Chris Satullo: Number cruncher looks like rookie of the year

The stat geek is

blowing up

.

In the last week, Nate Silver - a 30-year-old runner of algorithms - has:

Appeared on

The Colbert Report

and

Countdown with Keith Olbermann.

Done a featured interview in Time.

Been called "indispensable" by E.J. Dionne, grand sage of the Washington Post.

Seen his election projection Web site,

» READ MORE: www.fivethirtyeight.com

, get bombarded by 693,216 unique visitors in a day.

"It's been pretty surreal," Silver says.

All this for a guy who ventured his first political analysis anonymously, using the alias Poblano, on the Daily Kos blog just last fall.

Why is Nate Silver suddenly the It Boy of punditry?

Because, back when he was still Poblano, his projections nailed Super Tuesday's outcomes. Then, on the decisive night of the Democratic primaries, he lasered the North Carolina and Indiana results.

Now that Poblano has unmasked himself, baseball stat freaks are nodding and saying, "I knew it."

To them, Nate Silver was already a cult figure, guru of the Baseball Prospectus, inventor of PECOTA, an uncannily accurate system for predicting baseball performance.

How cool is PECOTA? Silver predicted last spring that the Tampa Bay Rays, who had had zero winning seasons in their sorry existence, would win 88 games. The Rays won 97 and made the playoffs.

Now, Silver is applying his just-the-stats-ma'am approach to politics. Here is some of what you'll find on

» READ MORE: www.fivethirtyeight.com

(the name refers to the total number of electoral votes):

A map with state-by-state projections and a likely electoral-vote tally (as of Friday, Barack Obama 347.6, John McCain, 190.4; 270 are needed to win). A listing of all legitimate polls in every state. A chart of the probabilities on how any given state will vote (Pennsylvania, 92 percent likely for Obama; New Jersey, 93 percent).

Silver's projection method includes a weighting of polls depending on sample sizes and each pollster's record of accuracy going back to 2000, and other factors such as a state's demographics.

Any baseball stat-head can tell you that the average fan holds in his head a bunch of wrongheaded notions. For example, wins don't measure a pitcher's worth (that's mostly luck), and the sacrifice bunt isn't smart baseball.

What similar political myths exasperate Silver?

"It still amazes me that people are shocked when the polls are off," he says. "They often are."

Another gripe is when pundits discount Obama's poll numbers due to the "Bradley Effect."

This refers to the alleged phenomenon, posited most famously in Tom Bradley's failed run for California governor in 1982, in which white voters tell pollsters they'll vote for a black candidate, then don't.

"Polling numbers from the [2008] primaries suggested no presence of a Bradley Effect," Silver wrote on his blog. "On the contrary, it was Barack Obama - not Hillary Clinton - who somewhat outperformed his polls on Election Days."

Yet talking heads flog this myth, and fretful Democrats make it grist for paranoia.

"The electorate has changed since '82," Silver says. "Polling has changed since then. A lot of polling is done robotically. Why would anyone lie to a robot?"

Silver stresses a basic, but often mangled, point: The Bradley Effect is not about whether some whites won't vote for a black candidate. He's sure that happens.

The Bradley Effect pertains only if white voters lie to pollsters about their intentions, or suffer widespread, last-minute changes of heart.

Instead, in primaries Obama outperformed his polls by an average of 3.3 points per state. This leads Silver to suspect there may be a "reverse Bradley effect." That might occur if pollsters' models - based on past elections - chronically underestimate turnout among two groups that are gaga for Obama, blacks and the young.

Part of this effect could stem from the cell-phone factor. Pollsters dial landlines; as many as 20 percent of Americans have only cell service.

"And this group tends to be young, urban, tech-savvy," Silver said. "In other words, Obama's base."

Silver is, he freely admits, an Obama supporter. But the first loyalty of this economics major and son of a political scientist remains the data. Data don't lie; they can even sing, if you know how to listen.

After the election, Silver expects to go on seeking the music hidden in the numbers. He wouldn't mind some down time ("I don't sleep much"), but he thinks tracking the how and why of votes in Congress, plus the 2008 mid-terms, could keep him busy.

"There's lots of unexplored territory," Silver says.