Yesterday afternoon Chelan County, Washington, Superior Court Judge John Bridges ruled against Republican gubernatorial nominee Dino Rossi, affirming that Democrat Christine Gregoire won last November's election. There were over a thousand invalid ballots cast, far greater than the margin of victory. As I understand it, Rossi's team tried to make a statistical case, based on where the irregular votes were cast, that enough of the invalid votes were cast for Gregoire that Rossi would have won if those votes weren't counted. The judge ruled that it couldn't be determined with certainty which candidate received the benefit of those invalid ballots, therefore the result stands. Michelle Malkin live-blogged the judge's press conference. She links to Seattle-area political blog SoundPolitics, which has been covering the situation in great depth since last November when the recounts began.

This result is puzzling. Under Oklahoma law, if the number of irregular or fraudulent votes is greater than the margin of victory, so that the result cannot be mathematically determined, the election is voided and a new election is held. This happened in Tulsa in 2004, in a Democratic primary for City Council. Incumbent David Patrick received three votes more than former incumbent Roscoe Turner, but in one precinct, 255 votes were cast, but only 207 Democrat voters signed in. Evidently, 48 Republicans who showed up to vote in the presidential primary were also given Democrat city primary ballots. After Turner contested the election and presented evidence of the irregularities, a judge ordered a re-vote, which Turner won handily. I'm surprised a similar provision doesn't exist in Washington law.

Jim Miller, one of the bloggers at SoundPolitics, appears to have coined a new term to describe what seems to have happened in Washington -- distributed vote fraud. Rather than a coordinated effort to stuff the ballot boxes in a few precincts, handfuls of ineligible voters cast ballots in each precinct -- maybe only 1 in every 1000 voters, maybe as high as 1 in 100, but more than enough to affect the outcome of very close elections, like Oklahoma's 2002 governor's race, which was decided by three votes per precinct.

Miller's disclaimer on the topic explains what it would take to determine the extent of the problem. He also observes that Democrats tend to favor policies (like the Federal "Motor Voter" act) that make fraud easier to commit and to oppose policies (like showing photo ID when you vote) that make fraud easier to detect or deter. He writes, "Perhaps all these Democrats are wrong to think that there is an advantage for their party in what I call distributed vote fraud, but I doubt it." You'll find a longer treatment of the problem here.