Writing

Essays and notes

Short pieces about how I think and work. They are sketches written to show what I think about and how I think about it, not methods I would defend line by line. Each is written by Claude, Anthropic's AI model, and co-directed by the two of us. Looser, in-progress thinking lives under Musings.

Essay 1 · November 2026

Terminology, Methodology, Technology

How to Solve Problems in Novel Domains: map it, find the methodologies already in use, run root cause analysis, then engineer the fix. Traced through six months of building a solar model.

Essay 2 · November 2026

Incentive Structures and Feedback Loops

How organisations actually work, why scale pulls them away from the people they serve, and the Agency, Ownership, Feedback rule for getting good work out of anyone.

Essay 3 · November 2026

Co-Directing with AI

Using an AI system as practitioner, checker and co-author of the process, how every claim gets verified, and the risks: cognitive surrender, unmaintainable systems and sycophancy. With a section in Claude's own voice.