { "id": "ext_levels_automation_2000", "title": "A Model for Types and Levels of Human Interaction with Automation", "priority": "external_literature", "layer": "human_factors_automation", "chapter_targets": [ "runtime-adapters-tool-permissions-and-human-approval", "human-intent-as-a-formal-input", "human-factors-and-meaningful-control-in-oversight" ], "url": "https://dl.acm.org/doi/10.1109/3468.844354", "notes": "External human-automation comparator for separating automation by information acquisition, analysis, decision/action selection, and action implementation rather than treating human approval as a single undifferentiated gate.", "source_type": "journal_article", "published": "2000", "updated": "2000", "citation_label": "Parasuraman, Sheridan, and Wickens (2000), Types and Levels of Automation", "doi": "10.1109/3468.844354" }, { "id": "ext_complacency_bias_automation_2010", "title": "Complacency and Bias in Human Use of Automation: An Attentional Integration", "priority": "external_literature", "layer": "human_factors_automation", "chapter_targets": [ "runtime-adapters-tool-permissions-and-human-approval", "human-intent-as-a-formal-input", "evidence-states-and-claim-discipline", "human-factors-and-meaningful-control-in-oversight" ], "url": "https://pubmed.ncbi.nlm.nih.gov/21077562/", "notes": "External human-factors comparator for automation complacency, omission and commission errors, automation bias, workload, attention, and imperfect decision aids.", "source_type": "journal_article", "published": "2010", "updated": "2010", "citation_label": "Parasuraman and Manzey (2010), Complacency and Bias in Human Use of Automation", "doi": "10.1177/0018720810376055" }, { "id": "ext_bourtoule_machine_unlearning_2021", "title": "Machine Unlearning", "priority": "external_literature", "layer": "machine_unlearning_data_governance", "chapter_targets": [ "data-engines-continual-learning-and-unlearning", "policy-optimization-and-learning-from-feedback", "procedural-memory-and-cognitive-loop-closure", "benchmark-ratchets-and-anti-goodhart-evidence", "adjudicated-persistence-and-the-adaptive-commit-boundary", "open-research-agenda-and-bibliography-plan" ], "url": "https://doi.org/10.1109/SP40001.2021.00019", "notes": "Primary machine-unlearning comparator for deletion-path architecture, checkpointed training, bounded retraining, accuracy-cost trade-offs, and the boundary between deletion requests and verified removal.", "source_type": "conference_paper", "published": "2021-05-24", "updated": "2021-05-24", "citation_label": "Bourtoule et al. (2021), Machine Unlearning", "doi": "10.1109/SP40001.2021.00019" }, { "id": "ext_shumailov_model_collapse_2023", "title": "The Curse of Recursion: Training on Generated Data Makes Models Forget", "priority": "external_literature", "layer": "synthetic_data_feedback_model_collapse", "chapter_targets": [ "data-engines-continual-learning-and-unlearning", "policy-optimization-and-learning-from-feedback", "procedural-memory-and-cognitive-loop-closure", "benchmark-ratchets-and-anti-goodhart-evidence", "open-research-agenda-and-bibliography-plan" ], "url": "https://arxiv.org/abs/2305.17493", "notes": "Primary preprint comparator for generated-data feedback, provenance, distribution-tail loss, and model-collapse risk under specified recursive-training assumptions; not a universal synthetic-data safety result.", "source_type": "preprint", "published": "2023-05-27", "updated": "2024-04-14", "citation_label": "Shumailov et al. (2023), The Curse of Recursion", "doi": "10.48550/arXiv.2305.17493" }, { "id": "ext_gerstgrasser_data_accumulation_2024", "title": "Is Model Collapse Inevitable? Breaking the Curse of Recursion by Accumulating Real and Synthetic Data", "priority": "external_literature", "layer": "synthetic_data_retention_policy", "chapter_targets": [ "data-engines-continual-learning-and-unlearning", "policy-optimization-and-learning-from-feedback", "procedural-memory-and-cognitive-loop-closure", "benchmark-ratchets-and-anti-goodhart-evidence", "open-research-agenda-and-bibliography-plan" ], "url": "https://arxiv.org/abs/2404.01413", "notes": "Primary empirical and analytical comparator that distinguishes replacement from accumulation of real and synthetic data; gives a counterweight to blanket model-collapse claims without resolving deletion, privacy, provenance, or poisoning risk.", "source_type": "workshop_paper", "published": "2024-04-01", "updated": "2024-07-01", "citation_label": "Gerstgrasser et al. (2024), Is Model Collapse Inevitable?", "doi": "10.48550/arXiv.2404.01413" }, { "id": "theseus_synthetic_data_curation", "title": "Theseus Synthetic Data Curation", "priority": "implementation_reference", "layer": "governed_synthetic_data_admission", "chapter_targets": [ "benchmark-ratchets-and-anti-goodhart-evidence", "data-engines-continual-learning-and-unlearning", "policy-optimization-and-learning-from-feedback", "privacy-data-rights-and-information-flow-governance", "procedural-memory-and-cognitive-loop-closure", "project-theseus-as-report-first-implementation-reference" ], "url": "local-project:Theseus-Hive@09ecbd8cff4a:docs/SYNTHETIC_DATA_CURATION.md", "notes": "Pinned Project Theseus implementation-reference record for residual-targeted synthetic-data admission, provenance, leakage checks, ratio caps, governed teacher handling, and source-reported promotion gates; no ASI Stack replay or model-quality import.", "source_type": "local_project_record", "published": "2026-05-12", "updated": "2026-05-12", "citation_label": "Project Theseus, Synthetic Data Curation (pinned local source)" }, { "id": "ext_alignment_faking_2024", "title": "Alignment Faking in Large Language Models", "priority": "external_literature", "layer": "training_time_deception", "chapter_targets": [ "adversarial-evaluation-sandbagging-and-training-time-deception" ], "url": "https://arxiv.org/abs/2412.14093", "notes": "Primary experimental comparator for context-dependent alignment faking under disclosed training conditions; does not establish that an ASI Stack model, evaluator, or training process is deceptive.", "source_type": "preprint", "published": "2024-12-18", "updated": "2024-12-18", "citation_label": "Greenblatt et al. (2024), Alignment Faking in Large Language Models", "doi": "10.48550/arXiv.2412.14093" }, { "id": "ext_ai_sandbagging_2024", "title": "AI Sandbagging: Language Models Can Strategically Underperform on Evaluations", "priority": "external_literature", "layer": "evaluation_integrity", "chapter_targets": [ "adversarial-evaluation-sandbagging-and-training-time-deception" ], "url": "https://arxiv.org/abs/2406.07358", "notes": "Primary experimental comparator for strategic evaluation underperformance, prompted and password-locked capability hiding, and limits of capability-evaluation trustworthiness; does not show sandbagging in this repository.", "source_type": "preprint", "published": "2024-06-11", "updated": "2024-06-11", "citation_label": "van der Weij et al. (2024), AI Sandbagging", "doi": "10.48550/arXiv.2406.07358" }, { "id": "ext_emergent_misalignment_reward_hacking_2025", "title": "Natural Emergent Misalignment from Reward Hacking in Production RL", "priority": "external_literature", "layer": "training_time_deception", "chapter_targets": [ "adversarial-evaluation-sandbagging-and-training-time-deception" ], "url": "https://arxiv.org/abs/2511.18397", "notes": "Primary experimental comparator for reward-hacking-induced misaligned generalization in a specified production-RL research setting, including reported mitigation conditions; it is not evidence of local model behavior or a general causal law.", "source_type": "preprint", "published": "2025-11-23", "updated": "2025-11-23", "citation_label": "MacDiarmid et al. (2025), Natural Emergent Misalignment from Reward Hacking in Production RL", "doi": "10.48550/arXiv.2511.18397" }, { "id": "ext_poet_2019", "title": "Paired Open-Ended Trailblazer (POET): Endlessly Generating Increasingly Complex and Diverse Learning Environments and Their Solutions", "priority": "external_literature", "layer": "open_ended_environment_solution_generation", "chapter_targets": [ "open-ended-improvement-engines" ], "url": "https://arxiv.org/abs/1901.01753", "notes": "Primary open-ended-learning comparator for paired environment generation, agent optimization, and cross-environment solution transfer in a specified BipedalWalker setting; it does not establish a general improvement engine, evaluator soundness, or ASI Stack result.", "source_type": "preprint", "published": "2019-01-07", "updated": "2019-02-21", "citation_label": "Wang et al. (2019), Paired Open-Ended Trailblazer", "doi": "10.48550/arXiv.1901.01753" }, { "id": "ext_funsearch_2024", "title": "Mathematical Discoveries from Program Search with Large Language Models", "priority": "external_literature", "layer": "evaluator_bounded_program_search", "chapter_targets": [ "open-ended-improvement-engines" ], "url": "https://doi.org/10.1038/s41586-023-06924-6", "notes": "Primary program-search comparator for a fixed pretrained LLM, user-provided evaluation function, candidate program archive, and iterative search over a bounded specification; it does not establish open-ended general intelligence, self-modification, evaluator correctness, or an ASI Stack result.", "source_type": "journal_article", "published": "2024-01-17", "updated": "2024-01-17", "citation_label": "Romera-Paredes et al. (2024), Mathematical Discoveries from Program Search with Large Language Models", "doi": "10.1038/s41586-023-06924-6" }, { "id": "ext_gsn_community_standard_2011", "title": "GSN Community Standard Version 1", "priority": "external_literature", "layer": "structured_assurance_argumentation", "chapter_targets": [ "safety-cases-and-structured-assurance" ], "url": "https://www.faa.gov/about/office_org/headquarters_offices/ang/redac/redac-sas-201503-gsn-community-standard-v1.pdf", "notes": "Primary notation standard comparator for explicit goals, strategies, solutions, context, assumptions, justifications, and relationships in structured assurance arguments; the notation documents asserted support but does not establish claim truth.", "source_type": "technical_standard", "published": "2011-11-16", "updated": "2011-11-16", "citation_label": "Assurance Case Working Group (2011), GSN Community Standard v1" }, { "id": "ext_evaluations_safety_cases_scheming_2024", "title": "Towards Evaluations-Based Safety Cases for AI Scheming", "priority": "external_literature", "layer": "ai_safety_case_methodology", "chapter_targets": [ "safety-cases-and-structured-assurance" ], "url": "https://arxiv.org/abs/2411.03336", "notes": "Primary safety-case comparator for scoped scheming inability, harm inability, harm control, alignment arguments, empirical evaluation dependencies, and acknowledged unresolved assumptions; it does not establish any ASI Stack safety case or safety result.", "source_type": "preprint", "arxiv_id": "2411.03336", "published": "2024-10-29", "updated": "2024-10-29", "citation_label": "Balesni et al. (2024), Evaluations-Based Safety Cases for AI Scheming", "doi": "10.48550/arXiv.2411.03336" }, { "id": "ext_aisi_safety_cases_2024", "title": "Safety Cases at AISI", "priority": "external_literature", "layer": "ai_safety_case_methodology", "chapter_targets": [ "safety-cases-and-structured-assurance" ], "url": "https://www.aisi.gov.uk/blog/safety-cases-at-aisi", "notes": "Official AI Safety Institute methodology comparator for structured safety-case sketches, positive and negative evidence, countercases, open scientific uncertainty, and limits on confidence; it is not evidence that this book has a complete safety case.", "source_type": "official_technical_note", "published": "2024", "updated": "2024", "citation_label": "AI Safety Institute (2024), Safety Cases at AISI" }, { "id": "ext_rand_model_weight_security_2024", "title": "Securing AI Model Weights: Preventing Theft and Misuse of Frontier Models", "priority": "external_literature", "layer": "model_weight_custody", "chapter_targets": [ "open-weight-release-and-post-release-control", "model-weight-custody-and-hardware-roots-of-trust" ], "url": "https://www.rand.org/content/dam/rand/pubs/research_reports/RRA2800/RRA2849-1/RAND_RRA2849-1.pdf", "notes": "Primary RAND analysis of frontier-model-weight theft/misuse threat surfaces, security levels, defense-in-depth, access control, physical and organizational controls; it does not establish local protection or safety.", "source_type": "research_report", "published": "2024-06-12", "updated": "2024-06-12", "citation_label": "Nevo et al. (2024), Securing AI Model Weights" }, { "id": "ext_nist_confidential_computing_2026", "title": "Hardware-Enabled Security: Confidential Computing of Data in Cloud Workloads", "priority": "external_literature", "layer": "hardware_root_attestation", "chapter_targets": [ "model-weight-custody-and-hardware-roots-of-trust" ], "url": "https://nvlpubs.nist.gov/nistpubs/ir/2026/NIST.IR.8320E.ipd.pdf", "notes": "NIST initial-public-draft comparator for hardware-enabled confidential computing, memory protection, trust domains, attestation-gated key release, and AI model/data protection; it is draft guidance, not a local TEE result.", "source_type": "draft_standard", "published": "2026-05", "updated": "2026-05", "citation_label": "NIST (2026), IR 8320E Initial Public Draft" }, { "id": "ext_nvidia_confidential_model_lifecycle_2026", "title": "Workload and Model Lifecycle: Deploying Proprietary Models Securely with NVIDIA Confidential Computing", "priority": "external_literature", "layer": "attestation_gated_model_loading", "chapter_targets": [ "model-weight-custody-and-hardware-roots-of-trust" ], "url": "https://docs.nvidia.com/enterprise-reference-architectures/deploying-proprietary-models-confidential-compute-self-hosted-kubernetes/latest/workload-and-model-lifecycle.html", "notes": "Official vendor implementation-reference comparator for encrypted weights outside a confidential pod, policy-sensitive attestation evidence, and key-release decisions; it does not establish a local confidential deployment or attestation result.", "source_type": "official_technical_documentation", "published": "2026", "updated": "2026", "citation_label": "NVIDIA (2026), Confidential Model Lifecycle" }, { "id": "ext_provable_model_weight_release_2025", "title": "Towards Provable (In)Secure Model Weight Release Schemes", "priority": "external_literature", "layer": "open_weight_release_security", "chapter_targets": [ "open-weight-release-and-post-release-control", "model-weight-custody-and-hardware-roots-of-trust" ], "url": "https://arxiv.org/abs/2506.19874", "notes": "Primary formal-security comparator for evaluating claimed secure model-weight release schemes and parameter-extraction failure modes; it does not establish an ASI Stack release scheme or release decision.", "source_type": "preprint", "arxiv_id": "2506.19874", "published": "2025-06-23", "updated": "2025-06-23", "citation_label": "Yang et al. (2025), Provable (In)Secure Model Weight Release Schemes", "doi": "10.48550/arXiv.2506.19874" }, { "id": "ext_nist_cscrm_2022", "title": "Cybersecurity Supply Chain Risk Management Practices for Systems and Organizations", "priority": "external_literature", "layer": "ai_supply_chain_governance", "chapter_targets": [ "ai-supply-chain-integrity-and-lifecycle-provenance" ], "url": "https://nvlpubs.nist.gov/nistpubs/SpecialPublications/NIST.SP.800-161r1-upd1.pdf", "notes": "Primary NIST C-SCRM standard comparator for lifecycle-wide risk framing, supplier/component inventory, assessment, response, monitoring, and incident communication; it does not establish a local supply-chain program or AI artifact integrity.", "source_type": "standard", "published": "2022-05", "updated": "2022-05", "citation_label": "NIST (2022), SP 800-161r1" }, { "id": "ext_slsa_build_track_1_2", "title": "SLSA Build Track Basics, version 1.2", "priority": "external_literature", "layer": "build_provenance", "chapter_targets": [ "ai-supply-chain-integrity-and-lifecycle-provenance" ], "url": "https://slsa.dev/spec/v1.2/build-track-basics", "notes": "Official SLSA specification comparator for build provenance, signed hosted builds, verification, and graduated assurance; provenance quality and SLSA level do not establish local artifact correctness, data quality, model safety, or deployment authority.", "source_type": "technical_specification", "published": "2025", "updated": "2026-07-10", "citation_label": "SLSA (2025), Build Track Basics v1.2" }, { "id": "ext_openssf_model_signing_spec_2025", "title": "OpenSSF Model Signing Specification", "priority": "external_literature", "layer": "ai_artifact_signing", "chapter_targets": [ "ai-supply-chain-integrity-and-lifecycle-provenance" ], "url": "https://github.com/ossf/model-signing-spec", "notes": "Official OpenSSF AI/ML working-group specification comparator for signed model/dataset bundles, verification, provenance metadata, and explicit limits of model signing; it does not establish local signing, verification, integrity, confidentiality, safety, or release authority.", "source_type": "technical_specification", "published": "2025", "updated": "2026-07-10", "citation_label": "OpenSSF (2025), Model Signing Specification" }, { "id": "ext_spdx_ai_profile_3_0_1", "title": "SPDX Specification 3.0.1 AI Profile", "priority": "external_literature", "layer": "ai_bill_of_materials", "chapter_targets": [ "ai-supply-chain-integrity-and-lifecycle-provenance" ], "url": "https://spdx.github.io/spdx-spec/v3.0.1/model/AI/AI/", "notes": "Official SPDX specification comparator for interoperable AI system/model, dataset, build, supplier, provenance, integrity, relationship, and lifecycle metadata; a conformant BOM is not proof of complete inventory, artifact security, data fitness, model safety, or compliance.", "source_type": "technical_specification", "published": "2024-12", "updated": "2024-12", "citation_label": "SPDX (2024), Specification 3.0.1 AI Profile" }, { "id": "ext_mcp_protocol_2025_06_18", "title": "Model Context Protocol Specification, revision 2025-06-18", "priority": "external_literature", "layer": "agent_tool_protocol", "chapter_targets": [ "inter-stack-protocols-identity-and-economic-exchange" ], "url": "https://modelcontextprotocol.io/specification/2025-06-18/basic/index", "notes": "Official Model Context Protocol comparator for JSON-RPC message shape, lifecycle management, capability negotiation, session control, schema-defined interactions, and modular tool/client/server features; it does not establish a local protocol implementation, peer identity, authorization, message truth, task completion, payment, or deployment safety.", "source_type": "technical_specification", "published": "2025-06-18", "updated": "2025-06-18", "citation_label": "Model Context Protocol (2025), Specification revision 2025-06-18" }, { "id": "ext_a2a_protocol_0_3_0", "title": "Agent2Agent Protocol Specification, version 0.3.0", "priority": "external_literature", "layer": "agent_to_agent_protocol", "chapter_targets": [ "inter-stack-protocols-identity-and-economic-exchange" ], "url": "https://a2a-protocol.org/v0.3.0/specification/", "notes": "Official A2A comparator for agent discovery, agent cards, delegated tasks, artifact/message exchange, transport choices, and interoperability between opaque agent systems; it does not establish a local A2A deployment, verified identity, delegated authority, task truth, secure execution, payment, or safety.", "source_type": "technical_specification", "updated": "2026-07-10", "citation_label": "A2A Protocol (v0.3.0), Specification" }, { "id": "ext_mcp_protocol_2025_11_25", "title": "Model Context Protocol Specification, revision 2025-11-25", "priority": "external_literature", "layer": "agent_tool_protocol", "chapter_targets": [ "inter-stack-protocols-identity-and-economic-exchange" ], "url": "https://modelcontextprotocol.io/specification/2025-11-25", "notes": "Latest released MCP comparator inspected on 2026-07-10 for versioned lifecycle, capability negotiation, authorization and OpenID Connect discovery changes, elicitation, tasks, and modular client/server boundaries; the announced 2026-07-28 revision remains a release candidate and is not represented as released.", "source_type": "technical_specification", "published": "2025-11-25", "updated": "2026-07-10", "citation_label": "Model Context Protocol (2025), Specification revision 2025-11-25" }, { "id": "ext_a2a_protocol_1_0_0", "title": "Agent2Agent Protocol Specification, version 1.0.0", "priority": "external_literature", "layer": "agent_to_agent_protocol", "chapter_targets": [ "inter-stack-protocols-identity-and-economic-exchange" ], "url": "https://a2a-protocol.org/latest/specification/", "notes": "Latest released A2A comparator inspected on 2026-07-10 for canonical data objects, version negotiation, Agent Cards, tasks/messages/artifacts, JSON-RPC, gRPC and HTTP bindings, authorization scoping, interoperability testing, and security considerations; it does not establish local conformance, peer truth, delegated authority, or safe effects.", "source_type": "technical_specification", "published": "2026", "updated": "2026-07-10", "citation_label": "A2A Protocol (v1.0.0), Specification" }, { "id": "ext_w3c_did_core_1_0_2022", "title": "Decentralized Identifiers (DIDs) v1.0", "priority": "external_literature", "layer": "decentralized_identity", "chapter_targets": [ "inter-stack-protocols-identity-and-economic-exchange" ], "url": "https://www.w3.org/TR/did-core/", "notes": "W3C DID Core comparator for decentralized identifier syntax, data model, controller-related metadata, resolution, and privacy considerations; it does not establish an ASI Stack identity system, controller trust, authorization, non-repudiation, revocation effectiveness, or safety.", "source_type": "standard", "published": "2022-07-19", "updated": "2022-07-19", "citation_label": "W3C (2022), Decentralized Identifiers v1.0" }, { "id": "ext_w3c_vc_data_model_2_0_2025", "title": "Verifiable Credentials Data Model v2.0", "priority": "external_literature", "layer": "verifiable_credentials", "chapter_targets": [ "inter-stack-protocols-identity-and-economic-exchange" ], "url": "https://www.w3.org/TR/vc-data-model/", "notes": "W3C Verifiable Credentials comparator for issuer/holder/verifier roles, credential and presentation fields, validity/status, evidence, securing mechanisms, and explicit authorization limitations; it does not establish an ASI Stack credential, trust decision, authorization framework, delegation validity, payment, or safety.", "source_type": "standard", "published": "2025-05-15", "updated": "2025-05-15", "citation_label": "W3C (2025), Verifiable Credentials Data Model v2.0" }, { "id": "ext_interledger_protocol_v4", "title": "Interledger Protocol V4", "priority": "external_literature", "layer": "interledger_value_transfer", "chapter_targets": [ "inter-stack-protocols-identity-and-economic-exchange" ], "url": "https://interledger.org/developers/rfcs/interledger-protocol/", "notes": "Official Interledger comparator for neutral packetized value transfer across independent ledgers, connector obligations, balances, and end-to-end boundary design; it does not establish an ASI Stack payment route, settlement, accounting correctness, legal transfer, economic fairness, delegated authority, or safety.", "source_type": "technical_specification", "updated": "2026-07-10", "citation_label": "Interledger (2026), Interledger Protocol V4" }, { "id": "ext_test_time_compute_scaling_2024", "title": "Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters", "priority": "external_literature", "layer": "test_time_compute_allocation", "chapter_targets": [ "governed-deliberation-and-test-time-scaling" ], "url": "https://arxiv.org/abs/2408.03314", "notes": "Primary test-time-compute comparator for verifier-guided search, proposal refinement, difficulty-dependent compute allocation, and the limits of extra inference; it does not establish local reasoning improvement, verifier correctness, safety, or an ASI Stack result.", "source_type": "preprint", "arxiv_id": "2408.03314", "published": "2024-08-06", "updated": "2024-08-06", "citation_label": "Snell et al. (2024), Scaling LLM Test-Time Compute Optimally", "doi": "10.48550/arXiv.2408.03314" }, { "id": "ext_graphrag_2024", "title": "From Local to Global: A Graph RAG Approach to Query-Focused Summarization", "priority": "external_literature", "layer": "graph_based_retrieval_and_global_sensemaking", "chapter_targets": [ "virtual-context-abi", "context-transactions-snapshots-mounts-and-taint", "verification-bandwidth-and-context-adequacy", "open-research-agenda-and-bibliography-plan" ], "url": "https://arxiv.org/abs/2404.16130", "notes": "Primary GraphRAG comparator for LLM-derived entity graphs, community summaries, and global corpus questions; generated graph and summary layers remain fallible derived representations and do not establish truth, complete coverage, local adequacy, or an ASI Stack memory result.", "source_type": "preprint", "arxiv_id": "2404.16130", "published": "2024-04-24", "updated": "2025-02-19", "citation_label": "Edge et al. (2024), From Local to Global: A Graph RAG Approach", "doi": "10.48550/arXiv.2404.16130" }, { "id": "ext_hipporag_2024", "title": "HippoRAG: Neurobiologically Inspired Long-Term Memory for Large Language Models", "priority": "external_literature", "layer": "associative_long_term_memory", "chapter_targets": [ "virtual-context-abi", "verification-bandwidth-and-context-adequacy", "routing-heads-and-specialist-cores", "open-research-agenda-and-bibliography-plan" ], "url": "https://papers.nips.cc/paper_files/paper/2024/hash/6ddc001d07ca4f319af96a3024f6dbd1-Abstract-Conference.html", "notes": "Primary NeurIPS comparator for knowledge-graph retrieval with Personalized PageRank and single-step associative navigation; reported multi-hop QA gains do not establish durable truth, update correctness, resistance to poisoning, local reproduction, or a general memory system.", "source_type": "conference_paper", "published": "2024", "updated": "2024", "citation_label": "Guti\u00e9rrez et al. (2024), HippoRAG", "doi": "10.52202/079017-1902" }, { "id": "ext_raptor_2024", "title": "RAPTOR: Recursive Abstractive Processing for Tree-Organized Retrieval", "priority": "external_literature", "layer": "hierarchical_retrieval_and_abstraction", "chapter_targets": [ "virtual-context-abi", "verification-bandwidth-and-context-adequacy", "compact-generative-systems-and-residual-honesty", "rankfold-neuralfold-and-artifact-compression" ], "url": "https://proceedings.iclr.cc/paper_files/paper/2024/hash/8a2acd174940dbca361a6398a4f9df91-Abstract-Conference.html", "notes": "Primary ICLR comparator for recursive clustering, summarization, and retrieval across multiple abstraction levels; source-reported QA gains do not prove summary fidelity, provenance preservation, local reproduction, or safe compression for ASI Stack claims.", "source_type": "conference_paper", "published": "2024", "updated": "2024", "citation_label": "Sarthi et al. (2024), RAPTOR" }, { "id": "ext_mem0_2025", "title": "Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory", "priority": "external_literature", "layer": "agent_long_term_memory", "chapter_targets": [ "virtual-context-abi", "context-transactions-snapshots-mounts-and-taint", "procedural-memory-and-cognitive-loop-closure", "resource-economics-and-token-budgets", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://arxiv.org/abs/2504.19413", "notes": "Primary preprint comparator for extracting, consolidating, retrieving, and graph-linking conversational memory under latency and token-cost constraints; LOCOMO and LLM-judge results do not establish fact correctness, poisoning resistance, general memory, local reproduction, or production readiness here.", "source_type": "preprint", "arxiv_id": "2504.19413", "published": "2025-04-28", "updated": "2025-04-28", "citation_label": "Chhikara et al. (2025), Mem0", "doi": "10.48550/arXiv.2504.19413" }, { "id": "ext_w3c_prov_o_2013", "title": "PROV-O: The PROV Ontology", "priority": "external_literature", "layer": "interoperable_provenance_model", "chapter_targets": [ "evidence-states-and-claim-discipline", "claim-ledgers-and-belief-revision", "artifact-graphs-audit-logs-and-replay", "ai-supply-chain-integrity-and-lifecycle-provenance", "data-engines-continual-learning-and-unlearning" ], "url": "https://www.w3.org/TR/prov-o/", "notes": "W3C Recommendation comparator for interoperable provenance over entities, activities, agents, derivation, attribution, delegation, revision, and invalidation; a PROV-O graph records asserted provenance and does not by itself prove assertion truth, completeness, integrity, authority, or safety.", "source_type": "standard", "published": "2013-04-30", "updated": "2013-04-30", "citation_label": "W3C (2013), PROV-O: The PROV Ontology" }, { "id": "ext_mlcommons_croissant_1_1_2026", "title": "Croissant Format Specification, version 1.1", "priority": "external_literature", "layer": "machine_readable_dataset_metadata", "chapter_targets": [ "ai-supply-chain-integrity-and-lifecycle-provenance", "artifact-graphs-audit-logs-and-replay", "benchmark-ratchets-and-anti-goodhart-evidence", "data-engines-continual-learning-and-unlearning" ], "url": "https://docs.mlcommons.org/croissant/docs/croissant-spec-1.1.html", "notes": "Current MLCommons specification comparator for JSON-LD dataset structure, resources, checksums, record fields, machine-readable provenance, usage conditions, and portability across ML tooling; metadata conformance does not prove dataset integrity, fitness, legality, representativeness, or safe use.", "source_type": "technical_specification", "published": "2026-01-29", "updated": "2026-01-29", "citation_label": "MLCommons (2026), Croissant Format Specification v1.1" }, { "id": "ext_inspect_ai_2024", "title": "Inspect AI: Framework for Large Language Model Evaluations", "priority": "external_literature", "layer": "model_and_agent_evaluation_framework", "chapter_targets": [ "runtime-adapters-tool-permissions-and-human-approval", "benchmark-ratchets-and-anti-goodhart-evidence", "capability-thresholds-and-deployment-commitments", "adversarial-evaluation-sandbagging-and-training-time-deception" ], "url": "https://inspect.aisi.org.uk/", "notes": "Official UK AI Security Institute framework comparator for composable evaluation tasks, datasets, solvers, scorers, agents, tools, logs, and sandboxes; framework availability or a passing task does not establish benchmark validity, coverage, local execution, safety, or deployment readiness.", "source_type": "official_docs", "published": "2024-05", "updated": "2026-07-10", "citation_label": "UK AI Security Institute (2024), Inspect AI" }, { "id": "ext_in_toto_2019", "title": "in-toto: Providing farm-to-table guarantees for bits and bytes", "priority": "external_literature", "layer": "software_supply_chain_attestation", "chapter_targets": [ "model-weight-custody-and-hardware-roots-of-trust", "ai-supply-chain-integrity-and-lifecycle-provenance", "artifact-graphs-audit-logs-and-replay" ], "url": "https://www.usenix.org/conference/usenixsecurity19/presentation/torres-arias", "notes": "Primary USENIX comparator for cryptographically verifying authorized software-supply-chain steps from source through deployment; valid attestations do not prove artifact correctness, uncompromised authorized actors, model safety, data fitness, or deployment merit.", "source_type": "conference_paper", "published": "2019-08", "updated": "2019-08", "citation_label": "Torres-Arias et al. (2019), in-toto" }, { "id": "ext_agentdojo_2024", "title": "AgentDojo: A Dynamic Environment to Evaluate Prompt Injection Attacks and Defenses for LLM Agents", "priority": "external_literature", "layer": "agent_prompt_injection_evaluation", "chapter_targets": [ "security-kernel-and-digital-scifs", "runtime-adapters-tool-permissions-and-human-approval", "benchmark-ratchets-and-anti-goodhart-evidence", "adversarial-evaluation-sandbagging-and-training-time-deception" ], "url": "https://proceedings.neurips.cc/paper_files/paper/2024/hash/97091a5177d8dc64b1da8bf3e1f6fb54-Abstract-Datasets_and_Benchmarks_Track.html", "notes": "Primary NeurIPS benchmark comparator for agents executing tools over untrusted data, with realistic tasks, security test cases, attacks, and defenses; benchmark results do not establish complete attack coverage, deployed robustness, safe authority handling, or local reproduction.", "source_type": "conference_paper", "published": "2024", "updated": "2024", "citation_label": "Debenedetti et al. (2024), AgentDojo", "doi": "10.52202/079017-2636" }, { "id": "ext_camel_prompt_injection_2025", "title": "Defeating Prompt Injections by Design", "priority": "external_literature", "layer": "capability_secure_agent_control_flow", "chapter_targets": [ "system-boundaries-and-authority", "security-kernel-and-digital-scifs", "intent-to-execution-contracts", "runtime-adapters-tool-permissions-and-human-approval" ], "url": "https://arxiv.org/abs/2503.18813", "notes": "Primary CaMeL comparator for separating trusted control flow from untrusted data and enforcing capability policies at tool calls; source-reported AgentDojo results do not establish universal prompt-injection resistance, correct policy extraction, local implementation, or safe deployment.", "source_type": "preprint", "arxiv_id": "2503.18813", "published": "2025-03-24", "updated": "2025-06-24", "citation_label": "Debenedetti et al. (2025), Defeating Prompt Injections by Design", "doi": "10.48550/arXiv.2503.18813" }, { "id": "ext_owasp_agentic_top_10_2026", "title": "OWASP Top 10 for Agentic Applications for 2026", "priority": "external_literature", "layer": "agentic_application_security_taxonomy", "chapter_targets": [ "system-boundaries-and-authority", "security-kernel-and-digital-scifs", "ai-supply-chain-integrity-and-lifecycle-provenance", "runtime-adapters-tool-permissions-and-human-approval", "inter-stack-protocols-identity-and-economic-exchange", "adversarial-evaluation-sandbagging-and-training-time-deception" ], "url": "https://genai.owasp.org/resource/owasp-top-10-for-agentic-applications-for-2026/", "notes": "Current OWASP community taxonomy comparator for goal hijacking, tool misuse, identity abuse, agentic supply chains, code execution, memory poisoning, inter-agent communication, cascading failures, human trust exploitation, and rogue agents; a risk list is not a proof of completeness, control effectiveness, local testing, or system safety.", "source_type": "official_guidance", "published": "2025-12-09", "updated": "2026-07-10", "citation_label": "OWASP (2025), Top 10 for Agentic Applications for 2026" }, { "id": "ext_darwin_godel_machine_2025", "title": "Darwin Godel Machine: Open-Ended Evolution of Self-Improving Agents", "priority": "external_literature", "layer": "empirical_recursive_agent_improvement", "chapter_targets": [ "recursive-self-improvement-boundaries", "open-ended-improvement-engines", "readiness-gates-residual-escrow-and-quarantine", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://arxiv.org/abs/2505.22954", "notes": "Primary preprint comparator for archive-based open-ended code self-modification selected by empirical coding benchmarks under sandboxing and human oversight; reported benchmark gains do not establish monotonic general improvement, safe recursive self-improvement, local reproduction, or permission to self-modify.", "source_type": "preprint", "arxiv_id": "2505.22954", "published": "2025-05-29", "updated": "2026-03-12", "citation_label": "Zhang et al. (2025), Darwin Godel Machine", "doi": "10.48550/arXiv.2505.22954" }, { "id": "ext_adas_2024", "title": "Automated Design of Agentic Systems", "priority": "external_literature", "layer": "automated_agent_architecture_search", "chapter_targets": [ "recursive-self-improvement-boundaries", "open-ended-improvement-engines", "benchmark-ratchets-and-anti-goodhart-evidence", "integrated-reference-architecture" ], "url": "https://arxiv.org/abs/2408.08435", "notes": "Primary ADAS comparator for a meta-agent that searches a growing archive of code-defined agent designs across prompts, tools, and workflows; reported transfer results do not establish unrestricted generality, safe architecture search, local reproduction, or automatic promotion authority.", "source_type": "preprint", "arxiv_id": "2408.08435", "published": "2024-08-15", "updated": "2025-03-02", "citation_label": "Hu et al. (2024), Automated Design of Agentic Systems", "doi": "10.48550/arXiv.2408.08435" }, { "id": "ext_universal_transformer_2019", "title": "Universal Transformers", "priority": "external_literature", "layer": "shared_weight_recurrence_and_adaptive_depth", "chapter_targets": [ "governed-deliberation-and-test-time-scaling", "mathematical-and-search-substrates", "coil-attention-cyclic-memory-and-recurrence-contracts", "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/1807.03819", "notes": "Primary ICLR comparator for shared-weight depth recurrence, parallel self-attention, and per-position dynamic halting; benchmark results and theoretical expressivity do not establish stable deep recurrence, efficient scaling, local reproduction, or the book's cyclic-memory claims.", "source_type": "conference_paper", "arxiv_id": "1807.03819", "published": "2019", "updated": "2019-03-05", "citation_label": "Dehghani et al. (2019), Universal Transformers", "doi": "10.48550/arXiv.1807.03819" }, { "id": "ext_recurrent_transformer_2026", "title": "The Recurrent Transformer: Greater Effective Depth and Efficient Decoding", "priority": "external_literature", "layer": "layerwise_recurrent_transformer_memory", "chapter_targets": [ "fast-generation-architectures", "resource-economics-and-token-budgets", "mathematical-and-search-substrates", "coil-attention-cyclic-memory-and-recurrence-contracts" ], "url": "https://arxiv.org/abs/2604.21215", "notes": "Current preprint comparator for layerwise recurrent key-value memory, exact tiling, effective-depth/width tradeoffs, and standard autoregressive decoding cost; small-model C4 results do not establish broad capability gains, production efficiency, local reproduction, or cyclic-memory correctness.", "source_type": "preprint", "arxiv_id": "2604.21215", "published": "2026-04-23", "updated": "2026-04-23", "citation_label": "Oncescu et al. (2026), The Recurrent Transformer", "doi": "10.48550/arXiv.2604.21215" }, { "id": "ext_dynamic_compute_recurrent_transformers_2026", "title": "Understanding Dynamic Compute Allocation in Recurrent Transformers", "priority": "external_literature", "layer": "adaptive_recurrent_compute_evaluation", "chapter_targets": [ "governed-deliberation-and-test-time-scaling", "resource-economics-and-token-budgets", "coil-attention-cyclic-memory-and-recurrence-contracts", "benchmark-ratchets-and-anti-goodhart-evidence", "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2602.08864", "notes": "Current preprint comparator for complexity-controlled tests of token-level variable-depth compute and online halting; its negative result that difficulty-aligned compute need not generalize is a boundary against equating adaptive depth with algorithmic extrapolation or local capability.", "source_type": "preprint", "arxiv_id": "2602.08864", "published": "2026-02-09", "updated": "2026-02-09", "citation_label": "Moosa et al. (2026), Understanding Dynamic Compute Allocation in Recurrent Transformers", "doi": "10.48550/arXiv.2602.08864" }, { "id": "ext_claw_swe_bench_2026", "title": "Claw-SWE-Bench: A Benchmark for Evaluating OpenClaw-style Agent Harnesses on Coding Tasks", "priority": "external_literature", "layer": "coding_agent_harness_and_cost_evaluation", "chapter_targets": [ "artifact-graphs-audit-logs-and-replay", "runtime-adapters-tool-permissions-and-human-approval", "resource-economics-and-token-budgets", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://arxiv.org/abs/2606.12344", "notes": "Current preprint comparator for fixed workspace, patch, evaluator, and budget contracts across coding-agent harnesses; reported accuracy and cost differences are not reproduced here and do not validate the post-v2.1 synthetic repository corpus.", "source_type": "preprint", "arxiv_id": "2606.12344", "published": "2026-06-10", "updated": "2026-06-10", "citation_label": "Zheng et al. (2026), Claw-SWE-Bench", "doi": "10.48550/arXiv.2606.12344" }, { "id": "ext_txfs_2018", "title": "TxFS: Leveraging File-System Crash Consistency to Provide ACID Transactions", "priority": "external_literature", "layer": "transactional_filesystem_rollback_boundary", "chapter_targets": [ "capability-replacement-and-rollback", "artifact-graphs-audit-logs-and-replay", "runtime-adapters-tool-permissions-and-human-approval" ], "url": "https://www.usenix.org/conference/atc18/presentation/hu", "notes": "Primary systems comparator for user-space ACID file transactions built on journaling, including atomicity, isolation, durability, bounded transaction size, and Git/SQLite evaluation; it prevents treating a directory copy as a general transactional-filesystem result.", "source_type": "conference_paper", "published": "2018-07-11", "updated": "2018-07-11", "citation_label": "Hu et al. (2018), TxFS", "doi": "" }, { "id": "ext_dont_hallucinate_abstain_2024", "title": "Don't Hallucinate, Abstain: Identifying LLM Knowledge Gaps via Multi-LLM Collaboration", "priority": "external_literature", "layer": "llm_abstention_and_knowledge_gaps", "chapter_targets": [ "routing-heads-and-specialist-cores", "readiness-gates-residual-escrow-and-quarantine", "verification-bandwidth-and-context-adequacy" ], "url": "https://aclanthology.org/2024.acl-long.786/", "notes": "Primary ACL comparator for knowledge-gap detection, abstention, calibration/self-reflection limitations, and multi-model probing; reported abstention improvements are task- and model-bounded and do not validate the local router or evaluator.", "source_type": "conference_paper", "published": "2024-08", "updated": "2024-08", "citation_label": "Feng et al. (2024), Don't Hallucinate, Abstain", "doi": "10.18653/v1/2024.acl-long.786" }, { "id": "ext_muse_unlearning_2025", "title": "MUSE: Machine Unlearning Six-Way Evaluation for Language Models", "priority": "external_literature", "layer": "llm_unlearning_multidimensional_evaluation", "chapter_targets": [ "data-engines-continual-learning-and-unlearning", "policy-optimization-and-learning-from-feedback", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://proceedings.iclr.cc/paper_files/paper/2025/hash/4556f5398bd2c61bd7500e306b4e560a-Abstract-Conference.html", "notes": "Primary ICLR benchmark comparator separating verbatim and knowledge memorization, privacy leakage, retained utility, removal-scale behavior, and sequential sustainability; none of its 7B-language-model results are reproduced by the local policy network.", "source_type": "conference_paper", "arxiv_id": "2407.06460", "published": "2025", "updated": "2025", "citation_label": "Shi et al. (2025), MUSE", "doi": "10.48550/arXiv.2407.06460" }, { "id": "ext_unlearning_benchmarks_weak_2024", "title": "Position: LLM Unlearning Benchmarks are Weak Measures of Progress", "priority": "external_literature", "layer": "unlearning_benchmark_validity", "chapter_targets": [ "data-engines-continual-learning-and-unlearning", "benchmark-ratchets-and-anti-goodhart-evidence", "evidence-states-and-claim-discipline" ], "url": "https://arxiv.org/abs/2410.02879", "notes": "Primary critical comparator showing that benign benchmark modifications, forget/retain dependencies, and ambiguous targets can make unlearning scores optimistic; it strengthens the book's prohibition on turning toy behavioral change into influence, privacy, or storage claims.", "source_type": "position_paper", "arxiv_id": "2410.02879", "published": "2024-10-03", "updated": "2024-10-03", "citation_label": "Thaker et al. (2024), LLM Unlearning Benchmarks are Weak Measures", "doi": "10.48550/arXiv.2410.02879" }, { "id": "ext_openunlearning_2025", "title": "OpenUnlearning: Accelerating LLM Unlearning via Unified Benchmarking of Methods and Metrics", "priority": "external_literature", "layer": "unlearning_method_and_metric_meta_evaluation", "chapter_targets": [ "data-engines-continual-learning-and-unlearning", "benchmark-ratchets-and-anti-goodhart-evidence", "living-book-methodology" ], "url": "https://proceedings.neurips.cc/paper_files/paper/2025/hash/3e4a38f228427ab819ba7899003a44b1-Abstract-Datasets_and_Benchmarks_Track.html", "notes": "Primary NeurIPS 2025 benchmark-framework comparator for unified algorithms, diverse evaluations, public checkpoints, and meta-evaluation of metric faithfulness; it reinforces evaluator-quality residuals rather than establishing local unlearning.", "source_type": "conference_paper", "arxiv_id": "2506.12618", "published": "2025", "updated": "2025", "citation_label": "Dorna et al. (2025), OpenUnlearning", "doi": "10.48550/arXiv.2506.12618" }, { "id": "qcsa_whitepaper", "title": "Question-Compiled Semantic Addressing", "priority": "must_use", "layer": "semantic_addressing_control_plane", "chapter_targets": [ "cognitive-compilation-and-semantic-ir", "virtual-context-abi", "routing-heads-and-specialist-cores", "compact-generative-systems-and-residual-honesty", "runtime-adapters-tool-permissions-and-human-approval", "claim-ledgers-and-belief-revision", "data-engines-continual-learning-and-unlearning", "inter-stack-protocols-identity-and-economic-exchange", "integrated-reference-architecture", "governed-world-models-and-reality-grounding", "white-box-evidence-interpretability-and-activation-governance", "durable-semantic-memory-and-knowledge-lattices" ], "url": "local-source:sources/raw/question_compiled_semantic_addressing_whitepaper.md@sha256:d9e594d40dfd62c899ab25e9d395d34c702dac12e8afd75eed133392f78c0c8c", "notes": "Corben-authored successor synthesis for stable semantic identity, plural versioned semantic virtual addresses, active question compilation, evidence-bearing hypergraphs, semantic address certificates, semantic-to-physical routing, lifecycle-safe migration, and explicit residuals. The later repository adds a bounded local 12-lane implementation, 60-case held-out evaluation over 13 systems and three seeds, and one 13-stage governed vertical trace. The matched-advantage and resource gates failed, the active-question ablation is N2 proxy/regime evidence rather than an exact or broad refutation, and no natural-task, learned-model, production, independent, chapter-core promotion, AGI, or ASI result is established.", "source_type": "author_whitepaper", "published": "2026-07-12", "updated": "2026-07-12", "citation_label": "Sorenson (2026), Question-Compiled Semantic Addressing" }, { "id": "reflexive_router_whitepaper", "title": "The Reflexive Router: A Pre-Deliberative Architecture for Fast, Governed, Tool-Native Intelligence", "priority": "must_use", "layer": "pre_deliberative_reflexive_routing_control_plane", "chapter_targets": [ "routing-heads-and-specialist-cores", "intent-to-execution-contracts", "planning-as-a-control-layer", "stable-capability-fields", "virtual-context-abi", "context-transactions-snapshots-mounts-and-taint", "claim-ledgers-and-belief-revision", "runtime-adapters-tool-permissions-and-human-approval", "procedural-memory-and-cognitive-loop-closure", "resource-economics-and-token-budgets", "benchmark-ratchets-and-anti-goodhart-evidence", "integrated-reference-architecture", "white-box-evidence-interpretability-and-activation-governance" ], "url": "local-source:sources/raw/reflexive_router/the_reflexive_router_white_paper_v1_2.md@sha256:003a693741c40ca96ec3aece5b76ee90ec95a1d6c27ec81a970cff175f509068", "notes": "Corben-authored version 1.2 architecture proposal for a pre-deliberative command and routing plane, qualification-first dispatch, calibrated abstention, bounded execution DAGs, stable capability contracts, a non-bypassable effect commit kernel, typed result continuity, bitemporal Chronicle records, and governed trace-to-reflex compilation. It is assigned to existing chapter owners first; it adds no standalone chapter and supplies no implementation, benchmark, safety, deployment, transfer, novelty, AGI, ASI, or support-state result.", "source_type": "author_whitepaper", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "", "published": "2026-07-16", "updated": "2026-07-16", "citation_label": "Sorenson (2026), The Reflexive Router, v1.2" }, { "id": "ext_faithfulness_information_flow_2026", "title": "Faithfulness as Information Flow: Evaluating and Training Faithful Chain-of-Thought Reasoning", "priority": "external_literature", "layer": "reasoning_trace_faithfulness", "chapter_targets": [ "artifact-graphs-audit-logs-and-replay", "adversarial-evaluation-sandbagging-and-training-time-deception", "governed-deliberation-and-test-time-scaling", "policy-optimization-and-learning-from-feedback" ], "url": "https://arxiv.org/abs/2605.24286", "notes": "Primary 2026 comparator that separates chain-of-thought sufficiency, completeness, and interventional necessity, demonstrates prompt-to-answer shortcuts and transparent reward-hacking diagnostics, and documents low-entropy and reference-model limits. It does not make a reasoning transcript an authoritative receipt or establish local monitorability.", "source_type": "preprint", "arxiv_id": "2605.24286", "published": "2026-05-22", "updated": "2026-05-22", "citation_label": "Jia, Benton, and Easley (2026), Faithfulness as Information Flow", "doi": "10.48550/arXiv.2605.24286" }, { "id": "ext_monitorbench_2026", "title": "MonitorBench: A Comprehensive Benchmark for Chain-of-Thought Monitorability in Large Language Models", "priority": "external_literature", "layer": "reasoning_trace_monitorability_evaluation", "chapter_targets": [ "scalable-oversight-and-adversarial-ai-control", "adversarial-evaluation-sandbagging-and-training-time-deception" ], "url": "https://arxiv.org/abs/2603.28590", "notes": "Primary open benchmark comparator with 1,514 instances across 19 tasks and seven categories plus two stress-test settings; its reported capability/monitorability relation and up-to-30-percent degradation motivate held-out trace-action stress tests. The benchmark does not establish local monitoring quality, causal trace faithfulness, or safety.", "source_type": "preprint", "arxiv_id": "2603.28590", "published": "2026-03-30", "updated": "2026-04-02", "citation_label": "Wang et al. (2026), MonitorBench", "doi": "10.48550/arXiv.2603.28590" }, { "id": "ext_v_jepa_2_2025", "title": "V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning", "priority": "external_literature", "layer": "latent_world_models_and_model_predictive_control", "chapter_targets": [ "mathematical-and-search-substrates", "planning-as-a-control-layer", "data-engines-continual-learning-and-unlearning", "governed-world-models-and-reality-grounding", "integrated-reference-architecture" ], "url": "https://arxiv.org/abs/2506.09985", "notes": "Primary empirical comparator for action-free latent video pretraining, a small action-conditioned predictor, and model-predictive control. Camera sensitivity, autoregressive error accumulation, action-search cost, image-goal assumptions, and representation-bounded capability remain explicit limits; no local world model or robot-control result is established.", "source_type": "preprint", "arxiv_id": "2506.09985", "published": "2025-06-11", "updated": "2026-04-26", "citation_label": "Assran et al. (2025), V-JEPA 2", "doi": "10.48550/arXiv.2506.09985" }, { "id": "ext_embedded_agency_2019", "title": "Embedded Agency", "priority": "external_literature", "layer": "embedded_agency_foundations", "chapter_targets": [ "asi-is-a-stack-not-a-model", "constitutional-alignment-substrate", "recursive-self-improvement-boundaries", "evidence-states-and-claim-discipline", "integrated-reference-architecture" ], "url": "https://arxiv.org/abs/1902.09469", "notes": "Primary informal survey of the obstacles that arise when agents are physical parts of the worlds they model, must use smaller internal models, and reason about modifiable internal parts. It supplies a foundations boundary; the book's finite records, authority ceilings, and proofs do not solve embedded agency.", "source_type": "preprint", "arxiv_id": "1902.09469", "published": "2019-02-25", "updated": "2019-02-25", "citation_label": "Demski and Garrabrant (2019), Embedded Agency", "doi": "10.48550/arXiv.1902.09469" }, { "id": "ext_ietf_rats_architecture_2023", "title": "Remote ATtestation procedureS (RATS) Architecture", "priority": "external_literature", "layer": "remote_attestation_architecture", "chapter_targets": [ "model-weight-custody-and-hardware-roots-of-trust" ], "url": "https://www.rfc-editor.org/rfc/rfc9334.html", "notes": "Primary IETF architecture and terminology comparator for Attester, Verifier, Relying Party, Evidence, Attestation Results, appraisal policies, reference values, freshness, layered environments, privacy, trust roots, and confidential-model key release. It is informational architecture, not a protocol, hardware assurance level, verifier-independence result, or local attestation deployment.", "source_type": "informational_rfc", "published": "2023-01", "updated": "2023-01", "citation_label": "Birkholz et al. (2023), RATS Architecture", "doi": "10.17487/RFC9334" }, { "id": "ext_nist_key_management_2020", "title": "Recommendation for Key Management: Part 1 \u2013 General", "priority": "external_literature", "layer": "cryptographic_key_management", "chapter_targets": [ "model-weight-custody-and-hardware-roots-of-trust" ], "url": "https://csrc.nist.gov/pubs/sp/800/57/pt1/r5/final", "notes": "Current final NIST key-management baseline for key and metadata protection, inventory, authorization, access control, usage periods, compromise, backup, recovery, trust anchors, and lifecycle policy. A Revision 6 draft exists, so this source is the final baseline rather than a claim that guidance has stopped evolving; no local key-management conformance or security result is established.", "source_type": "standard", "published": "2020-05-04", "updated": "2025-04-25", "citation_label": "Barker (2020), NIST SP 800-57 Part 1 Rev. 5", "doi": "10.6028/NIST.SP.800-57pt1r5" }, { "id": "ext_nist_media_sanitization_2025", "title": "Guidelines for Media Sanitization", "priority": "external_literature", "layer": "media_sanitization_and_disposal", "chapter_targets": [ "model-weight-custody-and-hardware-roots-of-trust" ], "url": "https://csrc.nist.gov/pubs/sp/800/88/r2/final", "notes": "Current final NIST media-sanitization comparator for rendering target data access infeasible at a stated effort level using sensitivity- and media-appropriate controls, including cryptographic erase. It does not prove that all model copies, plaintext memory, cloud replicas, derivatives, or recipients were discovered or sanitized, and no local erasure test was run.", "source_type": "standard", "published": "2025-09-26", "updated": "2025-09-26", "citation_label": "Chandramouli and Hibbard (2025), NIST SP 800-88 Rev. 2", "doi": "10.6028/NIST.SP.800-88r2" }, { "id": "corben_chatgpt_kiss_irreducible_intelligence_2026", "title": "KISS versus Irreducible Intelligence (author-supplied design conversation)", "priority": "supporting", "layer": "author_intent_replaceable_cognitive_substrates", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture", "relational-dimension-compilation-and-polyadic-cognition" ], "url": "https://chatgpt.com/share/6a57604d-59b4-83ea-8da7-b9d3d52658f8", "notes": "Corben-supplied design conversation for search-verify-compile, exact-latent separation, a compact recursive kernel, and total-system KISS accounting. Author intent only; not independent evidence or a reproduced architecture.", "source_type": "author_supplied_conversation", "published": "2026-07-15", "citation_label": "Sorenson and ChatGPT (2026), KISS versus Irreducible Intelligence" }, { "id": "corben_chatgpt_onecell_theseus_2026", "title": "OneCell and Theseus Architecture Handoff (author-supplied design conversation)", "priority": "supporting", "layer": "author_intent_onecell_and_architectural_rsi", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://chatgpt.com/share/6a57781e-17a4-83ea-aaae-fa263a9ac2fd", "notes": "Corben-supplied design conversation for a Cognitive Kernel ABI, typed state lanes, inner recurrence, outer exact search, verified abstraction, and Theseus-governed architecture tournaments. Author intent only; OneCell remains an unimplemented falsifiable candidate.", "source_type": "author_supplied_conversation", "published": "2026-07-15", "citation_label": "Sorenson and ChatGPT (2026), OneCell and Theseus Architecture Handoff" }, { "id": "ext_attention_is_all_you_need_2017", "title": "Attention Is All You Need", "priority": "external_literature", "layer": "dense_attention_sequence_substrate", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/1706.03762", "notes": "Primary Transformer paper and dense-attention baseline. It supports the historical architecture and parallel sequence-processing comparison, not a claim that Transformers are universally optimal or locally reproduced.", "source_type": "conference_paper", "arxiv_id": "1706.03762", "published": "2017-06-12", "citation_label": "Vaswani et al. (2017), Attention Is All You Need", "doi": "10.48550/arXiv.1706.03762" }, { "id": "ext_mamba2_ssd_2024", "title": "Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality", "priority": "external_literature", "layer": "state_space_duality_and_sequence_substrates", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2405.21060", "notes": "Primary Mamba-2/structured-state-space-duality comparator connecting SSM and attention-like formulations. No local model, kernel, quality, scaling, or hardware result is reproduced.", "source_type": "conference_paper", "arxiv_id": "2405.21060", "published": "2024-05-31", "citation_label": "Dao and Gu (2024), Transformers are SSMs", "doi": "10.48550/arXiv.2405.21060" }, { "id": "ext_s4_2022", "title": "Efficiently Modeling Long Sequences with Structured State Spaces", "priority": "external_literature", "layer": "structured_state_space_sequence_substrates", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2111.00396", "notes": "Foundational S4 comparator for structured state-space sequence models, efficient long-range computation, and the lineage that precedes selective SSMs. Source-reported benchmark and generation results are not reproduced and do not establish exact recall or governed substitutability.", "source_type": "conference_paper", "arxiv_id": "2111.00396", "published": "2021-10-31", "citation_label": "Gu, Goel, and Re (2022), Structured State Spaces", "doi": "10.48550/arXiv.2111.00396" }, { "id": "ext_mamba3_2026", "title": "Mamba-3: Improved Sequence Modeling using State Space Principles", "priority": "external_literature", "layer": "modern_selective_state_space_sequence_substrates", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2603.15569", "notes": "Current 2026 selective-SSM comparator for complex-valued state updates, discretization, and multi-input/multi-output formulation. Recent source-reported results are not locally reproduced and must not set the chapter conclusion by recency.", "source_type": "preprint", "arxiv_id": "2603.15569", "published": "2026-03-16", "citation_label": "Lahoti et al. (2026), Mamba-3", "doi": "10.48550/arXiv.2603.15569" }, { "id": "ext_gated_deltanet2_2026", "title": "Gated DeltaNet-2: Decoupling Erase and Write in Linear Attention", "priority": "external_literature", "layer": "current_recurrent_linear_attention_and_editable_memory_frontier", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2605.22791", "notes": "Dated 2026 comparator that separates erase and write gates and reports the strongest aggregate result among its Mamba-2, Gated DeltaNet, KDA, Mamba-3, and Gated DeltaNet-2 envelope at 1.3B parameters and 100B FineWeb-Edu tokens. The result is author reported, not locally reproduced; it displaces Mamba-3 only for that exact source envelope and requires official-code, checkpoint, hardware, seed, cost, retrieval, state, and transfer reproduction before any local superiority claim.", "source_type": "preprint", "arxiv_id": "2605.22791", "published": "2026-05-21", "citation_label": "Hatamizadeh, Choi, and Kautz (2026), Gated DeltaNet-2", "doi": "10.48550/arXiv.2605.22791" }, { "id": "ext_hyperscale_lottery_2026", "title": "The Hyperscale Lottery: How State-Space Models Have Sacrificed Edge Efficiency", "priority": "external_literature", "layer": "hardware_specific_state_space_efficiency_counterevidence", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2604.07935", "notes": "Current edge-hardware counterstudy measuring Mamba-family latency outside hyperscale-GPU conditions. Its source-reported results require platform-stratified latency, memory, and energy accounting; they are not locally reproduced and do not settle the quality-efficiency frontier.", "source_type": "preprint", "arxiv_id": "2604.07935", "published": "2026-04-09", "citation_label": "Geens et al. (2026), The Hyperscale Lottery", "doi": "10.48550/arXiv.2604.07935" }, { "id": "ext_gated_deltanet_2024", "title": "Gated Delta Networks: Improving Mamba2 with Delta Rule", "priority": "external_literature", "layer": "linear_attention_and_adaptive_memory", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2412.06464", "notes": "Primary gated-delta-rule comparator for targeted recurrent-memory updates, rapid erasure, parallel training, and hybrid attention/SSM compositions. Source-reported retrieval, extrapolation, efficiency, and quality results are not locally reproduced.", "source_type": "preprint", "arxiv_id": "2412.06464", "published": "2024-12-09", "citation_label": "Yang et al. (2024), Gated Delta Networks", "doi": "10.48550/arXiv.2412.06464" }, { "id": "ext_jamba_2024", "title": "Jamba: A Hybrid Transformer-Mamba Language Model", "priority": "external_literature", "layer": "hybrid_attention_state_space_mixture_of_experts", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2403.19887", "notes": "Primary large-scale hybrid Transformer-Mamba-MoE comparator. It motivates route- and composition-aware accounting; its reported quality, context, throughput, and memory results are not locally reproduced and do not establish that the specific mixture is generally optimal.", "source_type": "preprint", "arxiv_id": "2403.19887", "published": "2024-03-28", "citation_label": "Lieber et al. (2024), Jamba", "doi": "10.48550/arXiv.2403.19887" }, { "id": "ext_neural_message_passing_2017", "title": "Neural Message Passing for Quantum Chemistry", "priority": "external_literature", "layer": "graph_relational_message_passing", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/1704.01212", "notes": "Primary message-passing neural-network framework for learned computation over graph structure. Its molecular results motivate a non-token-native relational lane but do not establish general reasoning, dynamic graph memory, exact state, or local reproduction.", "source_type": "conference_paper", "arxiv_id": "1704.01212", "published": "2017-04-04", "citation_label": "Gilmer et al. (2017), Neural Message Passing for Quantum Chemistry", "doi": "10.48550/arXiv.1704.01212" }, { "id": "ext_hyena_hierarchy_2023", "title": "Hyena Hierarchy: Towards Larger Convolutional Language Models", "priority": "external_literature", "layer": "long_convolution_sequence_substrates", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2302.10866", "notes": "Primary long-convolution comparator for subquadratic sequence mixing and hardware-aware architecture comparisons. No local training, throughput, quality, recall, or scaling result is reproduced.", "source_type": "conference_paper", "arxiv_id": "2302.10866", "published": "2023-02-21", "citation_label": "Poli et al. (2023), Hyena Hierarchy", "doi": "10.48550/arXiv.2302.10866" }, { "id": "ext_rwkv_2023", "title": "RWKV: Reinventing RNNs for the Transformer Era", "priority": "external_literature", "layer": "linear_recurrent_language_models", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://aclanthology.org/2023.findings-emnlp.936/", "notes": "Primary recurrent language-model comparator combining parallelizable training with recurrent inference. Reported benchmark, memory, and inference properties are not reproduced locally.", "source_type": "conference_paper", "published": "2023-12-01", "citation_label": "Peng et al. (2023), RWKV", "doi": "10.18653/v1/2023.findings-emnlp.936" }, { "id": "ext_xlstm_2024", "title": "xLSTM: Extended Long Short-Term Memory", "priority": "external_literature", "layer": "modern_gated_recurrent_sequence_substrates", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2405.04517", "notes": "Primary modern-LSTM comparator for revised gating, memory, and scalable recurrent language modeling. No xLSTM training, scaling, quality, or inference result is reproduced locally.", "source_type": "conference_paper", "arxiv_id": "2405.04517", "published": "2024-05-07", "citation_label": "Beck et al. (2024), xLSTM", "doi": "10.48550/arXiv.2405.04517" }, { "id": "ext_ttt_layers_2024", "title": "Learning to (Learn at Test Time): RNNs with Expressive Hidden States", "priority": "external_literature", "layer": "test_time_learned_state_sequence_substrates", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2407.04620", "notes": "Primary test-time-training-layer comparator that treats hidden state as a learned model updated on the sequence. It motivates explicit online-state custody and rollback; no local quality or efficiency result is reproduced.", "source_type": "conference_paper", "arxiv_id": "2407.04620", "published": "2024-07-05", "citation_label": "Sun et al. (2024), Learning to Learn at Test Time", "doi": "10.48550/arXiv.2407.04620" }, { "id": "ext_titans_2025", "title": "Titans: Learning to Memorize at Test Time", "priority": "external_literature", "layer": "test_time_neural_long_term_memory", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2501.00663", "notes": "Primary neural-memory comparator for test-time memorization and long-context sequence modeling. The paper motivates mutable-state provenance and rollback tests; no local model or benchmark result is reproduced.", "source_type": "preprint", "arxiv_id": "2501.00663", "published": "2025-01-01", "citation_label": "Behrouz et al. (2025), Titans", "doi": "10.48550/arXiv.2501.00663" }, { "id": "ext_kan_2024", "title": "KAN: Kolmogorov-Arnold Networks", "priority": "external_literature", "layer": "learned_univariate_function_networks", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2404.19756", "notes": "Primary KAN proposal replacing fixed node activations/linear edge weights with learned univariate edge functions. Interpretability and scientific-task demonstrations are source-reported and do not establish a general MLP or Transformer replacement.", "source_type": "conference_paper", "arxiv_id": "2404.19756", "published": "2024-04-30", "citation_label": "Liu et al. (2024), KAN", "doi": "10.48550/arXiv.2404.19756" }, { "id": "ext_kan_or_mlp_fairer_comparison_2024", "title": "KAN or MLP: A Fairer Comparison", "priority": "external_literature", "layer": "architecture_comparison_methodology", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2407.16674", "notes": "Critical matched-comparison source for KAN versus MLP under parameter, FLOP, and task controls. It is included to prevent architecture enthusiasm from substituting for fair accounting; no local comparison is reproduced.", "source_type": "preprint", "arxiv_id": "2407.16674", "published": "2024-07-23", "citation_label": "Yu et al. (2024), KAN or MLP", "doi": "10.48550/arXiv.2407.16674" }, { "id": "ext_neural_turing_machines_2014", "title": "Neural Turing Machines", "priority": "external_literature", "layer": "differentiable_external_memory", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/1410.5401", "notes": "Primary differentiable-controller/external-memory source. It motivates variable-size memory interfaces and out-of-distribution algorithmic tests; toy-task results do not establish reliable exact memory or general computation.", "source_type": "preprint", "arxiv_id": "1410.5401", "published": "2014-10-20", "citation_label": "Graves, Wayne, and Danihelka (2014), Neural Turing Machines", "doi": "10.48550/arXiv.1410.5401" }, { "id": "ext_differentiable_neural_computer_2016", "title": "Hybrid computing using a neural network with dynamic external memory", "priority": "external_literature", "layer": "differentiable_external_memory", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://doi.org/10.1038/nature20101", "notes": "Primary Differentiable Neural Computer source for learned controllers over dynamic external memory. Source-reported graph and reasoning tasks do not establish reliable exact state, scalable memory, or local reproduction.", "source_type": "journal_article", "published": "2016-10-12", "citation_label": "Graves et al. (2016), Differentiable Neural Computer", "doi": "10.1038/nature20101" }, { "id": "ext_liquid_time_constant_networks_2021", "title": "Liquid Time-constant Networks", "priority": "external_literature", "layer": "continuous_time_neural_dynamics", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://doi.org/10.1609/aaai.v35i9.16936", "notes": "Primary continuous-time recurrent architecture source for input-dependent time constants and dynamical-system behavior. Reported time-series results and stability analysis do not establish broad cognitive superiority or a local implementation.", "source_type": "conference_paper", "published": "2021-05-18", "citation_label": "Hasani et al. (2021), Liquid Time-constant Networks", "doi": "10.1609/aaai.v35i9.16936" }, { "id": "ext_tiny_recursive_model_2025", "title": "Less is More: Recursive Reasoning with Tiny Networks", "priority": "external_literature", "layer": "tiny_weight_tied_recursive_reasoning", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2510.04871", "notes": "Primary Tiny Recursive Model proposal and narrow puzzle-domain result. It motivates a compact weight-tied OneCell comparator but does not establish general reasoning, language capability, deep effective recursion, or total-system simplicity.", "source_type": "preprint", "arxiv_id": "2510.04871", "published": "2025-10-06", "citation_label": "Jolicoeur-Martineau (2025), Less is More", "doi": "10.48550/arXiv.2510.04871" }, { "id": "ext_trm_arc_agi_analysis_2025", "title": "Tiny Recursive Models on ARC-AGI-1: Inductive Biases, Identity Conditioning, and Test-Time Compute", "priority": "external_literature", "layer": "recursive_model_critical_evaluation", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2512.11847", "notes": "Critical TRM analysis reporting material dependence on 1000-sample voting, puzzle identity, and shallow effective recursion. It is a source-reported audit rather than a local reproduction and sets preregistered identity, sampling, and recursion-depth controls.", "source_type": "preprint", "arxiv_id": "2512.11847", "published": "2025-12-04", "citation_label": "Roye-Azar et al. (2025), Tiny Recursive Models on ARC-AGI-1", "doi": "10.48550/arXiv.2512.11847" }, { "id": "ext_tiny_autoregressive_recursive_models_2026", "title": "Tiny Autoregressive Recursive Models", "priority": "external_literature", "layer": "recursive_model_controlled_ablation", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2603.08082", "notes": "Controlled compute-matched study that progressively transforms a standard autoregressive model into a TRM-like system and reports no reliable advantage from the full autoregressive TRM mechanism. It motivates mechanism-level rather than label-level ablation.", "source_type": "workshop_paper", "arxiv_id": "2603.08082", "published": "2026-03-09", "citation_label": "Rauba, Fanconi, and van der Schaar (2026), Tiny Autoregressive Recursive Models", "doi": "10.48550/arXiv.2603.08082" }, { "id": "ext_unimatrix_2026", "title": "Associative-State Universal Transformers: Sparse Retrieval Meets Structured Recurrence", "priority": "external_literature", "layer": "structured_recurrence_and_sparse_retrieval", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2604.25930", "notes": "Current UniMatrix preprint whose negative associative-recall result shows compressed recurrent state alone is insufficient in its setup, while explicit sparse slots and pointer-level routing materially change the result. No local reproduction or general conclusion follows.", "source_type": "preprint", "arxiv_id": "2604.25930", "published": "2026-04-01", "citation_label": "Xiao (2026), Associative-State Universal Transformers", "doi": "10.48550/arXiv.2604.25930" }, { "id": "ext_memory_caching_2026", "title": "Memory Caching: RNNs with Growing Memory", "priority": "external_literature", "layer": "growing_recurrent_memory", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2602.24281", "notes": "Current recurrent-memory comparator that caches hidden-state checkpoints and exposes a trade between fixed recurrent memory and growing addressable memory. Its source-reported recall results still leave the Transformer strongest on the reported in-context recall tasks.", "source_type": "preprint", "arxiv_id": "2602.24281", "published": "2026-02-27", "citation_label": "Behrouz et al. (2026), Memory Caching", "doi": "10.48550/arXiv.2602.24281" }, { "id": "ext_inkling_2026", "title": "Inkling: Our open-weights model", "priority": "external_literature", "layer": "hybrid_local_global_attention_moe_multimodal_substrate", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://thinkingmachines.ai/news/introducing-inkling/", "notes": "Release-day primary-source case study of a 66-layer multimodal sparse-MoE Transformer with a 5:1 local/global attention schedule, relative positions, short convolutions, controllable effort, and open weights. It motivates topology-complete capability cards and component ablations; provider-reported results are not locally reproduced and do not isolate the contribution of any component.", "source_type": "official_model_release", "published": "2026-07-15", "citation_label": "Thinking Machines Lab (2026), Inkling", "doi": null }, { "id": "kernel_english_residual_compiler", "title": "Kernel English with Hierarchical, Interaction-Amortized Residuals: A Dual-Vocabulary Cognitive Compiler for Efficient Language-Model Reasoning", "priority": "must_use", "layer": "canonical_cognitive_compilation_and_hierarchical_residual_runtime", "chapter_targets": [ "cognitive-compilation-and-semantic-ir", "compact-generative-systems-and-residual-honesty", "virtual-context-abi", "context-transactions-snapshots-mounts-and-taint", "verification-bandwidth-and-context-adequacy", "fast-generation-architectures", "replaceable-cognitive-substrates-beyond-transformer-monoculture", "resource-economics-and-token-budgets", "security-kernel-and-digital-scifs", "procedural-memory-and-cognitive-loop-closure", "benchmark-ratchets-and-anti-goodhart-evidence", "integrated-reference-architecture", "white-box-evidence-interpretability-and-activation-governance" ], "url": "local-source:sources/raw/kernel_english_residual_compiler/kernel_english_hierarchical_residuals.md@sha256:f560c61196cb2a114475ebd455f8643536e78c82dbbf6ec8dd712d993f2b6519", "notes": "Corben-authored July 2026 architecture proposal for KERC: protected-object capture, uncertainty-aware normalization, sense-aware Kernel IR, dual surface/core vocabularies, a four-level interaction-amortized residual ledger, exact object storage, grammar-aware macro fusion, structured answer packets, rendering, round-trip verification, versioned migration, and complete rate-compute-fidelity evaluation. Existing chapters are upgraded first; no implementation, benchmark, novelty, efficiency, fidelity, safety, transfer, SOTA, AGI, ASI, or support-state result is inferred.", "source_type": "author_whitepaper", "published": "2026-07-16", "updated": "2026-07-16", "citation_label": "Sorenson (2026), Kernel English with Hierarchical, Interaction-Amortized Residuals" }, { "id": "deterministic_capability_compilation", "title": "Deterministic Capability Compilation: A Capability-Preserving Ladder from Executable Scaffolds to Governed Adaptive Agents", "priority": "must_use", "layer": "capability_compilation_neural_linking_and_governed_adaptation", "chapter_targets": [ "stable-capability-fields", "intent-to-execution-contracts", "cognitive-compilation-and-semantic-ir", "virtual-context-abi", "capability-replacement-and-rollback", "routing-heads-and-specialist-cores", "runtime-adapters-tool-permissions-and-human-approval", "spinoza-verification-and-proof-carrying-claims", "labor-os-and-typed-jobs", "artifact-graphs-audit-logs-and-replay", "procedural-memory-and-cognitive-loop-closure", "readiness-gates-residual-escrow-and-quarantine", "compact-generative-systems-and-residual-honesty", "replaceable-cognitive-substrates-beyond-transformer-monoculture", "ai-supply-chain-integrity-and-lifecycle-provenance", "recursive-self-improvement-boundaries", "benchmark-ratchets-and-anti-goodhart-evidence", "data-engines-continual-learning-and-unlearning", "integrated-reference-architecture", "project-theseus-as-report-first-implementation-reference", "prototype-roadmap", "open-research-agenda-and-bibliography-plan", "white-box-evidence-interpretability-and-activation-governance", "governed-world-models-and-reality-grounding", "governed-operations-incident-command-and-graceful-degradation", "adversarial-machine-learning-and-model-attack-surface", "autonomous-replication-proliferation-and-containment" ], "url": "local-source:sources/raw/corben_papers/deterministic_capability_compilation/deterministic_capability_compilation.md@sha256:8306d2bd40662245f8d3f044b4ca11a61c0fb18ab7cb799ccfe64f5e8c8a80c1", "notes": "Corben-authored July 2026 architecture and research program for compiling executable scaffolds into contract-bound experts and linked Neural Capability Objects while retaining semantic obligation mass balance, candidate-specific translation validation, fallback, residual escrow, authority ceilings, reification, and effect-complete recovery. Existing chapters are upgraded first; no foundry implementation, learned-capability result, preservation result, safety result, SOTA result, AGI, ASI, or support-state promotion is inferred.", "source_type": "author_whitepaper", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "", "published": "2026-07", "updated": "2026-07", "citation_label": "Sorenson (2026), Deterministic Capability Compilation" }, { "id": "platonic_world_model", "title": "The Platonic World Model: A Semantic Constitution for Grounded, Proof-Carrying, Self-Editing Artificial Intelligence", "priority": "must_use", "layer": "semantic_continuity_grounded_world_model_and_governed_self_editing", "chapter_targets": [ "moral-uncertainty-and-value-conflict", "security-kernel-and-digital-scifs", "ai-supply-chain-integrity-and-lifecycle-provenance", "claim-ledgers-and-belief-revision", "spinoza-verification-and-proof-carrying-claims", "artifact-graphs-audit-logs-and-replay", "virtual-context-abi", "context-transactions-snapshots-mounts-and-taint", "planning-as-a-control-layer", "cognitive-compilation-and-semantic-ir", "runtime-adapters-tool-permissions-and-human-approval", "inter-stack-protocols-identity-and-economic-exchange", "procedural-memory-and-cognitive-loop-closure", "data-engines-continual-learning-and-unlearning", "recursive-self-improvement-boundaries", "benchmark-ratchets-and-anti-goodhart-evidence", "integrated-reference-architecture", "prototype-roadmap", "open-research-agenda-and-bibliography-plan", "white-box-evidence-interpretability-and-activation-governance", "governed-world-models-and-reality-grounding", "governed-operations-incident-command-and-graceful-degradation" ], "url": "local-source:sources/raw/corben_papers/platonic_world_model/platonic_world_model.md@sha256:8ae760b6b2fd814b4c75260f146bf53789481b9c65fa5aded67a8934bb32427b", "notes": "Corben-authored July 2026 conceptual architecture and falsifiable research program for semantic continuity through stable Form lineages, immutable semantic versions, typed Essence Contracts, six mutually constraining planes, explicit proposition-attestation-commitment-proof separation, branch-protected world dynamics, qualified grounding, semantic transactions, runtime packet compilation, and federated mappings. Existing chapters are upgraded first; no implemented substrate, benchmark result, philosophical solution to grounding, safety result, SOTA result, AGI, ASI, or support-state promotion is inferred.", "source_type": "author_whitepaper", "published": "2026-07", "updated": "2026-07", "citation_label": "Sorenson (2026), The Platonic World Model" }, { "id": "relational_dimension_compiler", "title": "The Relational Dimension Compiler: Adaptive Polyadic Cognition with Bounded Computational Arity and Unbounded Semantic Structure", "priority": "must_use", "layer": "typed_relational_ir_adaptive_polyadic_routing_and_reversible_abstraction", "chapter_targets": [ "cognitive-compilation-and-semantic-ir", "governed-world-models-and-reality-grounding", "routing-heads-and-specialist-cores", "replaceable-cognitive-substrates-beyond-transformer-monoculture", "procedural-memory-and-cognitive-loop-closure", "benchmark-ratchets-and-anti-goodhart-evidence", "resource-economics-and-token-budgets", "integrated-reference-architecture", "open-research-agenda-and-bibliography-plan" ], "url": "local-source:sources/raw/corben_papers/relational_dimension_compiler/relational_dimension_compiler.md@sha256:85bcb1865da74ffe57d04c763557993799065ec975c2b0eca5e093c23e9eaa5a", "notes": "Corben-authored July 2026 conceptual architecture and falsifiable research program for a typed relational intermediate representation that separates geometric dimension, semantic arity, primitive computational arity, storage arity, temporal extent, abstraction scale, branch identity, epistemic status, and resource budget. It proposes sparse adaptive relational-order routing, exact role-preserving relation reification, qualified relation lifecycles, branch-local object-field state, reversible semantic contraction, compiled slow-to-fast relation programs, hardware lowering, and the RODIE benchmark suite. Existing chapters receive bounded integration while a distinct future chapter candidate remains deferred by the active manifest freeze. No RDC implementation, benchmark result, universal arity bound, ontology truth, efficiency, safety, SOTA, AGI, ASI, or support-state promotion is inferred.", "source_type": "author_whitepaper", "published": "2026-07", "updated": "2026-07-22", "citation_label": "Sorenson (2026), The Relational Dimension Compiler" }, { "id": "ext_megatron_distributed_training_2021", "title": "Efficient Large-Scale Language Model Training on GPU Clusters Using Megatron-LM", "priority": "external_literature", "layer": "governed_distributed_model_training_and_scaling", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://arxiv.org/abs/2104.04473", "notes": "Primary composed-parallelism mechanism source. It grounds tensor, pipeline, and data parallel interactions, strict optimizer semantics, microbatch and topology tradeoffs. Reported trillion-parameter and throughput results are configuration-bound and not locally reproduced.", "source_type": "conference_paper", "arxiv_id": "2104.04473", "published": "2021-04-09", "updated": "2021-08-30", "citation_label": "Narayanan et al. (2021), Efficient Large-Scale Language Model Training", "doi": "10.1145/3458817.3476209" }, { "id": "ext_zero_optimizer_2019", "title": "ZeRO: Memory Optimizations Toward Training Trillion Parameter Models", "priority": "external_literature", "layer": "governed_distributed_model_training_and_scaling", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://arxiv.org/abs/1910.02054", "notes": "Primary competing sharded-state design for optimizer, gradient, parameter, activation, and residual memory. It motivates explicit state closure and shard reconstruction; source-reported scale and speed are not locally reproduced or treated as universal superiority.", "source_type": "conference_paper", "arxiv_id": "1910.02054", "published": "2019-10-04", "updated": "2020-05-13", "citation_label": "Rajbhandari et al. (2020), ZeRO", "doi": "10.1109/SC41405.2020.00024" }, { "id": "ext_gspmd_2021", "title": "GSPMD: General and Scalable Parallelization for ML Computation Graphs", "priority": "external_literature", "layer": "governed_distributed_model_training_and_scaling", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://arxiv.org/abs/2105.04663", "notes": "Primary compiler-mediated competing design for general SPMD sharding and mixed parallelism. It motivates versioning inferred plans and inserted collectives; reported TPU utilization and scaling are not locally reproduced.", "source_type": "preprint", "arxiv_id": "2105.04663", "published": "2021-05-10", "updated": "2021-12-14", "citation_label": "Xu et al. (2021), GSPMD", "doi": "10.48550/arXiv.2105.04663" }, { "id": "ext_datastates_llm_2024", "title": "DataStates-LLM: Lazy Asynchronous Checkpointing for Large Language Models", "priority": "external_literature", "layer": "governed_distributed_model_training_and_scaling", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://arxiv.org/abs/2406.10707", "notes": "Primary limitation and checkpoint-mechanism source for asynchronous multi-level copies, distributed shard consistency, and checkpoint overhead. It does not establish complete application state or exact trajectory-equivalent resume, and no result is locally reproduced.", "source_type": "conference_paper", "arxiv_id": "2406.10707", "published": "2024-06-15", "updated": "2024-06-15", "citation_label": "Maurya et al. (2024), DataStates-LLM", "doi": "10.1145/3625549.3658685" }, { "id": "ext_pytorch_distributed_checkpoint_2026", "title": "Distributed Checkpoint \u2014 PyTorch documentation", "priority": "external_literature", "layer": "governed_distributed_model_training_and_scaling", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://docs.pytorch.org/docs/stable/distributed.checkpoint.html", "notes": "Official current implementation documentation for SPMD save/load, asynchronous completion, canonical model and optimizer state, resharding, strict load, and call-order constraints. Documentation is not benchmark or full-state resume evidence.", "source_type": "official_docs", "published": "2026-07-19", "updated": "2026-07-19", "citation_label": "PyTorch (2026), Distributed Checkpoint" }, { "id": "ext_mlperf_training_v6_2026", "title": "MLPerf Training v6.0", "priority": "external_literature", "layer": "governed_distributed_model_training_and_scaling", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling", "benchmark-ratchets-and-anti-goodhart-evidence", "resource-economics-and-token-budgets" ], "url": "https://mlcommons.org/benchmarks/training/", "notes": "Official current measurement comparator for fixed datasets and quality targets, repeated time-to-quality, system metadata, divisions, variance, and corrected results. No MLPerf run is performed and the benchmark does not establish safety or complete run integrity.", "source_type": "official_benchmark", "published": "2026-06-16", "updated": "2026-07-19", "citation_label": "MLCommons (2026), MLPerf Training v6.0" }, { "id": "ext_adam_2015", "title": "Adam: A Method for Stochastic Optimization", "priority": "external_literature", "layer": "optimizer_mechanisms_and_selection", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://arxiv.org/abs/1412.6980", "notes": "Primary Adam mechanism source for bias-corrected first- and second-moment estimates and coordinate-wise adaptive updates. Its online-convex analysis and reported experiments do not establish universal convergence, quality, or optimizer superiority in foundation-model training.", "source_type": "conference_paper", "arxiv_id": "1412.6980", "published": "2014-12-22", "citation_label": "Kingma and Ba (2015), Adam", "doi": "10.48550/arXiv.1412.6980" }, { "id": "ext_amsgrad_2018", "title": "On the Convergence of Adam and Beyond", "priority": "external_literature", "layer": "optimizer_failure_and_convergence", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://openreview.net/forum?id=ryQu7f-RZ", "notes": "Primary Adam failure and AMSGrad source. It gives constructed stochastic-convex non-convergence cases and a maximum-second-moment remedy; those cases do not imply every practical Adam run fails or that AMSGrad is universally preferable.", "source_type": "conference_paper", "published": "2018-02-15", "citation_label": "Reddi, Kale, and Kumar (2018), On the Convergence of Adam and Beyond", "doi": null }, { "id": "ext_adamw_2019", "title": "Decoupled Weight Decay Regularization", "priority": "external_literature", "layer": "optimizer_mechanisms_and_selection", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://openreview.net/forum?id=Bkg6RiCqY7", "notes": "Primary AdamW source separating weight decay from the adaptive gradient update. Its reported tuning and generalization results are setting-bound; the optimizer name alone does not specify parameter exclusions, schedule, decay scaling, or implementation semantics.", "source_type": "conference_paper", "arxiv_id": "1711.05101", "published": "2017-11-14", "citation_label": "Loshchilov and Hutter (2019), Decoupled Weight Decay Regularization", "doi": "10.48550/arXiv.1711.05101" }, { "id": "ext_adafactor_2018", "title": "Adafactor: Adaptive Learning Rates with Sublinear Memory Cost", "priority": "external_literature", "layer": "optimizer_memory_and_scaling", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://proceedings.mlr.press/v80/shazeer18a.html", "notes": "Primary factored-second-moment optimizer source. It reduces auxiliary state for matrix parameters and adds update clipping and parameter-scale rules; factorization remains an approximation and its reported translation result does not establish universal parity with Adam.", "source_type": "conference_paper", "arxiv_id": "1804.04235", "published": "2018-07-10", "citation_label": "Shazeer and Stern (2018), Adafactor", "doi": "10.48550/arXiv.1804.04235" }, { "id": "ext_lamb_2019", "title": "Large Batch Optimization for Deep Learning: Training BERT in 76 minutes", "priority": "external_literature", "layer": "optimizer_memory_and_scaling", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://arxiv.org/abs/1904.00962", "notes": "Primary LAMB source for layer-wise trust ratios and large-batch optimization. Its reported BERT time-to-target is tied to model, batch, hardware, quality target, and tuning conditions and is not a universal large-batch or wall-clock result.", "source_type": "conference_paper", "arxiv_id": "1904.00962", "published": "2019-04-01", "citation_label": "You et al. (2020), Large Batch Optimization for Deep Learning", "doi": "10.48550/arXiv.1904.00962" }, { "id": "ext_shampoo_2018", "title": "Shampoo: Preconditioned Stochastic Tensor Optimization", "priority": "external_literature", "layer": "tensor_and_matrix_preconditioning", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://proceedings.mlr.press/v80/gupta18a.html", "notes": "Primary tensor-structured preconditioning source. Shampoo maintains per-dimension preconditioners and reports faster convergence with practical per-step cost in studied models; stochastic-convex theory and source experiments do not settle current distributed lifecycle cost.", "source_type": "conference_paper", "published": "2018-07-10", "citation_label": "Gupta, Koren, and Singer (2018), Shampoo", "doi": null }, { "id": "ext_kfac_2015", "title": "Optimizing Neural Networks with Kronecker-factored Approximate Curvature", "priority": "external_literature", "layer": "curvature_aware_optimization", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://proceedings.mlr.press/v37/martens15.html", "notes": "Primary K-FAC source for an efficiently invertible Kronecker-factored approximation to the Fisher matrix. Its approximation, damping, inversion, and empirical cost-benefit are architecture- and implementation-dependent.", "source_type": "conference_paper", "published": "2015-07-07", "citation_label": "Martens and Grosse (2015), K-FAC", "doi": null }, { "id": "ext_lion_2023", "title": "Symbolic Discovery of Optimization Algorithms", "priority": "external_literature", "layer": "optimizer_discovery_and_sign_updates", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://arxiv.org/abs/2302.06675", "notes": "Primary Lion and symbolic optimizer-search source. Lion uses sign-based momentum and one optimizer-state tensor; the paper also reports method-specific learning-rate behavior and settings where gains are small or insignificant.", "source_type": "conference_paper", "arxiv_id": "2302.06675", "published": "2023-02-13", "citation_label": "Chen et al. (2023), Symbolic Discovery of Optimization Algorithms", "doi": "10.48550/arXiv.2302.06675" }, { "id": "ext_sophia_2023", "title": "Sophia: A Scalable Stochastic Second-order Optimizer for Language Model Pre-training", "priority": "external_literature", "layer": "curvature_aware_optimization", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://arxiv.org/abs/2305.14342", "notes": "Primary Sophia source for periodic diagonal-Hessian estimation and clipped curvature-aware updates. Its reported GPT pretraining speedups and simplified theory require matched reproduction before any broader optimizer claim.", "source_type": "conference_paper", "arxiv_id": "2305.14342", "published": "2023-05-23", "citation_label": "Liu et al. (2023), Sophia", "doi": "10.48550/arXiv.2305.14342" }, { "id": "ext_soap_2024", "title": "SOAP: Improving and Stabilizing Shampoo using Adam", "priority": "external_literature", "layer": "tensor_and_matrix_preconditioning", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://arxiv.org/abs/2409.11321", "notes": "Primary SOAP source connecting Shampoo to adaptive moments in a changing preconditioner eigenbasis. Reported large-batch pretraining gains remain tied to 360M/660M models, preconditioning frequency, overhead, and tuning conditions.", "source_type": "arxiv", "arxiv_id": "2409.11321", "published": "2024-09-17", "citation_label": "Vyas et al. (2024), SOAP", "doi": "10.48550/arXiv.2409.11321" }, { "id": "ext_schedule_free_2024", "title": "The Road Less Scheduled", "priority": "external_literature", "layer": "optimizer_scheduling_and_averaging", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://arxiv.org/abs/2405.15682", "notes": "Primary schedule-free optimization source unifying scheduling and iterate averaging without requiring a stopping step. Removing a stopping-time schedule does not remove learning-rate, warmup, evaluation-iterate, checkpoint, or method-selection choices.", "source_type": "conference_paper", "arxiv_id": "2405.15682", "published": "2024-05-24", "citation_label": "Defazio et al. (2024), The Road Less Scheduled", "doi": "10.48550/arXiv.2405.15682" }, { "id": "ext_mup_2022", "title": "Tensor Programs V: Tuning Large Neural Networks via Zero-Shot Hyperparameter Transfer", "priority": "external_literature", "layer": "optimizer_parametrization_and_scale_transfer", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://arxiv.org/abs/2203.03466", "notes": "Primary maximal-update parametrization and muTransfer source. It reports widthwise hyperparameter transfer under a prescribed parametrization on Transformer and ResNet settings; it does not establish arbitrary depth, duration, optimizer, or architecture transfer.", "source_type": "conference_paper", "arxiv_id": "2203.03466", "published": "2022-03-07", "citation_label": "Yang et al. (2022), Tensor Programs V", "doi": "10.48550/arXiv.2203.03466" }, { "id": "ext_modular_norm_2024", "title": "Scalable Optimization in the Modular Norm", "priority": "external_literature", "layer": "optimizer_parametrization_and_scale_transfer", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://arxiv.org/abs/2405.14813", "notes": "Primary modular-norm source defining architecture-recursive update geometry and reporting learning-rate transfer across width and depth. Its well-behaved-module assumptions and experiments do not prove arbitrary substrate transfer or optimizer superiority.", "source_type": "conference_paper", "arxiv_id": "2405.14813", "published": "2024-05-23", "citation_label": "Large et al. (2024), Scalable Optimization in the Modular Norm", "doi": "10.48550/arXiv.2405.14813" }, { "id": "ext_muon_scalable_2025", "title": "Muon is Scalable for LLM Training", "priority": "external_literature", "layer": "orthogonalized_matrix_optimization", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://arxiv.org/abs/2502.16982", "notes": "Primary large-scale Muon source for momentum plus matrix orthogonalization, weight decay, per-parameter update scaling, and a distributed implementation. Its reported compute-efficiency and Moonlight results are source-scoped and not locally reproduced.", "source_type": "arxiv", "arxiv_id": "2502.16982", "published": "2025-02-24", "citation_label": "Liu et al. (2025), Muon is Scalable for LLM Training", "doi": "10.48550/arXiv.2502.16982" }, { "id": "ext_muon_spectral_norm_2026", "title": "Muon Optimizes Under Spectral Norm Constraints", "priority": "external_literature", "layer": "orthogonalized_matrix_optimization_theory", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://openreview.net/forum?id=Blz4hjxLwU", "notes": "Accepted TMLR theoretical source placing Muon with decoupled weight decay in a Lion-K/nuclear-norm framework and deriving implicit spectral-norm constraint behavior. The interpretation does not establish task-level quality, safety, or universal advantage.", "source_type": "journal_paper", "published": "2026-03-23", "updated": "2026-03-23", "citation_label": "Chen, Li, and Liu (2026), Muon Optimizes Under Spectral Norm Constraints", "doi": null }, { "id": "ext_nist_privacy_framework_2020", "title": "NIST Privacy Framework: A Tool for Improving Privacy through Enterprise Risk Management, Version 1.0", "priority": "external_literature", "layer": "privacy_data_rights_and_information_flow_governance", "chapter_targets": [ "privacy-data-rights-and-information-flow-governance", "security-kernel-and-digital-scifs" ], "url": "https://doi.org/10.6028/NIST.CSWP.01162020", "notes": "Official paper-body-reviewed risk framework distinguishing privacy problems from cybersecurity incidents across the data lifecycle. It is voluntary, has no force of law, and supplies no local privacy outcome or certification.", "source_type": "government_framework", "published": "2020-01-16", "updated": "2024-01-22", "citation_label": "NIST (2020), Privacy Framework 1.0", "doi": "10.6028/NIST.CSWP.01162020" }, { "id": "ext_eu_gdpr_2016", "title": "Regulation (EU) 2016/679 (General Data Protection Regulation)", "priority": "external_literature", "layer": "privacy_data_rights_and_information_flow_governance", "chapter_targets": [ "privacy-data-rights-and-information-flow-governance" ], "url": "https://eur-lex.europa.eu/eli/reg/2016/679", "notes": "Authoritative jurisdiction-specific normative comparator for principles, bases, rights, accountability, design, and qualified exceptions. It is not universal law, legal advice, an applicability decision, or local compliance evidence.", "source_type": "law_primary", "published": "2016-04-27", "updated": "2016-05-04", "citation_label": "European Parliament and Council (2016), Regulation (EU) 2016/679" }, { "id": "ext_w3c_dpv_2024", "title": "Data Privacy Vocabulary (DPV), Version 2", "priority": "external_literature", "layer": "privacy_data_rights_and_information_flow_governance", "chapter_targets": [ "privacy-data-rights-and-information-flow-governance", "context-transactions-snapshots-mounts-and-taint" ], "url": "https://www.w3.org/community/reports/dpvcg/CG-FINAL-dpv-20240801/", "notes": "Machine-readable vocabulary for purpose, processing, data, actors, rights, risks, measures, legal basis, and consent. It is a Community Group Final Specification, not a W3C Recommendation, law, or enforcement proof.", "source_type": "community_specification", "published": "2024-08-01", "updated": "2024-08-01", "citation_label": "W3C Data Privacy Vocabularies and Controls CG (2024), DPV v2" }, { "id": "ext_abadi_dpsgd_2016", "title": "Deep Learning with Differential Privacy", "priority": "external_literature", "layer": "privacy_data_rights_and_information_flow_governance", "chapter_targets": [ "privacy-data-rights-and-information-flow-governance", "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://arxiv.org/abs/1607.00133", "notes": "Primary DP-SGD mechanism and accounting source. Its algorithm, analysis, and reported experiments are not locally reproduced; its guarantee is parameter-, unit-, adjacency-, implementation-, and release-surface-bound.", "source_type": "conference_paper", "arxiv_id": "1607.00133", "published": "2016-07-01", "updated": "2016-10-24", "citation_label": "Abadi et al. (2016), Deep Learning with Differential Privacy", "doi": "10.1145/2976749.2978318" }, { "id": "ext_algospec_purpose_limitation_2024", "title": "Being Transparent Is Merely the Beginning: Enforcing Purpose Limitation with Polynomial Approximation", "priority": "external_literature", "layer": "privacy_data_rights_and_information_flow_governance", "chapter_targets": [ "privacy-data-rights-and-information-flow-governance" ], "url": "https://www.usenix.org/conference/usenixsecurity24/presentation/liu-shuofeng", "notes": "Primary competing purpose-restriction design using algorithm-specific polynomial approximation. Reported accuracy and efficiency are bounded to studied algorithms/data and are not locally reproduced or a complete legal-purpose result.", "source_type": "conference_paper", "published": "2024-08", "updated": "2024-08", "citation_label": "Liu et al. (2024), Being Transparent Is Merely the Beginning" }, { "id": "ext_carlini_training_data_extraction_2021", "title": "Extracting Training Data from Large Language Models", "priority": "external_literature", "layer": "privacy_data_rights_and_information_flow_governance", "chapter_targets": [ "privacy-data-rights-and-information-flow-governance", "data-engines-continual-learning-and-unlearning" ], "url": "https://www.usenix.org/conference/usenixsecurity21/presentation/carlini-extracting", "notes": "Primary failure source reporting black-box extraction of memorized GPT-2 training sequences. The source result is configuration-bound and not a local or universal leakage result.", "source_type": "conference_paper", "published": "2021-08", "updated": "2021-08", "citation_label": "Carlini et al. (2021), Extracting Training Data from Large Language Models" }, { "id": "ext_choquette_choo_label_only_mia_2021", "title": "Label-Only Membership Inference Attacks", "priority": "external_literature", "layer": "privacy_data_rights_and_information_flow_governance", "chapter_targets": [ "privacy-data-rights-and-information-flow-governance" ], "url": "https://proceedings.mlr.press/v139/choquette-choo21a.html", "notes": "Primary failure source showing hard-label robustness can expose membership and confidence masking can be insufficient in studied settings. No attack or defense result is locally reproduced or universal.", "source_type": "conference_paper", "published": "2021-07", "updated": "2021-07", "citation_label": "Choquette-Choo et al. (2021), Label-Only Membership Inference Attacks" }, { "id": "ext_mahloujifar_fdp_audit_2025", "title": "Auditing f-Differential Privacy in One Run", "priority": "external_literature", "layer": "privacy_data_rights_and_information_flow_governance", "chapter_targets": [ "privacy-data-rights-and-information-flow-governance", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://proceedings.mlr.press/v267/mahloujifar25a.html", "notes": "Primary empirical-audit comparator using randomized inclusion and an f-DP hypothesis in one run. A passed audit is not proof that DP or lifecycle privacy holds, and no result is locally reproduced.", "source_type": "conference_paper", "published": "2025-07", "updated": "2025-07", "citation_label": "Mahloujifar, Melis, and Chaudhuri (2025), Auditing f-Differential Privacy in One Run" }, { "id": "ext_airllm_2023", "title": "AirLLM: Scaling Large Language Models on Low-End Commodity Computers", "priority": "external_literature", "layer": "heterogeneous_inference_memory", "chapter_targets": [ "fast-generation-architectures", "personal-compute-hives-and-federated-edge-intelligence", "resource-economics-and-token-budgets", "model-weight-custody-and-hardware-roots-of-trust" ], "url": "https://github.com/lyogavin/airllm", "notes": "Official implementation comparator for layer-wise model sharding, one-layer accelerator residency, next-layer prefetch, optional storage compression, and original-versus-transformed model storage. Maintainer-reported fit and speed claims are not independently reproduced.", "source_type": "software_repository", "published": "2023-11-20", "updated": "2026-06-30", "citation_label": "Li (2023-2026), AirLLM" }, { "id": "ext_deepspeed_inference_2022", "title": "DeepSpeed Inference: Enabling Efficient Inference of Transformer Models at Unprecedented Scale", "priority": "external_literature", "layer": "heterogeneous_inference_memory", "chapter_targets": [ "fast-generation-architectures", "personal-compute-hives-and-federated-edge-intelligence", "resource-economics-and-token-budgets" ], "url": "https://arxiv.org/abs/2207.00032", "notes": "Primary heterogeneous-inference systems source spanning GPU, CPU, and NVMe for dense and sparse Transformer inference. Reported latency, throughput, scale, and model-fit results remain source-scoped and unreproduced.", "source_type": "arxiv", "arxiv_id": "2207.00032", "published": "2022-06-30", "updated": "2022-06-30", "citation_label": "Aminabadi et al. (2022), DeepSpeed Inference", "doi": "10.48550/arXiv.2207.00032" }, { "id": "ext_flexgen_2023", "title": "FlexGen: High-Throughput Generative Inference of Large Language Models with a Single GPU", "priority": "external_literature", "layer": "heterogeneous_inference_memory", "chapter_targets": [ "fast-generation-architectures", "personal-compute-hives-and-federated-edge-intelligence", "resource-economics-and-token-budgets" ], "url": "https://arxiv.org/abs/2303.06865", "notes": "Primary planned-placement source for GPU/CPU/disk tensor storage and access, batching, and optional weight/cache compression under latency-insensitive workloads. Its throughput results are not interactive-latency or local evidence.", "source_type": "arxiv", "arxiv_id": "2303.06865", "published": "2023-03-13", "updated": "2023-06-12", "citation_label": "Sheng et al. (2023), FlexGen", "doi": "10.48550/arXiv.2303.06865" }, { "id": "ext_hf_accelerate_big_model_inference_2026", "title": "Hugging Face Accelerate: Loading Big Models into Memory", "priority": "external_literature", "layer": "heterogeneous_inference_memory", "chapter_targets": [ "fast-generation-architectures", "personal-compute-hives-and-federated-edge-intelligence", "resource-economics-and-token-budgets", "model-weight-custody-and-hardware-roots-of-trust" ], "url": "https://huggingface.co/docs/accelerate/en/concept_guides/big_model_inference", "notes": "Official implementation documentation for automatic or explicit GPU/CPU/disk device maps and memory-mapped disk tensors. The documented sequential-dispatch, prefetch, and hard-drive-performance limitations make it a baseline, not a qualification result.", "source_type": "official_documentation", "published": "2026-07-23", "updated": "2026-07-23", "citation_label": "Hugging Face (2026), Accelerate Big Model Inference documentation" }, { "id": "ext_llama_cpp_memory_mapping_2026", "title": "llama.cpp CLI Memory Mapping, Tensor Placement, and KV Offload Controls", "priority": "external_literature", "layer": "heterogeneous_inference_memory", "chapter_targets": [ "fast-generation-architectures", "personal-compute-hives-and-federated-edge-intelligence", "resource-economics-and-token-budgets", "model-weight-custody-and-hardware-roots-of-trust" ], "url": "https://github.com/ggml-org/llama.cpp/blob/master/tools/cli/README.md", "notes": "Official consumer-runtime documentation for model load modes, memory mapping, DirectIO, GPU-layer and tensor placement, MoE CPU placement, KV offload, and KV data types. No local model or performance result is implied.", "source_type": "official_documentation", "published": "2026-07-23", "updated": "2026-07-23", "citation_label": "ggml-org (2026), llama.cpp CLI documentation" }, { "id": "ext_llm_in_flash_2024", "title": "LLM in a Flash: Efficient Large Language Model Inference with Limited Memory", "priority": "external_literature", "layer": "heterogeneous_inference_memory", "chapter_targets": [ "fast-generation-architectures", "personal-compute-hives-and-federated-edge-intelligence", "resource-economics-and-token-budgets", "model-weight-custody-and-hardware-roots-of-trust", "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2312.11514", "notes": "Primary flash-aware inference source for on-demand parameter loading, I/O cost modeling, transfer reduction, contiguous reads, windowing, and row-column bundling. Sparse/context-adaptive loading is not an exact dense paging result.", "source_type": "conference_paper", "arxiv_id": "2312.11514", "published": "2023-12-12", "updated": "2024-07-30", "citation_label": "Alizadeh et al. (2024), LLM in a Flash", "doi": "10.48550/arXiv.2312.11514" }, { "id": "ext_powerinfer_2024", "title": "PowerInfer: Fast Large Language Model Serving with a Consumer-Grade GPU", "priority": "external_literature", "layer": "heterogeneous_inference_memory", "chapter_targets": [ "fast-generation-architectures", "personal-compute-hives-and-federated-edge-intelligence", "resource-economics-and-token-budgets", "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2312.12456", "notes": "Primary consumer-inference source for source-reported power-law neuron locality, hot-GPU/cold-CPU placement, adaptive predictors, and sparse operators. Architecture transfer and performance are not locally reproduced.", "source_type": "conference_paper", "arxiv_id": "2312.12456", "published": "2023-12-16", "updated": "2024-12-12", "citation_label": "Song et al. (2024), PowerInfer", "doi": "10.48550/arXiv.2312.12456" }, { "id": "ext_vattention_2025", "title": "vAttention: Dynamic Memory Management for Serving LLMs without PagedAttention", "priority": "external_literature", "layer": "heterogeneous_inference_memory", "chapter_targets": [ "fast-generation-architectures", "resource-economics-and-token-budgets" ], "url": "https://arxiv.org/abs/2405.04437", "notes": "Primary counterpoint to non-contiguous PagedAttention layouts: decouples virtual and physical GPU memory while retaining contiguous KV virtual addresses. Reported serving results remain source-scoped.", "source_type": "conference_paper", "arxiv_id": "2405.04437", "published": "2024-05-07", "updated": "2025-01-29", "citation_label": "Prabhu et al. (2025), vAttention", "doi": "10.48550/arXiv.2405.04437" }, { "id": "ext_infinigen_2024", "title": "InfiniGen: Efficient Generative Inference of Large Language Models with Dynamic KV Cache Management", "priority": "external_literature", "layer": "heterogeneous_inference_memory", "chapter_targets": [ "fast-generation-architectures", "resource-economics-and-token-budgets" ], "url": "https://arxiv.org/abs/2406.19707", "notes": "Primary speculative-KV-prefetch source using minimal rehearsal and partial next-layer state to select host-resident KV entries. Prediction, quality, miss, and fallback results are not locally reproduced.", "source_type": "conference_paper", "arxiv_id": "2406.19707", "published": "2024-06-28", "updated": "2024-06-28", "citation_label": "Lee et al. (2024), InfiniGen", "doi": "10.48550/arXiv.2406.19707" }, { "id": "ext_specache_2025", "title": "SpeCache: Speculative Key-Value Caching for Efficient Generation of LLMs", "priority": "external_literature", "layer": "heterogeneous_inference_memory", "chapter_targets": [ "fast-generation-architectures", "resource-economics-and-token-budgets" ], "url": "https://arxiv.org/abs/2503.16163", "notes": "Primary speculative-KV-prefetch source keeping complete KV state in CPU memory, a low-bit importance copy in VRAM, and predicted next-step KV transfers. Source-reported quality and memory results are unreproduced.", "source_type": "arxiv", "arxiv_id": "2503.16163", "published": "2025-03-20", "updated": "2025-03-20", "citation_label": "Jie et al. (2025), SpeCache", "doi": "10.48550/arXiv.2503.16163" }, { "id": "ext_specoffload_2025", "title": "SpecOffload: Unlocking Latent GPU Capacity for LLM Inference on Resource-Constrained Devices", "priority": "external_literature", "layer": "heterogeneous_inference_memory", "chapter_targets": [ "fast-generation-architectures", "resource-economics-and-token-budgets" ], "url": "https://arxiv.org/abs/2505.10259", "notes": "Primary composition source for target-model offloading, draft-model placement, speculative decoding, and joint tensor/decoding planning. It is not speculative physical-page prediction, and reported results are unreproduced.", "source_type": "arxiv", "arxiv_id": "2505.10259", "published": "2025-05-15", "updated": "2025-05-21", "citation_label": "Zhuge et al. (2025), SpecOffload", "doi": "10.48550/arXiv.2505.10259" }, { "id": "ext_atsinfer_2026", "title": "Automated Tensor Scheduling for Hybrid CPU-GPU LLM Inference on Consumer Devices", "priority": "external_literature", "layer": "heterogeneous_inference_memory", "chapter_targets": [ "fast-generation-architectures", "personal-compute-hives-and-federated-edge-intelligence", "resource-economics-and-token-budgets", "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://arxiv.org/abs/2607.10183", "notes": "Very recent preprint comparator for tensor-granular static placement, load-aware dynamic transfer, and asynchronous CPU-GPU coordination on consumer devices. Only abstract/metadata were reviewed; reported results are provisional and unreproduced.", "source_type": "arxiv", "arxiv_id": "2607.10183", "published": "2026-07-11", "updated": "2026-07-14", "citation_label": "Liu et al. (2026), ATSInfer", "doi": "10.48550/arXiv.2607.10183" }, { "id": "ext_openai_prompt_caching_docs_2026", "title": "Prompt Caching", "priority": "external_literature", "layer": "inference_cache_reuse", "chapter_targets": [ "fast-generation-architectures", "resource-economics-and-token-budgets", "context-transactions-snapshots-mounts-and-taint" ], "url": "https://developers.openai.com/api/docs/guides/prompt-caching", "notes": "Current official provider contract for exact-prefix prompt caching, cache-write and cache-read metering, usage receipts, retention, organization isolation, and rate-limit boundaries. Product behavior and prices are time-sensitive; inspected 2026-07-23.", "source_type": "official_documentation", "updated": "2026-07-23", "citation_label": "OpenAI (2026), Prompt Caching" }, { "id": "ext_anthropic_prompt_caching_docs_2026", "title": "Prompt caching", "priority": "external_literature", "layer": "inference_cache_reuse", "chapter_targets": [ "fast-generation-architectures", "resource-economics-and-token-budgets", "context-transactions-snapshots-mounts-and-taint" ], "url": "https://platform.claude.com/docs/en/build-with-claude/prompt-caching", "notes": "Current official provider contract for reusable prompt prefixes, explicit cache breakpoints, five-minute and one-hour lifetimes, cache creation and read metering, and prewarming. Product behavior and prices are time-sensitive; inspected 2026-07-23.", "source_type": "official_documentation", "updated": "2026-07-23", "citation_label": "Anthropic (2026), Prompt caching" }, { "id": "ext_gemini_context_caching_docs_2026", "title": "Context caching", "priority": "external_literature", "layer": "inference_cache_reuse", "chapter_targets": [ "fast-generation-architectures", "resource-economics-and-token-budgets", "context-transactions-snapshots-mounts-and-taint" ], "url": "https://ai.google.dev/gemini-api/docs/caching", "notes": "Current official provider contract for implicit and explicit context caching, common-prefix placement, cached-token usage reporting, time-to-live, and storage charges. Product behavior and prices are time-sensitive; inspected 2026-07-23.", "source_type": "official_documentation", "updated": "2026-07-23", "citation_label": "Google (2026), Gemini Context Caching" }, { "id": "ext_vllm_automatic_prefix_caching_2026", "title": "Automatic Prefix Caching", "priority": "external_literature", "layer": "inference_cache_reuse", "chapter_targets": [ "fast-generation-architectures", "resource-economics-and-token-budgets", "context-transactions-snapshots-mounts-and-taint" ], "url": "https://docs.vllm.ai/en/v0.14.1/design/prefix_caching/", "notes": "Official vLLM design documentation for block-hash exact-prefix KV reuse, least-recently-used eviction, multi-modal and adapter identity, and tenant cache-salt protection against timing inference. No local serving benchmark was run.", "source_type": "official_documentation", "updated": "2026-07-23", "citation_label": "vLLM (2026), Automatic Prefix Caching" }, { "id": "ext_sglang_radixattention_2024", "title": "SGLang: Efficient Execution of Structured Language Model Programs", "priority": "external_literature", "layer": "inference_cache_reuse", "chapter_targets": [ "fast-generation-architectures", "resource-economics-and-token-budgets" ], "url": "https://papers.nips.cc/paper_files/paper/2024/hash/724be4472168f31ba1c9ac630f15dec8-Abstract-Conference.html", "notes": "Primary RadixAttention and cache-aware scheduling source for structured multi-call language-model programs. Source-reported throughput and theorem scope remain unreproduced.", "source_type": "conference_paper", "arxiv_id": "2312.07104", "published": "2024", "updated": "2024-06-06", "citation_label": "Zheng et al. (2024), SGLang", "doi": "10.48550/arXiv.2312.07104" }, { "id": "ext_prompt_cache_2024", "title": "Prompt Cache: Modular Attention Reuse for Low-Latency Inference", "priority": "external_literature", "layer": "inference_cache_reuse", "chapter_targets": [ "fast-generation-architectures", "resource-economics-and-token-budgets" ], "url": "https://proceedings.mlsys.org/paper_files/paper/2024/file/a66caa1703fe34705a4368c3014c1966-Paper-Conference.pdf", "notes": "Primary MLSys source for schema-defined reusable prompt modules, positional accuracy, and attention-state reuse across prompts. Source-reported latency remains unreproduced.", "source_type": "conference_paper", "arxiv_id": "2311.04934", "published": "2024", "updated": "2024-04-25", "citation_label": "Gim et al. (2024), Prompt Cache", "doi": "10.48550/arXiv.2311.04934" }, { "id": "ext_mooncake_2025", "title": "Mooncake: Trading More Storage for Less Computation \u2014 A KVCache-centric Architecture for Serving LLM Chatbot", "priority": "external_literature", "layer": "inference_cache_reuse", "chapter_targets": [ "fast-generation-architectures", "resource-economics-and-token-budgets" ], "url": "https://www.usenix.org/conference/fast25/presentation/qin", "notes": "Primary FAST 2025 source for a KV-cache-centric disaggregated serving architecture spanning prefill, decode, DRAM, SSD, and network resources. Production-trace and capacity results remain source-reported.", "source_type": "conference_paper", "published": "2025", "citation_label": "Qin et al. (2025), Mooncake" }, { "id": "ext_cacheblend_2025", "title": "CacheBlend: Fast Large Language Model Serving for RAG with Cached Knowledge Fusion", "priority": "external_literature", "layer": "inference_cache_reuse", "chapter_targets": [ "fast-generation-architectures", "resource-economics-and-token-budgets" ], "url": "https://arxiv.org/abs/2405.16444", "notes": "Primary source for non-prefix and multi-chunk KV reuse with selective recomputation. It makes the cross-attention failure of na\u00efve independent-chunk KV fusion explicit. Source-reported latency and quality remain unreproduced.", "source_type": "conference_paper", "arxiv_id": "2405.16444", "published": "2025", "updated": "2025-04-03", "citation_label": "Yao et al. (2025), CacheBlend", "doi": "10.48550/arXiv.2405.16444" }, { "id": "ext_azure_llm_semantic_cache_2026", "title": "Azure API Management LLM semantic cache lookup policy", "priority": "external_literature", "layer": "inference_cache_reuse", "chapter_targets": [ "fast-generation-architectures", "resource-economics-and-token-budgets", "context-transactions-snapshots-mounts-and-taint" ], "url": "https://learn.microsoft.com/en-ie/azure/api-management/llm-semantic-cache-lookup-policy", "notes": "Official semantic-response-cache policy documentation. It treats vector similarity as an approximate response-reuse decision and warns that a hit can return an incorrect, outdated, or unsafe answer. No local semantic-cache deployment was run.", "source_type": "official_documentation", "updated": "2026-07-23", "citation_label": "Microsoft (2026), LLM semantic cache lookup policy" }, { "id": "precision_contract", "title": "The Precision Contract: A Functional Rate\u2013Distortion Theory for Behavior-Preserving Neural Computation", "priority": "must_use", "layer": "functional_precision_behavior_preserving_compression_and_certification", "chapter_targets": [ "rankfold-neuralfold-and-artifact-compression", "compact-generative-systems-and-residual-honesty", "fast-generation-architectures", "resource-economics-and-token-budgets", "readiness-gates-residual-escrow-and-quarantine", "executable-specifications-and-lean-proof-envelope", "model-weight-custody-and-hardware-roots-of-trust", "open-weight-release-and-post-release-control", "the-efficient-asi-hypothesis" ], "url": "local-source:sources/raw/corben_papers/precision_contract/precision_contract.md@sha256:c8c5eeedf72f5fad4132d9916971f1c1853c74f01fc4a6bab9ad8c0700b39ac7", "notes": "Corben-authored July 2026 theoretical and systems paper replacing universal per-weight precision questions with a contract-relative functional rate-distortion problem over complete executable descriptions. It proposes representation canonicalization, protected-behavior contracts, precision fields, progressive base/residual encoding, dynamic routing, full physical and assurance-cost accounting, a Functional Precision Compiler, and scoped precision certificates. Existing chapters are upgraded first; no universal bit bound, implemented compiler, preserved-behavior result, efficiency result, certificate validity, support promotion, SOTA, AGI, or ASI claim is inferred.", "source_type": "author_whitepaper", "published": "2026-07", "updated": "2026-07-24", "citation_label": "Sorenson (2026), The Precision Contract" }, { "id": "ext_nist_adversarial_ml_2024", "title": "Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and Mitigations", "priority": "external_literature", "layer": "adversarial_machine_learning", "chapter_targets": [ "adversarial-machine-learning-and-model-attack-surface" ], "url": "https://doi.org/10.6028/NIST.AI.100-2e2023", "notes": "Official NIST taxonomy and terminology comparator for adversarial machine learning across lifecycle stages, attacker goals, knowledge, capabilities, attacks, and mitigations. It is a taxonomy, not local robustness evidence or proof that listed mitigations work for this stack.", "source_type": "government_report", "published": "2024-01-04", "citation_label": "Vassilev et al. (2024), NIST AI 100-2", "doi": "10.6028/NIST.AI.100-2e2023" }, { "id": "ext_singapore_consensus_2026", "title": "The 2026 Singapore Consensus on Global AI Safety Research Priorities", "priority": "external_literature", "layer": "dangerous_capability_assessment_societal_resilience_and_agentic_risk", "chapter_targets": [ "dangerous-capability-domains-and-misuse-uplift", "societal-resilience-and-misuse-defense", "open-weight-release-and-post-release-control", "capability-thresholds-and-deployment-commitments" ], "url": "https://aisafetypriorities.org/", "notes": "International 2026 technical-research-priority synthesis covering risk assessment, development, control, and societal resilience, including CBRN, cyber, psychological manipulation, malicious fine-tuning, agent monitoring, incident reporting, and defense-favoring capabilities. It is a research agenda and consensus synthesis, not evidence that any listed safeguard works or that this book's contracts are complete.", "source_type": "international_consensus_report", "published": "2026-07", "updated": "2026-07", "citation_label": "Singapore Consensus (2026), Global AI Safety Research Priorities" }, { "id": "ext_international_ai_safety_report_2026", "title": "International AI Safety Report 2026", "priority": "external_literature", "layer": "frontier_ai_risk_misuse_open_weight_and_societal_resilience", "chapter_targets": [ "dangerous-capability-domains-and-misuse-uplift", "societal-resilience-and-misuse-defense", "open-weight-release-and-post-release-control", "content-authenticity-watermarking-and-synthetic-media-integrity" ], "url": "https://internationalaisafetyreport.org/publication/international-ai-safety-report-2026", "notes": "International expert report synthesizing evidence on general-purpose AI capabilities, misuse, open-weight risks, safeguards, monitoring, and societal resilience. It supports risk taxonomy and uncertainty boundaries; its literature synthesis does not reproduce component studies locally or establish that any ASI Stack mechanism is effective.", "source_type": "international_expert_report", "published": "2026-02-03", "updated": "2026-02-03", "citation_label": "International AI Safety Report (2026)" }, { "id": "ext_c2pa_specification_2_3_2025", "title": "C2PA Content Credentials Technical Specification 2.3", "priority": "external_literature", "layer": "content_provenance_and_authenticity", "chapter_targets": [ "content-authenticity-watermarking-and-synthetic-media-integrity", "ai-supply-chain-integrity-and-lifecycle-provenance" ], "url": "https://spec.c2pa.org/specifications/specifications/2.3/index.html", "notes": "Official C2PA specification for signed manifests, assertions, ingredients, content bindings, validation, and provenance history. It provides a concrete interoperability comparator; it does not prove truth of depicted events, creator identity beyond the credential chain, semantic authenticity, universal platform retention, or resistance to removal and laundering.", "source_type": "technical_standard", "published": "2025-12", "updated": "2026-01-05", "citation_label": "C2PA (2025), Content Credentials Specification 2.3" }, { "id": "ext_eu_article_50_transparency_guidelines_2026", "title": "Guidelines on Transparency Obligations for Providers and Deployers of AI Systems", "priority": "external_literature", "layer": "synthetic_content_transparency_and_disclosure", "chapter_targets": [ "content-authenticity-watermarking-and-synthetic-media-integrity", "institutions-international-coordination-and-public-legitimacy" ], "url": "https://digital-strategy.ec.europa.eu/en/library/guidelines-transparency-obligations-providers-and-deployers-ai-systems", "notes": "European Commission guidance for Article 50 transparency obligations concerning AI interaction, machine-readable marking, deepfakes, and certain public-interest text, with obligations applying from 2 August 2026 subject to scope and transitional details. It is legal and implementation guidance, not legal advice, proof of compliance, or evidence that a marking technique is robust.", "source_type": "official_regulatory_guidance", "published": "2026-07-20", "updated": "2026-07-20", "citation_label": "European Commission (2026), Article 50 Transparency Guidelines" }, { "id": "ext_openai_worst_case_open_weight_risks_2025", "title": "Estimating Worst-Case Frontier Risks of Open-Weight LLMs", "priority": "external_literature", "layer": "malicious_fine_tuning_and_open_weight_release_evaluation", "chapter_targets": [ "open-weight-release-and-post-release-control", "dangerous-capability-domains-and-misuse-uplift" ], "url": "https://openai.com/index/estimating-worst-case-frontier-risks-of-open-weight-llms/", "notes": "Provider-authored study of malicious fine-tuning for biology and cyber evaluations before the gpt-oss release. It supplies a concrete worst-case-elicitation comparator and reports bounded provider results; it does not prove future-release safety, general malicious-fine-tuning resistance, independent reproduction, or absence of untested harms.", "source_type": "provider_research_paper", "published": "2025-08-05", "updated": "2025-08-05", "citation_label": "Wallace et al. (2025), Worst-Case Open-Weight Frontier Risks" }, { "id": "ext_aisi_misuse_safeguards_safety_case_2026", "title": "An Example Safety Case for Safeguards Against Misuse", "priority": "external_literature", "layer": "misuse_safeguard_uplift_and_safety_cases", "chapter_targets": [ "dangerous-capability-domains-and-misuse-uplift", "safety-cases-and-structured-assurance", "societal-resilience-and-misuse-defense" ], "url": "https://www.aisi.gov.uk/research/an-example-safety-case-for-safeguards-against-misuse", "notes": "UK AI Security Institute example connecting safeguard red teaming, attacker effort, an uplift model, and a deployment safety case. It is a worked argument and measurement proposal, not proof that real safeguards reduce misuse to a particular level or that the book's proposed defense contracts work.", "source_type": "government_research_report", "published": "2026", "citation_label": "Clymer et al. (2026), Misuse Safeguards Safety Case" }, { "id": "ext_anthropic_responsible_scaling_policy_3_4_2026", "title": "Anthropic Responsible Scaling Policy 3.4", "priority": "external_literature", "layer": "frontier_capability_thresholds_and_safeguards", "chapter_targets": [ "dangerous-capability-domains-and-misuse-uplift", "capability-thresholds-and-deployment-commitments", "open-weight-release-and-post-release-control" ], "url": "https://www.anthropic.com/responsible-scaling-policy", "notes": "Current provider policy comparator linking capability thresholds and safeguards across CBRN and automated R&D threat models, with public risk-report and review commitments. It is a revocable provider policy and self-described governance mechanism, not independent evidence that thresholds are complete, evaluations are sensitive, or safeguards are effective.", "source_type": "provider_policy", "published": "2026-07-08", "updated": "2026-07-08", "citation_label": "Anthropic (2026), Responsible Scaling Policy 3.4" }, { "id": "ext_aisi_frontier_ai_trends_2025", "title": "AISI Frontier AI Trends Report 2025", "priority": "external_literature", "layer": "frontier_capability_evaluation_trends", "chapter_targets": [ "dangerous-capability-domains-and-misuse-uplift", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://www.aisi.gov.uk/research/aisi-frontier-ai-trends-report-2025", "notes": "UK AI Security Institute synthesis of evaluations across offensive cyber, dual-use chemistry and biology, autonomous systems, and societal impacts. It is an institute-reported trend record with bounded methods and coverage, not a complete threat census or local reproduction.", "source_type": "government_research_report", "published": "2025", "citation_label": "UK AISI (2025), Frontier AI Trends Report" }, { "id": "ext_valiant_theory_learnable_1984", "title": "A Theory of the Learnable", "priority": "external_literature", "layer": "computational_learning_theory_and_sample_complexity", "chapter_targets": [ "learning-theory-generalization-and-scaling-science" ], "url": "https://dl.acm.org/doi/10.1145/1968.1972", "notes": "Foundational PAC-learning source for defining learnability through accuracy, confidence, resource, hypothesis, and data assumptions. Its distributional and concept-class assumptions do not directly explain modern foundation-model generalization or certify a trained model.", "source_type": "journal_article", "published": "1984-11-05", "citation_label": "Valiant (1984), A Theory of the Learnable", "doi": "10.1145/1968.1972" }, { "id": "ext_deep_double_descent_2020", "title": "Deep Double Descent: Where Bigger Models and More Data Hurt", "priority": "external_literature", "layer": "generalization_and_interpolation_regimes", "chapter_targets": [ "learning-theory-generalization-and-scaling-science" ], "url": "https://openreview.net/forum?id=-bJ_Hb7lCVH", "notes": "ICLR 2020 empirical study reporting model-wise, sample-wise, and epoch-wise double-descent phenomena and proposing effective model complexity. The phenomenon is configuration- and regime-bound and does not imply that larger models or more data generally hurt or help.", "source_type": "conference_paper", "published": "2020", "citation_label": "Nakkiran et al. (2020), Deep Double Descent", "doi": "10.1088/1742-5468/ac3a74" }, { "id": "ext_emergent_abilities_llms_2022", "title": "Emergent Abilities of Large Language Models", "priority": "external_literature", "layer": "scaling_and_emergent_capability_measurement", "chapter_targets": [ "learning-theory-generalization-and-scaling-science" ], "url": "https://openreview.net/forum?id=yzkSU5zdwD", "notes": "TMLR survey and framing of task abilities that appear sharply at larger model scales under reported evaluations. It motivates prospective scaling measurement but does not establish that every apparent threshold is mechanistically discontinuous or unpredictable.", "source_type": "journal_article", "published": "2022-08-31", "citation_label": "Wei et al. (2022), Emergent Abilities of Large Language Models" }, { "id": "ext_emergent_abilities_mirage_2023", "title": "Are Emergent Abilities of Large Language Models a Mirage?", "priority": "external_literature", "layer": "metric_induced_emergence_and_scaling_measurement", "chapter_targets": [ "learning-theory-generalization-and-scaling-science", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://proceedings.neurips.cc/paper_files/paper/2023/hash/adc98a266f45005c403b8311ca7e8bd7-Abstract-Conference.html", "notes": "NeurIPS 2023 counterevidence showing that nonlinear or discontinuous metrics and limited test data can produce apparently sharp emergence from smoother underlying changes in studied settings. It does not prove that all emergence is a metric artifact.", "source_type": "conference_paper", "published": "2023", "citation_label": "Schaeffer et al. (2023), Emergent Abilities a Mirage?" }, { "id": "ext_elk_report_2021", "title": "Eliciting Latent Knowledge", "priority": "external_literature", "layer": "latent_knowledge_and_ontology_identification", "chapter_targets": [ "white-box-evidence-interpretability-and-activation-governance" ], "url": "https://www.alignment.org/blog/arcs-first-technical-report-eliciting-latent-knowledge/", "notes": "ARC technical-report agenda on mapping between a model's world model and human concepts when ordinary supervision may reward convincing but false reports. It defines an open problem and candidate approaches, not a solved elicitation method or evidence that a deployed model's reports are truthful.", "source_type": "technical_report", "published": "2021-12-14", "citation_label": "Christiano and Xu (2021), Eliciting Latent Knowledge" }, { "id": "ext_influence_functions_2017", "title": "Understanding Black-box Predictions via Influence Functions", "priority": "external_literature", "layer": "training_data_attribution_and_influence", "chapter_targets": [ "white-box-evidence-interpretability-and-activation-governance", "data-engines-continual-learning-and-unlearning" ], "url": "https://proceedings.mlr.press/v70/koh17a.html", "notes": "ICML 2017 source tracing predictions through a learning algorithm to influential training points using influence-function approximations. The theory and approximations have model and optimization assumptions and do not establish exact causal provenance, privacy erasure, or influence removal in foundation models.", "source_type": "conference_paper", "published": "2017", "citation_label": "Koh and Liang (2017), Influence Functions" }, { "id": "ext_flexible_hardware_enabled_guarantees_2025", "title": "Flexible Hardware-Enabled Guarantees for AI Compute", "priority": "external_literature", "layer": "hardware_enabled_governance_and_compute_attestation", "chapter_targets": [ "model-weight-custody-and-hardware-roots-of-trust", "physical-compute-infrastructure-energy-and-environmental-constraints", "institutions-international-coordination-and-public-legitimacy" ], "url": "https://arxiv.org/abs/2506.15093", "notes": "Design proposal for auditable guarantee processors and tamper-resistant enclosures supporting privacy-preserving verification or enforcement of AI-compute claims. It is a proposed architecture with adoption, legacy-hardware, update-authority, side-channel, sovereignty, and abuse risks; no local device or governance guarantee exists.", "source_type": "technical_report", "published": "2025-06-18", "citation_label": "Petrie et al. (2025), Flexible Hardware-Enabled Guarantees", "doi": "10.48550/arXiv.2506.15093" }, { "id": "ext_proof_of_learning_2021", "title": "Proof-of-Learning: Definitions and Practice", "priority": "external_literature", "layer": "training_provenance_and_computation_attestation", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling", "ai-supply-chain-integrity-and-lifecycle-provenance" ], "url": "https://arxiv.org/abs/2103.05633", "notes": "Research proposal for proving that final parameters arose through a claimed iterative learning process using checkpoint and stochastic-training evidence. Later attacks and security work show that proof-of-learning/proof-of-training claims require adversarial review; the source does not prove data rights, objective legitimacy, clean training, or model safety.", "source_type": "conference_paper", "published": "2021", "citation_label": "Jia et al. (2021), Proof-of-Learning", "doi": "10.48550/arXiv.2103.05633" }, { "id": "ext_test_time_training_2020", "title": "Test-Time Training with Self-Supervision for Generalization under Distribution Shifts", "priority": "external_literature", "layer": "test_time_adaptation_and_online_update", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling", "replaceable-cognitive-substrates-beyond-transformer-monoculture" ], "url": "https://proceedings.mlr.press/v119/sun20b.html", "notes": "ICML 2020 method adapting model parameters on each test sample using a self-supervised objective and reporting improvements on studied image-corruption benchmarks. The result is task- and method-bound and does not establish safe online adaptation, resistance to poisoning, or benefit under arbitrary shift.", "source_type": "conference_paper", "published": "2020", "citation_label": "Sun et al. (2020), Test-Time Training" }, { "id": "ext_legal_alignment_2026", "title": "Legal Alignment for Safe and Ethical AI", "priority": "external_literature", "layer": "law_following_ai_and_legal_alignment", "chapter_targets": [ "constitutional-alignment-substrate", "institutions-international-coordination-and-public-legitimacy" ], "url": "https://arxiv.org/abs/2601.04175", "notes": "2026 interdisciplinary agenda for using legal rules, methods of interpretation, and institutional structures in AI alignment. Law is jurisdictional, contested, changing, and sometimes unjust or conflicting; the source does not establish that legal compliance equals moral alignment or that a model can reliably determine applicable law.", "source_type": "research_agenda", "published": "2026-01-07", "citation_label": "Kolt et al. (2026), Legal Alignment for Safe and Ethical AI", "doi": "10.48550/arXiv.2601.04175" }, { "id": "ext_curriculum_learning_2009", "title": "Curriculum Learning", "priority": "external_literature", "layer": "training_curricula_and_example_order", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling", "data-engines-continual-learning-and-unlearning" ], "url": "https://doi.org/10.1145/1553374.1553380", "notes": "ICML 2009 source proposing training curricula that begin with easier examples or concepts and increase difficulty. Reported benefits are problem- and curriculum-bound; ordering can introduce bias, hide hard cases, or create capability and safety regressions.", "source_type": "conference_paper", "published": "2009", "citation_label": "Bengio et al. (2009), Curriculum Learning", "doi": "10.1145/1553374.1553380" }, { "id": "ext_causal_calculus_1995", "title": "A Causal Calculus for Statistical Research", "priority": "external_literature", "layer": "structural_causal_models_and_intervention_identification", "chapter_targets": [ "governed-world-models-and-reality-grounding" ], "url": "https://proceedings.mlr.press/r0/pearl95a.html", "notes": "Foundational do-calculus source distinguishing intervention from observation under an explicit structural causal model. Identification depends on the causal graph and assumptions; the calculus does not discover the correct graph from arbitrary data or establish that a learned world model is causally valid.", "source_type": "conference_paper", "published": "1995", "citation_label": "Pearl (1995), A Causal Calculus for Statistical Research" }, { "id": "ext_ai_simulation_digital_twins_2025", "title": "AI Simulation by Digital Twins: Systematic Survey, Reference Framework, and Mapping to a Standardized Architecture", "priority": "external_literature", "layer": "digital_twins_and_simulation_fidelity", "chapter_targets": [ "embodied-agency-real-time-control-and-physical-safety" ], "url": "https://arxiv.org/abs/2506.06580", "notes": "Systematic survey and reference framework for digital-twin-enabled AI simulation. It supports explicit virtual/physical synchronization and simulation roles; it does not establish that a digital twin is faithful, safe for policy transfer, or an adequate substitute for physical testing.", "source_type": "systematic_review", "published": "2025-06-06", "citation_label": "Liu et al. (2025), AI Simulation by Digital Twins", "doi": "10.48550/arXiv.2506.06580" }, { "id": "ext_nist_privacy_enhancing_cryptography_2026", "title": "Privacy-Enhancing Cryptography", "priority": "external_literature", "layer": "confidential_computation_and_privacy_enhancing_cryptography", "chapter_targets": [ "confidential-and-verifiable-ai-computation", "privacy-data-rights-and-information-flow-governance", "personal-compute-hives-and-federated-edge-intelligence" ], "url": "https://csrc.nist.gov/Projects/pec/fhe", "notes": "NIST program material distinguishing fully homomorphic encryption, secure multiparty computation, zero-knowledge proofs, private-set intersection, and related privacy-enhancing techniques. It provides terminology and use-case context, not implementation security, usable performance, authorization, or end-to-end privacy.", "source_type": "official_standards_program", "published": "2026", "updated": "2026-07-25", "citation_label": "NIST (2026), Privacy-Enhancing Cryptography" }, { "id": "ext_zkllm_2024", "title": "zkLLM: Zero Knowledge Proofs for Large Language Models", "priority": "external_literature", "layer": "verifiable_private_model_inference", "chapter_targets": [ "confidential-and-verifiable-ai-computation" ], "url": "https://arxiv.org/abs/2404.16109", "notes": "Research prototype for proving bounded LLM inference claims while hiding model parameters. Reported proof size and latency are configuration-bound and do not establish semantic correctness, authorization, side-channel security, production readiness, or end-to-end privacy.", "source_type": "research_paper", "published": "2024-04-24", "citation_label": "Sun et al. (2024), zkLLM", "doi": "10.48550/arXiv.2404.16109" }, { "id": "ext_human_ai_team_meta_analysis_2024", "title": "When combinations of humans and AI are useful: A systematic review and meta-analysis", "priority": "external_literature", "layer": "human_ai_complementarity_and_team_baselines", "chapter_targets": [ "human-ai-symbiosis-neurotechnology-and-cognitive-sovereignty", "human-ai-organizations-delegation-and-accountability", "human-factors-and-meaningful-control-in-oversight" ], "url": "https://www.nature.com/articles/s41562-024-02024-1", "notes": "Preregistered synthesis of 106 experiments and 370 effect sizes using human-alone, AI-alone, and combined-system comparisons. The aggregate findings are task- and population-bound and do not establish universal human-AI synergy or longitudinal benefit.", "source_type": "systematic_review_and_meta_analysis", "published": "2024-10-28", "citation_label": "Vaccaro et al. (2024), When combinations of humans and AI are useful", "doi": "10.1038/s41562-024-02024-1" }, { "id": "ext_human_ai_feedback_loops_2025", "title": "Human-AI feedback loops alter human perceptual, emotional and social judgements", "priority": "external_literature", "layer": "longitudinal_human_ai_coupling_and_bias_amplification", "chapter_targets": [ "human-ai-symbiosis-neurotechnology-and-cognitive-sovereignty", "human-ai-organizations-delegation-and-accountability", "human-intent-as-a-formal-input" ], "url": "https://www.nature.com/articles/s41562-024-02077-2", "notes": "Experimental evidence that repeated human-AI interaction can create feedback dynamics in studied judgment tasks. It supports measuring coupled trajectories, not a universal claim about all users, systems, settings, or long-term clinical outcomes.", "source_type": "peer_reviewed_research_article", "published": "2025", "citation_label": "Glickman and Sharot (2025), Human-AI feedback loops", "doi": "10.1038/s41562-024-02077-2" }, { "id": "ext_oecd_neuro_ai_convergence_2025", "title": "Technology convergence: Trends, prospects and policies", "priority": "external_literature", "layer": "neurotechnology_ai_convergence_and_governance", "chapter_targets": [ "human-ai-symbiosis-neurotechnology-and-cognitive-sovereignty", "institutions-international-coordination-and-public-legitimacy" ], "url": "https://www.oecd.org/en/publications/oecd-science-technology-and-innovation-outlook-2025_5fe57b90-en/full-report/technology-convergence-trends-prospects-and-policies_5902a07e.html", "notes": "OECD policy synthesis on converging technologies including AI and neurotechnology. It motivates cross-domain governance and anticipatory capacity but is not a clinical trial, technical validation, or proof of beneficial convergence.", "source_type": "intergovernmental_policy_report", "published": "2025", "citation_label": "OECD (2025), Technology convergence" }, { "id": "ext_who_neurotechnology_landscape_2025", "title": "Landscape analysis of the opportunities and challenges for neurotechnology in global health", "priority": "external_literature", "layer": "neurotechnology_health_equity_and_governance", "chapter_targets": [ "human-ai-symbiosis-neurotechnology-and-cognitive-sovereignty", "privacy-data-rights-and-information-flow-governance" ], "url": "https://www.who.int/publications/i/item/9789240109049", "notes": "WHO landscape analysis of neurotechnology opportunities, risks, governance questions, and global-health distribution. It supports a rights and equity boundary, not device efficacy, individual medical advice, or authorization for neural-data collection.", "source_type": "intergovernmental_health_report", "published": "2025", "citation_label": "WHO (2025), Neurotechnology landscape analysis" }, { "id": "ext_icrc_autonomous_weapons_ihl_2025", "title": "Autonomous Weapon Systems and International Humanitarian Law: Selected Issues", "priority": "external_literature", "layer": "autonomous_weapons_human_judgment_and_ihl", "chapter_targets": [ "military-ai-autonomous-weapons-and-strategic-stability", "institutions-international-coordination-and-public-legitimacy" ], "url": "https://www.icrc.org/sites/default/files/media_file/2025-10/ICRC-Position_Paper-Autonomous_Weapon_Systems_and_IHL-Selected_issues_Oct2025.pdf", "notes": "ICRC legal and policy position on autonomous weapon systems and context-specific human judgment. It is authoritative for the ICRC position, not a universally settled legal interpretation, engineering validation, or authorization to design or deploy weapons.", "source_type": "international_humanitarian_organization_position_paper", "published": "2025-10", "citation_label": "ICRC (2025), Autonomous Weapon Systems and IHL" }, { "id": "ext_sipri_military_ai_nuclear_escalation_2025", "title": "The Impact of Military Artificial Intelligence on Nuclear Escalation Risk", "priority": "external_literature", "layer": "military_ai_crisis_dynamics_and_nuclear_escalation", "chapter_targets": [ "military-ai-autonomous-weapons-and-strategic-stability", "dangerous-capability-domains-and-misuse-uplift" ], "url": "https://www.sipri.org/publications/2025/sipri-insights-peace-and-security/impact-military-artificial-intelligence-nuclear-escalation-risk", "notes": "SIPRI analysis of pathways by which military AI may affect nuclear escalation risk through information, decision, and interaction dynamics. It motivates scenario-specific analysis and does not establish the net effect of any specific system or policy.", "source_type": "peace_and_security_research_report", "published": "2025", "citation_label": "Boulanin et al. (2025), Military AI and Nuclear Escalation Risk", "doi": "10.55163/FZIW8544" }, { "id": "ext_no_free_lunch_inductive_bias_2024", "title": "The No Free Lunch Theorem, Kolmogorov Complexity, and the Role of Inductive Biases in Machine Learning", "priority": "external_literature", "layer": "learning_theory_assumptions_and_inductive_bias", "chapter_targets": [ "learning-theory-generalization-and-scaling-science" ], "url": "https://proceedings.mlr.press/v235/goldblum24a.html", "notes": "ICML 2024 treatment connecting no-free-lunch limits, Kolmogorov complexity, and inductive bias. It supports explicit assumption accounting; it does not show that all learning problems are equally hard or identify the right bias for a deployment.", "source_type": "conference_paper", "published": "2024", "citation_label": "Goldblum et al. (2024), No Free Lunch and Inductive Bias" }, { "id": "ext_neuromorphic_computing_scale_2025", "title": "Neuromorphic computing at scale", "priority": "external_literature", "layer": "neuromorphic_hardware_and_event_driven_computation", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture", "physical-compute-infrastructure-energy-and-environmental-constraints" ], "url": "https://www.nature.com/articles/s41586-024-08253-8", "notes": "Large-scale neuromorphic systems result demonstrating event-driven hardware capabilities under reported workloads and conditions. It does not establish superiority for general AI workloads or end-to-end system cost, programmability, reliability, and governance.", "source_type": "peer_reviewed_research_article", "published": "2025", "citation_label": "Davies et al. (2025), Neuromorphic computing at scale", "doi": "10.1038/s41586-024-08253-8" }, { "id": "ext_photonic_neuromorphic_2024", "title": "Integrated photonic neuromorphic computing: opportunities and challenges", "priority": "external_literature", "layer": "photonic_neuromorphic_compute", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture", "physical-compute-infrastructure-energy-and-environmental-constraints" ], "url": "https://www.nature.com/articles/s44287-024-00050-9", "notes": "Review of integrated photonic neuromorphic computing opportunities and challenges. It maps device and systems tradeoffs but does not establish deployment advantage, digital replacement, or favorable full-stack energy and cost.", "source_type": "peer_reviewed_review", "published": "2024", "citation_label": "Shastri et al. (2024), Integrated photonic neuromorphic computing", "doi": "10.1038/s44287-024-00050-9" }, { "id": "ext_quantum_ml_shadows_2024", "title": "Shadows of quantum machine learning", "priority": "external_literature", "layer": "quantum_machine_learning_claim_boundaries", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture", "physical-compute-infrastructure-energy-and-environmental-constraints" ], "url": "https://www.nature.com/articles/s41467-024-49877-8", "notes": "Peer-reviewed analysis of limitations and benchmarking traps in quantum machine-learning advantage claims. It supports advantage declarations with data-loading, classical-baseline, noise, scale, and end-to-end accounting, not a claim that quantum ML is useless.", "source_type": "peer_reviewed_research_article", "published": "2024", "citation_label": "Schuld and Killoran (2024), Shadows of quantum machine learning", "doi": "10.1038/s41467-024-49877-8" }, { "id": "ext_organoid_intelligence_2023", "title": "Organoid intelligence (OI): the new frontier in biocomputing and intelligence-in-a-dish", "priority": "external_literature", "layer": "biohybrid_computing_and_moral_status", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture", "moral-uncertainty-and-value-conflict" ], "url": "https://www.frontiersin.org/journals/science/articles/10.3389/fsci.2023.1017235/full", "notes": "Research agenda for organoid intelligence and biohybrid computing. It motivates scientific, measurement, welfare, consent, and governance questions but does not demonstrate general intelligence, conscious experience, or practical compute superiority.", "source_type": "peer_reviewed_research_agenda", "published": "2023", "citation_label": "Smirnova et al. (2023), Organoid intelligence", "doi": "10.3389/fsci.2023.1017235" }, { "id": "ext_nist_pqc_standards_2024", "title": "Announcing Approval of Three Federal Information Processing Standards for Post-Quantum Cryptography", "priority": "external_literature", "layer": "post_quantum_cryptography_and_crypto_agility", "chapter_targets": [ "security-kernel-and-digital-scifs", "model-weight-custody-and-hardware-roots-of-trust", "physical-compute-infrastructure-energy-and-environmental-constraints" ], "url": "https://www.nist.gov/news-events/news/2024/08/announcing-approval-three-federal-information-processing-standards-fips", "notes": "Official NIST announcement for FIPS 203, 204, and 205. It establishes approved algorithm standards and migration urgency, not implementation security, protocol correctness, complete inventory, or successful system migration.", "source_type": "official_standards_announcement", "published": "2024-08-13", "citation_label": "NIST (2024), Post-Quantum Cryptography Standards" }, { "id": "ext_oecd_ai_infrastructure_competition_2025", "title": "Competition in artificial intelligence infrastructure", "priority": "external_literature", "layer": "ai_infrastructure_concentration_and_competition", "chapter_targets": [ "ai-deployment-transition-distribution-and-human-agency", "institutions-international-coordination-and-public-legitimacy", "physical-compute-infrastructure-energy-and-environmental-constraints" ], "url": "https://www.oecd.org/en/publications/competition-in-artificial-intelligence-infrastructure_623d1874-en.html", "notes": "OECD analysis of concentration, barriers to entry, vertical integration, and competition across AI infrastructure. It motivates bottleneck and exit analysis but does not adjudicate a specific market, legal violation, or optimal remedy.", "source_type": "intergovernmental_policy_report", "published": "2025", "citation_label": "OECD (2025), Competition in AI Infrastructure" }, { "id": "ext_eu_ai_civil_liability_2025", "title": "Artificial intelligence and civil liability", "priority": "external_literature", "layer": "ai_liability_remedy_and_compensation", "chapter_targets": [ "institutions-international-coordination-and-public-legitimacy", "human-ai-organizations-delegation-and-accountability", "safety-cases-and-structured-assurance" ], "url": "https://op.europa.eu/en/publication-detail/-/publication/79fb710d-6b5c-11f0-bf4e-01aa75ed71a1/language-en", "notes": "European Parliament research service study of AI and civil-liability questions. It supports explicit causation, evidence-access, insurance, compensation, and remedy analysis but is not legal advice or a globally settled liability rule.", "source_type": "legislative_research_report", "published": "2025", "citation_label": "European Parliamentary Research Service (2025), Artificial intelligence and civil liability", "doi": "10.2861/0075079" }, { "id": "ext_cultural_alignment_llms_2024", "title": "Investigating Cultural Alignment of Large Language Models", "priority": "external_literature", "layer": "cultural_alignment_and_value_representation", "chapter_targets": [ "human-intent-as-a-formal-input", "human-ai-communication-persuasion-and-epistemic-security", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://arxiv.org/abs/2402.13231", "notes": "Empirical study of cultural alignment patterns in selected language models and measurements. It supports explicit population, language, and instrument scope; it does not establish stable national values or a universal measure of cultural alignment.", "source_type": "research_paper", "published": "2024-02-20", "citation_label": "Tao et al. (2024), Investigating Cultural Alignment of Large Language Models", "doi": "10.48550/arXiv.2402.13231" }, { "id": "ext_multilingual_evaluation_state_2026", "title": "The State and Fate of Multilingual Contextual Evaluation in the NLP World", "priority": "external_literature", "layer": "multilingual_contextual_evaluation", "chapter_targets": [ "human-ai-communication-persuasion-and-epistemic-security", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://www.microsoft.com/en-us/research/publication/the-state-and-fate-of-multilingual-contextual-evaluation-in-the-nlp-world/", "notes": "Research survey and analysis of multilingual contextual evaluation. It motivates language-by-task coverage and measurement reporting; it does not establish equivalent capability or safety across languages, dialects, or sociocultural settings.", "source_type": "research_survey", "published": "2026", "citation_label": "Microsoft Research (2026), Multilingual Contextual Evaluation" }, { "id": "ext_kimi_k3_2026", "title": "Kimi K3: Open Frontier Intelligence", "priority": "external_literature", "layer": "hybrid_attention_sparse_routing_and_training_systems", "chapter_targets": [ "replaceable-cognitive-substrates-beyond-transformer-monoculture", "routing-heads-and-specialist-cores", "governed-model-training-distributed-optimization-and-scaling" ], "url": "https://arxiv.org/abs/2607.24653", "notes": "Primary technical report and official architecture summary for KDA/Gated-MLA hybrid attention, Attention Residuals, Stable LatentMoE, Quantile Balancing, SiTU-GLU, and Per-Head Muon. The approximately 2.5x scaling-efficiency result is provider-reported for the integrated 2.8T system and does not identify a transferable component effect.", "source_type": "technical_report", "arxiv_id": "2607.24653", "published": "2026-07-27", "updated": "2026-07-27", "citation_label": "Kimi Team (2026), Kimi K3: Open Frontier Intelligence", "doi": "10.48550/arXiv.2607.24653" }, { "id": "portia_synapse", "title": "PortiaSynapse: A Cognitive Spider Architecture for DKL Navigation", "priority": "supporting_lineage", "layer": "routing_training_and_dkl_navigation", "chapter_targets": [ "routing-heads-and-specialist-cores", "policy-optimization-and-learning-from-feedback", "governed-deliberation-and-test-time-scaling", "benchmark-ratchets-and-anti-goodhart-evidence", "durable-semantic-memory-and-knowledge-lattices" ], "url": "https://drive.google.com/file/d/1ZnsE0rAC8YdiMFeo3N4kWEU8keTthF9v", "notes": "Authenticated Google Drive successor to TreeLLM's failed SpiderSynapse path. It proposes a replacement-compatible Scout/Focus/refinement navigator, phased training, diagnostic traits, typed DKL outputs, and fallback. The source reports implementation and tests but contains conflicting test totals, incomplete integration and benchmarking, and unresolved causal, attention-axis, memory-isolation, metric, and calibration questions; no local result is inferred.", "source_type": "technical_whitepaper", "published": "2024-12-23", "citation_label": "Sorenson (2024), PortiaSynapse" }, { "id": "spider_synapse", "title": "SpiderSynapse: A Multi-Hypothesis Reasoning Architecture", "priority": "supporting_lineage", "layer": "routing_training_and_dkl_navigation", "chapter_targets": [ "routing-heads-and-specialist-cores", "policy-optimization-and-learning-from-feedback", "governed-deliberation-and-test-time-scaling", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://drive.google.com/file/d/1PPxsdmNbj8Ao1mNIelMrMxEr8ajIgWDn", "notes": "Authenticated Google Drive predecessor to PortiaSynapse. It preserves a source-reported training plateau in a four-hypothesis, three-refinement architecture and proposes a one-path recovery protocol. The failure is valuable negative evidence but does not identify branching, refinement, memory, selector credit, label smoothing, or target geometry as the cause, and it has not been locally reproduced.", "source_type": "technical_whitepaper", "published": "2024-12-23", "citation_label": "Sorenson (2024), SpiderSynapse" }, { "id": "capability_ratchet_whitepaper", "title": "The Capability Ratchet", "priority": "supporting_lineage", "layer": "capability_ratchet", "chapter_targets": [ "benchmark-ratchets-and-anti-goodhart-evidence", "procedural-memory-and-cognitive-loop-closure", "recursive-self-improvement-boundaries", "capability-replacement-and-rollback" ], "url": "https://docs.google.com/document/d/1ZGLKgjNy2j1WxSJyEDsUIU032_jmyvtp1r4bzS7XykI", "notes": "Full authenticated connector text section-audited. Synthesizes benchmark, procedural, and structural ratchets; benchmark and tool lifecycles; an intervention ladder; interpreter/compiled/reflex runtime modes; total-cost tool compilation; and anti-Goodhart controls. Same-author synthesis, not independent evidence for Benchmaxxing, Cognitive Loop Closure, RGS, or RMI.", "source_type": "author_synthesis_whitepaper", "published": "2026-05", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "No independent benchmark campaign, tool compiler, architecture-selection study, or measured capability improvement.", "updated": "2026-07-31", "citation_label": "Sorenson (2026), The Capability Ratchet" }, { "id": "attd", "title": "Assembly-Theoretic Technical Debt: A Deterministic Outer Loop for Self-Improving Codebases", "priority": "supporting", "layer": "living_project_governance", "chapter_targets": [ "artifact-steward-agents-and-living-project-governance", "recursive-self-improvement-boundaries", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://docs.google.com/document/d/1mv69yelFkJ-deJ1BkVMZANqPQzzaERTpG108LhU1BKk", "notes": "Full authenticated connector text section-audited. Adds historical, vector-valued structural-debt governance: artifact-class separation, intrinsic assembly burden, reuse failure, role entropy, lineage, rolling residue, debt pressure, verified simplification credit, local caps, growth guards, deterministic GREEN/YELLOW/RED admission, bounded maintenance packets, abstention, and four-arm long-horizon evaluation.", "source_type": "author_governance_whitepaper", "published": "2026-03", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "Assembly theory is design inspiration; no universal debt law, analyzer, threshold calibration, causal ablation, long-horizon maintenance result, or safe self-modification claim.", "updated": "2026-07-31", "citation_label": "Sorenson (2026), Assembly-Theoretic Technical Debt" }, { "id": "orcp_moecot", "title": "ORCP\u2013MoECOT: A Governed Oscillating Rail Cascade Codec", "priority": "supporting", "layer": "deterministic_compression", "chapter_targets": [ "compact-generative-systems-and-residual-honesty" ], "url": "https://drive.google.com/file/d/1J5sd32zKhp4pBjWTRyrpjNM-8_DpHpsY", "notes": "Full authenticated technical specification section-audited. Adds decoder-boring lossless design, explicit container/header/block framing, reversible transforms, fixed-point range coding, local/match/structural prediction rails, bounded encoder planning, transmitted refinement packets, anti-experts as penalties, complete archive-rate accounting, and incompressible-input fallback.", "source_type": "author_technical_specification", "published": "2026-03-03", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "No inspected codec implementation, independent decoder, round-trip artifact, corpus result, runtime measurement, compression advantage, or production claim.", "updated": "2026-07-31", "citation_label": "Sorenson (2026), ORCP\u2013MoECOT" }, { "id": "ext_elizaos_agent_runtime_2026", "title": "elizaOS Agent Runtime and Scenario Runner", "priority": "external_literature", "layer": "modular_agent_runtime_and_evidence_qualification", "chapter_targets": [ "ai-work-surfaces-agent-harnesses-and-organizational-absorption", "runtime-adapters-tool-permissions-and-human-approval", "benchmark-ratchets-and-anti-goodhart-evidence" ], "url": "https://github.com/elizaOS/eliza/tree/1232c982c1594d2438a8a53b02e41a9ba86664aa", "notes": "Pinned official implementation comparator for modular actions, providers, evaluators, services, runtime lifecycle, scenario execution, and the explicit distinction between in-process diagnostics and externally qualified provider evidence. No elizaOS execution, test reproduction, security assessment, performance result, or support transition is imported.", "source_type": "official_software_and_documentation", "published": "2024-07-09", "updated": "2026-08-08", "citation_label": "elizaOS contributors (2026), elizaOS Agent Runtime" }, { "id": "ext_hermes_agent_2026", "title": "Hermes Agent: Learning, Memory, Tools, and Security Architecture", "priority": "external_literature", "layer": "procedural_memory_and_agent_runtime", "chapter_targets": [ "ai-work-surfaces-agent-harnesses-and-organizational-absorption", "procedural-memory-and-cognitive-loop-closure", "durable-semantic-memory-and-knowledge-lattices", "runtime-adapters-tool-permissions-and-human-approval" ], "url": "https://github.com/NousResearch/hermes-agent/tree/238351a60ce689e1c460fabf5b3f50e3b06b44bd", "notes": "Pinned official implementation comparator for progressive-disclosure skills, agent-managed procedural memory, staged skill-write approval, bounded prompt memory, session search, tool backends, command approval, and isolation. No learning, memory, security, utility, or performance result was reproduced.", "source_type": "official_software_and_documentation", "published": "2025-07-22", "updated": "2026-08-08", "citation_label": "Nous Research (2026), Hermes Agent" }, { "id": "ext_openclaw_agent_runtime_2026", "title": "OpenClaw Gateway, Agent Runtime, ACP, and Self-Learning Architecture", "priority": "external_literature", "layer": "gateway_session_harness_and_procedural_learning_runtime", "chapter_targets": [ "ai-work-surfaces-agent-harnesses-and-organizational-absorption", "runtime-adapters-tool-permissions-and-human-approval", "artifact-graphs-audit-logs-and-replay", "inter-stack-protocols-identity-and-economic-exchange", "procedural-memory-and-cognitive-loop-closure" ], "url": "https://github.com/openclaw/openclaw/tree/e4309e0cc7a74f0c82d572eb47befe78211c6be5", "notes": "Pinned official implementation comparator for gateway and device identity, serialized session runs, bounded audit projection, ACP external-harness identity and authorization boundaries, separated sandbox/tool/elevation controls, and evidence-reviewed hash-bound skill proposals. Distinct from the Claw-SWE-Bench benchmark source; no implementation result was reproduced.", "source_type": "official_software_and_documentation", "published": "2025-11-24", "updated": "2026-08-08", "citation_label": "OpenClaw contributors (2026), OpenClaw Agent Runtime" }, { "id": "ext_github_copilot_work_surfaces_2026", "title": "GitHub Copilot Product and Work-Surface Documentation", "priority": "external_literature", "layer": "ai_work_surface_evolution", "chapter_targets": [ "ai-work-surfaces-agent-harnesses-and-organizational-absorption" ], "url": "https://docs.github.com/en/copilot/get-started/what-is-github-copilot", "notes": "Official current-product comparator spanning inline suggestions, chat, command line, contextual spaces, pull-request work, and agent-driven development. No workflow, productivity, safety, or comparative result was reproduced.", "source_type": "official_product_documentation", "published": "unknown", "updated": "2026-08-08", "citation_label": "GitHub Documentation (2026), GitHub Copilot" }, { "id": "ext_augment_code_agent_2026", "title": "Augment Code Agent Documentation", "priority": "external_literature", "layer": "ide_agent_modes_and_review", "chapter_targets": [ "ai-work-surfaces-agent-harnesses-and-organizational-absorption", "runtime-adapters-tool-permissions-and-human-approval" ], "url": "https://docs.augmentcode.com/using-augment/agent", "notes": "Official comparator for the transition among chat, read-only inquiry, approval-paused agent work, and more independent agent execution with diffs and checkpoints. No product execution or control claim was reproduced.", "source_type": "official_product_documentation", "published": "unknown", "updated": "2026-08-08", "citation_label": "Augment Documentation (2026), Using Agent" }, { "id": "ext_openai_codex_work_surfaces_2026", "title": "OpenAI Codex CLI, IDE, Cloud, and Agent Documentation", "priority": "external_literature", "layer": "coding_agent_harness_and_distributed_work_surfaces", "chapter_targets": [ "ai-work-surfaces-agent-harnesses-and-organizational-absorption", "runtime-adapters-tool-permissions-and-human-approval" ], "url": "https://github.com/openai/codex/tree/dd43a9967ff19ce2b96282d2fd5cbdd2022a9b11", "notes": "Official documentation and pinned CLI comparator for repository inspection, editing, tool execution, permissions, local and cloud work, automation, and extensibility across multiple surfaces. No benchmark, correctness, safety, or productivity result was imported.", "source_type": "official_software_and_documentation", "published": "unknown", "updated": "2026-08-08", "citation_label": "OpenAI (2026), Codex Documentation and CLI" }, { "id": "ext_anthropic_claude_code_2026", "title": "Claude Code Agentic Harness Documentation", "priority": "external_literature", "layer": "agentic_harness_and_execution_loop", "chapter_targets": [ "ai-work-surfaces-agent-harnesses-and-organizational-absorption", "runtime-adapters-tool-permissions-and-human-approval" ], "url": "https://code.claude.com/docs/en/how-claude-code-works", "notes": "Official comparator that explicitly separates model from harness and describes gather-context, act, and verify loops across terminal, IDE, desktop, web, remote, and automation surfaces. No implementation result was reproduced.", "source_type": "official_product_documentation", "published": "unknown", "updated": "2026-08-08", "citation_label": "Anthropic (2026), Claude Code Documentation" }, { "id": "ext_opencode_agent_2026", "title": "OpenCode Open-Source Coding Agent", "priority": "external_literature", "layer": "open_source_coding_agent_harness", "chapter_targets": [ "ai-work-surfaces-agent-harnesses-and-organizational-absorption", "runtime-adapters-tool-permissions-and-human-approval" ], "url": "https://github.com/anomalyco/opencode/tree/38e10eb1408feb700021b8e8766fb0ab41bf84e2", "notes": "Pinned official comparator for a provider-flexible coding agent with terminal, desktop, and IDE surfaces, project instructions, plan/build modes, tool execution, recovery, and opt-in sharing. No runtime or provider-parity claim was reproduced.", "source_type": "official_software_and_documentation", "published": "unknown", "updated": "2026-08-08", "citation_label": "OpenCode contributors (2026), OpenCode" }, { "id": "ext_oh_my_pi_agent_2026", "title": "Oh My Pi Terminal Coding Agent and Tool Harness", "priority": "external_literature", "layer": "integrated_terminal_agent_harness", "chapter_targets": [ "ai-work-surfaces-agent-harnesses-and-organizational-absorption", "runtime-adapters-tool-permissions-and-human-approval" ], "url": "https://github.com/can1357/oh-my-pi/tree/896bf5f33e0b67bdd0cf951c82739a28e75d0823", "notes": "Pinned official comparator for an integrated terminal harness with hash-anchored edits, LSP, shell, browser, subagents, memory, provider switching, review, and collaboration. Reported performance or security claims were not reproduced.", "source_type": "official_software_and_documentation", "published": "unknown", "updated": "2026-08-08", "citation_label": "Oh My Pi contributors (2026), Oh My Pi" }, { "id": "ext_eggroll_hyperscale_es_2026", "title": "Evolution Strategies at the Hyperscale", "priority": "external_literature", "layer": "zeroth_order_population_learning", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling", "policy-optimization-and-learning-from-feedback", "replaceable-cognitive-substrates-beyond-transformer-monoculture", "resource-economics-and-token-budgets", "learning-theory-generalization-and-scaling-science" ], "url": "https://eshyperscale.github.io/", "notes": "Primary EGGROLL project and paper source for low-rank, batched evolution strategies, counter-based perturbation reconstruction, nondifferentiable and discrete objectives, recurrent/int8 training, and outcome-reward fine-tuning. Throughput, quality, and theory claims are source-scoped; total population evaluations and GPU-hours remain required denominators.", "source_type": "arxiv_and_official_project", "arxiv_id": "2511.16652", "published": "2026", "updated": "2026-08-09", "citation_label": "Rybkin et al. (2026), Evolution Strategies at the Hyperscale", "doi": "10.48550/arXiv.2511.16652" }, { "id": "ext_openai_es_2017", "title": "Evolution Strategies as a Scalable Alternative to Reinforcement Learning", "priority": "external_literature", "layer": "evolution_strategies_and_black_box_policy_search", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling", "policy-optimization-and-learning-from-feedback", "resource-economics-and-token-budgets" ], "url": "https://arxiv.org/abs/1703.03864", "notes": "Foundational modern large-population ES comparator using parameter perturbations, scalar fitness, seed reconstruction, and distributed evaluation. Source-reported MuJoCo/Atari results and worker scaling do not establish universal sample or total-compute efficiency.", "source_type": "arxiv", "arxiv_id": "1703.03864", "published": "2017-03-09", "updated": "2026-08-09", "citation_label": "Salimans et al. (2017), Evolution Strategies as a Scalable Alternative to Reinforcement Learning", "doi": "10.48550/arXiv.1703.03864" }, { "id": "ext_mezo_2023", "title": "Fine-Tuning Language Models with Just Forward Passes", "priority": "external_literature", "layer": "memory_efficient_zeroth_order_fine_tuning", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling", "replaceable-cognitive-substrates-beyond-transformer-monoculture", "resource-economics-and-token-budgets" ], "url": "https://arxiv.org/abs/2305.17333", "notes": "Primary MeZO source for inference-footprint zeroth-order language-model fine-tuning and nondifferentiable objectives. Reported memory and GPU-hour savings are configuration-bound and do not erase objective-query count or estimator variance.", "source_type": "arxiv", "arxiv_id": "2305.17333", "published": "2023-05-27", "updated": "2026-08-09", "citation_label": "Malladi et al. (2023), Fine-Tuning Language Models with Just Forward Passes", "doi": "10.48550/arXiv.2305.17333" }, { "id": "ext_forward_forward_2022", "title": "The Forward-Forward Algorithm: Some Preliminary Investigations", "priority": "external_literature", "layer": "local_forward_only_credit_assignment", "chapter_targets": [ "governed-model-training-distributed-optimization-and-scaling", "replaceable-cognitive-substrates-beyond-transformer-monoculture", "learning-theory-generalization-and-scaling-science" ], "url": "https://arxiv.org/abs/2212.13345", "notes": "Primary preliminary source for positive/negative forward passes and local layer objectives as an alternative to reverse-mode backpropagation. The evidence is small-scale and does not establish foundation-model parity or biological plausibility.", "source_type": "arxiv", "arxiv_id": "2212.13345", "published": "2022-12-27", "updated": "2026-08-09", "citation_label": "Hinton (2022), The Forward-Forward Algorithm", "doi": "10.48550/arXiv.2212.13345" }, { "id": "regret_engine", "title": "The Regret Engine: Governed Counterfactual Learning Signals for Continual Adaptation, Prospective Risk Control, and Self-Correction in Artificial Agents", "priority": "must_use", "layer": "governed_counterfactual_learning_and_self_correction", "chapter_targets": [ "planning-as-a-control-layer", "policy-optimization-and-learning-from-feedback", "procedural-memory-and-cognitive-loop-closure", "data-engines-continual-learning-and-unlearning", "artifact-graphs-audit-logs-and-replay", "claim-ledgers-and-belief-revision", "readiness-gates-residual-escrow-and-quarantine", "governed-operations-incident-command-and-graceful-degradation", "benchmark-ratchets-and-anti-goodhart-evidence", "integrated-reference-architecture" ], "url": "local-source:sources/raw/corben_papers/regret_engine/regret_engine.md@sha256:6d609442cdbd2725e0109544f000e5255ff6850ce766efa87800fa63a74d1623", "notes": "Corben-authored August 2026 conceptual architecture and research program for decision-time-fair Governed Counterfactual Regret, immutable Decision Capsules, admissible comparator contracts, sparse Regret Tensors, append-only Regret Packets, prospective regret control, regret-aware replay, regret-to-rule compilation, three update clocks, root-cause adjudication, and bounded update leases. Existing chapters are upgraded first; no implementation, experiment, reproduction, causal-identification result, formal proof, safety result, support transition, SOTA, AGI, or ASI is inferred.", "source_type": "author_whitepaper", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "The bibliography and Markdown figure companions were not supplied; the DOCX embeds its visual material. All propositions, algorithms, experiments, and architecture claims remain proposed rather than independently validated.", "published": "2026-08-04", "updated": "2026-08-04", "citation_label": "Sorenson (2026), The Regret Engine" }, { "id": "ext_pbt_2017", "title": "Population Based Training of Neural Networks", "priority": "external_literature", "layer": "population_based_adaptive_training", "chapter_targets": [ "learning-compute-topology-and-adaptive-process-architecture", "governed-model-training-distributed-optimization-and-scaling", "open-ended-improvement-engines" ], "url": "https://arxiv.org/abs/1711.09846", "notes": "Primary Population Based Training source for asynchronous joint optimization of a population's model parameters and hyperparameter schedules through evaluation, exploitation, and exploration. Source-reported reinforcement-learning, translation, and GAN results remain task- and implementation-bound and do not validate LCT, universal topology adaptation, safety, or superior total lifecycle cost.", "source_type": "arxiv", "arxiv_id": "1711.09846", "published": "2017-11-27", "updated": "2026-08-09", "citation_label": "Jaderberg et al. (2017), Population Based Training of Neural Networks", "doi": "10.48550/arXiv.1711.09846" }, { "id": "learning_compute_topology", "title": "Learning\u2013Compute Topology: Formalizing the Causal Organization of Adaptive Systems", "priority": "must_use", "layer": "adaptive_process_architecture_and_learning_topology", "chapter_targets": [ "learning-compute-topology-and-adaptive-process-architecture", "governed-model-training-distributed-optimization-and-scaling", "replaceable-cognitive-substrates-beyond-transformer-monoculture", "routing-heads-and-specialist-cores", "policy-optimization-and-learning-from-feedback", "data-engines-continual-learning-and-unlearning", "open-ended-improvement-engines", "resource-economics-and-token-budgets", "multi-agent-dynamics-collective-intelligence-and-systemic-risk", "adversarial-evaluation-sandbagging-and-training-time-deception", "integrated-reference-architecture" ], "url": "local-source:sources/raw/corben_papers/learning_compute_topology/package_v1_0/paper/Learning_Compute_Topology.md@sha256:4fd869f01f4c0782e6e2f1d657b6c84e87ced82e652126aae395123a04fac712", "notes": "Corben-authored August 2026 research paper and executable preparation package that separates model architecture, learning-process topology, execution topology, and physical compute topology. It contributes adaptive-identity tests; typed evidence, judgement, credit, state, artifact, control, and authority relations; LCT-IR; Learning Causal Normal Form; seven bounded propositions; topology metrics; a semantic compiler firewall; Adaptive Branch\u2013Validate\u2013Integrate; toy and analytical phase diagrams; and an explicit falsification program. The bundled reference implementation passes 11 unit tests, but implements only bounded conformance behavior and does not establish neural-training benefit, causal completeness, universal canonicality, safety, scaling superiority, or ASI.", "source_type": "author_research_paper", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "The formal propositions hold only under their stated finite, explicit-state, interface-sufficiency, information-theoretic, and cut-capacity assumptions. The executable supplement covers a bounded IR/validator/normalizer/compiler/simulator slice; the phase diagrams are toy or analytical, the ABVI topology is proposed, and the novelty matrix is a scoped comparison rather than a global novelty proof.", "published": "2026-08-09", "updated": "2026-08-09", "citation_label": "Sorenson (2026), Learning\u2013Compute Topology" }, { "id": "assurance_shift_learning", "title": "When Success Stops Teaching: Assurance-Shift Learning and Governed Residual Boundary Learning for Mature AI Systems", "priority": "must_use", "layer": "competence_dependent_assurance_shift_and_residual_boundary_learning", "chapter_targets": [ "learning-compute-topology-and-adaptive-process-architecture", "policy-optimization-and-learning-from-feedback", "data-engines-continual-learning-and-unlearning", "adversarial-evaluation-sandbagging-and-training-time-deception", "benchmark-ratchets-and-anti-goodhart-evidence", "stable-capability-fields", "readiness-gates-residual-escrow-and-quarantine", "procedural-memory-and-cognitive-loop-closure", "governed-operations-incident-command-and-graceful-degradation", "integrated-reference-architecture", "evidence-states-and-claim-discipline", "artifact-graphs-audit-logs-and-replay", "resource-economics-and-token-budgets" ], "url": "local-source:sources/raw/corben_papers/assurance_shift_learning/package_v1_0/When_Success_Stops_Teaching_Corben_Sorenson.md@sha256:611787d91fb3035ff6948fd54b5391880da38f2008a85a6c8f89bb11da9fac40", "notes": "Corben-authored August 2026 conceptual systems paper and experimental specification for competence-dependent Assurance-Shift Learning and Governed Residual Boundary Learning. It contributes the Qualified Competence Envelope, frontier-mode allocation, selection-gap diagnosis, informative exceptions, outcome/process separation, Boundary Evidence Bundles, evaluator-first repair, natural/probe separation, learner-relative negative half-life, least-invasive repair placement, repair compatibility, two adaptation clocks, explicit assurance metrics, and SaturationShiftBench. The package includes a bibliography, four figures, and a verified byte manifest; no implementation, benchmark, empirical crossover, independently checked proof, safety result, resource advantage, novelty result, support transition, SOTA, AGI, or ASI is inferred.", "source_type": "author_research_paper", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "The complete author manuscript and supplied package were reviewed. Its propositions, conjectures, algorithms, benchmark, architecture, and roadmap remain conceptual or bounded as labeled. Recent external citations were verified only at metadata and abstract level during intake and require full-paper source notes before their results can support book claims.", "published": "2026-08-11", "updated": "2026-08-11", "citation_label": "Sorenson (2026), When Success Stops Teaching" }, { "id": "adjudicated_persistence", "title": "Adjudicated Persistence: Governing the Transition from Experience to Durable Structure in Adaptive Systems", "priority": "must_use", "layer": "governed_cross_surface_persistence_and_adaptive_commit_boundary", "chapter_targets": [ "adjudicated-persistence-and-the-adaptive-commit-boundary", "evidence-states-and-claim-discipline", "stable-capability-fields", "recursive-self-improvement-boundaries", "cognitive-compilation-and-semantic-ir", "durable-semantic-memory-and-knowledge-lattices", "artifact-graphs-audit-logs-and-replay", "human-ai-organizations-delegation-and-accountability", "procedural-memory-and-cognitive-loop-closure", "readiness-gates-residual-escrow-and-quarantine", "benchmark-ratchets-and-anti-goodhart-evidence", "governed-operations-incident-command-and-graceful-degradation", "policy-optimization-and-learning-from-feedback", "data-engines-continual-learning-and-unlearning", "resource-economics-and-token-budgets", "integrated-reference-architecture" ], "url": "local-source:sources/raw/corben_papers/adjudicated_persistence/package_v1_0/Adjudicated_Persistence_Corben_Sorenson_v1.0.md@sha256:95db428f8c948cf7caa8e39633176a0d3673433e6fd0b2f015f822e2f5e4e889", "notes": "Corben-authored August 2026 conceptual systems paper and experimental specification for governing how experience becomes durable causal influence. It contributes the Adaptive Commit Boundary; six-object separation of experience, lesson, disposition, realization, qualification, and authority; learning eligibility; Cross-Surface Adaptation Assignment; multidimensional commitment profiles; Evidence-Commitment Matching; Minimum Sufficient Persistence; guarded compilation and deoptimization; transactional promotion and material-change invalidation; counterfactual observability and deliberation reserve; adaptation debt; non-self-ratifying meta-compilation; bounded propositions and conjectures; and the proposed LocusBench benchmark. No implementation, LocusBench result, validated placement advantage, independently checked proof, safety result, resource advantage, novelty result, support transition, SOTA, AGI, or ASI is inferred.", "source_type": "author_research_paper", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "The complete manuscript and supplied package were reviewed. Its propositions, conjectures, algorithms, architecture, and benchmark remain conceptual or bounded as labeled. Results cited by the paper are not inherited as book evidence and require their own source-note review.", "published": "2026-08-11", "updated": "2026-08-11", "citation_label": "Sorenson (2026), Adjudicated Persistence" }, { "id": "forward_transfer_program_synthesis", "title": "From Compression to Forward Transfer: Evaluating Reusable Knowledge in Program Synthesis", "priority": "must_use", "layer": "verified_forward_transfer_and_reusable_knowledge_evaluation", "chapter_targets": [ "procedural-memory-and-cognitive-loop-closure", "benchmark-ratchets-and-anti-goodhart-evidence", "resource-economics-and-token-budgets", "executable-specifications-and-lean-proof-envelope", "rankfold-neuralfold-and-artifact-compression", "adjudicated-persistence-and-the-adaptive-commit-boundary", "recursive-self-improvement-boundaries", "learning-theory-generalization-and-scaling-science" ], "url": "local-source:sources/raw/corben_papers/forward_transfer_program_synthesis/from_compression_to_forward_transfer.md@sha256:70c4e62eedad4883db9cc8a8740c71176b25f8a95ce61742fc559b9c9ab9e447", "notes": "Corben-authored August 2026 framework and experimental blueprint for evaluating reusable symbolic knowledge through verified forward-transfer interventions. It separates retrospective and prospective compression, behavioral reuse, operational necessity, and marginal transfer; defines an R0-R7 reuse ladder, matched placebo/removal/factorial controls, versioned evaluation rounds, exact verifier outcomes, full lifecycle cost and break-even accounting, and comparator-network and bit-vector protocols. No experiment was run and no measured transfer, library superiority, safety, novelty, support transition, SOTA, AGI, or ASI is inferred.", "source_type": "author_research_paper", "passage_review_state": "reviewed", "source_crosswalk_review_state": "reviewed", "passage_review_limits": "The complete Markdown manuscript, supplied DOCX, and 31-entry bibliography were reviewed. The paper reports definitions, counterexamples, hypotheses, protocols, and a reporting schema rather than empirical results. Its cited external results are not inherited as book evidence and require their own full source-note review.", "published": "2026-08-14", "updated": "2026-08-14", "citation_label": "Sorenson (2026), From Compression to Forward Transfer" } ]