Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Strategic Cost Selection in Participatory Budgeting
Authors: Piotr Faliszewski, Łukasz Janeczko, Andrzej Kaczmarczyk, Grzegorz Lisowski, Piotr Skowron, Stanisław Szufa, Mateusz Szwagierczak
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We also provide an experimental study of cost selection on real-life PB election data. |
| Researcher Affiliation | Academia | Piotr Faliszewski AGH University Krak ow, Poland EMAIL Łukasz Janeczko AGH University Krak ow, Poland EMAIL Andrzej Kaczmarczyk University of Chicago Chicago, USA EMAIL Grzegorz Lisowski University of Groningen Groningen, The Netherlands EMAIL Piotr Skowron University of Warsaw Warsaw, Poland EMAIL Stanisław Szufa CNRS, Universit e Paris Dauphine-PSL Paris, France EMAIL Mateusz Szwagierczak AGH University Krak ow, Poland EMAIL |
| Pseudocode | Yes | Algorithm 1: Finding a Nash equilibrium for AV/cost. |
| Open Source Code | Yes | Our code is publicly available at https://github.com/Project-PRAGMA/strategic-cost-selection-in-PB--Neur IPS2025. |
| Open Datasets | Yes | We conduct such an analysis based on the Pabulib dataset [Faliszewski et al., 2023]. |
| Dataset Splits | No | The paper uses real-life PB instances from the Pabulib dataset, such as Wesola and Kleine Wereld, and analyzes them with their original project costs. It does not describe any explicit train/test/validation splits or methodologies for partitioning these datasets for model training or evaluation. The analysis involves simulating dynamics on these complete instances. |
| Hardware Specification | Yes | None of the experiments required extensive computational resources. All the experiments can be feasibly computed on a standard Mac Book Air with M1 chip. |
| Software Dependencies | No | The paper does not explicitly mention specific software dependencies with version numbers (e.g., programming languages, libraries, or frameworks with their versions) that would be needed to replicate the experiments. |
| Experiment Setup | Yes | Given a PB instance, our dynamics go as follows. First, each proposer reports the same cost as was originally chosen for his or her project. Then, in each iteration, one of them, selected uniformly at random, either slightly increases or decreases his or her project s cost. Specifically, the proposer chooses a number x between 0 and cost/10 uniformly at random (where cost is the current project s cost) and if their project was losing in the previous iteration, then the proposer decreases its cost by x, and if it was winning, then he or she increases its cost by x (but if this action would have caused the project to lose, then the proposer does not change the project s cost). We expect to converge to an NE, if one exists, after running sufficiently many iterations. In Figure 2 we present the results of the dynamics, that is, the obtained strategy profiles, after 10 000 iterations. |