Why I deleted my second brain

I spent two years building a linked note system. It was beautiful. I never once used it to find something I did not already remember.

The graph flatters you

A dense graph looks like knowledge. What it actually measures is how much time you spent making connections, which is not the same as how much you understood.

A note you cannot find is not a note. It is a diary entry about having had a thought.

Folders have edges

A folder forces a decision: does this belong here or somewhere else? That decision is the filing. Links defer the decision forever, which feels generous and turns out to be the problem.

What I actually lost

Almost nothing, which was the uncomfortable part. I exported everything, moved it into dated folders, and in the following year I went looking for a specific old note about eleven times. Every one of those times I found it with full-text search in under ten seconds.

Search was doing the job I believed the link graph was doing. The graph was doing a different job: making me feel like the reading had been productive.

The productivity of feeling productive

Building the system is enormously satisfying. It has all the texture of real work — decisions, structure, visible progress — without the risk of being wrong in public. That is exactly what makes it dangerous.

Any activity that feels like thinking and cannot fail is probably not thinking.

The test I use now: has this note changed anything I did? Not “could it”. Did it. Notes that pass are rare, and they are almost always short and written in my own words rather than quoted.

What I kept

• A single file per project, appended to, never reorganised
• A reading log with one sentence per book about what I disagreed with
• Dated folders, because a date is a fact and a category is an opinion

The reading log is the only part that surprised me. Recording disagreement rather than summary forces a position, and a position is the thing that turns out to be worth having later.