Main idea

A network of atomic notes linked to one another propagates what it contains. A false note contaminates its neighbors, an out-of-date nuance ends up cited in ten places, and noise circulates as readily as everything else. This flaw is either disqualifying or harmless depending on the material you put in.

On a product opportunity — a customer need, a market on the move, a positioning hypothesis — the information holds for one file and not for the next. There, the network propagates errors faster than it consolidates anything at all.

On the "by the book" knowledge of a trade, the balance reverses. The NF, ISO and EN standards of industrial maintenance, regulations, technical vocabulary, processes documented for decades: this information is stable, verifiable against a reference text, and does not change value depending on which case you bring it into. The same setup that fails on one succeeds on the other, and the difference does not come from the method.

Why it matters

This shifts a badly posed question. "Does the second brain work?" has no general answer: it depends on the life expectancy of what you file in it. Before building a base, the useful sorting is therefore not thematic but temporal — what, in this domain, will still be true on the next case?

It also explains why the same practitioner can have abandoned a method on one terrain and taken it up again on another without contradicting themselves.

Nuances and limits

A standard gets revised, and a "by the book" base then ages all at once, across every note that depends on it — stability is not immobility.

And the boundary is not that of the official text: an internal business rule that was never standardized can be perfectly stable, while a published "best practice" remains debatable.

Open questions

• By what sign do you recognize that knowledge you believed stable in fact depended on the context of the case where you met it?