Stewart Mills is Wasting His Money on a Recount

Stewart Mills, a repeatedly unsuccessful candidate for Minnesota’s 8th Congressional District, has decided to pay for a recount of his race, which was not close enough to qualify for a free recount. MPR’s Brian Bakst reported that Mills said “we are well within the margin where a hand recount, where human eyeballs on every ballot could make the difference between victory and defeat.” Bakst also reported that “he’s not alleging fraud or impropriety. But he said people might not have properly darkened ovals or could have left stray marks on their ballots that confused the counting machines.” In other words, Mills is making three claims:

  1. The recount will not find signs of fraud or improper tabulation. Instead, the problems found will be limited to those that affect individual ballots, such as improperly filled ovals.
  2. The recount does stand a good chance of finding such ballots.
  3. The number of such ballots found might be “well within” the range that would make a difference in the election outcome.

Of these three claims, the first two are based in fact. The third is not.

There are no large tabulation errors to find

Because the total number of ballots cast in each precinct is matched to the number of voters who signed in, the only way for the tabulation to be wrong is by assigning the votes to the wrong candidate. This can happen even without fraudulent tampering. If we look for historical precedents across the nation, rather than only in Minnesota, scanners have on rare occasion been misconfigured so that two candidates’ positions on the ballot are swapped, resulting in their vote counts being attributed to each other. Pre-election testing is supposed to catch such errors, but on rare occasion (again, outside Minnesota) the testing has been negligently performed such that the error slipped through.

However, even if we aren’t willing to credit Minnesota’s election officials with being more careful than their peers, or we are concerned the same effect could be achieved through skillful tampering, we can be sure that thousands of votes haven’t been shifted in this way. There are two options for how a large shift could occur: by shifting a few votes in each of a large number of precincts, or by shifting a large number of votes in a few precincts. (The former would include swapping the vote counts in lots of close precincts, while the latter would include swapping the counts in a few precincts that Mills won by a large margin.) We can rule out each of these two possibilities, for different reasons.

Mis-tabulation spread thinly over many precincts is precisely the kind of problem that Minnesota’s post-election review (audit) is well equipped to detect. A random sample of precincts in each county already had their ballots hand counted for comparison with the scanners’ counts. If large numbers of precincts had been mis-tabulated, the chances are very high that one of those precincts would have been chosen for review.

A few precincts with major shifts, on the other hand, would have attracted attention. For example, if some heavily DFL-leaning Duluth precincts had gone lopsidedly for Mills rather than Nolan, everyone would have noticed the oddity.

Some non-machine-readable votes will likely be discerned

Minnesotans have looked at millions of optical scan ballots in the 2008 recount of the Coleman/Franken race for US Senate and the 2010 recount of the Dayton/Emmer race for Governor. We know that some voters indicate their intent in non-standard ways. If enough ballots are looked at, some of these non-standard ones will surely be found. In addition to the incompletely filled in ovals that Mills mentions, I wouldn’t be surprised if a ballot turned up where the voter had circled the candidate’s name rather than marking the oval at all. We might even see a case where someone filled in both candidates’ ovals, put an X through one of them, and put an arrow pointing at the other with the words “this one.” These are not far-fetched conjectures. They are actual experience from past recounts.

The number of such ballots will, however, be too small to matter

The prior recounts also give us an indication of how common it is to find voter intent expressed in a non-machine-readable fashion. The answer may depend to some degree on how hard the candidates are looking for votes. In 2008, more were found than in 2010. However, even in 2008 the numbers were in the hundreds, not thousands. For example, in Saint Louis County (which provides about 1/3 of the 8th District’s votes), Franken picked up 112 votes in the recount and Coleman gained 53. A third-party candidate also gained 8 votes.

Extrapolating from the 2008 experience, neither Nolan or Mills can pick up more than 350 votes in the recount. Based on the 2010 experience, the actual numbers may be substantially smaller.

Chances are that the net change in the vote margin will be smaller yet, because the two candidates extra votes will tend to cancel out. (For example, in the 2008 numbers I quoted above, Franken netted only 59 votes, as 53 of his 112 were offset by Coleman’s gains.) However, even if it is only Mills’s supporters, and not Nolan’s, who fill out their ballots in non-machine-readable ways, nothing in the historical record suggests he could pick up enough new votes to overcome his deficit of 2,009 votes. That number just isn’t “well within” the range, whatever he may say.

Minnesota’s threshold for free recounts was carefully chosen. I was among the experts who helped shape the legislators’ deliberation on the bill that established that threshold. They strongly valued ensuring public confidence in election outcomes, so they set the threshold conservatively high, to be absolutely sure that any election with a legitimately doubtful outcome would qualify. Any candidate who chooses to fund a recount above that threshold is apt to be wasting their money. Stewart Mills, whose margin of defeat is more than twice the threshold, is surely doing so.


Max Hailperin is Professor Emeritus of Mathematics, Computer Science, and Statistics at Gustavus Adolphus College; he earned his Ph.D at Stanford and S.B at the Massachusetts Institute of Technology. In 2014, he was awarded the National Association of Secretaries of State (NASS) Medallion “in recognition of his service and contributions to election-related technology and legislation.” He was appointed by Governor Mark Dayton to the Electronic Roster Task Force in 2013.