Just to play devil's advocate...if we look at the overheating theory, it would seem that the interior of the set screw -- in other words, the recessed part of it closest to the wire -- would be the first to overheat. Right?
If so, why would it not be natural color with no signs of overheating? Why would any overheating discoloration be present only on the top surfaces of the screw -- furthest from the point of contact/arcing? The thickness of the screw at the end contacting the wire is not much, if any, greater than that of the sides of the screw. It would seem to me that any overheating would start at the contact end of the screw and radiate out from there. Any discoloration should do so also.