I think pointing out a logical fallacy is not a refutation. Rather than saying "your argument is wrong", it says "your argument is not well-formed". This is much like the difference between a program that has a bug and a program that has a parse error.
To take the most glaring example: "my point is true because my point is true". Obviously, nobody actually says this seriously. But some arguments are isomorphic to this. And any argument like that is not wrong per se--it is literally meaningless.
Other fallacies are more subtle and complicated, but the idea remains the same: using a fallacy does not invalidate your point but merely renders your argument meaningless. I can say that "the sky is blue because everybody says so" and I would not be wrong--given the sky is actually blue, of course--but the argument would still be a fallacy. I think to point this out is valuable even (perhaps especially) if you do not disagree with the point in question.
As long as you say that this what you are doing, I agree. Too often, however, people content themselves with just pointing out the flaw or fallacy. This gives the impression that they mean to argue with the central point.
Fallacious arguments (of which circular arguments are a subset) aren't literally meaningless. "My point is true because my point is true" is perfectly comprehensible, it just fails to establish the fact that "my point is true". You can disagree with it reasonably by pointing out the fallacy, so it has linguistic substance, and it is not meaningless.
It is very different from, say, "colorless green ideas sleep furiously" which is literally meaningless.
A logical fallacy does imply that an argument is wrong. It doesn't mean that the conclusion is wrong, but it means the argument is wrong. If an argument rests on a logical fallacy, then it does not provide evidence for its conclusion.
As humans, we can do better than a compiler that gives up at the first missing semicolon. We can keep reading and respond to the rest of the comment with more than a link to the wikipedia article on ad hominem.
Sure, if there is a 'rest of the comment'. Usually, there isn't anything else to go by. There's just the fallacious argument. It doesn't matter whether you can think of all kinds of ways to support the point of view the other espouses: you're not in his head and you can't know whether he thought all the things you can think of in favour of his position. There's only the fallacious argument and that needs to be struck down. If someone else wants to support the position, he'll come along.
That's certainly reasonable. But this does not mean you should ignore fallacies either. Usually when I see people talk about fallacies it's part of a point-by-point rebuttal against some post--you refute some of the points in order, perhaps agree with others and also point out any points that are actually fallacies. I think this is perfectly reasonable, but I can also see how solely pointing out a fallacy would usually not be terribly useful.
Also, ad hominem is a special case: not only is it a complete fallacy but it is also often quite rude. Minimizing personal attacks is important in keeping discussion pleasant and civil even ignoring their logical implications.
If I could only have one rule for some discussion forum, that would probably be it: no personal attacks.
I have to take issue with this. I think it's an important and useful rhetorical tool that, while sometimes can be used in a petty way to score points, can elevate the debate. If half of your counter-argument is "compiler warnings" for your competitor's argument, I think that's valuable.
I, for one, would wildly prefer a world where every debater was a nit-picking SOB when it comes to logical fallacies and the average debate "compiled cleanly without warnings". It would be like heaven compared to where we are today.
If we're talking about the formal fallacies, I agree, but my experience with people name-dropping compiler errors for informal fallacies is that they're often not very useful or accurate diagnoses. Ad hominem is one that is often useful, mostly as a sort of meta-rule about debate etiquette. But things like "slippery slope" are used to characterize a wide range of arguments of which only a subset are strictly fallacious. And some informal fallacies are only fallacies in the specific context of purported logical proofs, but not necessarily in other kinds of arguments. For example, it's true that an "appeal to authority" in an argument means it's not a valid logical argument that proves its conclusions from its premises. But people sometimes raise that objection in contexts where the authority is being invoked as an epistemic authority that raises the likelihood of its conclusion being true (because we have reason to believe that the authority's judgments have non-zero epistemic value), not as part of a logical proof. For example, while it's true that "the consensus of physicists is that X" doesn't prove anything about X, if I'm not a physicist and have a background belief that the consensus of physicists is generally a good guide on what is likely to be true about physics, it's a good inference that I should at least provisionally believe X.
In fact, even computers can often do better these days. Classical logic as implemented by theorem-provers is not anywhere near the state of the art in AI argumentation systems.
To take the most glaring example: "my point is true because my point is true". Obviously, nobody actually says this seriously. But some arguments are isomorphic to this. And any argument like that is not wrong per se--it is literally meaningless.
Other fallacies are more subtle and complicated, but the idea remains the same: using a fallacy does not invalidate your point but merely renders your argument meaningless. I can say that "the sky is blue because everybody says so" and I would not be wrong--given the sky is actually blue, of course--but the argument would still be a fallacy. I think to point this out is valuable even (perhaps especially) if you do not disagree with the point in question.