Nate Silver’s Talent Isn’t Predicting, It’s Measuring
November 18, 2012 Leave a comment
The internet misunderstands Nate Silver. The joke hashtags and pundit-mocking based on the success of Silver’s model all focus on his ability to make predictions. And Silver does do a great job making predictions, but that’s only because he’s great at measuring. In the case of elections, Silver is an expert at measuring what the polls are really saying about public opinion. That’s why he left Romney a 9% chance of winning. If the polls were wrong, his measurement of what they said would be wrong. The same is true of Silver’s work on PECOTA at Baseball Prospectus. His system allowed you to accurately predict how Mark Gudzielanek would perform in the 2004 season, but that’s only because the system created a better way to measure Grudzielanek’s *true* performance from 2001-2003.
Is this a minor semantic distinction? Yes. But it’s an important one for the media to make when it comes to teaching the lay person about social science. Rather than convey the notion that some mathematician is in a lab creating an incomprehensible set of equations that can predict the future, a focus on measurement emphasizes data collection and all the moving moving pieces that are required to make an accurate prediction.