Bookshelf

Systems over goals, and the habit is getting into the cab rather than the gym. It became a habit tracker in 2019 and a habits database in 2026. What it does not cover is the threshold: environment design alone never got me over the starting line.

A mental tool is a procedure for a situation, and it only exists once it is written down. That became a playbook in 2019, another in 2020, and a routing table of fifty-nine protocols in 2026.

The brain is for having ideas, not storing them. It became a single capture queue and a weekly review.

Without metrics you drift toward visible busyness. By 2026 my version was that deep work comes from environment design, not from pushing.

The voice target for my technical writing, and a digest of all forty-nine concepts.

Problem, discussion, solution, in full sentences. The shape I want a reference to have.

Recognizing that a thing I was writing had a math book's structure is what organized it.

The reference for deciding that what I was writing was a field manual and not a book.

It keeps catching me designing things that are not necessary. The nuance I added in 2026 is that human formalization still tames problems brute search cannot.

I was not ready for it the first time. It works after about five years of programming, and it is my best argument for reading a book twice.

What stayed by 2025 is the plain message: stay in constant contact with reality and gather real data. That phrase, reality contact, runs most of what I do and is the name of a company.

A procedural field manual for extracting information from the world. It became thirty-plus interviews that autumn and the structure my agents use.

A startup is a search for a repeatable business model. The first chapters are the useful ones.

Driving with constant feedback from the road, not planning a rocket launch.

Read in French at seventeen. Eleven years later the same idea held: the fear points at the thing you know you must do.

The book that made me feel understood in 2024.

I made my own annotated edition over about ten hours. The wheelwright story is the source of a distinction I use constantly: some knowledge only comes from contact.

Reading it at eighteen made me want to write something like it. Fourteen years of journaling followed.

At seventeen I set myself a two-week trial of satisfying only natural and necessary desires and promised to report. The bounded trial with a report is still the unit I use.

A leader is like a parent. It landed when a team's sense of why had gone fuzzy while I was not present enough.

It made me want to work with the evolutionary architecture of my own cognition rather than against it.

It engaged my mind in a way I had missed, and taught me that without an external force I would not learn on my own.

A method for creating and evolving reframes over a long period, which is different from reading a book once.

On Lisp informed the bottom-up design of my own language in 2024.

Re-read with notes on the commute. Mastery requires emotional mastery rather than intellect.

A taxonomy of patterns. It became my example that a taxonomy is not a theory.

I outgrew it rather than disliked it.

It is not the way I learn.

It was not what I was looking for.

A philosophy that keeps having to correct its own mischaracterization has too few error-correcting codes.

It has been on the list since 2024.

I started it as an audiobook.

I started it.

It is in progress, with a model as tutor.