Quality on the Web
This essay was written in 2007
In all likelihood, if you’re reading this, you’ve heard the joke “On the Internet, nobody knows you’re a dog.” In his 1993 New Yorker cartoon, Peter Steiner inadvertently highlighted a strength – and a dilemma – for the new universe of anonymous interactions. Interestingly, Mr Steiner himself did not intend on any particular message. “I did the drawing of these dogs at the computer like one of those make-up-a-caption contests,” he said in a New York Times interview in 2000 [3xxh3v]. ”There wasn’t any profound tapping into the zeitgeist. I guess, though, when you tap into the zeitgeist you don’t necessarily know you’re doing it.”.
As long as you’re a smart dog contributing in a positive way to the discussion, then whether or not you belong to the Canis lupus familiaris subspecies doesn’t matter much. Problems arise, however, if you act like an idiot. Regardless of your species. As I labeled a BOF session at the Web 2.0 conference, “on the internet, everybody can see if you’re an idiot, they just can’t do much about it.”
Enthusiastic though we all are about the notions of long tails, the wisdom of crowds, and emergent behavior, the reality is that the quality of online discourse has progressed little from the days of ARPANET mailing lists, USENET FAQs and The Well in 1980s. Good structure still requires editors, good content still requires writers, and good discussions still require moderators.
I discovered this myself when I coded one of the first online reader-comment pages to accompany my (printed) column back in 1994. Initially intended to be “uncensored”, I beat a hasty retreat when anonymous posters discovered that they could post foul language, no doubt to their own great mirth. I’m hardly alone. The Los Angeles times had a slightly more (in)famous experience in their 2005 “wikitorial” project. That experiment lasted two days [22vbeq].
Various flavors of moderation, participant editing and/or voting, and variations of reputation systems are being used today to leverage the crowd without falling victim to it’s vices.
But history would teach us that this isn’t so simple. Past efforts to tame the crowd, to encourage and coerce it to only yield “good” results (in some sense), have met with limited success. The dilemma lies in the subtleties of group behavior. As critics of Surowiecki’s book have pointed out, it’s easy to find anecdotal evidence to fit your point of view — efficient pricing markets at one end of the spectrum and witch-hunts at the other (there are a number of quality postings in the Amazon feedback forum [yqdhbz]).
Let me say as an aside that this is not about the controversies surrounding Wikipedia. They target an eminently solvable problem: to gather in a single searchable online asset all known non-controversial facts. The key is non-controversiality, or as wikipedians would label the notion: maintaining a neutral point of view. It might not be Wikipedia per se that solves the problem, it may instead be Larry Sanger’s Citizendium [2pfxpy] with its emphasis on embracing participant credentials. But regardless, it’s a solvable problem. Non-controversial facts are relatively easy to sort out.
Problems arise in other areas. Most notably, the wiki principle breaks down in the face of controversy. There are simply no mechanisms for moderating a discussion. It also breaks down in the face of systematic gaming or prejudice. And it is also weak in the face of a steady stream of information.
So today there are multiple efforts to define sets of checks and balances. But these easily become complex, and they also easily become essentially a political system. Orlowski pointed this out recently with respect to Wikipedia, which has perhaps some of the most complicated self-regulation policies, when he likened the non-apologies in the Essjay scandal to that of British Prime Minister Tony Blair [24wsca].
In a political system, being right or wrong doesn’t matter, all that matters is staying in your position of influence. When your rating/voting system becomes a social group, then social dynamics and organizational psychology kick in. And they quickly become a game of social position, not of optimizing the quality of the result. Anybody who has worked for a large organization knows exactly what I’m talking about.
And that, in a nutshell, is the key challenge for the next generation of online discourse. We must find a better way.