Computational social choice is concerned with the design and analysis of mechanisms for collective decision making. This course provides a thorough introduction to both classical social choice theory, originating in Economics and Political Science, and modern computational social choice, emphasising its interface with Computer Science and AI. The intention is to enable students to conduct independent research in this exciting and fast-moving field. This is an advanced, research-oriented course in the Master of Logic that is also listed as an elective for the MSc AI and the MSc Computational Science. Students from other programmes, such as Computer Science, Mathematics, Economics, or Philosophy, are equally welcome (contact me if you are unsure about your qualifications).
This page will be updated throughout the course, with links to the slides used in class, literature recommendations, and homework assignments. Please check it regularly. You may also be interested in previous editions of the course.
Nature of the Course and Prerequisites: The focus of this course is on theory: we will mostly be concerned with developing formal models of collective decision making and with proving theorems about those models. Therefore, I will expect what sometimes is called mathematical maturity, meaning that you need to have prior experience with working out and writing up mathematical proofs.
Special Theme for 2026: The special theme for this edition of the course will be voting theory. We will see how to analyse the question of how to choose an adequate voting rule from a wide variety of methodological angles. We will also briefly discuss a few other topics, notably judgment aggregation and fair allocation.
Literature: We will mostly rely on original research papers, which I will post on this website as we go along. Beyond these papers, the main reference for this course is the Handbook of Computational Social Choice, the electronic edition of which is freely available online.
Research Seminars: You are always welcome to attend our local COMSOC Seminar as well as the international COMSOC Video Seminar (currently dormant but hopefully coming back soon).
| Date | Content | Homework |
|---|---|---|
| Tuesday 27 October 2026 |
This first lecture has been a bit of a sightseeing tour of computational social choice, with various examples for problems addressed, results obtained, and techniques employed in different subfields of this research area. Below are links to some of the papers cited during the lecture. Together they illustrate the breadth of the field. I recommend reading the introductions of these papers to get some idea of what people have been working on.
|
Read the historical (not the technical) parts of Chapter 1 of the Handbook and prepare the presentation of your assigned voting rule. |
| Wednesday 28 October 2026 |
We started with student presentations of a large number of voting rules, illustrating some of the many different ideas people have had over the centuries for how to run an election. We then tried to put some order into this large space, first by identifying the positional scoring rules and the Condorcet extensions, and then by classifying rules in terms of the information they need access to and the computational hardenss of their winner determination problems. We also saw that some (classes of) rules necessarily violate some seemingly mild axiomatic properties. Some of this material (and a lot more) is covered in Chapter 2 of the Handbook:
|
Homework #1 |
| Friday 30 October 2026 |
This lecture was dedicated to different approaches one might take to characterise a given voting rule. First, using the axiomatic method, we proved the seminal result of May, providing a characterisation of the simple majority rule in terms of three simple and convincing axioms. I recommend having a look May's original paper, which is a nice example for an early application of the axiomatic method and still highly accessible today.
|
|
| Tuesday 3 November 2026 |
||
| Wednesday 4 November 2026 |
Strategic Behaviour |
Homework #2 |
| Friday 6 November 2026 |
Coping with Strategic Behaviour |
|
| Tuesday 10 November 2026 |
Information and Communication |
Homework #3 |
| Friday 13 November 2026 |
Discussion: Project Startup |
|
| Tuesday 17 November 2026 |
Guest Lecture by Patrick Lederer on Approval-based Committee Voting |
|
| Friday 20 November 2026 |
Discussion: Project Progress |
|
| Tuesday 24 November 2026 |
Automated Reasoning 1 |
|
| Wednesday 25 November 2026 |
Automated Reasoning 2 |
Homework #4 |
| Friday 27 November 2026 |
Automated Reasoning 3 |
|
| Tuesday 1 December 2026 |
Guest Lecture by Théo Delemazure on Experiments in Computational Social Choice |
|
| Wednesday 2 December 2026 |
Discussion: Paper Writing and Reviewing |
Homework #5 |
| Friday 4 Decdember 2026 |
Judgment Aggregation |
|
| Tuesday 8 December 2026 |
Fair Allocation |
|
| Wednesday 9 December 2026 |
Discussion: Presenting |
|
| Friday 11 Decdember 2026 |
Buffer |
During the second part of the course you will work on a small research project on a topic related to the special theme, write a short paper about it, and present your insights and results in a talk. The purpose of this exercise is to give you some exposure to what it is like to do research. Projects are to be worked on in groups of three people each. Your paper must be 4-5 pages long, excluding references, using the IJCAI LaTeX style (IJCAI is the International Joint Conference on Artificial Intelligence, the kind of conference where a lot of work on computational social choice gets published). I have a strong preference for your paper not having an appendix, but if you feel that you need to put some material in an appendix, get in touch to discuss this well before you finalise your paper. Finally, you must meet all of the deadlines below (what these involve exactly will get announced in due time):
Identifying a Topic. Finding a good topic for a project is difficult and you should set aside a significant amount of time for this task. You have a lot of freedom about what to work on, but your project must be related to the special theme of voting. My advice is to start thinking about possible ideas early, but to not settle on a specific topic until we are a few weeks into course. The first project discussion session in particular is intended to help you generate ideas.
One possible approach is to look through the literature and see whether you can extend an existing paper in some way. For example, maybe you can come up with better proofs of the main results, maybe you can generalise or extend some of the results, maybe you can strengthen some of the results by restricting attention to interesting special cases, or maybe you can obtain new insights by varying the methods used in the original paper (e.g., by adding an algorithmic perspective to a paper that is largely axiomatic, or vice versa).
Check the COMSOC community site for suggestions on which conference proceedings and journals to consult when looking for relevant literature. While this is not a hard rule, I recommend that you limit your search to papers that have been published fairly recently. This makes it more interesting and gives you a better chance managing to do something original. Please note that my own papers are off limits for this exercise (because I will have to grade your project and naturally will be somewhat biased when it comes to my own work).
If you choose to embark on a project that involves an implementation or experiments, check the list of tools mainated by the COMSOC community to help you get started.
Requesting approval of your project. Start talking to me about your tentative ideas early on. To formally request approval of your project, upload a one-page PDF on Canvas in which you answer the following three questions:
Drafts and reviewing. As an intermediate step, each group will produce a 2-page draft, consisting of most of Section 2 (often called "The Model" in social choice papers), a few sentences of motivation (from Section 1), and a tentative outline (in the form of a bullet list) of what they still hope top do. Each draft will be anonymously reviewed by several students from the other groups (see how to write a review).
Talks. In the final week of the course, we are going to organise a mini-workshop where each group will present their work (20 minutes talk + 10 minutes for questions). Every group member should contribute in some form. Every course participant will be expected to attend all talks.