We're trying to assign a static attribute to explain behaviors which shake out to a particular (and highly individual) emotional response. That's not quite the Worst Argument -- though it is related -- but it is a very bad habit of argument.

We're never going to find a "creepyp" type predicate attached to anyone. It may be that some subset of LWers exhibit behavior which reliably tends to alienate certain groups we'd be interested in hearing more from, though, and if so it should be possible for us to describe this behavior and try to develop group norms to exclude it: as a community we're pretty good at analyzing that sort of thing, and it certainly beats spiraling further into semantic fail.

On the other hand, I can see some potential for close examination of the problem to lead into gender fail -- something that we've historically been very poor at dealing with.