Bayesian game
A game in which each player privately holds one of several possible characteristics, drawn from a distribution everyone knows.
Often misread asTreating the uncertainty as uncertainty about what the other player will do, so the model has one opponent behaving randomly rather than several possible opponents each choosing deliberately.
How to see it
You know the odds on who you face, not the person
Each player privately holds one of several possible situations, and everybody knows the chances of each. So you are not choosing against a known opponent, you are choosing against a known spread of possible opponents, every one of whom is working the same problem from their own side.
The checkThe uncertainty is about which opponent you face. Each possible opponent has their own best response, and you are averaging over those, not over noise.
Try itIn a sealed bid auction, write down what you think your rival will do. If the answer is a single number, you have modelled a known opponent. If it is a rule turning their valuation into a bid, you have modelled a Bayesian game.
Know these first
Leads on to
Easily confused with
Sources
- Harsanyi, J. C. (1967-68). Management Science 14(3), 159-182; 14(5), 320-334; 14(7), 486-502.
- Fudenberg, D. & Tirole, J. (1991). MIT Press.
- Gibbons, R. (1992). Harvester Wheatsheaf. Published in the United States as 'Game Theory for Applied Economists', Princeton University Press.