I agree that the negative's are a very difficult case, but in my experience is that most people don't cover the easier case of candidates they can evaluate. Further, evaluation techniques themselves are pretty error prone. But without collecting data and experimenting with it, how do we improve the problem?

But that means that you don't actually know if this extra signal of quality (or at least, notability) has any value at all. That tells me that you shouldn't be including it in your analysis of a candidate if you don't know what it actually represents.