a Bayesian approach would point to probabilistic likelihood of deliberate use of provocative ad hominem language as a trolling technique.

LOL. You're on LW, y'know, not Lifehack. People like me will look at you taking a random collection of keywords, throwing them all into a blender set on high, and then regurgitating them onto a page -- and be not impressed.

To quote My Cousin Vinny, "...everything that guy just said is bullshit" X-)

Why?

It seems to me that there are plenty of other obvious reasons why people might make ad hominem attacks. For instance:

• They might expect them to influence other people in a direction they like. (I would guess that this motivation is substantially more common than desire to provoke an angry response.)
• They might be angry (or otherwise upset) themselves, for whatever reason, and respond as angry people often do by attacking in any way that presents itself.
• They might be failing to distinguish between an idea and the person presenting it, and be only dimly aware that what they're doing is an ad hominem attack at all.
• They might be hoping to provoke (not an angry response but) reflection on the part of the person being attacked ("wow, I had no idea anyone would react so strongly to what I said; could there be something wrong with it?"). I doubt this works very often, but there might be cases where it's a reasonable last resort.

In this particular case I think the first and last of those are the most likely motives.