Ad hominem is a fallacy when the motives or nature of the arguer are brought up inappropriately. If it is relevant, it is not ad hominem, usually because it expands into something else if you really think about it. As long as we've suspended Godwin's Law for the day...
"The sky is blue." "He's a Nazi!" is ad hominem.
"Has anyone noticed the Jews have an awful lot of banking jobs?" "He's a Nazi!" is, if true, very relevant. The first statement is likely not just a mere observation that may be true or may be false but is historically associated with antisemitism and is probably being used as the thin end of the wedge and not a neutral observation. It's relevant here because it expands into an argument where the origins and motivations of the arguer are in fact very relevant.
Similarly, when you see a new physics theory coming from a known crackpot, while the fact they are a crackpot may technically have no effect on the truth value of the theory, when it comes time for you to evaluate the theory you are justified in noticing it's coming from a crackpot.
Bayesian logic lets you express it even more cleanly; in Aristotelian logic the motivations of the speaker are irrelevant to the abstract truth of the statement; in more practical Bayesian logic, as you personally examine the likely truth value of somebody else's statement (with no presumption of access to the abstract truth) their motivations can be very relevant. "ad hominem" is over-weighting their motivations or falsely weighting their motivations, and under this formulation you also get reverse ad hominem, in which you under-weight their motivations, which is every much a fallacy as the original ad hominem, and which Zed talks about in his post.
All things considered, people being hyper about ad hominem is probably better than letting it slide too often, because it is certainly true that on average it is not a valid attack. But he's absolutely correct that it is sometimes valid, and while I'm not sure Zed's necessarily correct about Tor, I do think he's a lot more justified in his opinions than the pileon here consensus is indicating.
"The sky is blue." "He's a Nazi!" is ad hominem.
"Has anyone noticed the Jews have an awful lot of banking jobs?" "He's a Nazi!" is, if true, very relevant. The first statement is likely not just a mere observation that may be true or may be false but is historically associated with antisemitism and is probably being used as the thin end of the wedge and not a neutral observation. It's relevant here because it expands into an argument where the origins and motivations of the arguer are in fact very relevant.
Similarly, when you see a new physics theory coming from a known crackpot, while the fact they are a crackpot may technically have no effect on the truth value of the theory, when it comes time for you to evaluate the theory you are justified in noticing it's coming from a crackpot.
Bayesian logic lets you express it even more cleanly; in Aristotelian logic the motivations of the speaker are irrelevant to the abstract truth of the statement; in more practical Bayesian logic, as you personally examine the likely truth value of somebody else's statement (with no presumption of access to the abstract truth) their motivations can be very relevant. "ad hominem" is over-weighting their motivations or falsely weighting their motivations, and under this formulation you also get reverse ad hominem, in which you under-weight their motivations, which is every much a fallacy as the original ad hominem, and which Zed talks about in his post.
All things considered, people being hyper about ad hominem is probably better than letting it slide too often, because it is certainly true that on average it is not a valid attack. But he's absolutely correct that it is sometimes valid, and while I'm not sure Zed's necessarily correct about Tor, I do think he's a lot more justified in his opinions than the pileon here consensus is indicating.