Main idea

A page that assembles notes has two ways of ceasing to be one, and they are symmetrical. Downwards, it starts developing for its own sake one of the ideas it links: the paragraph stretches, the argument settles in, and what is growing there is a full note lodged in the wrong place — it has to be extracted. Upwards, it shrinks to an enumeration of pointers: nothing explains any more what those notes do together, and what you get is a navigation table, useful for moving around, incapable of making anything understood.

Between the two, the band is narrow: enough text to state what the set gives you to understand, not enough to redo the work of the assembled notes.

The two drifts are not corrected the same way. The first is corrected by extraction — the idea goes off to live elsewhere. The second is corrected by writing — you have to produce the synthesis that was missing, or give up the page.

Why it matters

This gives two observable signals instead of a quality judgement. You look at whether a passage could be read alone, outside the page: if so, it is turning into a self-contained idea. You look at whether, with the links removed, anything is still being said: if not, there was no synthesis.

It also avoids the reflex correction. Faced with an overlong page, you cut; but cutting a synthesis that goes deep destroys the idea instead of moving it, and cutting a list of links merely makes it shorter.

Nuances and limits

The band is not equally wide everywhere. An entry page meant to direct owns being closer to a navigation table; a page preparing an article accepts going deeper, since the development will soon move into the text.

And the passage from one drift to the other is not linear: the same page can go deep on an idea in one paragraph and be nothing but a list of pointers in the next.

Open questions

• Should an idea extracted from a synthesis leave a trace of its passage there, or only a link?