By now it’s well-known that most Internet forums devolve over time, even when the people running the forum take concrete steps to avoid devolution. But the main problem is not necessarily the trolls who deliberately attempt to degrade the quality of the conversation. It’s low-quality comments that aren’t necessarily malicious or even mean-spirited but do reflect shallow knowledge. Not only that, but such comments are often designed to appeal to groupish belief or to raise the status of the commenter, rather than sharing information and asking genuine questions.

Kens offered this insightful observation on HN:

My theory (based on many years of Usenet) is that there are three basic types of online participants: “cocktail party”, “scientific conference”, and “debate team”. In “cocktail party”, the participants are having an entertaining conversation and sharing anecdotes. In “scientific conference”, the participants are trying to increase knowledge and solve problems. In “debate team”, the participants are trying to prove their point is right.

Unfortunately, the people in scientific conference mode attract the cocktail people, but the latter don’t tend to attract the former. Debate team-types tend to be attracted to both—they’re the people exhibiting groupish and status-based behavior. In HN land, I’m probably closer to cocktail mode people than the scientific conference mode people, though I want to act more like a conference person.

Still, it’s worth looking more carefully at what the scientific conference-mode means. I don’t think scientific conference means a literal presentation of new results, but I think it does mean that the people commenting are deeply informed, deeply curious, reasonably respectful, and work to speak from a position of knowledge, rather than ignorance, about a subject. In this sense I fit the scientific conference mode when I discuss a small but real number of issues related to teaching, urban planning / development, and grant writing / government practices. The second one relates least to my day-to-day life but is a personal interest about which I’ve read a fair amount. Towards this end, I suspect a lot of people could improve the quality of the conversation simply by not commenting.

I distill this general idea to a simple behavior heuristic that might be valuable to others: don’t comment unless you have a special, unusual, or well-informed viewpoint. Many of my comments link to books and/or articles I’ve read that elaborate on whatever point I’m making or trying to make (here’s one example, linking to Bryan Caplan’s Selfish Reasons to Have More Kids, and here’s another, citing Edward Glaeser’s The Triumph of the City; in response to the second, someone even said, “I got these books simply because of this recommendation,” which makes me feel warm and fuzzy inside). Some of the ideas contained in the books or articles I cite might be wrong or badly argued, but at least I’m basing my comments on something specific rather than some general philosophical point. Too many people argue based on first principles or unsourced speculation. The latter isn’t always bad; for example, someone might work in a field and know something deep and important about it without having a link to a specific discussion of the idea being discussed.

Many of my comments that don’t link to books or articles still deal with specific issues in which I have above average expertise, knowledge, or experience. This comment discusses how I deal with a student who asks a question relating solely to his or her individual issue in a large group without sounding like a jerk (I think, anyway), this comment is about a specific product I’ve used (the Unicomp Customizer), and this comment is about specificity in writing and thinking. Again, I might be wrong, but in each case I’m writing based on experience.

You can find some exceptions to the principles I’ve discussed above. You should ask logical or reasonable follow-up questions, especially if you’d like more information (here’s one sample; here’s another.) Succinct is often beautiful. Focus on genuine questions, rather than challenging people because their beliefs don’t match yours.

You don’t always have to follow these rules—I don’t—but if you’re debating about whether you should post a comment, you should probably err on the side of silence and not intruding on other people’s time. Unfortunately, the kind of people who most need such internal self-restraint are probably also the ones least likely to use it, and I doubt anything can be done to solve this problem, which seems like a variant of the Dunning-Kruger Effect.

Despite all this, I don’t see a solution to the fundamental problems, at least beyond the person at the margin who might read this and change his or her behavior slightly.

In other words, one can appeal to community rules and norms, or resort to meta-posts (like this one).* Such an appeal shouldn’t be done too often, or a community will spend more time discussing its own rules and norms than it does discussing and reading the material that should be the purpose of its existence (I last wrote a post like this in January; that January post still seems relevant, but I feel like enough time has passed and that I’ve observed enough behavior to make this post relevant too). But perhaps the occasional reminder will, as I said, help at the margins.

Oh, and the other rule for commenting? When you’re done with substantive content, stop.

* Granted, these kinds of posts and comments can make a community deteriorate. For example, there are a set of overly long comments by jsprink_banned and josteink that fail to distinguish between an argument and how the argument is presented: I suspect they’re unhappy with the mod functions mostly because they haven’t focused on how an argument is delivered. Civility counts for a lot, at all levels of debate; see, for example, Tyler Cowen’s comments on civility and Paul Krugman (and his comments on the limits of binary, good versus evil thinking in general).

There are also comments like this, in which the poster argues from nothing, attempts to activate an anti-corporate ideology, and ignores the obvious, abundant evidence of the continued importance of firms. Alex-C, fortunately, did reply: “I almost can’t tell if this comment came from some sort of Markov text generator.” HN used to have many fewer of those kinds of comments, and when it did get those kinds of comments, they were much less likely to rise. It takes more effort than it should for me not to respond to them directly. Incidentally, the comment Alex-C was replying to meant to say something like this.