Main idea

Faced with an almost finished book about a method built over months, the natural slope is to collect whatever confirms that it is good, different or original. That collection always succeeds: there is always something to find that goes the way you hope.

The opposite operation consists of stating the implicit hypothesis — "maybe this part is new" — then methodically looking for what could take it apart. When a counter-example appears, you do not argue with it: you revise the hypothesis. The claimed territory shrinks with each iteration, first the building blocks, then certain assemblies, then the idea of making everyday work a mechanism for capitalization, then part of the grammar itself.

The result looks counter-intuitive: as the claim narrows, the proposition becomes clearer. What remains is what resisted.

Why it matters

This gives an operational definition of rigour that depends on no discipline: the effort goes into the counter-example, not into the proof. A hypothesis nobody has tried to break has not been tested, whatever the volume of what supports it.

And it makes the exercise bearable by turning it around: each counter-example found is not a loss, it is a clarification of what you are asserting.

Nuances and limits

The stance has a real psychological cost, and it is asymmetrical: it asks you to scuttle what you have just built, at the moment you are most attached to it.

And it does not hold indefinitely — pushed too far, it turns into the manufacture of counter-examples, which is the same bias with the opposite sign.

Open questions

• Who can help state an author's implicit hypothesis, given that it is precisely what he does not see?