How Accurate Is Our Election Apparatus? In Cairo last week, President Obama promoted the "ideal of democracy" abroad. But here at home, commentator Leonard Mlodinow says our process of counting votes is far from perfect. Just look at Minnesota's U.S. Senate race.
NPR logo How Accurate Is Our Election Apparatus?

How Accurate Is Our Election Apparatus?

Commentator Leonard Mlodinow is a writer and teaches probability and random processes at the California Institute of Technology. Mlodinow wrote for the television series Star Trek: The Next Generation. His most recent book is The Drunkard's Walk: How Randomness Rules Our Lives. Courtesty of Leonard Mlodinow hide caption

toggle caption
Courtesty of Leonard Mlodinow

Al Franken just turned 58, and he's hoping, before his next birthday, to start a new career representing Minnesota in the United States Senate. In the past several months, there has been a lot of wrangling about the vote count, but is there any reason to believe the final result will reflect the will of the electorate?

In elections with millions of voters, a tied vote count is unlikely, but it is a mistake to think that a very close vote count means anything more than a tie would mean. A vote is a kind of measurement, in which we are measuring how many people support each candidate and care enough to take the trouble to cast a ballot. And, as in any measuring process, there is a certain degree of error.

The difference between the candidates in the Minnesota race is about 300 votes, so it sounds like 300 more Minnesotans voted for Franken than for his opponent, but that is probably not true. Yes, after all the recounts, the arrow on the vote-measuring instrument was left pointing a tiny bit to the Democratic side. But is the instrument that is the Minnesota state election process fine enough to reliably measure a margin of 300 votes out of 3 million — or 1 out of every 10,000? Or are we just fooling ourselves, trying to measure inches with a stick calibrated in miles?

What kind of apparatus can measure something with an error less than 1 in 10,000? In science, those are pretty fancy instruments with names like the Wilkinson Microwave Anisotropy Probe. And so the question is: Is our government's election apparatus as accurate as a $150 million satellite?

There are many reasons for error in vote counts. Mistakes happen in sorting and in counting — just imagine trying to count to 10,000 without making an error. Regular ballots can get lost, but then there are the absentee ballots. The post office can lose them. Or they can be wrongly rejected by election officials.

No one seems to have noticed that the various recounts in the Minnesota election have varied by almost 1,000 votes — three times as many votes as Franken's winning margin. So how should we decide close races? In New Hampshire's U.S. Senate race of 1974, when after two recounts the Republican candidate appeared to have won by just two votes, a new election was called. It resulted in record-breaking turnout, with the Democrat winning by 27,000 votes.

Or we could save time and money by doing what a Japanese executive did a few years ago when he couldn't decide whether Christie's or Sotheby's should be the auction house to sell his company's $20 million art collection: We could make the candidates play rock, paper, scissors.