Three years, in public

I started this journal at the bottom of a hill in San Francisco and I am publishing it from a kitchen table in South Carolina, which is about the right distance for three years.

Here is what is on the page. The research entries are the science, written plainly, with an evidence label on every one so you can tell a dataset from a design from a hypothesis without reading the fine print. The journal entries, like this one, are me. They are dated by when the thing happened, they are in my own words, and they are lightly edited from what I actually typed at the time, because the version of me that was standing in ice rain is a better witness than the version writing this.

Three rules I kept, and will keep.

Date everything. A record without a date is a story. A story is fine, but this is not a story site.

Label the claim. Atom, signal, pattern, or hypothesis. Evidence, hypothesis, simulation, or observation design. If a number came out of a model, it says so. If a number was measured with an uncalibrated microphone, it says that too.

Keep the record when the thing is gone. Buildspace closed in 2024 and the dissertation about it is stronger for having watched that happen. StormNet was retired as a name last week and every entry that used the name still uses it.

What I want from the next year is small and specific. A handful of people running a phone through a storm, anywhere, so that the next dataset has more than one address on it. The open-source platform out the door so those people can do it without me. And the Lab doing what it was named to do in April 2025, which is to be the place every research thread passes through on its way to a paper, a dataset, or a release.

If you got here from a search, welcome. The journal is in order. The first entry is about why a sidewalk. It still is.