The methodology was to start with best-seller lists--and there's the rub. Books which are best-sellers are generally crafted to appeal to a specific audience. It's no surprise, then, that publishers know their trade, or that sales patterns reflect that knowledge. What would a different methodology--one designed to study books lower on the list--tell us?

What do people who read Republic.Com read? What do people who read A Problem from Hell read? I wouldn't describe either of those books as a polemic, or as a best-seller.

Most importantly, what are the books that do cross over? What sort of books are they? How do they help form peoples' ideas? Who's the better propagandist, Ann Coulter or Tom Clancy?

So much study of this sort focuses on the "big picture" and leaves one disheartened. This one is a picture of the status quo--a "You Are Here" mark on a blank roadmap.

Of course people who read partisan books tend to be partisan, and of course such people are unlikely to change their minds because of a partisan argument from their opponents. If I spent my time thinking about that, I'd be a very unhappy and ineffective person. I'd rather think about shifting that center Valdis points to (okay, it's a cut set, but I think of it as a center).

Valdis sneaks up on this thought, but from a very pragmatic, tactical angle: "So, if you are working a 2004 political campaign what do you do with this information? Obviously you will not be successful in removing a reader from deep in one cluster and transplanting them into the other cluster. All you can do is focus on the edge nodes and the bridges."

That's great campaign advice, but not much of a long-term strategy, for politics or for democracy.