David Brat described his GOP primary victory over Eric Cantor as "a miracle from God."

So maybe the Almighty saw it coming. 

But virtually nobody else did. Cantor's own polls had him leading the Tea Party-backed Brat by 34 points, reported the Washington Post. Candidates' internal polling is notoriously unreliable. But a more recent sounding by Daily Caller, of  583 primary voters, also had Cantor winning, 52 percent to 40 percent with 9 percent undecided. 
The about-to-be-former Majority Leader is not a Republican In Name Only: He's a dependable conservative who has voted with his party 95 percent of the time. (Not surprising, given that he was a party leader.) He ran hard, not taking the race for granted. And he spent about $2 million, compared to Brat's $123,000. Some estimates put Cantor's spending as high as $5 million. 

But as Brat said, "The good news is, dollars don't vote. People do." 

In the primary in Virginia's conservative 7th district on Tuesday, 65,000 people voted--about 20,000 more than turned out for the primaries two years ago. Those were Brat's people, it's clear--energized by his outsider status, his hard-line stance against immigration reform, his success in tying Cantor to "big business," or all of the above.

Estimating turnout is a key part of projecting elections. In the 2012 presidential race, most pollsters assumed that African-American voters wouldn't turn out at the high percentage seen in 2008's historic election. Wrong! In the 2008 election, while pollsters figured that minorities would show up in unusual numbers, they underestimated the level of participation by young, first-time voters. 

Cantor's blow-out poll projections may have made his supporters complacent. They didn't show up, and Brat's backers did. No polling outfits or pundits foresaw that.

The one prediction that held up? Nate Silver's analysis, three weeks ago, that it was inaccurate to say the Tea Party is in decline.

But as the stunning results came in on Tuesday evening, even Silver had to admit his surprise. 

"Like earthquakes, primary challenges are hard to predict," he said