philcrissman.com

Paul Graham just posted his thoughts on the 2-year anniversary of HackerNews.

It’s interesting for a number of reasons.

One thing that stands out is the implicit understanding that many mainstream communities of a similar nature have suffered what Paul refers to as “dilution”. The audience of HackerNews is likely to be familiar with this idea, as from what I’ve seen, it is largely made up of people who arrived independently at the same conclusion; basically that many sites (Digg, Reddit) have been ruined by their own immense popularity.

It calls to mind the old quote attributed to Yogi Berra:

Nobody goes there anymore, it’s too crowded.

What’s interesting is that “ruined” is necessarily a subjective thing. Clearly someone is visiting Digg and Reddit; many someones, to be exact. Just as clearly, most of them must not think that Digg and Reddit are “broken” or “diluted” (and might even argue that assertion). So, what’s the difference? Why is it broken for some, and not for others?

My feeling is that it has a lot to do with what the individual user values. If all you want is a reasonably current list of “top news” stories in given area, Digg/Reddit may do just fine. But I’ve tried, in the past, to restrict myself to just Digg/technology or programming.reddit, and have still found that I gradually lost interest in ever returning. HackerNews, on the other hand, continues to deliver interesting, valuable content and discussion.

I really hesitate to frame the difference in elitist terms, the intelligentsia vs. the hoi polloi [1], because I really don’t think that’s the real reason for the disparity. Still, there is evidently a certain quality of news and discussion that is valued in forums like HackerNews and (apparently?) not valued in some other communities. The point is not that the uber-communities are “bad” and the niche-communities are “good” (though it’s possibly true); the point is more, for those who prefer the one type of community, how to prevent the dilution from occurring.

Slashdot

Like most from my generation (?) my real introduction to online communities was Slashdot. I’m too young to have really been active on usenet, and too old not to remember when the Slashdot effect was known for bringing web-servers to their knees.

Slashdot was, and is, run by editors and policed by its own community in occasional roles as moderators. Anyone can submit a story, but it’s a Slashdot editor who decides if that story is published to the front page. Anyone can comment, but only a moderator (in days of yore, any logged in user might randomly be selected to have several moderation points; I don’t know if this has changed) could raise or lower the score of any given comment.

One of the interesting things about the moderation of comments on Slashdot is that there have always been (as far as I can remember) specific categories of moderation. That is, you could moderate a comment +1 Insightful or +1 Funny, but not both.

I was reminded of this by Paul’s mention of “stupid” comments in his essay, noting that they are quite often jokes. This is part of what started me thinking about the concept of culture for online communities.

By having a positive moderation category for “Funny”, Slashdot was explicitly endorsing the idea that a comment could be valuable simply by being humorous. There are a lot of different sorts of humor. In some cases someone might say something that is genuinely funny, and often this is a worthwhile comment; humor can sometimes be extremely insightful, especially as it tends to pair unexpected concepts, or (often) to highlight pain and/or suffering. On the other hand, there are memes that tend to go in and out of style, which are “funny” only in the sense that they are a shibboleth, of sorts. If you “get” the joke, it would mean you are a part of an invisible community (made up of everyone else who “gets” the joke). Abuse of this sort of meme tended to get squashed, even on a community which explicitly valued humor as commentary. There are only so many times that “all your X are belong to us” and “in Soviet Russia, [noun] [verb]s you!” are really funny, and after that, they start to seem like a cheap way to ask for a +1 moderation. Accordingly, such derivative jokes tended to get down-modded severely once they had passed the point of being novel.

Modern community sites, like Reddit and HackerNews, tend to have simply an up vote option, or possibly an up OR down vote option. So you might vote a comment up because you think it’s funny, or because you think it’s insightful (occasionally both); but in the end, it is just a +1 to the comment (or news item). No one knows why you voted for it, unless you also comment on it.

Interestingly FriendFeed has a similar ambiguity with its “Like” paradigm. You can “Like” an entry on FriendFeed, whether that is a delicious bookmark, a twitter/friendfeed message, a blog post: anything. That is the only “voting” mechanism available. You often see people remark on this when there is a controversial or appalling story; people want to highlight it, to bring it more notice, but it is somehow awkward to “Like” a story about genocide in Darfur, for example. I can see an advantage to a simple up or down vote in comparison, but the “Like” feature is nonetheless an integral part of FriendFeed, and certainly adds something (if I’m not mistaken, Facebook recently copied it for its wall items).

I really hate that I have to refer to Slashdot as a “legacy” community. It isn’t gone, by any means; it’s still quite active. But still, it represents a different era of the web. It has editors, rather than a completely user-generated content model. It is built on Perl [2]. It has definitely taken a hit in traffic, though it still maintains steady, slow growth. Google’s data shows what happened:

 slashdot   digg   reddit

(This is a comparison of search terms, not traffic per se, but it’s still a decent picture: I can’t find compete or alexa graphs which go far enough back to show when slashdot reigned supreme.)

Toward Intelligent Conversation

Of course, there are problems with explicitly valuing humor as a positive contribution. It’s a lot easier to try to be funny than insightful, especially if you are allowed to hijack a common meme (e.g., I, for one, welcome our meme-flogging overlords) or simply disparage someone or something that makes an easy target. I can see why pg, and others who have fled the Diggs and Reddits of the web to HackerNews (at least for now), tend to view the tendency toward making “joke” comments as a net negative. Anyone could have said it, and it really contributes nothing to the conversation.

The nice thing about simple voting without categories is that it discourages knee-jerk joking without prohibiting humor completely. After all, if you say something humorous that is relevant (or at least, not obnoxious) it’s may get voted up anyway. But it’s not necessarily likely to get voted up simply because your referenced the Simpsons, the Family Guy, or some obscure video game that everyone pretends to remember.

Enter karma. Karma was around on Slashdot as well, though they didn’t show a score; your karma on Slashdot was given a name (Excellent, Poor, etc.) based on an internal algorithm on how your comments had been moderated. In the current voting model of most community sites, karma is simply a number.

I don’t know if there have been studies made on this, but I imagine the presence of karma influences users in different ways. For some it is a motivator to try to “game” the system; to submit items or comments which are likely to be up-voted, thus making the increase of karma a goal in and of itself. For others, it may actually be a deterrent to comment or posting; if you know that whatever you submit or say will affect your “karma” on the site, you might second guess every comment you are about to make, and may ultimately not say much at all. Those are both extreme poles of reaction to a karma-based system, and neither one is what a karma system is intended to encourage. For the first group: people who want to game the system will (probably) always try to game the system. Over time, hopefully, we make smarter systems which prevent this from happening as easily. For the latter; I suppose what you have to do is make karma important—but not too important. That, and probably just encourage people to get over themselves and just take part in the discussion.

Conclusions

I tend to agree with Paul’s conclusion: it isn’t people you want to exclude, it’s certain types of behavior. Judging from Digg and Reddit, behavior does seem to deteriorate when the crowd explodes in numbers. But that doesn’t necessarily mean it must always deteriorate; there may be a way to allow for growth, even huge growth, and still maintain the sort of forum that users want to participate in.

But part of it will simply be individuals’ preference. This may be a function of average age of the user, occupation, outlook, values, but it will always exist. Some people will always prefer Digg, or Reddit. Some will never leave Slashdot, and will keep going for that +5 Funny on every comment. And hopefully the communities that want to be different are able to keep the character that they’ve valued from their inception.

[1] You can’t really even use the term hoi polloi with out sounding somewhat condescending toward “the masses.”

[2] I’m sorry. I couldn’t resist.