Lectures
You can download the lectures here.
This class is partially based on the following excellent courses and textbook, with many of the readings coming from their resources:
- Tim Roughgarden’s course, Incentives in Computer Science. Labeled as “R” in the suggested readings below.
- David Easley and Jon Kleinberg’s course/textbook Networks, Crowds, and Markets. Labeled as “EK” in the suggested readings below.
-
-
-
Lecture 3 - Intro to mechanism design and auctions
tl;dr: Intro to mechanism design and auctions.
[slides] [handwritten]
-
Lecture 4 - Games and auctions continued
tl;dr: Games and auctions continued.
[handwritten]
Reading R Lecture 2
Optional Reading
-
Lecture 5 - Intro to centralized markets
tl;dr: Intro to centralized markets.
[slides] [handwritten]
-
-
Lecture 7 - Matching markets in practice
tl;dr: Matching markets.
[handwritten]
Optional Reading
- The Redesign of the Matching Market for American Physicians: Some Engineering Aspects of Economic Design
- The New York City High School Match
- Deferred acceptance algorithms: History, theory, practice, and open questions, international Journal of game Theory 36 (3), 537-569
- Modeling Assumptions Clash with the Real World: Transparency, Equity, and Community Challenges for Student Assignment Algorithms
-
Guest 1 - Parag Pathak Guest Lecture
tl;dr: Parag Pathak (MIT).
-
Lecture 8 - Intro to networks
tl;dr: Intro to networks.
[handwritten]
Reading EK Chapter 2
Optional Reading
-
-
NO CLASS FEBRUARY BREAK
tl;dr:
-
-
-
Guest 2 - Ana-Andreea Stoica Guest Lecture
tl;dr: Ana-Andreea Stoica (Max Planck Institute for Intelligent Systems).
Title: Algorithm design for social good: fair design and strategic interactions
Abstract: In this talk, I will present my recent work on social aspects of algorithm design, encompassing two lines of work. In one line of work, we study methods for diagnosing when and why an algorithm is biased against societal minority groups, with applications to ranking algorithms and social influence. In another line of work, we study the impact of strategic interactions in social contexts, with a focus on precise error estimation of treatment effects in the presence of competition. In this talk, I will present theoretical challenges in designing algorithms that (1) do not amplify bias against minority groups and (2) are aware of strategic feedback from a population that aims to maximize its utility.
Papers:
- Stoica, Ana-Andreea, Nelly Litvak, and Augustin Chaintreau. “Fairness Rising from the Ranks: HITS and PageRank on Homophilic Networks.” To appear at The Web Conf’24.
- Stoica, Ana-Andreea, Jessy Xinyi Han, and Augustin Chaintreau. “Seeding network influence in biased networks and the benefits of diversity.” In Proceedings of The Web Conference 2020, pp. 2089-2098. 2020.
-
Student paper presentations
tl;dr: Student paper presentations
-
Student paper presentations
tl;dr: Student paper presentations
-
Student paper presentations
tl;dr: Student paper presentations
-
Student paper presentations
tl;dr: Student paper presentations
-
Guest 3 - Manish Raghavan Guest Lecture
tl;dr: Manish Raghavan (MIT).
-
Guest 4 - Meena Jagadeesan Guest Lecture
tl;dr: Meena Jagadeesan (UC Berkeley).
-
Student paper presentations
tl;dr: Student paper presentations
-
Guest 5 - Akshaya Suresh Guest Lecture
tl;dr: Akshaya Suresh (Yale).
Title - Redesigning Recommendation on VolunteerMatch: Theory and Practice
Abstract – This work is the result of a multi-collaboration with VolunteerMatch (VM), the largest platform in the United States connecting volunteers to organizations, to help them redesign their recommendation system. VM has helped to facilitate over 10 million connections and would ideally like to help all of its organizations find enough connections to fulfill their needs. However, in practice many opportunities for volunteering get no connections and there is a significant number of “wasted” or excess connections. Reallocating interest from opportunities with too many volunteers to accommodate towards those not getting enough can significantly improve welfare for all agents. We studied how VM can improve its recommendation system to accomplish this reallocation and maximize useful sign-ups. Our contributions are both theoretical and applied. On the theory side, we model a key feature of many online platforms – multi- channel traffic – and design a new algorithm for this context with provable guarantees that are near-optimal in some regimes. On the applied side, we designed and implemented a new recommendation algorithm, SmartSort, which led to significant improvements in equity on VM’s platform and provided insights into the equity-efficiency trade-off in the context of VM.
Bio Akshaya Suresh is PhD candidate in Operations at Yale University at the School of Management (SOM). Her research uses tools from statistical modeling and data-driven optimization for non-profit and public sector applications, including collaborations with partners like VolunteerMatch, the Florida Center for Reading Research, and the New Haven Lexinome Project. Prior to attending Yale SOM, Akshaya received a BS in Astrophysics from Yale University, and an MA in Social Science from the University of Chicago. She has professional experience in data science consulting and public policy, and served as a Presidential Management Fellow at the Internal Revenue Service and the Department of Housing and Urban Development. In her free time, she enjoys needle crafts, learning new languages, and playing violin.
-
Student paper presentations
tl;dr: Student paper presentations
-
Student paper presentations
tl;dr: Student paper presentations
-
Student paper presentations
tl;dr: Student paper presentations
-