"It's so crowded, nobody goes there anymore." The paradox is attributed to Yogi Berra, and despite its naked illogic you know exactly what he meant.

The Browser braintrust has been thinking about that quote for a few weeks, since reports began to surface that traffic at social networking sites like MySpace and Facebook is beginning to trail off. Why should that be?

It's well-known that users of sites like these can be as fickle as the teenagers that many of them are; just ask the good people behind erstwhile category leader Friendster. The hard question is: why does a site go from hot to not?

In Friendster's case, as Gary Rivlin masterfully documented a couple of months ago, the site had slow-loading pages and failed to move quickly enough into the music space that made MySpace a hit.

But it's also a possibility that scale alone can take the community out of community sites. That is, at a certain number of members and active participants, a community Web site qualitatively changes into something different than its original organizing principle.

So for example: based on number of digs, the most important stories to Digg members are consistently stories about technology, and fairly geeky technology at that. A desire to know about technology news and opinion clearly motivates people to visit Digg.com and to register their opinions. However, the overwhelming tech focus is also a function of scale; Linux stories, for example, perform well on Digg because Linux programmers and devotees are greatly overrepresented on Digg.com. As the user population of Digg.com increases and begins, statistically speaking, to more closely represent the population as a whole presumably the proportion of Linux-related stories among the "most Dugg" will fall.

We got a sense of this this week, when Digg started ranking videos according to user preference; the list is profoundly different from the list of Dugg stories, and is relatively tech-free. Maybe that indicates nothing more than the fact that what makes a good video is different from what makes a good Web story, but it got us thinking. If the entire American population were active Digg members, the most Dugg stories would presumably begin to look a lot more like the home page of CNN.com, or the "most e-mailed" page on Yahoo! News - no Linux there!

The question is: would that represent a success or a failure for Digg? On the one hand, as a real-time snapshot of the "wisdom" of the American crowd, it represents a triumph, a kind of democratic media mirror. On the other hand, it's a tower of Babel, a high-tech tautology that says little more than "People are interested in what people are interested in" (and we're back to Yogi Berra).

The ultimate significance of Digg, then, has arguably not been its heralded voting mechanism, but rather its ability to keep community defined at a particular size and relevance. Which means that the imperative of Web 2.0 companies to get as big as possible as fast as possible may in fact be a mistake.

Not to say that there is a fundamental flaw in the wisdom of crowds thesis per se. But as The Browser's friend Ben Schott once eloquently put it, "Anyone who's ever actually been in a crowd knows that it isn't wise". That's not quite right, but what does make sense is that the crowd's wisdom is a function of its size and the people who constitute it.