Today at Peet’s in Evanston I was trying to work out a model for this idea Sandeep and I are working on related to the game theory of torture.  I started drawing a litle graph and then got lost in thought.  I must have looked a little weird (nothing unusual there) because the woman next to me started asking me what was up with this squiggly plot on my pad of paper.

Most economists dread these moments when someone asks you what you do and you have to tell them you’re an economist and then prepare to deflect the inevitable questions and/or accusations “what’s going to happen with interest rates?”  “when’s the economy going to turn around?”  Usually I just mumble and wait for the person to get bored and go on with her reading.  For some reason I was talkative today.

I told her I was a game theorist.  “What’s that for?” I told her I was working on a theory of torture.  She looked horrified.  “How do you make a theory of torture?”

I told her that using game theory is a lot like screenwriting.   Imagine you were a film-maker and you wanted to make a point about torture.  You would invent characters and put them in the roles of torturer and torturee and you would describe the events.  You would depict how the torturer would plan his torture and how he would the torturee would react and how this would lead the torturer to adjust his approach.  If the film was going to be effective it would have believable characters and it would have to show the audience a plausible hypothetical situation and what happens when these characters act out their roles in that situation.  In short, its a model.

(As I was saying this I remembered that I learned to think of economics and literature in this way about 20 years ago from Tyler Cowen.  And he has a nice paper on it here.)

She looked even more horrified.  But I was pretty pleased.  I started thinking about Resevoir Dogs (nsfw).

While scripts and models are constrained by a similar requirement of coherence between character and events, there are differences and this makes them complementary.  A model necessarily maps out the entire game tree, while a script describes just one path.  In a model every counterfactual is analyzed and we see their consequences and this explains why those paths are not taken, but a film is a far more vivid account of the path taken.  In a model the off-equilibrium outcomes are the results of mistakes while a well-conceived script can bring in plausible external developments to place the characters in unexpected situations.

Of course film-makers get invited to better parties.