Blog¶

Writing

Causal analysis you can argue with.

Posts are structured around variables, counterfactuals, mechanisms, and interventions — with explicit uncertainty and falsification ideas.

Post	Notes
Why LLMs Fail at Knowledge Graph Extraction	Asserted graphs, identity resolution, constrained augmentation, provenance
Temporal Memory Beats Bigger Context	Temporal validity, supersession, hybrid retrieval, governance
Mechanism-Executable Causal GraphRAG	Executable clauses, EMCT primitives, governance loop, interventions
HRM’s Latent Reasoning Still Needs Verification	Adaptive depth, latent reasoning, harness verification, deployment risk
Agent Memory Needs Identity, Governance, and Decay	Identity, temporal validity, provenance, decay, governance
AI Coding Assistants: A Causal Framework for Productivity and Risk	Harness vs model, tests, measurement, Goodhart loops
From SWOT to Testable Causal Hypotheses	Translate narrative into falsifiable mechanisms + governance
Why Small Frictions Cascade into System Failure	Mechanisms, feedback loops, leverage points

A quick map of the structure: model → diagrams → evidence → interventions.

Curated entry points that scale better than a huge navigation tree.

If you want the infrastructure behind the writing, start here.