Friday, October 29

Halloween game theory on The Office

You know you think too much like an economist when…

I was watching The Office last night, and I knew Oscar was going to win the Halloween contest as soon as he threw in the towel and traded in his disco dude outfit for a "rational consumer" costume. For those who didn't see the episode, a bunch of employees were competing for the highly coveted Wilkes-Barre/Scranton coupon book in a Halloween costume contest, and some employees were taking things a little too seriously in hopes of snagging the big prize. Oscar decided everyone was getting too carried away, so to show how ridiculous he thought the whole thing was, he took off his costume and put on his normal clothes, claiming to be dressed as a rational consumer.

But this was a brilliant (if accidental) strategy by Oscar. The rules for the contest were simple; each employee got one vote for best costume, and you couldn't vote for yourself. The person receiving the most votes wins the coupon book.

So if your goal is to win the coupon book, who do you vote for? It's one big simultaneous-move game theory problem. Voting for who you honestly believe has the best costume (other than your own) is dumb, since you are helping out your biggest competitor. The better move is to vote for whomever you think has the worst costume, and Oscar clearly had the worst. So simple game theory says Oscar should get the most votes. And he did.


Of course, taking our game theory analysis a little bit further, if the characters on The Office were fully rational, they would foresee the mass push to vote for Oscar. Knowing Oscar will clearly win because he has the worst costume, people won't vote for him en mass. Rather, I think the dominant strategy would be to vote randomly, which would result in each person in the office receiving an expectation of one vote. Alternatively, people could sell or auction off their votes, and the person who put the highest valuation on the coupon book would eventually win it. But this would make for boring TV for non-economists.

2 comments:

  1. Nice analysis and thanks especially for the link to "The economics of orgasms." It may be the most exciting part of my Friday night.

    re: The Office, someone thinking "two moves ahead" might consider wearing an obviously "second worst" costume ("It's a sheet; I'm a ghost") on the premise that one-move ahead players will avoid voting for the clearly worst costume, but won't consider the possibility that they might all consequently dump their votes on the second-worst costume.

    It's a bit like what actually happens when experienced game theorists play the "Traveler's Dilemma" game. Rather than bet on the Nash Equilibrium result, they seem to bet based on an intuition about how many "moves ahead" the average player will analyze the game.

    ReplyDelete
  2. Good point Robert. I guess in a one-shot game like the one above, it's difficult for a player to guess how many moves ahead his colleagues would look. A game theorist would tend to assume they'd look infinitely far ahead, but in the real world (or The Office world) that's probably a bad assumption.

    ReplyDelete