David Brooks cites research looking at whether the Internet worsens political polarization. He finishes up suggesting that it doesn't.

If this study is correct, the Internet will not produce a cocooned public square, but a free-wheeling multilayered Mad Max public square. The study also suggests that if there is increased polarization (and there is), it’s probably not the Internet that’s causing it.

I think he's missing two things. 1) The Long Tail. As Chris Anderson's book details, the multifarious channels offered by the Web, cable TV etc. have the effect of moving significant amounts of behavior to extremes of the traditional bell-curve distribution. It's the same in politics. More of us are ordering weird Swedish movies on Netflix. And more of us are gravitating toward nutty right-wing and left-wing beliefs, although the right-wing ones seem to be attracting more adherents.

2) The impersonality of the Web. The Web causes people to behave in ways and say things that they would never dream of doing face to face. Humanity has evolved a set of instinctive responses for face-to-face interaction that allow people to deal with each other in person without killing each other too often. People are programmed not to give offense face-to-face.

But once we're dealing with people at a distance, those behavior governors disappear. When it's some faceless stranger who's challenging you, you respond in ways that you would shudder to use when, say, talking to somebody at a restaurant. Same thing happens when we're driving.