Bush at War" and "Sleeping With the Devil" are just two of the political books that have dominated the best-seller list of The New York Times in recent months. But according to Valdis Krebs, a social-network analyst in Cleveland, these volumes — the first a blow-by-blow account of White House deliberations in the aftermath of Sept. 11, the second an exposé of corruption and hypocrisy in American-Saudi relations — share an unusual distinction. They occupy a sparsely populated middle ground, rare titles that have been bought by people who generally tend to shop for much more partisan polemics.

Could it be that readers of these books represent some of the coveted undecided voters in the November election? That's one possible implication of Mr. Krebs's "network map" of popular political books. Curious to learn something about consumers of political titles in a tense election year, in January he analyzed purchase patterns for such books, using the Times's list and the Internet. First, he plugged titles from the list of The Times into Amazon.com and Barnes &Noble.com. Then, thanks to a "customers who bought this book also bought" feature on the Web sites, he was able to determine what other titles buyers had purchased at the same time. Following the links between titles, Mr. Krebs ended up with a list of 66 books.

His map showing how the titles are connected by buyers reveals a readership — or at least a book buyership — as fiercely polarized as the national electorate is said to be. On the left is a cluster of several dozen liberal polemics (the blue nodes) linked by a dense thicket of crisscrossing gray lines. On the right is a nearly identical cluster of conservative tracts (the red nodes). Connecting the blue and red sides of the map are just a few gray lines and gray nodes, all politically moderate or nonpartisan titles, including "Sleeping With the Devil" by Robert Baer, "Bush at War" by Bob Woodward and "All the Shah's Men" by Stephen Kinzer, a cultural correspondent for The New York Times.

Mr. Krebs, who got similar results when he conducted the same experiment last year, calls this pattern the "echo chamber" effect: for the most part, he found, buyers of liberal books buy only other liberal books, while buyers of conservative books buy only other conservative books. This finding appears to buttress the argument made by Cass Sunstein, a law professor at the University of Chicago, in his influential study "Republic.com" (Princeton University Press, 2001) that contemporary media and the Internet have abetted a culture of polarization, in which people primarily seek out points of view to which they already subscribe.

"These are people who are ardent supporters of either the left or the right," Mr. Krebs said in a telephone interview, referring to the book buyers represented on his map. "Through these books each side is solidifying its base."

The exception appears to be buyers of titles in the gray zone — what Mr. Krebs calls "bridging books" — which led him to this tantalizing, though admittedly untested, speculation: "It's through those books that people can get to both sides." Or, as he puts it on his Web site, www.org net.com: "See someone reading `Sleeping With the Devil'? That is someone you can talk to about your candidate." (Of course, it is always possible, he concedes, that undecided voters aren't reading political books at all, that they simply "can't stomach either side.")

Even with the bridging books, the average distance between the map's left and right clusters is still four links — or in network theory parlance, "four degrees" — Mr. Krebs said. Given that the clusters represent ideological extremes, he reasoned that if he expanded his book sample to include nonpolitical best sellers like "The Da Vinci Code" and "The South Beach Diet," the distance between left and right would be reduced. To his surprise, that turned out not to be the case, though what, if anything, this means he is not entirely sure.

"It's possible there is some book or video or CD that everybody buys that would bring the two sides closer together," he said. "But when I did a search using the top five best sellers at the time, it didn't bring the sides closer together."

Mr. Krebs is more confident about his initial results. "I like to think that this is just another way of showing that this is going to be a closely divided election and that each of the bases is armed to the teeth," he said. "By spending a couple of hours on the Net, I was able to show that same pattern as marketers spending thousands of hours and dollars getting people's opinions over the phone."

Copyright 2004 The New York Times Company. Reprinted from The New York Times, Arts & Ideas, of Saturday, March 13, 2004.