It's primary day in Virginia, where the state's Democrats are set to pick from three candidates for governor: State Sen. Creigh Deeds, who is now the frontrunner in all the polls; former state Del. Brian Moran; and former DNC chairman Terry McAuliffe, the colorful personality who campaigned for so long on Hillary Clinton's behalf, and who later became the frontrunner in this race for quite a while, but may have now blown it.

To be sure, Terry had some weaknesses all along. First of all, his tenure as DNC chairman was a period of one Dem failure after another, regardless of whether that was his own fault or due to circumstances beyond his control. And once it became clear that Hillary Clinton wasn't going to the Democratic nominee, his advocacy of her reached newer (and stranger) heights, with talk show hosts openly joking that he might have been on drugs. You got the feeling along the way that he was deliberately turning this into performance art -- such as when he appeared on Morning Joe in a Hawaiian shirt, waving around a bottle of Bacardi to celebrate Hillary's win in the Puerto Rico primary.

But in many respects, McAuliffe went into the race with all the big advantages. First was money. McAuliffe was the big-time leader in the money race: The most recent figures show he took in a total of $6.9 million for this race, tapping into his contacts from the business world, his time as DNC chairman, and his connections from the Clintons. In distant second was Moran with $3.8 million, and Deeds at $3.4 million. This money advantage allowed McAuliffe to advertise on TV for months now, while Moran and Deeds only just recently went on the air.

Terry's lead in the polls turned out to be quite fragile, however, as soon as he came under sustained attack. Moran began aggressively attacking him, which appears to have worn him down. Only what hurt McAuliffe didn't help Moran. The real beneficiary was Deeds, who for much of the race was treated like something of an also-ran by the media.

Deeds is considered to be the most conservative Democrat in the race on issues like guns, hails from the less Democratic southwest region of the state, and previously lost the 2005 election for state Attorney General (though it was a super-close result that required a recount to make sure). At this point the big newspaper endorsements, notably the Washington Post, have combined with the negativity between McAuliffe and Moran to put him to the front as the positive choice.

One caveat though: primary polling in Virginia can be very unpredictable because the state's primaries typically have astoundingly low turnout, and on top of that there's a severe thunderstorm today. In the 2006 Senate primary between Jim Webb and businessman Harris Miller, only about 155,000 people voted -- compared to 1.17 million people who voted for Jim Webb in the general election. The 2008 presidential primary between Barack Obama and Hillary Clinton saw a healthier 986,000 voters. But the state's normal down-ticket behavior soon reasserted itself -- a contested Congressional primary for a seat that the Dems were poised to pick up attracted less than 25,000 people.

But insofar as polls are the only measurement we have to rely on at this point -- and thus are still quite valuable -- take a look at the turn of events in the Pollster.com graph:

Just check out these polls from April through May:

SurveyUSA: McAuliffe 38%, Moran 22%, Deeds 22% (April 28)

PPP (D): McAuliffe 30%, Moran 20%, Deeds 14% (May 5)

SurveyUSA: McAuliffe 37%, Deeds 26%, Moran 22% (May 20)

Kos/R2K: McAuliffe 36%, Moran 22%, Deeds 13% (May 20)

PPP (D): McAuliffe 29%, Moran 20%, Deeds 20% (May 22)

And now look at some of the latest polls:

Kos/R2K: Deeds 30%, Moran 27%, McAuliffe 26% (June 3)

PPP (D): Deeds 40%, McAuliffe 26%, Moran 24% (June 7)

SurveyUSA: Deeds 42%, McAuliffe 30%, Moran 21% (June 8)

It's quite a difference. McAuliffe has tanked, Deeds has risen, and Moran's position isn't significantly different from where it was before. We'll find out tonight, at 7 p.m. ET, whether the surveys have accurately predicted the will of the voters -- and just how many voters show up.