Main idea

An honestly conducted prior-art search looks for counter-examples: it wants to bring down the hypothesis "this part of my method is new". As long as it succeeds, it is working.

The tipping point is recognizable by a precise sign: you now have to spend ten lines explaining why a system might vaguely be comparable. When an analogy needs a paragraph of justification, it is no longer the source that resembles, it is the researcher who brings it there. You are no longer putting the hypothesis to the test — you are producing the result you came for.

The mechanism is the exact mirror of the confirmation bias you thought you had avoided. At the start, you unconsciously look for what confirms the idea is original; at the end, for what confirms it is not. In both cases, you have stopped looking.

Why it matters

This gives an observable warning signal instead of a general vigilance. "Stay critical" is not actionable; "if the analogy needs ten lines to hold, stop" is.

And it says why a prior-art search needs a closing rule written in advance: without one, the only thing that stops the search is the moment you find what you wanted to find.

Nuances and limits

Some real kinships do genuinely require a long explanation, because the two fields share no vocabulary. The length of the justification is an indication, not a proof.

And the signal arrives late: it shows up once you have already spent time building the analogy, that is, at the moment you are least inclined to throw it away.

Open questions

• Who can spot that tipping point in someone who has been searching for weeks, when the person concerned is the last to see it?