Borrowed product sense

Nielsen Norman Group published a piece defining "product sense" this year: the ability to recognise when a current problem matches a past success or failure, and to know when that match is reliable and when it isn't. The second half is the interesting part. They call a case that looks exactly like a familiar pattern but isn't a "wicked situation": the most dangerous kind, because it doesn't feel uncertain the way a genuinely unfamiliar case does. It feels like confidence, right up until it's wrong.

That's a good, precise name for something worth naming. It's also, structurally, a description of what a model can't do.

A model doesn't hesitate the way an expert is supposed to when a pattern quietly stops applying. It has no mechanism for the doubt a wicked situation is meant to trigger. Handed a situation that resembles something familiar, it matches and answers, with the same fluency whether the match is sound or not.

Which means the judgement NN/g is describing, knowing when experience does and doesn't transfer, has to be supplied from outside the model if it's going to exist at all. Someone has to have already done the work of deciding what's actually comparable to what, before the model ever sees the brief. That's not a new idea on this site. It's context debt again, looked at from the other direction: the backlog isn't just missing facts, sometimes it's missing the knowledge of when a rule of thumb quietly stops being one.

This is NN/g independently naming a second piece of the same territory, a few months after "UX-context design." Worth noticing, not because it proves anything new, but because two unrelated pieces converging on the same shift from different directions is stronger evidence than one.