CSCI 357 - Fall 2020

Algorithmic Game Theory

Home | Lectures | Assignments | Resources | CS@Williams

Home

Course Description

This course focuses on topics in game theory and mechanism design from a computational perspective. We will explore questions such as: how to design algorithms that incentivize truthful behavior, that is, where the participants have no incentive to cheat? Should we let drivers selfishly minimize their commute time or let a central algorithm direct traffic? Does Arrow’s impossibility result mean that all voting protocols are doomed? The overarching goal of these questions is to understand and analyze selfish behavior and whether it can or should influence system design. Students will learn how to model and reason about incentives in computational systems both theoretically and empirically.

Objectives By the end of the course, the students should be able to:

• model strategic interactions in games and reason about them using appropriate solution concepts
• design tractable, yet effective, agent strategies for participants in a mechanism
• analyze properties of a mechanisms such as strategyproofness and pareto efficiency
• understand the design behind online markets such as ad markets, labor and dating markets

Syllabus & Textbook

The course syllabus is here.

Readings will be given from a collection of several different textbooks, and appropriate chapters will be provided via GLOW.

• Twenty Lectures in Algorithmic Game Theory by Roughgarden
• Algorithmic Economics: A Design Approach by Parkes and Sueken (Unpublished)
• Networks, Crowds and Markets by Easley and Kleinberg
• A Course in Game Theory by Osborne and Rubinstein
• Mechanism Design and Approximation by Hartline (Chapters available online.)
• Algorithmic Game Theory by Nisan, Roughgarden, Tardos and Vazirani (The entire book is available online via the link; look under Resources tab.)

Readings for each lecture will be linked to from the Lectures page.

Course Calendar (Office hours)