For now, let's disregard the cynical response: "If you're arguing online, you've already lost". Sure, there's no shortage of unproductive flaming happening on the internet, but let's assume for the moment that it's possible for people to disagree constructively.
The question then is that in the fortuitous event of such a disagreement, how should it be resolved? In a perfect world, sooner or later one party would be convinced and accept...
Ok, I'll wait until you've stopped laughing.
So, it's exceedingly rare for a participant in a debate to admit error and concede the point to his opponent. Nevertheless it was my first instinct for resolving debates at GreaterDebater. For in what way can you say you've really succeeded in a debate if your opponent remains unconvinced? I figured debates would go on until either one party conceded or a certain period of inactivity had passed. I soon realized that it would be very easy for a debate with this scheme to degenerate into a war of attrition. If the only requirement is to keep making replies until your opponent has stopped, this favors persistence over a valid argument. Idealism is all well and good, but we can't always expect the rest of the world to live up to our ideals. It's the nature of a debate that there will always be people who merely want to win, rather than to seek the truth.
In other formats, debates are often moderated by a panel of judges. Ideally, these judges are experts in the subject matter being debated and/or the art of debating. In theory, this sounds like the best solution, but it's difficult to implement. GreaterDebater isn't limited to any one subject, so a wide variety of experts would be needed. There will (hopefully) one day be a large volume of debates. Given the unpredictability of the subject matter and the scale at which I hope to operate, arranging expert judges for every debate would impose a crushing amount of overhead on the process. Maybe this is something that could be arranged under special circumstances, but the level of coordination required would greatly hamper the spontaneous nature of posting comments on the internet.
Finally, I came up with the approach of general voting. With a diverse enough user base any biases the individual voters might have should cancel each other out. What's left should be a reliable indicator of the how convincingly each side argued his or her case. Voting has proven to be effective at promoting insightful comments at sites like Reddit and Hacker News, so it should be too much of a stretch to suppose it could also pick a quality argument. With a time limit to ensure debates don't go on forever, a debater with a solid case should be able to win over a simple majority of disinterested third parties, even if he can't convince his opponent.
Originally published 2010-02-15 13:21:22