till date only theorists have really understood the potential that experiments hold for Game Theory. experimentalists just don't get it...
Schelling even as far back as the 60s said the following. that too in a tiny lil chapter titled 'Game Theory and Experimental Research'.
1. "Mathematical structure of the payoff function should not be permitted to dominate the analysis."
Experimentalists do exactly that. their subjects are shown mostly only the payoffs with the story or the game being hidden in 'neutral' language in order not to suggest any play to them. thus, hopelessly eliminating Schelling's elegant theory of the power of suggestion. reducing the game to a mathematical problem, if the subject wants to solve it; which very often he does not. what is this a test of? not of Game Theory as I see it and as it pretends to be.
2. "There is a danger in too much abstractness: we change the character of the game when we drastically alter the amount of contextual detail that it contains or when we eliminate such complicating factors as the players' uncertainties about each others' value systems."
Related to 1. above. the laboratory hides the game in what is shown to the subjects.
3. "Some essential part of the study of mixed-motive games is necessarily empirical."
This, the most important point experimentalists almost seem to never understand. that purely analytical means a-priori will never fully be able to predict what people will actually do. analytical methods rather predict what is stable play, or what is equilibrium play, or in simple words, what is clever play. whether this clever play is perceived by people or even when it is, whether they choose to play it or not is unknown till it happens.
especially with regard to the third point, Schelling points out how experiments could help Game Theory by exploring when and how do people actually play (or not) the stable strategy.
but its been wasted on researchers who simply do not understand the intent of the effort of Theory...
1 comment:
point number 3 is pretty powerful. it sounds interesting - how/when/why do people play pareto optimally
to do this without using payoffs.
I recall an "experiment" by Scott Barrett who used a set of cards - some black some red - to show how public goods can be underprovided. If you had a red card, you would get 5$ if you kept it to yourself but if you "contributed" it then everyone would get 1$ each. Or something like that. So if everyone gets a red card then they are all better off if they contribute
If there are 50 people, all have a red card, and all contribute then everyone gets 50$ but if all hold on to it then everyone gets 5$. But most chose to hold on their card.
schelling of course is all powerful.
Post a Comment