Main idea

Tooling questions are inexhaustible: which app for web pages, which for quick notes, which setup for PDFs, which automation from the phone, which filing scheme, which tags. They are sometimes useful. Above all they become an elegant way to avoid the work — you can spend more time improving the capture than transforming what you capture.

The criterion that settles it is therefore not "does my system cover every case?" but "does it help me produce knowledge?". An incomplete system that genuinely feeds the writing is worth more than a complete system you maintain.

And the consequence is unusual: an elegant system can be too heavy. Elegance doesn't redeem the cost of upkeep, it only makes it more pleasant to pay.

Why it matters

This gives a way out of a loop that has none of its own. As long as the criterion is completeness, every hole discovered justifies an improvement, and there will always be a hole. The production criterion, by contrast, saturates: past a certain point, adding produces nothing more.

It also reorients the diagnosis. When nothing comes out of an intelligence system, the spontaneous answer is to change tools; the first question to ask is whether the tool was the cause.

Nuances and limits

A tool can genuinely be responsible for a failure — a setup that requires being installed misses the sources you meet on the move, and no production criterion will correct that.

And "producing knowledge" isn't measured day by day. The criterion is good at the scale of a few months, not of one thin week.

Open questions

• How do you recognize, before losing months, that improving the tooling has become an escape?