Postdoc Jeremy Vollen Wins Distinguished Paper Award at IJCAI-ECAI 2026
Vollen and the research team introduced incentive ratios to public decision problems and the design of fair collective decision-making algorithms
Northwestern Engineering’s Jeremy Vollen earned a Distinguished Paper Award at the 35th International Joint Conference on Artificial Intelligence and the 29th European Conference on Artificial Intelligence (IJCAI-ECAI 2026), held August 15-21 at the Congress Centrum Bremen and University of Bremen in Bremen, Germany. IJCAI-ECAI is a premier international gathering of researchers and practitioners in AI.
The winning paper, titled “Approximate Strategyproofness in Approval-based Budget Division,” was among the more than 700 papers accepted to the main track, out of more than 5,200 total submissions. Three papers received Distinguished Paper Awards. Vollen’s coauthors include Haris Aziz (University of New South Wales) and Patrick Lederer (University of Amsterdam’s Institute for Logic, Language and Computation).
“Receiving the Distinguished Paper Award is a great source of pride for me and my collaborators,” said Vollen, a postdoctoral scholar in computer science advised by Edith Elkind, Ginni Rometty Professor of Computer Science at the McCormick School of Engineering. “What has been most meaningful, though, is the opportunity for the award to bring the paper to a wider audience. I hope that this recognition leads to greater interest in computational social choice and its potential to help design better collective decision-making systems.”
In this work, Vollen and the team introduced incentive ratios to the design of fair collective decision-making algorithms applied to public decision problems, such as participatory budgeting. In that instance, a city council decides how to spend a public budget through a direct democratic process. Citizens submit ballots expressing their preferences over various community priorities, such as affordable housing trusts, small business grants, and subsidized youth programs.
Studying scenarios in which the use of a divisible resource should reflect many people’s priorities, the researchers aimed to determine how much participants could gain by strategically misrepresenting their preferences. Termed the “free-rider problem,” donors distributing contributions among charities, for instance, might understate their support for a popular charity in the hope that others will fund it anyway, allowing more money to be directed to other priorities.
“While no rule can prevent every possible manipulation, we provide a starting point for studying the tradeoffs among fairness, efficiency, and resistance to manipulation in a more precise way,” Vollen said. “Our work presents a framework for asking not merely whether manipulation is possible, but whether it is consequential.”
The team found that a method that seeks to balance the interests of all agents, called the Nash product rule, is robust to manipulation, producing efficient outcomes and satisfying strong fairness guarantees across a wide range of settings. They proved that a participant cannot increase their benefit by more than a factor of two by misreporting their preferences.
“We also show that several other natural decision rules can be manipulated much more severely, and that the factor of two achieved by Nash is essentially the best possible among several broad classes of fair and efficient rules,” Vollen said. “Taken together, our results suggest that Nash is a simple but unusually robust rule for dividing a shared budget.”
As a next step, Vollen and his collaborators aim to investigate whether stronger protection against manipulation can be achieved by realigning expectations with respect to fairness or efficiency. For certain applications, decision-makers may reasonably prefer a rule that sacrifices a small amount of fairness in exchange for making manipulation substantially less rewarding.
To better model decision-making and strategic behaviors, the team will also study how often meaningful manipulation arises in realistic environments in which participants lack or have incomplete information and develop incentive concepts that account for this uncertainty about other participants’ preferences.