Actually, I've used this as a fairly effective technique for arresting flamewars. It doesn't always work, but when it does, subsequent posts are usually at the DH level of your own response rather than that of the parent comment.
The key seems to be drawing 3rd-party onlookers in before the original flamer responds. If someone else responds to your comment at a high DH level, the original flamer has a choice. He can respond to your comment with another flame, which makes him look stupid and petty because there's a sibling comment that's much more well-reasoned. He can respond with a real argument, in which case you've raised the level of discourse. Or he can go away, which seems to be what happens most of the time. Regardless of what he does, you're free to ignore his reply and continue responding to the person who engages you with actual arguments.
A corollary is that educated, rational lurkers hold a lot of power on discussion boards. If you don't get directly involved in arguments but instead cherry-pick the comments you respond to, you can set the whole tone for a community.
BTW, the same trick works in face-to-face conversations, as long as there are more than 2 people involved. The person who asks the questions controls the conversation, by virtue of which questions they choose to ask. And the person who sits back and shuts up controls whose ideas get developed, by virtue of who they choose to respond to. That's why the quietest person at a meeting usually controls it, as long as they're not just a passive onlooker.
1. Original flamebait
2. Congent response to flamebait (posted by you using username 1)
3. Congent response to post 2 (posted by you using username 2)
This might be slightly unethical, but it means that the original flamer will always be dissuaded from posting again, and you're more likely to get more congent discussion participants.
If using multiple usernames bothers you, you could probably just check yourself with a small counterargument at the end of post 2, and it seems likely that the original flamer would still be dissuaded.
The proven technique for dealing with flamebait is to tag it immediately for what it is, and not waste any reader's time with a detailed discussion of an idea that is not even worthy of refutation.
And yes, even Aristotle in the Topics agreed that there are propositions/arguments that are so silly or vain, they are not worthy of a serious refutation. Flamebait certainly counts as one of these. (BTW: I see there are Aquinas fans here, so I will point out that Aquinas expressed his agreement with this in, of all places, his commentary on the Ethics: I think it was Lecture I).
The only problem is that there really are people who are too quick to tag someone else's post as 'flamebait', even when it is not. This happens on those very topics that most need to be seriously discussed, simply because they are so important, yet so many people hold strong opinions on them, opinions that are not founded on sound logic, but on passionate attachment of one kind or another. So they are strong opinions, but they are wrong opinions. These are the discussions that are the most difficult. They should not be attempted by amateurs. But there is no enforcing of _this_ principle on the Internet;)
Unfortunately this rarely works in the sense of eliciting a reply. But when it doesn't, it does at least tend to shut them up.
(Admittedly, this is not always possible or desirable.)
The problem on those sites is that their is not only poor comments but also tons of users who upmod them.
This is a legitimate problem, and ironically it is a symptom that a news concentrator is becoming a victim of its own success. As the signal-to-noise ratio attracts a progressively larger crowd, the types of information that are viewed as 'signal' broaden to appeal to the lowest common denominator. It's similar to hiring in a small company -- but at least in that situation you have control over the purse strings. (Even the trolls get dumber as time goes by and a site grows more popular.) Thus the comments on an arbitrary story suffer, and it requires more work to pick through them. However, if you 'listen' to subcommunities (eg. economics, programming, math on reddit) you find that (unsurprisingly) the baseline rises again. If you have no interest in those subsets, however, you're kind of out of luck. (But if you're here you already knew that)
You could probably model this as a logistic process, provided that you account for renormalizations (such as the literal renormalization that took place on reddit recently).
but also tons of users who upmod them.
This part (or at least its independent, non-interaction effect) is irrelevant; if you focus solely on people who have strong and demonstrated critical thinking skills, you can learn a great deal, and also gain a different perspective. Their input is much more harmful when it comes to filtering in or out the news to be concentrated.
Most of the interesting threads in which I have participated eventually lost the interest of the masses because they had descended into minutiae. But that's where the interesting bits lay, so that's where they went.
Ignore the riffraff. Lord knows they'll usually ignore you if you're presenting anything challenging. Even in a swamp like Slashdot, there is occasionally a perspective-changing comment that is worth reading.
If you want to expose yourself to a wide variety of opinions, you have to be willing to do a bit of work yourself, and determine which ones hold merit. It's a two-way street, in many respects, and your priorities (in addition to your time pressures) will determine how far you are willing to take it.
It appears to me that the 'social news' sites on the Web are simply recapitulating the arc of the special-interest BBS nodes and Usenet cliques of years past. Everything old is new again.
Seasoned trolling pros are usually very good at this. I think it helps a lot to look at their underlying motivation and to crank up the bullshit detector.
edit: btw my experience is that more often than not i get downvoted if i do this
The new category brings to mind an interesting property of this hierarchy though, namely that it seems to be difficult to respond to an "argument" with a higher order technique than that which was originally presented. For example, if one were disagreeing with a purely ad hominem rant, one would find it futile and nearly impossible to disagree with the rant at any higher level; how can one truly refute an argument that has no central point? It's somewhat depressing because, if true, it implies that a discussion is upper bounded at the level of discourse that initiated the discussion.
The way out of this, as far as I can tell, is recognizing that statements are not made in a vacuum. Sometimes the only way to raise the level of discourse is to engage in a sort of meta-disagreement. In other words, sometimes the point of a statement is not the content, but the context. While it works to classify criticisms of tone, for example, as weak if one were dealing with a substantive argument, the metadata (tone, speaker, etc) around the content of a non-argument may allow one to say something intelligent about the statement by tying it to the broader context in which it was made.