A Tunable Incentive Mechanism for Binary Aggregation Without Verification
摘要
Binary aggregation without verifiable ground truth arises when agents' reports must be aggregated without access to gold-standard labels. This paper studies a tunable reward--penalty mechanism for binary aggregation without verification. Agents choose between a conforming strategy, which reports an informative private signal, and a non-conforming strategy, which follows a deterministic prior-informed report rule. For this mechanism, we derive cost-adjusted sufficient conditions for incentive compatibility and individual rationality as bounds on the reward--penalty ratio. The analysis identifies feasible ratio regions, cases in which ratio adjustment restores feasibility, and parameter regimes in which no ratio satisfies both constraints under the modeled construction. We also state a conditional all-conforming Nash equilibrium result within the restricted strategy set. Entropy-based scaling and stake-weighted redistribution are treated as extensions, with stake-weighted redistribution inducing agent-specific incentive constraints. Numerical checks support the closed-form Tier 1 quantities and illustrate threshold sensitivity.
引用
@article{arxiv.2606.30974,
title = {A Tunable Incentive Mechanism for Binary Aggregation Without Verification},
author = {Chien-Chih Chen and Wojciech Golab},
journal= {arXiv preprint arXiv:2606.30974},
year = {2026}
}
备注
Preprint. 17 pages, 2 figures