Workshop Details

Social choice has a rich history tracing back to the work of Borda, Condorcet, Laplace and others in the 18th century, and the work of Kenneth Arrow in the 1950s that led to the Nobel Prize in Economics. A current working definition might say that social choice theory studies how to aggregate individual preferences toward making a collective choice.

Beginning in the late 1980s social choice theory touched theoretical computer science, and questions arose about the computational difficulty of calculating the winner of an election or of manipulating the outcome through strategic voting. The field of computational or algorithmic social choice has exploded since then. In this workshop, we explore some of the key new directions of research, emphasizing two directions: (1) developing new algorithmic methods for analyzing social choice rules/mechanisms and constructing new ones; and (2) applying concepts of social choice to issues of interest in modern computer science, such as crowdsourcing, group recommender systems, and information retrieval systems. As Brandt, et al (2016) observe, today there is “increasing awareness of the great potential of computational social choice for important applications of decision-making technologies, in areas as diverse as policy making (e.g., matching junior doctors to hospitals), distributed computing (e.g., allocating bandwidth to processes), and education (e.g., aggregating student evaluations gathered by mean of peer assessment methods).”

The workshop will explore new concepts of design of auctions, voting rules, or matching mechanisms that perform well when the individual “voters” are behaving strategically – an old topic with new twists (Nisan, 2007). Here we will study such topics as “approximate mechanism design without money” (Procaccia and Tennenholtz, 2013) and “incentive compatible machine learning” (Dekel, et al, 2010).

One focus will be methods to scale up collaboration and decision making, enabling large communities to deliberate and brainstorm with one another on important issues with the aid of intelligently designed algorithms and digital communication platforms. The goal is to design tools that not only aid decision-makers in harnessing collective intelligence, but also create community members who are more informed and engaged in making their communities better. An example is the Stanford Participatory Budgeting Platform that allows cities, municipalities, states, foundations, and other organizations to offer a form of direct democracy where members vote to determine how the budget is allocated. Recent new algorithms for participatory budgeting show promise (Goel et al., 2019; Freeman et al., 2019).

Another theme will be randomized social choice. We will study preferences over probability distributions and in particular issues of strategyproofness in this context (Aziz, Brandt, and Brill, 2013b, Aziz, Brandl, and Brill, 2014); randomization in social choice (Kavitha, Mestre, and Nasre, 2011, Aziz, Brandt, and Stursberg, 2013); and concepts of random serial dictators (Aziz, Brandt, and Brill, 2013a, Aziz and Mestre 2014).

The workshop will also explore iterative voting, where voters cast votes in rounds, observing the outcomes of earlier rounds. An old topic in this area has to do with the Delphi method developed at the RAND Corporation in the 1960s (see, e.g, Dalkey, 1969). Here we study questions such as: For what voting rules and iterative voting models does the process converge? When does it converge to an outcome with desirable properties? For instance, Meir, et al. (2010) studied this problem for the plurality rule and showed that under some process assumptions the process converges to a Nash equilibrium; and Airiau and Endriss (2009) study the process where at each step a voter is randomly selected to propose a new alternative to the current winner. Other questions are: what can be concluded if voters have incomplete knowledge of others’ voting intentions (Reijngoud and Endriss, 2012), or if there is uncertainty in information about knowledge of others’ intentions (Meir, Lev, and Rosenschein, 2014).