Should Pundits Publicly Bet?

Former baseball statistics analyst Nate Silver, now blogging at the New York Times on elections, gives Obama an 86.3% chance of winning the Electoral College as of today. (I love the decimal point.)

Recently, Silver offered an even money $1,000 bet to TV talking head Joe Scarborough that Obama would win, for which he has widely praised by the increasingly dominant-in-the-media sabermetrics boys. (You know, the baseball analysts who lauded stats-savvy baseball executives like Billy Beane, Bill James, and Sandy Alderson for their brilliant stats-driven insights that they could win baseball games with formerly skinny guys like Jose Canseco, Mark McGwire, the Giambi Brothers, David Justice, David Ortiz, and Manny Ramirez who, remarkably, filled out into homer-hitting behemoths, thus proving Science! is always right.)

Yet, shouldn't Silver have defended the honor of his probability percentage forecast by offering Joe Scarborough odds of, say, 3 to 1? In general, shouldn't Silver present his odds not in scientific-sounding percentages all the way to the first decimal point, but as old-fashioned race track fraction odds like 3 to 1 or 11 to 2?

By the way, I have some sympathy for the NYT Public Editor's discomfort with public bets involving pundits, the best known of which was John Tierney winning $5,000 in a 2005 to 2010 bet with oil price forecaster Matthew Simmons that oil would not rise in five years from $65 per barrel to $200. As I wrote in 2010:

I certainly would have bet against Mr. Simmons, as well, but he didn't offer to bet me.  

My suspicion at the time was that Simmons was, more or less, writing a $5,000 check postdated January 1, 2011 to buy his book publicity right now on the NYT op-ed page. That might explain why he didn't haggle and try to split the difference with Tierney, such as putting the win-lose line halfway between $65 and $200, the way a real betting man trying to make money would. But a public bet that the price of oil would be, say, $132.50 or higher would have been kind of a boring story. In contrast, a man confident enough to put $5,000 on the round number of $200 is a man who is acting like he might know something, and thus you'd better buy his book to find out what it is.

Still, there is a lot to be said for public bets. I offered to bet pundit Michael Barone $1,000 in early 2004 that Hispanic turnout in the fall of 2004 would be closer to my estimate of 6.1% of the total vote than his 9% speculation (according to the Census Bureau, it was 6.0%), but Barone prudently shied away. Yet, Barone appears to have learned from the experience, so it was a net benefit for quality of discourse.

Moreover, I've long offered to publicly bet that various notorious Gaps in real world performance among demographic groups are not going to vanish over any tractable time frame.

In general, however, nobody is interested in betting against me that, say, school test scores in San Marino will continue to be higher than in Compton, or however you'd want to specify a bet over The Gap of gaps. That's too boring and depressing to bet over. People are interested in betting as action, not in betting as a tool for learning about reality.

What if in 1969, Arthur Jensen had made public bets with leaders of the conventional wisdom, with Jensen taking the unpopular side that in 2012, 43 years later, The Gap would be about the same as in 1969. Would that have made him more popular at his death?