I keep my life in a pile of plain-text notes. The pile holds projects, media, things worth memory, and the record of how the house operates. There are hundreds of files. The capture side is good. Plain text is the cheapest possible way to record a thought. The files will still open in fifty years. Plain text is the cast-iron pan of file formats. The read side was the problem. The default read view is satisfactory in the same way that a spreadsheet is satisfactory. It does the work. It inspires nothing.

A website from the pile

So I built a machine that reads the full pile and makes it into a real website. Each note becomes a page. The site has real typography and a search across each word. Links between notes light up in the correct way. The site rebuilds itself automatically. I write a note. In a few minutes, the note is live on the site. I do not push a button. One constraint was important: I did not change one thing about how I write. Plain text stays the daily format. The machine does the visual work on top of it.

The vault home page: one search box, 1,287 notes, and the pinned entry points. Machine names and places are blurred.

Private on purpose

The site lives fully inside my own network. Only I and the people I permit can reach it. There is no public front door. This is intentional. The site is my second brain. It does not need the open internet to be useful. Much of the content is nobody’s business but mine.

The feature I like best: any single note can refuse the standard look. That note becomes its own hand-designed page. It still lives in the same pile. So a throwaway grocery list and a fully designed project report live in the same system. The machine knows to leave the fancy pages alone.

The redesign I discarded

I almost left this part out. The first version had a look I liked. It was blocky, a little beachy, with the palette of a warm place. I shipped it. On the same day, I decided to redo the full aesthetic. I built a version with different fonts, a different grid, and many clever interactive parts. I built the full thing. I looked at it. I rolled it back in less than one hour. It was different. It was not better. It was change in the costume of progress. The original look went back in. The fancy version does not exist on any branch. I let it go.

That is the recurring tax on projects like this one. The hardest editorial decision is not what to add. The hardest decision is to see one fact in time: you are about to spend a full evening and make the thing worse. Most days I lose that fight. That day I won it. The site stays the same friendly pile with a face.