# The Complexities of Voting

Posted on: October 6, 2010

In a recent conversation with my math professor regarding a possible topic for my Capstone, I commented that I really enjoyed studying linear algebra. I told him I might be interested in doing some type of real-world application involving linear algebra for my project, and he began to tell me about Donald Saari. Mr. Saari is a mathematician who studies what he calls “the paradoxes and problems of voting procedures,” and analyzes voting methods using linear algebra. I am currently looking at two books he wrote: “Chaotic Elections!” and “Basic Geometry of Voting.”

In “Basic Geometry” Mr. Saari makes it clear that when there are only two choices up for a vote, there is generally no difficulty in determining the winner. But when there are more than two choices, things can get rather interesting. In order to demonstrate some of the challenges that can arise,  Mr. Saari opens the book with the story of a fictional academic department chair who finds himself in hot water as a result of a departmental vote. The problem is that the same vote can be interpreted different ways, depending on which voting method is used. Even if a method ranking preferences is used, various outcomes can result depending on which way one counts the rankings. In fact, each of the options can be deemed the winner, depending on the method used.

Now things can become even more complicated if a group of voters attempts to manipulate the process. For example, assume Al, Bob, and Chuck are candidates for the new Chair position. Of the 15 people voting, 7 are for Al, 7 are for Bob, and the only one pulling for Chuck is, well, Chuck. A ranking system is used that gives 2 points to one’s first choice, 1 point for one’s second choice, and 0 points for the third. Al receives 14 points from his group, 7 points from Bob’s, and 1 from Chuck. Bob receives 14 points from his group, 7 points from Al’s, and 0 from Chuck. So Al wins, 22-21-2, with Chuck’s vote determining the winner.

Let’s assume now that Bob’s supporters predict this outcome and decide to ensure Bob’s success. They each vote for Bob as their first choice, Chuck for their second, and Al for their third. Now Bob wins 21-15-9. Al’s supporters see this coming, so they decide to be “strategic” and vote for Al as their first choice, Chuck as their second, and Bob as their third. With this turn of events, that is, both groups voting for Chuck as their second choice, even though he was the first choice for only 1 out of the 15 voters, Chuck wins 16-15-14.

With this simple story, Mr. Saari demonstrates that it is not difficult for a candidate to win an election, even though that person was not the first choice of the majority of voters. As it turns out, this is not an anomaly. The outcome of a vote does not necessarily reflect the will of the people. As Mr. Saari states, his intent is simply to share what can go wrong in elections and why, in the hopes that voting errors can be prevented in the future. All in all, I found this to be a very intriguing topic, and with a little more research, it could turn into my Capstone project.

### 2 Responses to "The Complexities of Voting"

Nicely explained.

Eventually, Saari gives a pretty good way to think about elections using linear algebra. I think you could follow up and use this pretty well. You could run a few “elections” with some friends or students here at school and see how you can use linear spaces to manipulate the “winner.”

If this is a way you choose to go, I am happy to discuss it with you further, but maybe you will continue with your interest in Diophantine Equations (coming up soon in Abstract…)

Whatever happened to punching out the tab for the ONE person that you would like to vote for? I wonder how often this takes place. I’m surprised that this doesn’t happen more during the primaries. I’m also surprised that one political party doesn’t vote for the opposing political parties candidate that isn’t suppose to win. For example, Candidate A and B are running for the X party. Candidate A has a very good chance of winning the overall election in November so during the primary election, party Y votes for candidate B.
I’m curious to see what you find out if you do this for your capstone. There will definitely be a lot of data for you next month.