I was on MSNBC’s “Up With Chris Hayes” yesterday doing my Ancient Mariner harangue in favor of the National Popular Vote, along with Yale’s Akhil Reed Amar, one of the intellectual fathers of that ingenious plan, which would allow us to elect our Presidents the same way we elect governors and senators and everybody else, i.e., the candidate with the most votes wins—and would do it without messing with the Constitution. You can watch the relevant three segments. (One follows another, with unavoidable commercials.) Akhil and I managed to squeeze in most of our arguments, but right at the end Chris brought up a question we didn’t have time to fully answer: What about recounts? What if Florida 2000 were reproduced on a national scale?
Chris said he finds this a “fairly persuasive” point. “When we kind of cabin things off in one state,” like Ohio, he said,
“then at least it’ll just be in Ohio. But if you had a national popular vote that was within a very small, small margin, running a fifty-state recount seems like that would really be chaotic and catastrophic.”
I hear this worry a lot, even from people who, like Chris, are supporters of NPV. And, at first, it does seem intuitively troubling. But it’s unwarranted, for many reasons. Here are three.
1. It’s a wildly far-fetched scenario. The bigger the pool of voters, obviously, the longer the odds of a tie or near tie. If you do the math, extrapolating from the frequency of statewide recounts for offices like governor and senator, you find that the likelihood of a nationwide election that’s anything remotely as close as Florida 2000—not just in absolute numbers (537 votes) or percentagewise (9,492 votes)—is preposterously low. How low? Well, in a national popular-vote election, a disputable result—one close enough to be theoretically reversible via a recount—could be expected to occur once every 1,328 years.