Science

We Got Game Theory

I don’t believe in synchronicity, but game theory keeps popping up in my regular reading, in seemingly unrelated sources.

First, Amanda posted a link to a blog about game theory in the Quickies. Then, Michael Shermer wrote an interesting article about game theory in sports in Scientific American. And then NPR ran an article about how game theory just may decide the democratic nomination, however indirectly.

I started to think I should look into this.

Game theory is the process of strategically maximizing your own personal gain in a situation, based on the possible actions of the others involved. Shermer uses the classic “prisoner’s dilemma” as an example:

You and an accomplice have been arrested for a crime. The police take you both downtown for questioning and you are held in separate rooms, unable to communicate. You are both offered the same options:

  • If you both confess, you both will go to prison for 4 years
  • If neither of you confess, you both will go to prison for 2 years
  • If one of you confesses, and the other does not, the confessor goes free and the other goes to prison for 5 years

(The specific prison terms vary in different versions. I used the Princeton version; link below.)

The “game” then becomes guessing your accomplice’s strategy, and choosing a response that you think will minimize your prison time. You can test your skills at said game by playing against a computer here.

Now, as fun as theoretical examples are, real life is even more intriguing, if not messy.

Shermer writes about game theory in reference to cycling, however, cycling becomes a metaphor to sociology and human nature. He describes how the choices of a few cyclists to “dope” in order to gain competitive advantage, force the other cyclists to comply in order to stay competitive. For them, cycling is a career, and many have no academic background to fall back on like Shermer did. They have to stay competitive.

The sad thing, of course, is that if everyone is doping, no one is better off than when no one was doping, and everyone is less healthy. Doping only works when it gives you an advantage over those who are not.

Shermer offers specific suggestions for changing the game to eliminate the payoffs for doping, including no punishment and/or title revoking for prior use, regular testing going forward, public announcement of positive results, and a swift no-tolerance policy.

Changing the payoffs changes the game. Any game.

The NPR article points out that it is likely that neither Clinton or Obama will have the necessary delegates to win the democratic nomination, which will leave the decision in the hands of the superdelegates. Jim Miller, author of Game Theory At Work: How To Use Game Theory Outthink And Outmaneuver Your Competition offers a potential winning strategy using the concept of game theory.

Miller shows shameless cynicism for the concept that the superdelegates will “do what’s right for the party”. He points out that they are individuals with families and careers, many of which may be at stake with regard to the election. The superdelegates may have the power now, but once the election is over, the new President will have the power. “It’s a delicate dance”, as Miller puts it.

The game of the candidates is to act strategically based on what’s truly important to the superdelegates – their interests and futures – by convincing them that s/he will win the nomination (or even better, the election). The game of the superdelegates is to secure their futures by showing support and favor to the candidate they think will win, or at least not piss him or her off.

The factors in this case, as Miller outlines them are:

  • Clinton is perceived as tougher and more vindictive than Obama, and may receive more superdelegate votes out of fear
  • Clinton has stated that Obama would make an excellent vice-president, which Miller perceives as an attempt to undermine him
  • Obama will likely have the regular delegate lead

Miller suggests that Obama take a hard approach by stating that he will not accept the vice-presidential spot and that the nomination will not be legitimate if the superdelegate votes overturn the results of all the primaries. After all, the regular delegate count is indicative of the will of the states (not to mention their likelihood to support the nominated candidate, which is of most interest to the superdelegates). Miller even suggests that Obama might go so far as to threaten to run as a third party candidate. As if this year’s election could get any more interesting.

He also suggests that Clinton’s best bet to compete with this strategy would be to engineer a plan to re-run the Florida and Michigan primaries, awarding all votes to the winner. That would, of course, combat Obama’s illegitimacy claim.

Detractors of Miller’s theory point out that this sort of cut-throat political strategy goes against the heart of Obama’s message and appeal, and in fact, would put him on par with Clinton. So, sometimes even playing the game well changes things in ways you might not anticipate.

The whole thing makes me wonder how government reform could change the payoffs to create a more honest election process.

What I got most out of these articles is that game theory is fascinating because it applies to all of us in so many areas of our lives. Crime, politics, sports, ambition, relationships, etc…whatever is important to us. It’s interesting to predict strategies and analyze outcomes, whether of our own games or those of others. And it would be such useful information if we were able to figure out what works, and what doesn’t.

Ah, game theory…

Related Articles

7 Comments

  1. I always thought the irrationality of individual degrees of “trust” (i.e., in the other individual) ultimately undermine the predictive benefit of this type of analysis as applied to individuals. Also, I do not think game theory at a level of simplicity that is subject to comprehension (even to an MBA student, for example, as opposed to people who actually live and breath this stuff) is particularly useful with respect to a much broader population of actors. In a sufficient sized “free market,” where certain players can get together with certain other players and split the differences in the potential outcomes, it seems to become unwieldy. Doesn’t it? On the other hand, as I remember from my own MBA class on decision theory, you have to hang your hat on SOMETHING …

    As for the current political situation, regardless of what Clinton should be doing vis a vis Obama, and vice versa, I would suggest that the Democratic Party consider the following for the next election:

    1. Was it smart for the Democratic Party to choose (for the 2008 election) to select proportional primaries for the states, while the Republican Party allows states (including little ones like New York) to let winners take all? This is why I seriously doubt any speculation that there will be a revote in Michigan and Florida – especially an all-or-nothing. (What’s wrong with Florida anyway, Stacey? This is like two election in the last eight years you folks are screwing up!)

    2. Is it advantageous or disadvantageous for the Democratic Party to have a larger group of unaccountable “superdelegates” than the Republican Party?

    3. Will the Democratic Party candidate ultimately benefit from this prolonged battle all the way to the convention floor? (It didn’t help the Republican Party in the 1976 Ford/Reagan convention.) Is this helping or hurting McCain?

    The Democratic Party has obviously chosen a different path than the Republican Party.

    In terms of government reform to make the system more “honest,” I am not sure that the federal government could do anything (constitutionally speaking). In my opinion, we would certainly benefit from having a national one-day primary (to avoid front-end loading) and having a uniform open/closed primary system (and how to treat “independents”). I think alot of this is up to the national and state political parties.

    So how exacly does the Princeton version of game theory work in the context of a relationship?

    (a) If I ask her to marry me and she says yes, we both get 4 years of marital bliss (i.e., prison);

    (b) If I ask her to marry me and she says no, I assume 5 years of the lingering baggage of rejection and she goes free; or

    (c) If I do not ask her to marry me, we end up waddling around in a non-relationship for 2 years …

    Does that sound right? I suppose I need to account for the possibility that she might ask me to marry her (which would assume, of course, that she is not in fact reading this blog entry).

  2. SkepticalMale,

    You’re correct about trust being the lynchpin, but it’s not beyond the purvey of relatively simple game theory-a step beyond the classic four-square prisoner dilemma, but not much. Games like the Traveller’s Dilemma suggest that people have enough of a game theory sense to understand the mutual value of trust and trustworthiness-it’s not so much irrational as operating on a level that understands the payoff matrix and the first-approximation behavior of the other players-which includes knowledge of that payoff matrix and of Player 1.

    I think the real electoral game theory conclusion might just be that superdelegates are a bad idea-either they work as elected representatives, or they don’t work at all-bestowing an aristocratic power on party insiders is a situation itching for a self-interest-soaked showdown.

  3. Synchronicity is really just that you notice what you are already paying attention to.

    And game theory is fascinating and fun, especially when you can watch it play out in real world examples.

    Regarding the politics, I also think that superdelegates are a bad idea, just like I think that the electoral college has outlived its usefulness. Anything that takes power away from the individual voters should be removed from the system.

  4. Ginarly, thank you for adding that link!

    SkepticalMale Yes, individualistic qualities make generalized theories hard to study. I’m sure you recall from your own economics classes the concept of ceteris parabis and the assumption that people act rationally. Although these assumptions aren’t always true, accounting for every possible nuance of character makes the study of economics infinite and useless. So does the fact that the assumptions don’t apply in every case make the study of such theories useless? Economists say no. I say no.

    writerdd said:

    Synchronicity is really just that you notice what you are already paying attention to.

    I totally agree.

  5. Yeah, I began fondly using the more colloquial term “all other things being equal” long before I took an economics class (or a latin class), and like I said, as a decision-maker facing uncertainty, you have to hang your hat on something (e.g., I still use archaic decision trees and expected values on a weekly basis) … And it’s not that I don’t think it’s useful or fun to ponder and observe, it’s just that I don’t think one should put too much faith in the outcomes it suggests (oops, I did it again – used the word “faith”)

  6. Interesting poitn about the Prisoner’s Dilemma that is often overlooked: games theory also encompasses the idea of teams. If you impress pon the criminals in question that it is not an individual risk & reward situation, but a criminal team vs law-enforcement team risk & reward situation, it’s a no-brainer. No one talks, law-enforcement team loses. Team Cop understands this and makes a point to play team vs team while trying to influence the criminals to play against one another. You can expand upon this easily into our political situation now. Playing Obama v Clinton right now might help whichever one plays that particular game better, but is it acually better for Team Not Republicas Again? All the crap they (or their camps) throw at one another WILL be remembered come election day, so both need to be very careful how they play this, lest they wind up handing the win to Team Four More Years of Idiocy.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Back to top button