No I'm saying every trait must be weighed with a complex interaction of known traits, other immeasurable traits, and the changing environment a person is in. So yes there could be a purely negative trait but it is very hard to tell without testing that combination out. Just detecting a trait, I think is insufficient.

I think the chaotic domain of the Cynefin framework describes this situation the best. https://en.wikipedia.org/wiki/Cynefin_Framework

Probing or sensing is insufficient, acting must occur.

Maybe a simulation could be sufficient way of acting?