Do the math while doing the recount

The race for Senator Norm Coleman’s seat (R-MN) is an amazingly close one. Writer and comedian Al Franken, the Democratic challenger, has come very close to unseating the incumbent, but the margin — whichever way you look at the counts — is about as thin as it could be. The official tally before the recount put Senator Coleman ahead by only 215 votes, out of around 2.9 million, an edge of about .0074%.

Even that edge has gone down in the recount, but exactly how much is still in question, what with missing and disputed votes. The “search” for the missing votes has been interesting, with people literally looking in closets and desk drawers for them. Maybe the key ballot fell behind the sofa. In any case, the search has now officially stopped (one wonders what would happen if someone happened to move the right sofa tomorrow), and it’s now down to an analysis of the data and of thousands of disputed ballots (where the question is whether they’re filled out correctly, and whether the official count has correctly considered these voters’ intents).

But let’s take a moment to do a little long division. We’ll use the numbers from the NY Times for convenience. At this writing, it shows a difference of 206 votes between Senator Coleman and Mr Franken:

Candidate	Party	Votes	%
Norm Coleman	Republican	1,211,565	42.0%
Al Franken	Democrat	1,211,359	42.0%
Deak Barkley	Independent	437,389	15.2%
Charles Aldrich	Libertarian	13,916	0.5%
James Niemacki	Constitution	8,908	0.3%
(Total)		2,883,137

Now, a difference of 206 votes in 2,883,137 means that if one ballot in 14,000 was miscounted, the election could turn around. One in fourteen thousand. Do you think you could tally fourteen thousand ballots without making a single mistake? I don’t think I could.

And if you do think so, have a look at these images of actual challenged ballots (also linked from the image above). I find some (#2, for instance) to be straightforward, but some are really questionable, and certainly could have been miscounted. How many others were confusing to the counters, but not challenged? Would that make the difference in the election? One in fourteen thousand.

And remember that it isn’t just a question of tallying votes for Coleman and Franken: there are three other candidates too. They aren’t in the running, of course, but votes for them do affect the count, and the recount. Votes could have been misattributed for them or against them.

The right answer here is a problematic one: re-run the election, as a run-off between Mr Coleman and Mr Franken, with no other candidates. If that re-vote is too close to call, use the state’s tie-breaking rule — every state has one, from a coin toss to a selection by the Governor; I’m not sure what Minnesota’s is, but at least it’s codified.

It’s a problematic move for a couple of reasons. First, and most obviously, it’s expensive. But besides that, it’s also not the same thing. The momentum of the Presidential election is no longer there, a difference that’s likely to favour Senator Coleman, both as the incumbent and as the Republican candidate (Democrats who flocked to the polls to vote for President Obama won’t have that incentive now). We saw that when Georgia re-ran its Senate election and handily retained Senator Chambliss.

But it’s more fair that trying to decide a 200-vote margin based on ink-stained, smeared ballots. And it tells the people that we really care what vote they want to cast, rather than disputing ballots because of stray pencil marks.
 

---

As a side note, a look at these challenged ballots really highlights why we have to move away from voting by pencil (did you imagine that the tiniest pencil-tap could call your heavy, black, fully filled bubble into question?), or, indeed, by other means that allows for misinterpretation (you’ll recall the “hanging” and “pregnant” chads, from Florida in 2000).