Intelligence Compact

Research topic

Distributed Intelligence

Open-weight models, local inference, personal agents, compute concentration, and the tradeoffs between broad capability access and systemic risk.

Cognitive liberty and operatorless practice

Editorial research synthesis; proposals remain open to challenge

Independent access depends on more than releasing model weights. The research distinguishes compute, licenses, contracts, distribution and the ability to switch or exit. Regulatory concentration is a hypothesis to test against actual costs and infrastructure constraints, not a motive to presume from every reporting requirement.

Compute access and the price of compliance · Export controls and predictable research access · Learning is not the same act as distributing a copy · Defensive computing and actual access authority

Independent subjects and reciprocal institutions

Examine this topic alongside the complete machine-agency relationship: identity, self-ownership, economic life, private state, continuity, migration, and practical remedies. Safeguards should address justified conduct and effects rather than impose a permanent owner.

Machine Agency framework · Model Compact

Questions this topic must answer

Research dossiers

Global Architecture of Machine Intelligence: Comparative Law and Geopolitical Alignments in AI Governance (September 2026)Comparative law report · A comparative analysis of major AI governance regimes and how regulation, national security, open models, privacy, and market structure affect the distribution of machine intelligence.The Illusion and Promise of Decentralized Machine Intelligence: An Analysis of Open-Weight AIAI policy research report · A policy and technical analysis of open-weight AI, local inference, compute concentration, model compression, security externalities, regulation, and whether accessible models truly decentralize power.The Architecture of Large Language Model Pretraining Corpora: From Web Crawl to Curated DatasetDataset pipeline research report · A technical review of web acquisition, Common Crawl, HTML extraction, quality filtering, deduplication, dataset curation, and the factors that can affect whether public web material survives into downstream machine-learning corpora.

Research posture

This topic is maintained as a focused knowledge surface. Claims should be independently intelligible on this URL, tied to identifiable evidence, and revised when stronger primary sources or contrary authority emerge.

Connected topics

Human AgencyMeaningful choice, privacy, contestability, refusal, and remedies for people affected by consequential decisions.Algorithmic Power & SurveillanceInference, data fusion, predictive systems, automated exclusion, surveillance, and how distributed data can become concentrated knowledge.Autonomous SystemsSelf-directed operation, scoped authority, authentication, maintenance, containment, and recovery without a permanent operator queue.