Voter ID, Absentee Ballots, and Voter Fraud

I know, I know, I'm obsessed with Minnesota's voter ID amendment. Three new relevant pieces from today's Star Tribune:

• an op-ed by Independence Party guys Tim Penny and Tom Horner called Pass on Voter ID, but Add Ranked-Choice made a lot of good points. 
• Even better was a letter by David Smith of Minneapolis, who took on an earlier op-ed by former Senator Norm Coleman. Coleman made the mistake of comparing the problem of voter fraud with our societal response to drunk driving. Smith responded: "That's an analogy easily turned on its head. Although a case of drunken driving can lead to tragedy, we do not have zero-tolerance policies. We could lower the acceptable blood-alcohol level to zero, increase random police stops and require breathalyzer lockouts on all cars -- but we don't, because the expense and inconvenience is not considered acceptable. Why should we endure longer and more expensive voting processes in order to counter an insignificant problem? It's an expansion of government that the libertarian faction in the Republican Party should be opposing much more vehemently." Smith's argument reminds me of my earlier post on the problem of setting policy based on edge cases.
• Most important to read, though, was a New York Times story on the problems with absentee ballots. Increasingly popular because they cost less to administer than in-person voting, absentee ballots now account for 20 percent of votes cast. But there is a much higher rate of problems with absentee ballots, as well as a much higher fraud rate (which voter ID advocates don't seem to be concerned about for some reason). 

Continuing from the New York Times story: In the 2008 presidential race, "35.5 million voters requested absentee ballots, but only 27.9 million absentee votes were counted, said a study by Charles Stewart III, a political scientist at the Massachusetts Institute of Technology. He calculated that 3.9 million ballots requested by voters never reached them; that another 2.9 million ballots received by voters did not make it back to election officials; and that election officials rejected 800,000 ballots. That suggests an overall failure rate of as much as 21 percent."

Ballots are rejected mostly because there is no signature on the outside of the envelope, or the signature is thought not to match the one on file. The story tells of one county in Florida that has improved its process to decrease these types of problems:

In Leon County, Fla., the rejection rate for absentee ballots is less than 1 percent. The instructions provided to voters are clear, and the outer envelope is a model of graphic design, with a large signature box at its center.

The envelope requires only standard postage, and Sancho has made arrangements with the post office to pay for ballots that arrive without stamps.

All good ideas. But none of that will help prevent fraud by absentee ballot. The story concludes with this:

Election law experts say that pulling off in-person voter fraud on a scale large enough to swing an election, with scores if not hundreds of people committing a felony in public by pretending to be someone else, is hard to imagine, to say nothing of exceptionally risky.

There are much simpler and more effective alternatives to commit fraud on such a scale, said Heather Gerken, a law professor at Yale.

"You could steal some absentee ballots or stuff a ballot box or bribe an election administrator or fiddle with an electronic voting machine," she said. That explains, she said, "why all the evidence of stolen elections involves absentee ballots and the like." 

A lot of the people I've been talking to about the voter ID amendment plan to vote by absentee ballot, and quite a number of small Minnesota towns have gone all-absentee. Four Western states have turned their entire election system over to absentee votes. Are 20 percent of those voters being disenfranchised?

How can we both make sure legitimate votes get counted, and illegitimate votes are stopped?