What Osborne Actually Covers and How It Actually Works
Martin Osborne's An Introduction to Game Theory is probably the most used graduate-level textbook in the field right now. It covers non-cooperative game theory from the ground up — normal form games, extensive form games with perfect and imperfect information, repeated games, bargaining, evolutionary game theory, and Bayesian games. The writing is dense but precise, and it assumes you already know basic calculus and some linear algebra. If you don't, you'll spend more time flipping back to the math appendix than actually learning game theory. I picked this book up around 2014 when I was trying to model competitive pricing strategies for a SaaS product. The chapter on Nash equilibrium in continuous strategies turned out to be exactly what I needed, but not before I wasted three days misinterpreting the best response function derivations because Osborne never explicitly states the first-order condition assumptions in plain language. He just writes them out. You're expected to catch that he's assuming interior solutions and differentiable payoff functions without him saying so.
Osborne An Introduction To Game Theory — How People Actually Use It
Most people read this as a reference or a problem-solving companion, not cover-to-cover. The exercises are where the real teaching happens. Some are straightforward applications. Others, particularly in the extensive form and bargaining chapters, are genuinely tricky and will make you sit with them for an hour or two. The solutions manual exists separately and is worth getting if you're working through it alone. Here's the thing nobody tells you: Osborne's treatment of perfect Bayesian equilibrium in Chapter 6 is technically correct but practically underwhelming for anyone who's ever tried to use it in a real strategic situation. The conventions he uses — like the requirement that beliefs be derived via Bayes' rule wherever possible — work cleanly in textbook examples but collapse in edge cases where information sets aren't reached with positive probability. I ran into this when modeling a two-stage entry deterrence game where the entrant's beliefs about the incumbent's type depended on an off-equilibrium path observation. The textbook derivation gave a clean answer, but the equilibrium wasn't robust to small perturbations in the prior. I ended up using a trembling-hand refinement approach and recalibrating the beliefs manually, which Osborne barely addresses past a few paragraphs.
The Structural Walkthrough
Chapter 1 starts with basic concepts — players, strategies, payoffs, and the Nash equilibrium. It moves quickly. Chapter 2 is dominated by mixed strategies and the minimax theorem. Chapter 3 covers dominance solvability and rationalizability. Chapters 4 and 5 handle extensive form games, subgame perfection, and backward induction. This is where the book gets useful for actual analysis. Chapter 6 on Bayesian games and perfect Bayesian equilibrium is where most readers hit their first wall. The notation is heavy. The distinction between sequential equilibrium and perfect Bayesian equilibrium gets blurred because Osborne doesn't spend much time contrasting them. If you're coming from a economics background this might be fine. If you're coming from mathematics or computer science, you should probably supplement this with Fudenberg and Tirole for the refinements Osborne skirts around. Chapters 7 and 8 cover repeated games and evolutionary game theory. The Folk Theorem section is thorough. The replication dynamics and evolutionarily stable strategy material is where Osborne adds value beyond what you'd get from a survey paper — he derives things rather than just stating results.
Get the Full Details
The later chapters on bargaining (Chapter 9) and social choice (Chapter 10) are shorter and feel somewhat perfunctory compared to the depth of the earlier material. The bargaining chapter in particular could have used more on the Rubinstein model with asymmetric discounting, which Osborne only briefly touches.
Practical Tips That Actually Matter
Don't skip the math appendix. It covers fixed point theorems, optimization under constraints, and basic dynamic programming — all things you'll need without realizing you need them until you're stuck on a proof in Chapter 3 or 4. The exercise difficulty jumps significantly around Chapter 5. Problem sets on subgame perfection in infinite horizon games require a comfort with recursive methods that the main text doesn't fully develop. I found it helpful to work through the examples in Kihlstrom, Roth, and Schmeidler's Game Theory as a bridge, even though that book isn't as comprehensive. When Osborne introduces perfect Bayesian equilibrium, pay close attention to the difference between system of beliefs and strategy profiles. Beginners consistently conflate the two. The equilibrium concept requires both to be specified and mutually consistent, and the consistency condition is where most mistakes happen in practice.
For the evolutionary game theory section, the connection to dynamical systems is implicit. If you want to actually simulate replicator dynamics rather than just analyzing them on paper, you'll need to code the differential equations yourself. Osborne provides the equations but no implementation guidance. A quick Python script with SciPy's odeint handles most of the standard cases in under thirty lines.
Where This Book Falls Short
Osborne doesn't cover mechanism design in any substantive way. If your interest in game theory leans toward auction theory or contract design, you'll need something else — Milgrom's Putting Auction Theory to Work or Krishna's Auction Theory fill that gap. The book also has almost nothing on computational game theory, which matters if you're working with large strategy spaces or multi-agent systems. Nash equilibrium computation in general games is PPAD-complete, and Osborne doesn't address what that means for practical applications. The second edition, which is the one currently in circulation, corrected several errors from the first but introduced a few new ones. Check the errata page on Osborne's website before submitting homework or building analysis on specific derivations. I caught a sign error in Proposition 23.1 (the characterization of equilibrium in a particular bargaining game) that propagated through the proof of the corollary. There's also the question of whether Osborne's formalism matches how game theory is actually practiced outside academia. In industrial organization, for instance, researchers often use simpler equilibrium concepts or focus on comparative statics rather than the full perfection refinements Osborne emphasizes. The book is rigorous, but rigor isn't always the right tool for the job.
How to Get a Copy
The book is published by Cambridge University Press and is available through most academic retailers and library systems. The PDF circulates widely in university circles, though I'd recommend supporting the author and publisher if you can. Osborne has also posted lecture notes and supplementary material on his personal website at the University of Warwick, which complement the text well — particularly the notes on equilibrium refinements, which go further than the book does. If you're starting out and find Osborne too dense, Gibbons' Game Theory for Applied Economists covers roughly the same material at a lower barrier to entry. But if you're serious about the subject and want a reference you'll keep for years, Osborne is the one to get. It's not the most readable game theory book ever written, but it's accurate, complete within its scope, and the problems force you to actually think rather than just follow examples. The real utility comes from doing the exercises. Reading the chapters passively will give you a false sense of comprehension. The difference between understanding Nash equilibrium in normal form and being able to derive it in a specific continuous strategy game with asymmetric information is entirely in the problem sets. That's where the actual learning happens.
