A Course in Game Theory

Martin J. Osborne, Ariel Rubinstein

How can we model decisions by several people whose outcomes depend on one another's choices, and what do such models predict?

Get the app Read on the web

At a glance

Author: Martin J. Osborne, Ariel Rubinstein

Why read it

You will see why choosing at random, as when bluffing in poker, can be a deliberate strategy, and how to model it
You will know what common knowledge means: everyone knows, and knows that others know, and so on
You will weigh the trade off in abstract models: wide reach, but few conclusions until you get specific

Who it is for

A graduate student in economics meeting game theory formally
A strategist who wants rigor behind talk of bargaining and conflict
A mathematically trained reader curious how rivals decide

The idea people argue about

The textbook's most debatable commitment is to explain strategic interaction through sharply specified rational choice models. A common objection is that real people misunderstand games, care about fairness, follow habits, and make mistakes, so equilibrium can look like a prediction built from unrealistic agents. Osborne and Rubinstein answer by treating a model as a precise analytical object whose assumptions and implications must be stated, not as a claim that every person calculates perfectly. They also develop several solution concepts because different timing, information, and belief structures support different analyses. Equilibrium disciplines reasoning about mutual responses, but it does not guarantee that a predicted outcome is fair, efficient, unique, or empirically common. The remaining dispute concerns use: an abstract model can reveal strategic logic, yet its value in a particular setting depends on whether the selected players, actions, information, and preferences capture what matters there.

Three common misunderstandings

A Nash equilibrium is not automatically the best possible result for the players as a group. It only rules out a profitable unilateral change, so an equilibrium can be inefficient or leave every player worse off than another feasible outcome.
Game theory is not limited to zero sum conflict between enemies with entirely opposed interests. The textbook includes coordination, bargaining, repeated interaction, and other games in which interests partly coincide and the total payoff is not fixed.
A mixed strategy does not require a player to feel random or to imitate a coin in every real decision. It is a probability distribution over actions that can represent deliberate randomization or the beliefs needed to make opponents indifferent.