Nate Silver says his goal at the new FiveThirtyEight is to be a fox, not a hedgehog. That is, to know lots of little things, not just one big thing. Paul Krugman is not impressed with how that's turned out so far:
You can’t be an effective fox just by letting the data speak for itself — because it never does. You use data to inform your analysis, you let it tell you that your pet hypothesis is wrong, but data are never a substitute for hard thinking. If you think the data are speaking for themselves, what you’re really doing is implicit theorizing, which is a really bad idea (because you can’t test your assumptions if you don’t even know what you’re assuming.)
There are a whole bunch of new media ventures being launched right now: FiveThirtyEight, Ezra Klein's Vox.com, Glenn Greenwald's Intercept, and several others. That's sparked obsessive chattering among media types, which is entirely understandable. I've mostly stayed quiet because I figure there's really no point in saying anything until these sites have been fully up and running for a few months. Until then, things are just shaking out.
And I hope FiveThirtyEight shakes out. But at the risk of jumping the gun, its first day didn't do much for me, and I think the reason is pretty fundamental. My basic take is that Silver's data-driven approach to journalism works well with subjects that satisfy two criteria:
- They lend themselves to analysis via number crunching.
- They are currently underserved by serious number crunchers.
Both sports and poll aggregation fit this model, and Silver made a reputation with both of them. But they're the exceptions, not the rule. Economics? It doesn't satisfy #2. Science? Ditto. "Life"? That's a pretty broad category, but I suspect it mostly fails #1. This story about the search for MH370, for example, uses the totem "Bayesian" a lot, but basically just tells me that when the searchers get new information, it affects where they search next. And politics? Aside from poll aggregation, it's hard to say. There are plenty of political scientists who already crunch political numbers, but maybe they do it badly or present their results poorly. Maybe FiveThirtyEight can do better. However, this strained effort at analyzing gubernatorial elections doesn't seem like a promising start. Nor this article making the obvious point that polling in Crimea doesn't mean much these days.
To the extent that FiveThirtyEight is providing a quantitative take on ordinary news, I suspect its audience is small. Most people just don't want lots of numbers in their news. And to the extent that it's providing a quantitative take on quantitative subjects, it's competing with lots of people who are already doing it. So what's its niche?
We'll see. Maybe I'm too blinkered to see it. Or maybe I'm blinkered by the fact that I've been immersed in the semi-quantitative blogging world for over a decade. I already see a ton of this kind of stuff, so it doesn't seem to me like it's in short supply. But the supply is probably a lot shorter in big, highly-publicized mainstream sites like ESPN. So maybe that's what FiveThirtyEight is doing: bringing the wonkosphere to the masses. Maybe it will even work.