So, on the front page this morning there is the usual pundit round-up post, subtitled Quants vs. Pundits and the State of the Race. Quants vs. Pundits is a theme throughout the columns presented, but it is also the explicit subject of one of them, this piece on the Atlantic's website by James Fallows. It's a well-written column that dives headlong into the whole Quants vs. Pundits conflict, telling us that the "Republican pundits and political 'pros'" say that Romney is either winning or rapidly closing the gap, while the stat guys are saying that Obama's lead never went away, and that it is widening again. This, argues Fallows, is the political version of Moneyball. In his own words...
Apart from Silver's own background as a sports-stats analyst, we have an exceptionally clear instance of people judging from their experience, their "bones," their personal instinct, etc that things are going one way (like veteran scouts saying that a prospect "looks like a Big Leaguer"), while data (on-base efficiencies in one case, swing-state polls in another) point in the opposite direction.Now, the Moneyball comparison isn't awful, and as a proud member of the reality-based community, I'm glad that it's our guy who is relying on the evidence, as opposed to someone's gut feelings, but there is a fundamental flaw in the article: at no point does Fallows even attempt to go into the underlying motivations of the actors in this conflict.
The thing about Moneyball is, whether you're an old-school scout or a new-school stat analyst, or some kind of hybrid who tries to weigh both points of view, your goal - your agenda - is the same. You want to identify those players who are best able to help your team win as many baseball games as possible. You can question the wisdom of the people who use either approach, but the fact of the matter is they are all looking for the same result. And that is where the analogy falls apart on the political side.
Because the fact of the matter is, partisan pundits, 'neutral' pundits, and stat analysts do not have the same goals, or the same agenda.
Let's start with the partisan punidts - the Dick Morris/Karl Rove types. Their goal is a GOP Presidency. That's what they want. And how does Romney win this election? With an energized base, which will make GOTV efforts more successful, it'll make campaign contributions easier to coax from donors (not that that really matters at this point), basically it's like Kos says, over and over - an energetic campaign is only possible if people think that their guy is going to come out on top. And Morris and Rove know this. So of course they are going to say that all the signs are pointing towards Romney. Because if all Romney's people are hearing is that their candidate is doomed, they're going to stop putting out the effort that is needed to win. And the same goes for Democratic partisan pundits. Whether it's Rove, or Morris, or Gillespie, or Preibus, or Dean, Shalala, or whoever - if they are affiliated with a party or a candidate, as soon as they start making predictions, that's when it's time to stop paying attention. Because accuracy is not their goal.
What about neutral pundits? What is their agenda? Well, I'll give you a clue - it's starts with "R" and ends with "-atings." If they are going around telling their audience that, while Romney's first debate bump was significant, he was never able to completely erase Obama's lead and, in fact, now that lead is widening again, people are going to change the channel, because the drama is gone. Once again, the agenda is something other that accuracy. They need to tell people that the race is thiiiiiis close, and that any second some candidate may make the gaffe or great point or whatever that could swing the election at the last second, so stay tuned for live coverage in case that happens today. Accuracy is a nice bonus, if it reinforces that narrative. If it doesn't, it is sacrificed to the God of Ratings.
So, does anybody have accuracy as an end - THE end - to itself? The answer is yes. Specifically, accuracy is how one Mr. Nate Silver earned his reputation, his quasi-celebrity status, and, I'd be willing to postulate, a good deal of his current net worth. With the exception of Barack Obama, nobody had a better night, from a professional standpoint, on Election Day four years ago than Silver. His blog, which was then known to only hardcore political junkies, correctly picked every state except I believe Missouri and Indiana, got the national percentages right, correctly pinned down which states were going to be close and which weren't, and to what degrees. He pretty much nailed everything, and his reward was a spot on the New York Times webpage, a publishing deal, and TV spots. All of this dependent entirely on the accuracy of his forecast, and nothing else.
Dick Morris can get every prediction wrong from here to the apocalypse, and it won't matter a smidgen. I know this, because he has spent the better part of the last 6 years being wrong about virtually everything, and he's still on TV every day, spouting his nonsense. But Nate Silver knows that the Times bought him out because he made accurate predictions in the past, they expect him to make accurate predictions in the future, and if they can't count on him to deliver, they're going to find someone on whom they can count.
The best place to find good predictions is to start by looking for someone who is TRYING to make good predictions.