James Fallows explains why he doesn't allow comments on his blog at The Atlantic ("On Comments and Community: A New Plan?", 12/21/2010):
Unless a comment stream is actively moderated, it inevitably is ruined by bullies, hotheads, and trolls. If you feel otherwise, fine. This is what I think.
Corollary: The comment-communities that flourish, notably the Golden Horde of TN Coates, require real-time, frequent intervention by a moderator not afraid to put his stamp on the discussion.
To the list of "bullies, hotheads, and trolls" we need to add a number of other categories, such as "lazy narcissists who think that everyone should have the benefit of their uninformed off-the-cuff reactions", and "annoying obsessives who relate every topic to their fixed idea". And of course there's the problem of dealing with ever-more-resourceful spammers.
In my opinion, blog comments are on balance worthwhile, especially since the WordPress Akismet system is good at catching spam without too many false positives. However, I'm towards the more liberal end of LL contributors on this question — so please be careful out there, and help me with my end of the internal discussion.
It seems to me that there are two very different sorts of reasons why blog comments can go bad. One is that in an open forum, the participants will inevitably include people that you'd never choose to have a real-world conversation with, at least not twice. But the other problem is that different participants often have inconsistent ideas about what sort of interaction is happening.
In a real-world conversation, for example, it's normal that some of the participants lack crucial background information and therefore ask for basic explanations, or make a contribution whose presuppositions need to be corrected by others. But as the size of the group increases, this becomes less and less appropriate. A request for clarification is perfectly in order in a conversation among two or three people, but is likely to be a strange breach of etiquette if it comes from one of a thousand audience members in large lecture hall.
And in a conversation where all participants are using networked computers, and interactions are spread out over hours or days, how much link-following and independent look-up do we expect people to do before asking questions or expressing opinions? It's clear that blog commenters often assume conversational rules of engagement, where it's appropriate to ask or respond quickly; but authors typically find it annoying that someone asks a question or makes a comment that's dealt with at length in a prominently-featured link (or in a later paragraph of the post).
There are many large-scale web forums where things generally seem to work out positively: metafilter, reddit, the comments sections at unfogged, to name just a few of the many (very different) examples that I encounter from time to time. But what I see in the comments sections of newspapers and magazines tends to lend support to the negative attitudes of James Fallows and some of my Language Log colleagues.