Q&A: Nate Silver on China and the New FiveThirtyEight; how he uses data to eliminate bias, and how data from China can come from unexpected sources; I flew through Beijing [on the way to Hong Kong] – there was less physical brightness coming from Beijing than you would have seen from a comparable American city or European city

October 21, 2013, 3:40 PM

Q&A: Nate Silver on China and the New FiveThirtyEight

Nate Silver, founder of the FiveThirtyEight blog, is tired of politics.

Mr. Silver is renowned for his work on big data that led him to accurately predict the winner of the U.S. presidential election – twice. In the last U.S. election, he correctly declared the outcome in all 50 states, well before voters had headed to the polls. The writer and statistician has taken on a varied list of projects—he started in 2003 with a baseball analysis system and also achieved moderate success applying statistics modeling to World Series poker.  In April, Mr. Silver announced the end of his blog’s relationship with the New York Times, and a move over to ESPN—an opportunity to widen his appeal. In an interview, the FiveThirtyEight Editor-in-Chief talked about how he uses data to eliminate bias, and how data from China can come from unexpected sources. Edited excerpts:

How much of your predictions are your intuition, versus pure data analysis?

Once the model is designed there are no subjective tweaks made to it. In any type of complex system, there is judgment involved in the way you design a model. I don’t say, “I don’t like this result, so let me change the model” – it’s a matter of being completely disciplined in how you apply it. There is science, judgment, and experience – however you want to put it – in the principles of model design.

Can you apply good data analysis to poor data, for example, in China?

People in the United States and the United Kingdom overestimate the quality of economic data. Even if people are above board, it is simply hard to estimate something like the American economy. With China, you would have even more difficulty. I think the general lesson is that by looking at a broader consensus of indicators, you do well than just looking at one indicator or one sector. It is problematic to think about “how do you measure Chinese growth”. One way [is to look at] more public facing measures – by looking, for example, at the amount of light output emanating from China. I flew through Beijing [on the way to Hong Kong] – there was less physical brightness coming from Beijing than you would have seen from a comparable American city or European city.Is your style of analysis becoming mainstream?

The book Moneyball was written almost exactly 10 years ago now, and in baseball that fight has been won. Teams now realize that here is the value that statistical analysis can provide. They didn’t displace the scouts, but they may have displaced bad decision makers who had no criteria for making decisions at all.

In a lot of other fields, analysis is still quite primitive. I think there is more attention to it now – you hear a lot about big data, you read trade periodicals and every other ad is for big data. I’m not sure people really know what that means exactly.

It’s going to take a while…You usually encounter a lot of resistance.

Change might occur a lot more slowly than we might expect.

What’s your end game?

Personally I’m a little exhausted by politics. I think the point has been proven there as well as it could.

It’s more expanding out in to different topic areas, different verticals that FiveThirtyEight will cover. We are hiring additional reporters to cover politics; it will still be a big focus, but sports obviously with ESPN, looking at economics, maybe some lifestyle issues – travel, for example.

Almost any part of the world you can apply a more critical, empirical, statistically driven approach toward things. Finding people who can write about it compellingly is a little more of a challenge potentially, but it’s going to be a mix of serious stuff and fun stuff.

There is nothing about my philosophy that would only apply to politics or sports.