I think the point is to never communicate in evaluations. Observations are not codes for evaluations. Observations force discussion of concrete actions, which h

I think the point is to never communicate in evaluations. Observations are not codes for evaluations. Observations force discussion of concrete actions, which have a bunch of benefits over evaluations. Let's say X and Z are coworkers. Z messed up some report numbers. X could say "Z, you are sloppy." Z could easily respond "No I am not, how could you say that?" By focusing on a concrete example, X and Z can both discuss the actual problem (inaccurate report numbers) instead of bickering about whether or not Z is sloppy. NVC is also a check on the person initiating the communication. It's possible X's view of Z as a slob is not based in reality. If X can't find a specific thing, but just thinks Z is a slob, NVC makes it hard for X to say anything until they can point to a specific event/action. It's not perfect (X can just look for the first example that proves their belief) but it does prevent totally baseless claims of "you are a slob."