How much order does knowledge need?

The running experiment

Knowledge needs more than filing. A search function finds words, but it does not explain why a note was kept, what it hangs from, or whether I would still stand behind its sentence today. Too little order turns an archive into a digital attic.

Too much order is the other form of the same problem. Then I no longer spend time with the material, but with its system: cleaning up categories, carrying metadata forward, defending folders, explaining exceptions. That feels like maintenance. Sometimes it is only work produced by the order itself.

This website is a running test of a middle answer. The vault stays Markdown. Publication hangs on one field: public: true. The website does not read the vault, but an exported projection. If the website disappears, no knowledge is lost. If the vault is not available, the website should still be buildable from the export.

What has not been proven yet

That sounds reasonable while the collection is small and the rules are fresh. After a year, more is visible. Then it becomes clear whether links still help, or whether they are only traces of earlier attention. It also becomes clear whether status fields really describe a position, or whether they have to be maintained afterwards so the system looks orderly.

The question of visible mechanism belongs here. I want to be able to see which decision makes a note public, which links disappear during export, and which traces remain. Precisely a private reference like a private runway plan tests this boundary: publicly, only the visible text remains, not the internal record behind it.

What I would measure it by

The usable point is probably movable. Today, a flat structure with a few types, explicit publication and checkable boundaries is enough. Later, that may be too little.

Then the order should be allowed to grow. But it has to explain which friction it removes. Otherwise, it becomes the object.