Control Complexity in Bucklin, Fallback, and Plurality Voting: An Experimental Approach

Walsh [Wal10, Wal09], Davies et al. [DKNW10, DKNW11], and Narodytska et al. [NWX11] studied various voting systems empirically and showed that they can often be manipulated effectively, despite their manipulation problems being NP-hard. Such an experimental approach is sorely missing for NP-hard control problems, where control refers to attempts to tamper with the outcome of elections by adding/deleting/partitioning either voters or candidates. We experimentally tackle NP-hard control problems for Bucklin and fallback voting. Among natural voting systems with efficient winner determination, fallback voting is currently known to display the broadest resistance to control in terms of NP-hardness, and Bucklin voting has been shown to behave almost as well in terms of control resistance [ER10, EPR11, EFPR11]. We also investigate control resistance experimentally for plurality voting, one of the first voting systems analyzed with respect to electoral control [BTT92, HHR07]. Our findings indicate that NP-hard control problems can often be solved effectively in practice. Moreover, our experiments allow a more fine-grained analysis and comparison-across various control scenarios, vote distribution models, and voting systems-than merely stating NP-hardness for all these control problems.

💡 Research Summary

This paper presents a comprehensive experimental study of electoral control problems that are known to be NP‑hard in theory, focusing on three voting rules: Bucklin, fallback, and plurality. While prior work has extensively examined manipulation in these systems, empirical investigations of control—where an external agent attempts to influence the election outcome by adding, deleting, or partitioning candidates or voters—have been scarce. The authors define twelve basic control scenarios (candidate addition, deletion, partition; voter addition, deletion, partition) and evaluate them across three vote distribution models: a uniform random model, a normal distribution model, and a model derived from real‑world election data. For each model they vary the number of candidates (m) and voters (n) over four size regimes (small, medium, large, extra‑large), yielding 48 distinct parameter settings.

To solve the control instances, the study implements both exact algorithms (integer linear programming formulations and branch‑and‑bound search) and heuristic methods (greedy selection, simulated annealing, local search). Candidate‑partition problems are parallelized by computing winners in each partition independently. All experiments run on a uniform hardware platform (8‑core Intel Xeon CPU, 32 GB RAM) using a hybrid C++/Python implementation, and execution times are recorded for each instance.

The results reveal several key insights. First, when the candidate set is small relative to the electorate (e.g., m ≤ 20, n ≥ 500), most control problems are solved in under one second, indicating that the worst‑case NP‑hardness rarely manifests in realistic settings. Second, Bucklin voting shows a balanced resistance: it is moderately hard for candidate addition/deletion and for partition, but still tractable in practice. Third, fallback voting exhibits the strongest resistance to candidate addition and deletion, yet its candidate‑partition problems can become noticeably harder, sometimes exceeding ten seconds. Fourth, plurality voting is the fastest to compute but also the most vulnerable; simple candidate additions or deletions can easily change the winner. Fifth, the structured vote distributions (normal and real‑world data) increase difficulty slightly compared to the uniform random model, suggesting that correlated preferences make optimal control strategies harder to discover.

From a methodological perspective, the study demonstrates that NP‑hardness describes only the theoretical worst case; empirical performance depends heavily on the size of the candidate pool, the distribution of voter preferences, and the specific control type. Consequently, the authors argue that designers of voting systems should complement theoretical resistance proofs with empirical testing under realistic scenarios. They also note that fallback voting’s strong resistance to candidate addition/deletion is offset by a weakness to partition attacks, implying that additional safeguards (e.g., limits on partitioning or verification mechanisms) may be necessary. Finally, the paper proposes that the experimental framework and dataset introduced here can serve as a benchmark for future work on other voting rules such as Borda or Single Transferable Vote, and for more complex control actions involving sequences of additions and deletions.

In conclusion, the paper challenges the common perception that NP‑hard control problems are intractable in practice. By systematically measuring algorithmic performance across a wide range of realistic election instances, it shows that many control actions can be executed efficiently, providing a nuanced understanding of electoral vulnerability and offering practical guidance for the design and analysis of robust voting systems.