Of course I feel silly pointing to the windsock only after the Hindenberg has crashed, but I need to tell you that I have been avidly tracking the US presidential election via Race2004 for a couple of months now.
Their final projection, based on a continual aggregation of just about every opinion poll taken in the states since the DNC, turned out to be stunningly accurate. The only state Stephen missed was Wisconsin, and only by a handful of votes.
Opinion polling data is inevitably flawed by sampling errors and idiosyncratic methodology, but time-weighted aggregation of the data reduces the former and averages out the latter, and at the same time enables pollster-independent trend analysis. It therefore produces a pretty useful projection. Especially in a nation that so relentlessly pesters its people for their opinions as the United States.
On the radio I hear Americans—especially of the Democratic persuasion—reflecting on the figures and asking "why is our country so divided?" In fact, barring significant anomalies, America is likely to remain "divided" (in this numerical sense) well beyond the ambit of its immediate political context. The reason for that is precisely the kind of information found on Race2004—which both major parties surely accumulate for themselves. They pick their "battleground states" on the basis of the map that emerges, strategically ceding unlikely pickups, in search of the magic number: 270. 270 or 483, it means the same thing, so why expend money and energy where you're sure it's not needed? Both parties now separately draw lines down the demo-geographical middle of the nation—not the ideological middle. It is this that made the policy stances of Bush and Kerry seem mildly incoherent to external observers (though I suppose that in Bush an innate tendency is at least partly responsible).
Anyway, is it just me, or have you also noticed that this creepy, persistent sheen of unreality is suddenly a little more opaque?
Joseph | 3 Nov 2004