Customer Reviews

The Theory of Learning in Games (Economic Learning and Social Evolution)

5 star:		(0)
4 star:		(2)
3 star:		(0)
2 star:		(0)
1 star:		(1)

 

 

During the work on my master thesis ("Learning in strategic games") i bought several books about the topic. This is the one of them. Chapters 1 and 2 (Introduction, Fictitious Play) are really good introduction into the subject. The following chapters evolve the theory further giving some good ideas for practical implementation (I was writing a C program which had to be able to play the game and to learn). I would recommend this book to anyone interested in relatively new field - Learning in games.

An excellent treatise on some important work in the theory of learning in games. Fudenberg and Levine provide a good coverage of standard myopic play dynamics with a special emphasis on ficticious play and replicator dynamics. I particularly liked the sections going through the Kandori, Mailath and Rob (1993) model as well as Young (1993) on the evolution of convention.

The treatments of dynamic systems analysis, elementary game theory, stochastic approximation theory, etc., are necessarily short. The appendices do not suffice for a reader without a reasonable background.

Nonetheless an essential read for anybody doing serious work in learning, or wanting to know what all the fuss is about.

This book does not provide valuable information about learning systems. It demonstrates, that nonlinear dynamics can be used to describe a subclass of learning. I personally doubt, that this subclass is of great interest, because it neglects completely heuristic strategies in game playing. Besides this, nonlinear dynamics is only useful if the number of parameter of the system is small. I doubt, that these toy examples are sufficient to describe reality, e.g., economics.

Moreover, the organization of the book and the style it is written in, is in my view not favorable.

I guess, this book is for a very small readership that does not have to worry about the correspondence of a model with nature. But also from this perspective it can not be recommended, because it is not written well. Both thumbs down!