Barry Nalebuff: My own work in political science applies game theory to elections. And one of the things that we discovered there is that our [i.e. US] electoral system really doesn’t work when you have more than two candidates. We saw that if you go back, in terms of Ross Perot having a big influence in the [Bill] Clinton election.
We saw that inadvertently, with Pat Buchanan in the Florida election.
And we saw that in the primaries, where you might very well have imagined that either Mitt Romney, or others that were more in the conservative side of the Republican wing, might have won if their teams had been pooled, but there was one person who is a little bit separate, and in this case was [John] McCain who ended up getting the plurality.
And the problem is that, I have to vote not just for the person who I like the most, but also the one who I think may have the momentum. And it becomes a strategic question, as opposed to a preference question.
So how can you design electoral systems that allow people to do a better job expressing their true preferences, as opposed to their strategic preferences. And probably we have a chapter of this on the "Art of the Strategy" [book].
This is an old problem and it goes back actually to the French Revolution. A gentleman named Marquis de Condorcet was frustrated by the possibility that you get voting cycles, that you can have one person A who beats B, you’ve got B who beats C, and C who beats A. And it turns out it’s not that hard to come up with this, and so then you’ve got this challenge which is, who you end up picking or what policy you end up choosing, depends entirely on the order or the agenda, as opposed to what the will of the people are.
And therefore we hold up this national democracy as sort of the be all and end all. But actually that doesn’t tell you how to vote and how to express or aggregate people’s preferences.
Ken Arrow won a Nobel Prize for telling us that there is no perfect way of doing this; that each way has its problems. The way that I like to do it, my preference here is, that literally to have people vote on every possible pair. So if you’d like they could vote on Hillary [Clinton] versus [John] McCain, Hillary versus [Mitt] Romney, [Barack] Obama versus [Mike] Huckabee, Obama versus McCain and so on.
Then you can say, okay, this election was 70-30, this one was 65-35. Was there one person who could just beat everyone by a majority? It’s so great, that’s the one we want and that’s actually what it was now called a Condorcet winner.
But if there isn’t, is there someone who comes pretty darn close, always winning? So I don’t mind the fact that this person lost one election 51-49, as long as he beat everybody else by a healthy margin, and the person who they lost to at 51 lost to somebody else by more than 51.
And you say, well, okay. That’s really complicated. How could I get people to go and vote 26 different times or 100 times? And the answer is simply ask them to rank their candidates. So they’ll say this person is first, second, third and fourth, and having rank the candidates where a computer can vote for them.
Because it was between number 2 and 6, well, you vote for 2 over 6. With between 3 and 7, well then number 3 gets the nod. And so, in fact, by having somebody’s preferences, I know exactly how they would vote in any election that would be chosen and then we can offer them really a way to express their preferences even if they don’t get their first choice candidate.
Question: Are there alternatives to the Electoral College?
Barry Nalebuff: There is a group of people who think that Electoral College makes no sense and put me on that list.
The question is, how do you change it? And nobody is going to say, well, I want to give up my state’s votes just to go with the majority. But there is this movement where states say, if a number of states whose votes add up to over 270 electoral votes [IB] this petition, then all of those states will cast their electoral votes with the nationwide plurality winner.
Guess what? Without a constitutional amendment, you have moved away from the Electoral College. So essentially we’ll all agree to do something when enough of us get together so that actually we’ve implemented it and that would work.
Back in the 2000 [US presidential] election, I actually had a game theory proposal that unfortunately Ralph Nader didn’t take, but it turns out that a peculiarity of our electoral system is that two candidates could have picked the same electors.
In that way, a vote for Nader could have been a vote for [Al] Gore as long as they have the same electors. So Nader could have run and said vote for me, I’ll get my matching grants; should be able to show the world how many votes I have, but I realized that my votes actually, if I had to choose between Gore and [George W.] Bush, well, yeah, I don’t like either, but I like Gore a little bit more, so that’s how I like them to go, and he didn’t take that.
Question: How are the campaigns utilizing game theory?
Barry Nalebuff: Well, elections are pretty much a zero sum game. And so there is a question of, I have limited resources of the candidates time, and I have to think about these 50 states, and some of them are pretty obviously Democratic.
So Obama doesn’t spend much time in Connecticut and McCain doesn’t spend much in Alaska. Then you think about, what are the pivotal states? And then there is the, can I confuse you and make you think that Virginia is actually going to go the other way if you don’t spend anytime there and so [that] distracts you?
There’s the question of you want to bring this message and if I can throw enough of fear and uncertainty delaying confusion in, then you don’t get to bring your message out.
And so what usually, in most business strategies, is that zero sum or negative strategies, if you like don’t work.
If McDonalds says to the world hamburgers are carcinogenic, but ours are less carcinogenic than Burger King’s because they are flame broiled and ours are fried. And it turns out that McDonalds sales go down by a million and Burger Kings go down by 5 million, this doesn’t really help McDonald shareholders.
On the other hand, if you play a negative campaign and it turns out that you lose a million votes but your rival losses 5 million, you get elected.
So one of the things we appreciate, in terms of game theory, is in the zero sum games, winning votes or causing your rival to lose votes can be equally valuable in terms of winning elections. But it turns out the negative campaigns are less valuable, ultimately, in helping you run once you’ve won.
I think there’s an issue, in analogy, to selling Christmas trees right before Christmas. The day after Christmas, those trees aren’t worth very much. So you have to have a strategy to make sure that this is, again, look forward and reason backward.
I think it will tell us something about what we are going to see in terms of the race card strategy. That’s a challenging strategy. It’s something that is highly inflammatory. And we saw in the Jesse Helms, Harvey Gantt race that right before the election, Jesse Helms had this famous ad of a pair of white hands clumping up a job, a rejection slip and the message says, you deserve that job, you were qualified, but it went to adversity or an affirmative action person.
Essentially it was just a complete racist and outrageous ad, but because it ran 48 hours before the election, there wasn’t really time for the whole cycle to go through.
Again, this is a little bit of the look forward, reason backward. You are not going to see the race card really being played now because then it could be the backlash, they could be the response. The time you’ll see it happen is right near the end.
Then the question is, if you really anticipate that US or the Obama camp have to then have the conversation about it to inoculate themselves, because if they don’t, and it’s played, it puts them in a very difficult position.
Recorded on: Oct 2, 2008