The Wisdom of Browse
Just a thought from 20 April 2006 about Design, Geek, Internet.

There's some drama afoot lately as bloggers pick apart Digg's user-controlled editorial system, looking for evidence of editors lurking in the darkness. But much of the conversation is overlooking a crucial nuance when it comes to authentic media and democratic editorial systems.

For the uninitiated, Digg is a tech news site, where the members post links to interesting stuff, and then the community chooses which links get promoted to the front page. For readers, this means they see stuff that a lot of people think is interesting, which is why the site is so popular.

Digg's members influence which stories get promoted by "digging" those stories. A digg is like a vote, and everybody gets one. How, exactly, a story winds up on the Digg homepage is never explicitly disclosed, but most people have assumed that the front page is simply a collection of the stories with the most votes. And, certainly, Digg has encouraged this view. Their about page simply says, "Once a story has received enough diggs, it is instantly promoted."

But a simple voting system is not necessarily the best way to provide an interesting experience for users. In fact, it might very well be the worst.

It's been said that wisdom is the ability to hold two completely opposing thoughts in your head at the same time. So try being wise with me and see if you can hold both of these ideas in your head for a moment: When it comes to group behavior, people are both stupid and brilliant.

People are stupid in groups. Think about popularity for a moment. You know what the most popular TV show of all time is? Baywatch – a show with the primary distinguishing feature of slow-motion shots of Pamela Anderson in a swimsuit running down a beach. Think about mob behavior and our basest instincts. Think about the prom king and queen. If we learned anything in high school, it's that there's no better way for the crap to rise to the top than a popularity contest.

People are brilliant in groups. Crowds can also be wise. The famous example is to ask a group of people to guess the number of coins in a jar. Any single individual's answer will most likely be wrong. But if you take all the answers and average them together, most of the time, they're right (give or take a percent or two). Groups are also the foundation of democracy, which, for all it's problems, is pretty much a great thing.

So what's going on here? How can people in groups be both brilliant and idiotic? One word: Interface.

In the television example, people are just acting on their own, engaging their baser instincts, making selfish decisions. But in the coin-guess example, there's a structured interface in place – a moderator asking a specific question and then working with the data to produce the result. Bottom line: Crowds are not wise by default – they're wise when guided with a specific task and given specific moderation.

Back to Digg. First of all, it's their site, and they can do whatever they want. They have as much right to exclude sites or links for editorial reasons as Flickr has to exclude illustrations because they're not photos. And I applaud their efforts to keep their site on target. In the end, it's a site for readers, and the readers want an interesting front page to read every day. It's Digg's job to serve them up a great experience, or they're gone.

But Digg has fallen victim to its own system of mob rule. While a most-votes-first, tyranny of the masses approach works when your community is small, as you get bigger it falls apart as members gang up to game the system for their own purposes.

This will lead some to abandon the idea of giving the users a vote at all. A similar news site, Fark, experimented in letting their members vote on story placement, but abandoned the idea when they didn't like the results. But that, too, is a mistake.

The problem is not the voting, the problem is the mob rule. The crowd needs more guidance to be wise. Sorting the content by votes is often the first approach on sites like this, but it's not the only way to sort data. The site could just as easily experiment with other sorting methods. For example:

• Voting Velocity. Who cares how many total votes a story got? Instead, pay attention to the timeframe in which the votes came in. A story that gets a lot of votes very quickly would outrank a story that got more votes over a longer period of time. This sort would favor hot news over the slow burners.
• Member Karma. The system could assign "karma" points to members that consistently participate in quality ways (as judged by the site founders, naturally). The contributions by members with high karma could then comes in higher than the contributions of members with low karma. (Slashdot has a version of this running now, but it only applies to comments and is mostly to combat spam.) This may sound exclusive, but remember, anyone can earn karma – they just have to participate, and participate well, for a period of time.
• Other Metrics. Digg has other metrics to use besides voting: Number of clicks on the link, number of comments, number views on the link's conversation page. Digg could add them all up, or factor them in to an overall rating, which would be more nuanced than a simple vote.

And then there are the other methods that other sites use. They could editorially select what goes on the homepage (like Slashdot or every newspaper out there), or just go newest first (like MetaFilter or every blog out there).

Note that all of these ideas could be inverted, too. You could just as easily use voting velocity to discount the fast-rising stories in favor of the slow burners. Or use member karma to promote contributions by low-karma members who are brand new instead of the oldies who have accumulated a lot of karma already.

These sorting methods are all editorial decisions. You select the one that best produces the kind of contributions you want to see, and engenders the kind of community you want to host. And here's the really cool part: The default sort doesn't have to be the same for everybody. Instead of looking at everyone's metrics, it could just look at mine. Instead of looking at everyone's votes, it could just consider the votes by my friends. The result would be a very different site: more personal filter, less New York Times.

A great case in point is Flickr's Interestingness. Flickr could have just counted up how many times a photo has been favorited, and been done with it. But they realized that would have just produced a crass vote-getting game culture that wouldn't have actually produced an interesting selection of photos. So, instead, they came up with an Interestingness algorithm. How it works is a secret, but it takes into account all kinds of user behavior, above and beyond simple voting, to determine what photos are interesting.

Is it perfect? Hell, no. So they're constantly tweaking it to make it better (much to the chagrin of the Interestingness-watchers who are always trying to decompile the system in order to game it). This is reminiscent of Google's closely guarded and oft-tweaked PageRank algorithm that determines the order of their search results. Both companies had the same problem that Digg is starting to have: Too much data, a community of contributors with a huge incentive to game the system, and an audience looking for a quality experience.

As the web matures, and we get better at developing member-driven media sites like Digg, we have to look beyond simplistic majority-rule popularity contests if we ever want to take on traditional editor-driven media. People are complicated, and we're going to need complicated systems to really draw the wisdom out of the crowd.

In the end, sites like Digg are under no obligation to blindly follow the will of the people. They're not democracies and they're not nations, they're websites. And if you don't like the way they're doing things, don't give them your clicks. Or, even better, start your own and show 'em how it's done.

More thinking about Digg elsewhere:
» Digg Corrupted: Editor's Playground, not User-Driven Website
» Digg.com Is Silently Moderating Legitimate Submissions
» Digging The Madness of Crowds
» Digging Fraud (Kevin Rose Responds)