Here's another potentially big problem with the Big Sort (again, the phenomenon, not the book). Sorting can make the electorate, as a whole, look more radicalized and polarized than it really is.

Is the public more politically polarized, across the board, than it used to be? Could be, and I wish I knew. That question is at the center of a big and unsettled debate within political science. But this much is certain: just because you and I are sorted into more homogeneous communities, it doesn't mean that you and I, as individuals, are more likely to disagree, or that we disagree more strongly. Maybe these things are true, but sorting, by itself, doesn't make them so.

At the same time, sorting can create an illusion of greater overall disagreement. Why? The simple reality of aggregating preferences and attributing, ultimately, one preference to one location. Which, for many purposes in any democracy, is just unavoidable. You can only elect one House member. And the media legitimately need some kind of handle to describe you.

Suppose, for just a moment, that every person in America could safely be labeled "a moderate liberal" or "a moderate conservative". Then suppose that, over time, the moderate liberals all wind up in the same place; likewise with the moderate conservatives.

Almost any way that you choose to average the preferences, you will see trends that suggest that "liberal town" and "conservative town" are moving ideologically apart from each other. And you can't easily disprove those claims. If those claims work to your political advantage, you use them. Under these circumstances, it's indisputable that the mythical "average" resident of "conservative town" is more conservative than she used to be.

While this is going on, it's possible that not one person has changed his or her mind about anything. And they're saying, "geez, I don't feel more radical".

Look at the graph published by Morris Fiorina and Samuel Abrams earlier this year*:

This graph reports percentages of different responses to political ideology questions on a seven-pont scale using the General Social Survey.

The graph is a little hard to read, but notice that more people call themselves "moderate" than any other ideological category, by a wide margin. You probably expect as much. But it's also rock steady, even as The Big Sort has unfolded.

Now, this self-identification could misrepresent what people really believe, but it's a puzzle. Especially given what we know about the increased polarization and increased ideological distance between elites and elected officials. Not to mention all the evidence about sorting from Bill.

Here's a test you might think about in your own legislative or congressional district: if you wanted to win office, which is more decisive: the primary or the general election? At the point when the primary election matters most, a lot changes. Even if voters haven't become more extreme.

Finally, it's still entirely plausible that sorting often, and maybe always, provokes greater disagreement because it kicks off a chain reaction of social processes. Bill discusses the ways that social networks, groupthink, and the spiral of silence among ideological minorities, can all radicalize individuals.

My suspicion is that those additional social forces are really important, but they unfold more in some places than others. And if you look for other places that are both homogeneous and where people are legitimately wondering, "OMG, it doesn't seem like we're that extreme", you can find them.

* Fiorina, Morris and Samuel J. Abrams, 2008. Political Polarization in the American Public. Annual Review of Political Sciencee, v. 11, p. 571.