Just as happened with the Web, the natural tendency of Twitter will be to eventually whittle down the number of feeds devoted to corporate proselytizing and self-indulgent moment-in-the-life catalogs. That process is already happening but not fast enough. What will accelerate it will be methods of automatic filtering, whether developed by Twitter or others, that will push what is most interesting to the top. Using programs such as TweetDeck, it's already possible to categorize and filter Twitter feeds.
That alone, though, is not enough to channel the Twitter torrent. There is just too much material in 200 feeds; a few people will make the effort to elaborately categorize it all, but most will not. Somehow, for Twitter to remain viable, this will have to be automated. That could be done by harnessing the power of groups, in the manner of Digg, with tools that will push up the tweets that other followers have found worthwhile. It could also be done (I suspect more effectively) by programs that let users mark the most important tweets and learn to look for and highlight similar material.
Whether people will be willing to leave this kind of selection to automatic agents is something that we'll have to see. In the case of e-mail, we are not usually willing to do that. We want our e-mail systems to get rid of obvious spam, but while the technology is there, very few folks seem willing to let a bot guide their e-mail reading. With Twitter things may be different; for most people, Twitter feeds fall into the "nice to keep up with" drawer and so are different from e-mail, in which we absolutely, positively don't want to miss anything important.
More Web 2.0 Coverage from The Big Money:
The disappearance of pointless feeds, the narrowing of truly useful ones, and the emergence of systems that will automate the process of judging what's important and highlight it are the baseline for what Twitter needs to avoid collapsing under its own weight. It's certainly possible, though, that all that won't be enough. Twitter is an extremely clever idea, and it will continue to exist in some form. But that form may be very different from what we see now. It may, for instance, eventually turn into a way of pushing information to Web sites and Facebook pages; we will then get the feeds in our ordinary use of the Web, without specifically choosing to become "followers" and accumulating an ever-growing collection of feeds.
Certainly the folks who run Twitter may have new tricks up their sleeves. Not having been clever enough to invent Twitter, I would be loath to try to advise the people who did on what they should do to manage Twitter overload. But a prediction I am willing to make is that just as the growth of Twitter has been so rapid that it caught almost everyone by surprise, the problem of containing and managing the overload will become paramount for Twitter sooner, more forcefully, and more suddenly than folks expect. A lot of people have already had the experience of seeing Twitter abruptly go from engaging to essential to all-consuming. But it's become a little bit like the map that Lewis Carroll once envisioned: having grown to be as big as the world itself, it ceases to be valuable as a map. The question now is whether Twitter can be tamed before it too has grown too world-encompassing to be useful, and the cacophony of signals turns into pure noise.