The vast majority of the comments attached to “Thoughts on James Cameron’s Avatar and Neal Stephenson’s ‘Turn On, Tune In, Veg Out’” are terrible. They tend toward mindless invective and avoid careful scrutiny of what I actually wrote; they’re quite different from the comments this blog normally gets, which is largely because I submitted the Avatar post to Slashdot, home of the trolls. One friend noted the vitriol and in an e-mail said, “Okay, the Slashdot link explains the overall tone of the comments your “Avatar” post is attracting.”

Part of the reason the comments are so bad is the hit and run nature of comments, especially on larger sites. If you have something substantial to say, and particularly if you regularly have something substantial to say, you tend to get a blog of your own. I wrote about this phenomenon in “Commenting on comments:”

In “Comment is King,” Virginia Heffernan writes in the New York Times, “What commenters don’t do is provide a sustained or inventive analysis of Applebaum’s work. In fact, critics hardly seem to connect one column to the next.” She notes that comments are often vitriolic and ignorant, which will hardly surprise those used to reading large, public forums.”

Furthermore, it’s easier and demands less thought to post hit and run comments than it is to really engage an argument. I deleted the worst offenders and sent e-mails to their authors with a pointer to Paul Graham’s How To Disagree; none responded, except for one guy who didn’t understand the point I was trying to make even after three e-mails, when I gave up (“never argue with fools because from a distance people can’t tell who is who”). The hope is that by consciously cultivating better comments and by not responding to random insults, the whole discussion might improve.

(Paul Graham has given the subject a lot of thought too: he even wrote an essay about trolls. As he says, “The core users of News.YC are mostly refugees from other sites that were overrun by trolls.”)

Not every comment I got one was terrible—this one, from a person named “Dutch Uncle,” was probably the best argued of the lot, and it mostly avoided ad hominem attacks. It, however, was very much the exception.

Most comments tended to deal in generalities and not to cite specific parts of my argument. In this respect, they have the same problems I see in freshmen papers, which often want to make generalizations and abstractions without the concrete base necessary. This happens so often that I’ve actually begun a keeping a list of all the things freshmen have told me are “human nature,” with a special eye toward placing contradictory elements next to each other, and in class I now ceaselessly emphasize specifics in arguments.

Since I’ve see this disease before, I’ve already thought about it, and I think the generalization problem is linked to the problem of close reading, which is a really hard skill to develop and one I didn’t develop in earnest till I was around 22 or 23. Even then it was only with a tremendous amount of effort and practice on my part. Close reading demands that you consider every aspect of a writer’s argument, that you pay attention to their word choices and their sentences, and that you don’t attribute to them opinions they don’t necessarily hold. Francine Prose wrote a whole book on the subject called Reading Like a Writer, but the book is a paradox: in order to develop the close reading skills she demonstrates, you have to be able to closely read her book in the first place, which is hard without good teaching.

Mentioning Francine Prose brings up one other common point I saw in the comments: few pointed to sources or ideas outside themselves, and allusions were rare. In the best writing I see, such elements are common. That isn’t to say every time you post a comment, you should cite four peer-reviewed sources and a couple of blog posts, but ideas are often stronger when they show evidence of learning and synthesis from others. In my Avatar post, I brought together Greg Egan, a New Yorker article, Alain de Botton citing Wilhelm Worringer, Robert Putnam’s Bowling Alone, the Neal Stephenson essay, and Star Trek. Now, my argument about Avatar could still be totally wrong, like an essay with hundred citations, but at the very least other writers’ thoughts usually show that more thought has gone into an essay, or a comment. Almost every article in every newspaper and magazine piece worth reading cites at least half a dozen and often many more sources: quotes, other articles, journals, books, and more. That’s part of what make The Atlantic and The New Yorker so worth reading.

Citations area common because things that are really worth arguing about require incredible background knowledge to say anything intelligent. The big response I’ve had to many of the comments, especially the deleted ones, are suggestions to read more: read How Fiction Works, The Art of Criticism, and Reading Like a Writer, then post your angry Internet screeds after you’ve thought more about what you’re arguing. These kinds of pleas probably fall on the proverbially deaf ears, but at least with this post now I have somewhere to point bad commenters in the future.

I think one reason I find Slashdot conversations much less interesting than I did as a teenager isn’t because the nature of the site has changed, but because I’ve learned enough to have learned how hard it is to really know about something. Now I’m often more engaged by pure information and less often in invective and pure opinion, especially when that opinion isn’t backed up by much. The information/opinion binary is of course false, especially because the kind of information one presents often leaves pointers to one’s opinion, but it’s nonetheless useful to consider when you’re posting on Internet forums—or writing anywhere.

Incidentally, one reason I like reading Hacker News so much is that the site consciously tries to cultivate smarter, deeper conversation, much as I wish to; it’s trying to meld technical and cultural forces into a system that rewards and encourages high-content comments of the sort I mostly didn’t get regarding Avatar. I submitted the Avatar post to Hacker News before Slashdot, and the first, relatively good comment came from a Hacker News reader.

The problem of trolls is also very old, and probably goes back to the Internet’s beginnings—hence the need for a word like “troll,” with a definition in the Jargon File. As a result, I’m probably not going to change much by writing this, and to judge from my e-mail correspondent, trying to do so via e-mails and blog posts is mostly hopeless. But a part of me is an optimist who thinks or hopes change is possible and that by having a meta conversation about the nature of trolling, one can avoid the behavior in general, at least on a small scale. At Slashdot or Reddit scales, however, the hope fades, and one simply experiences the tragedy of the commons.

EDIT: Robin Hanson has an interesting alternate, but not mutually incompatible, theory in Why Comments Snark:

Comments disagree more than responding posts because post, but not comment, authors must attract readers. Post authors expect that reader experiences of a post will influence whether those readers come back for future posts. In contrast, comment authors less expect reader experience to influence future comment readership; folks read blog posts more because of the post author than who they expect to author comments there.