Mickey Kaus writes here:

The Schwartzman Test: A man named George B. Schwartzman finished a suprising ninth in the California recall race with 10.945 votes, right behind Gary Coleman. It’s not hard to figure out why. Blogger Matt Wall (a judge on a local election board) says this shows “the voting system has systematic error built into it”–big enough, he notes, to have decided Florida in 2000. … But was this an error that could be fixed by any kind of voting system? Did voters try vote for Schwarzenegger but accidentally vote for Schwartzman because of a confusing ballot? Or did they not remember Schwarzenegger’s name and actually vote for the name they intended (misguidedly) to vote for? The second type of error can’t be blamed on punch cards, or ballot design, etc. … The interesting study to make–attention, Rick Hasen–would be a comparison of Schwartzman’s showing in areas with different voting systems. It’s pretty clear that a higher Schwartzman vote means a higher error rate. If he did better in punch card districts, that would be evidence against punch cards. …

I’m not able to do such a study–at least not in the near term–but would be interested in hearing from anyone who looks into this issue.