---

LLM-generated skills work, if you generate them afterwards

February 17, 2026 │ ai, steering

Two different tricks for fast LLM inference

February 15, 2026 │ ai

On screwing up

February 11, 2026 │ good engineers, tech companies

Large tech companies don't need heroes

February 8, 2026 │ tech companies

Getting the main thing right

February 5, 2026 │ good engineers, tech companies

How does AI impact skill formation?

January 31, 2026 │ ai, papers

You have to know how to drive the car

January 26, 2026 │ tech companies, good engineers, cynicism

How I estimate work as a staff software engineer

January 24, 2026 │ tech companies, how to, large codebases

I'm addicted to being useful

January 20, 2026 │ good engineers, tech companies

Crypto grifters are recruiting open-source AI developers

January 17, 2026 │ ai, crypto

The Dictator's Handbook and the politics of technical competence

January 5, 2026 │ book reports

2025 was an excellent year for this blog

January 3, 2026 │ yearly wrapup, meta

Grok is enabling mass sexual harassment on Twitter

January 2, 2026 │ ai, ethics, alignment failures, policy, model personality, ai safety

Software engineers should be a little bit cynical

December 28, 2025 │ tech companies, cynicism

You can't design software you don't work on

December 27, 2025 │ software design, tech companies, large codebases