Main idea

Saving an article, setting a newsletter aside, sending yourself a link: none of these gestures produces knowledge. Each preserves a possibility — the source may one day feed a note, a text, a decision.

Three levels separate the possibility from the result: capture preserves, distillation transforms, action produces. Value starts at the second. A setup that stops at the first hasn't built capital, it has accumulated.

The gap is easy to miss because the two gestures look alike from outside: in both cases there is one more file. But the source that was never picked up again is exactly what it was at the moment you met it, whereas knowledge presupposes that you have rephrased it, connected it, set it against what you already knew.

Why it matters

This gives the right indicator for an intelligence or reading system. It isn't the number of sources coming in, it's the number of sources coming out transformed. A To read box that keeps growing describes a system that is failing, not one that is working.

It also explains why the feeling of learning can be entirely uncorrelated with learning. Capturing gives the satisfaction of a completed gesture without its return.

Nuances and limits

Some captures aren't meant to be distilled: keeping a reference so you can cite it, or a source to preserve traceability, is a legitimate documentary use that demands no transformation.

And picking something up again isn't always explicit: a source read attentively may already have done its work without leaving a note, even if that is rarer than we like to think.

Open questions

• Can you measure a capture system's pickup rate without instrumenting every source?