Why We Started With Micro Machines
Micro Machines were not a novelty wedge. They were the right first world for a system built around canon, missingness, and continuity.
Notes from building live systems, and from building our own house around collecting, provenance, and demand. Some are technical. Some are philosophical. All of them are attempts to make the work and the instinct behind it more legible.
Micro Machines were not a novelty wedge. They were the right first world for a system built around canon, missingness, and continuity.
Trying to make six government APIs agree taught me that institutions reveal themselves most clearly in their identifiers, omissions, and assumptions.
The boring choice is often the one that leaves fewer regrets six months later. These are the deliberately boring bets we keep making, and why.
I do not think clarity is a soft cultural virtue. I think it is infrastructure, and I think it compounds.
The domains vary. The instinct does not: take something ambiguous, structure it, and make it trustworthy enough to act on.
This started as a weird question and became a real experiment in persona, interface, and the difference between generic AI and situated judgment.
Code running is not the same thing as a system holding. Most failures happen in the space between those two facts.
The portfolio only looks incoherent if you look at the domains instead of the instinct behind them.
I do not name things for vibes. The names are meant to constrain the work, not decorate it.
We did not choose a Lorenz attractor because it looked cool. We chose it because it says something true about the kind of systems we work on.
I think writing is one of the main ways a system resists becoming oral tradition.
I wanted a prompt that would resist flattering the user and tell the truth instead, even when the flattering version would read better.