Winnipeg Free Press - PRINT EDITION
Noted predictor embraces uncertainty as virtuous
WHEN American statistician Nate Silver's The Signal and the Noise came out last fall, it quickly assumed a prominent spot on bestseller lists.
But even Silver, a reigning and still-rising star of American geekdom, would probably not have predicted the dramatic upturn in his book's popularity (reported to be up more than 800 per cent) after his widely visited political website, FiveThirtyEight.com, posted bang-on state-by state forecasts of the outcome of last November's U.S. presidential vote.
The Signal and the Noise is about predictions -- where they have been working and where and why they so often fail. Understanding how our public and private predictions can be improved is, according to Silver, a matter of distinguishing "signals," or patterns pointing to objective truth, from "noise," which is essentially false interpretations or inferences founded upon inadequate data or research bias.
Taking the reader on a tour of the state of contemporary disciplines from the natural and social sciences, Silver consults with a range of leading experts on the predictive accuracy of everything from the likelihood of a major terrorist attack to baseball-player performance.
Silver first earned notoriety for his statistical analysis in the late 1990s with the Oakland Athletics under general manager Billy Beane. Their story was told in the 2003 book Money Ball, which was adapted into a hit movie in 2011 starring Brad Pitt and Jonah Hill.
Weather forecasting, according to Silver, is pretty good and consistently improving. Economic projections, on the other hand, are characterized as so unreliable that many of those practising this dark science are incapable of acknowledging recessions -- even as they are occurring.
The prognostications of most American TV political pundits (think Fox News or the McLaughlin Group) should be regarded as little more than theatre and, apparently, any seismologist worthy of a Richter scale will concede that it is all but impossible to say in advance where and when an earthquake will occur.
Silver hits close to home a couple of times -- first with his review of the 1997 Flood of the Century and the inability of officials in Grand Forks, N.D., to predict the extent of the deluge that was about to overpower our neighbours to the south.
Overconfidence, and a failure to fully communicate the extent to which even the slightest variation of estimated Red River crest levels could devastate the city, were at least partially to blame for the ultimate flood toll.
A second local angle involves a profile of Winnipeg-born and raised sports bettor Haralabos "Bob" Volgaris.
Volgaris now resides in mansion high atop the Hollywood Hills, thanks to his invention of a proprietary wagering system said to be fastidiously balanced by statistical and intuitive inputs.
But despite the (so far) indisputable success of Volgaris as a professional gambler, Winnipeg unfortunately comes out on the bottom. Silver describes our dear home as a "hardworking but frostbitten city" and a place to be escaped if at all possible.
At the end of the day, Silver wants prediction to get better. He urges forecasters of all types to adopt a more probabilistic approach, to recognize and tame our unavoidable biases and to regard at least some uncertainty as a virtuous requirement for all good predictions.
Scott MacKay is the president of Probe Research, a Winnipeg-based public opinion and marketing research firm.
Republished from the Winnipeg Free Press print edition January 12, 2013 J9
Have you found an error, or know of something we’ve missed in one of our stories? Please use the form below and let us know.