(1) I've noted before that with ad hominem, ad verecundiam, and the like that it is important to distinguish two things, for both historical and practical reasons. First, there is what could be called the tactic or approach to argument. This is the original kind of idea that went with these labels. The original point, which we find most clearly in the Logic of Isaac Watts, is that these are argument-building approaches; they designate a field within which you can draw the middle terms that allow you to draw conclusions. This has to be distinguished from the error, the way in which this approach to argument goes wrong so as to create a fallacy. In general with these "ad" fallacies, the actual error is ignoratio elenchi, or at least something close to it; they are generally recognized as fallacies of irrelevance. The tactic in such cases is not producing something that actually addresses the supposed point of argument.
Mizrahi's argument, to the extent that it is right, can be translated into these terms by saying that not every instance of the tactic involves an error. This is very definitely true, and some of the reasons Mizrahi gives are certainly right for the right reason: you can identify cases that are tactically ad hominem that are not ad hominem fallacies because they are provably relevant. Arguing against appeal to authority is a good example, fairly straightforward.
(2) However, Mizrahi's failure to be entirely clear about the distinction between tactic and error leads him to make a mistake. He says,
When an appeal to authority is made, it’s reasonable to respond by pointing out that the authority appealed to is acting in a manner that is inconsistent with her advice. Such practical inconsistency provides a good reason to think that refusing to follow the authority’s advice wouldn’t be imprudent. It’s important to note that this sort of ad hominem argumentation is legitimate only as a rebuttal to appeals to authority.
This is certainly not true, however, because we can run arguments analogous to those touching on the appeal to authority with other cases. For instance, it is reasonable to argue against someone who denies that we can communicate truths, or know what's right or wrong, or know about the world, by noting ways in which their own life and practice fails to bear out their claim. This is indeed the most respectable form of ad hominem tactic; it is a staple of philosophy literally since ancient days. It is never so useful as it is when it is used against sloppy debunkers or skeptics, people who debunk or object so badly that their debunking or objection would redound on them as well as the object.
His mistake in restricting ad hominem, even of the structure he is considering, to appeal to authority, leads him to characterize ad hominem argument as "defeasible". Ad hominem as a tactic is not a specific argument, so it is not the sort of thing that can itself be defeasible; there is good reason to deny that ad hominem arguments are defeasible generally, if constructed properly -- it's just that, as with other approaches, an error can insinuate itself, and the ad hominem argument can fail to be relevant to the particular point at issue, even despite appearances.
(3) On the basis of his argument, Mizrahi continues:
If I’m right, rebellious children are on firm ground argumentatively when they challenge their parents’ advice on smoking with ‘You use tobacco, so why shouldn’t I?’ By being smokers themselves, and thus failing to set a positive example, the parents have undermined their status as authorities whose advice should be followed.
Mizrahi correctly notes that this is not the end of the story, because there are other reasons not to smoke. However, I think he is simply wrong on this conclusion for two reasons: (1) it does not follow from his actual argument; and (2) it does not follow from the correct analysis of ad hominem.
First, it does not follow from his own account because he has not actually established that the inconsistency is specifically of the kind that undermines the authority of the parent. All that the child has noted is that the parent fails to follow their own advice. But this on its own does not undermine authority. Imagine a stronger case, a parent who is a drug addict, counseling her child to stay far away from the drug to which she is addict. Is her authority in any way undermined by her addiction? Not in the least; indeed, if anything, it's obvious that you should take their advice very seriously. It's not the relevant kind of practical inconsistency. It is not enough to catch people out in failing to follow their own advice -- everyone fails to follow their own advice on all sorts of things for all sorts of reasons. What you need for the authority-undermining is exactly that: something that suggests that their status as an authority on the question is not to be taken seriously, either because they are clearly not the authorities they claim to be or because there is reason to think they are being dishonest in their advice or because there is reason to think the kind of principle they are denying is the kind that is necessary for them to have any authority at all. The only one that could be relevant here is if the parent is being dishonest; but this is not something that can be known from the facts that Mizrahi has given, and is not generally true of smoker parents who advise their children not to smoke.
Second, by not properly distinguishing the tactic and error, Mizrahi has also apparently assumed that, because the error does not always arise in using the tactic against appeals to authority, that it therefore always does not arise in using the tactic against appeals to authority. Mizrahi has not established that ad hominem is good (even if defeasible) in the context of appeals to authority, but only that appeals to authority provide some cases in which the ad hominem tactic can be seen not to involve the error of irrelevance that would make them a fallacy. In reality, the child is apparently guilty of irrelevance here -- the failure of the parents to comply with their own advice is not due to anything inconsistent with their authority in being able to assess whether smoking is a good thing, nor with anything that gives the child a reason to defer to parental authority on this matter.
As I noted, Mizrahi's conclusion would be a disaster for moral education, and this is for the reason that I noted earlier: everyone fails to follow their own advice quite a bit, for all sorts of reasons. The advice, however, should not be regarded as in any way contaminated by these things, unless it can actually be established that the practice is inconsistent specifically for reasons that would call into question the advice itself. And since advice is a major part of the foundation of moral education, not recognizing this is the kind of thing that can poison moral education.