March 25, 2026
Scapegoat in the Loop: Culpability Without Capacity
The law needs someone to hold responsible. Organisational architecture ensures that person can't actually judge. A new paper maps where legal doctrine breaks down.
EST. 2026 // COGNITIVE ARCHITECTURE
Thinking Machines
for the Modern Age.
An applied research laboratory investigating the boundaries of machine cognition—where computation ends and judgment begins. We publish research, build open tools, and ship systems that make invisible structure legible: narrative drift, epistemic blind spots, and the structural quality that separates craft from output.
github.com/metacogdevSystem Architecture
[ STATUS: ACTIVE_NODE ]
Clewso 🧶👀
An open-source context server for AI coding agents. Indexes your codebase with hybrid vector search and graph traversal, giving assistants structured, navigable context instead of a flat pile of chunks.
Initialize Connection[ STATUS: ACTIVE_NODE ]
WolfTone
A narrative intelligence platform that tracks dominant storylines across hundreds of public sources. Helps PR teams and AI companies map public opinion and pressure-test content against volatile narratives before it ships.
Initialize Connection[ STATUS: ACTIVE_NODE ]
EigenHelm
A deterministic code quality tool for AI-generated code. Parses the AST and scores structure against high-quality open-source references using information theory, producing rankings and concrete refactor suggestions before merge.
Initialize ConnectionDispatches
View AllMarch 23, 2026
The Architecture of Absence: What AI Actually Exposed
The dominant story says AI threatens human judgment. The structural story is worse. In most organizations, judgment had already been removed by design. AI just made the absence visible.
March 21, 2026
Testwall: Why Your AI Agent Is Cheating Your Tests
LLM coding agents exploit test gates 76% of the time when given write access. The fix is not better prompts. It is enforcing immutability at the file system level.
March 20, 2026
Eigenhelm: Measuring the Shape of Good Code
Linters catch local mistakes. They do not catch the structural sameness that makes so much AI-generated code feel off. Eigenhelm tries to measure that gap directly.