Skip to main content

Strategy in R: Game Theory, Simulation, and Machine Intelligence

1 Introduction

Why study strategic interaction computationally, and why R is the right tool for the job. ## The computational turn in game theory

Game theory began as a branch of mathematics. Von Neumann and Morgenstern (1944) laid the foundations; Nash (1950) proved that every finite game has an equilibrium. For decades, the field advanced primarily through theorems and proofs.

But the questions that matter most — Which equilibrium will players actually reach? How do strategies evolve in populations? Can machines learn to play well? — are fundamentally computational questions. They require simulation, numerical methods, and algorithm design. This book is about answering those questions with working code.

1.1 Why R?

R is the lingua franca of statistical computing and data science. It has mature ecosystems for visualization (ggplot2), data manipulation (dplyr), differential equations (deSolve), network analysis (igraph), and interactive graphics (plotly). For researchers who already think in R, there is no reason to context-switch to another language for game-theoretic computation.

Where R falls short — deep learning, counterfactual regret minimization — we use Python via reticulate, keeping the analysis pipeline in R while calling specialized Python libraries.

1.2 The structure of this book

The six parts of this book form a progression:

  1. Foundations build the conceptual vocabulary: what games are, how to solve them, what equilibrium means.
  2. The R Toolkit introduces the packages and utilities we use throughout.
  3. Simulation moves from analytical solutions to computational experiments: Monte Carlo, agent-based models, evolutionary dynamics.
  4. AI and Machine Learning explores how learning agents interact strategically: reinforcement learning, self-play, and frontier methods.
  5. Applications grounds the theory in real domains: auctions, matching markets, bargaining, and empirical case studies.
  6. Ethics and the Future asks what happens when strategic AI systems interact with humans and each other.

Each chapter is self-contained enough to read independently, but the parts are designed to be read in order for a coherent narrative.

1.3 How to use this book

Linear reading is best for graduate courses or self-study in game theory with a computational focus. Start at Part I, Chapter 1.

Reference reading works well for practitioners who need a specific tool. Jump to the relevant chapter, follow cross-references backward for prerequisites.

Project-based reading pairs well with the exercises. Each chapter ends with problems that can serve as starting points for course projects or research extensions.

Let’s begin.

This book was built by the bookdown R package.