The folks at FiveThirtyEight.com do the best possible job in predicting unimportant things like who'll be our next president. But now they're getting into the critical matters…like who'll win the Academy Awards. I'll predict they'll eventually realize they're trying to predict the unpredictable.
This is not to say they won't get some forecasts right. Any of us could just going by the "buzz" and the nature of some nominations. I would say for instance that Sylvester Stallone is quite likely going to win Best Supporting Actor for Creed but I'm not basing that on a point system based on other awards the way the FiveThirtyEight folks are doing. I just think the "story" there — Stallone, who he is, what that win would mean dramatically, etc. — makes voters want to see him snag the statuette.
But that's not a scientific formula, it's just a hunch. As even the FiveThirtyEight experts note…
We can't poll the 7,000-odd people who vote on the Academy Awards. We don't even have a good idea of who exactly they are. This means that unlike the methods we use to predict, say, the Iowa caucuses, the problem we're trying to solve is pretty much stripped of input data.
So there's no input data but they're going to try to formulate predictions anyway. At some point, they're going to figure out that won't work. Or so I predict.