Research map / Multi-agent systems and games / Games and equilibria
Preventing Arbitrage from Collusion When Eliciting Probabilities
Rupert Freeman, David M. Pennock, Dominik Peters, Bo Waggoner
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| colluding agents all report their group mean instead of truthful reports | quadratic scoring rule · 3 agents, reports (0.2,0.4,0.6) | total coalition payment: $1.84 -> $1.92 if rain; $2.44 -> $2.52 otherwise | Improved |
| set exponent k > (n+2)/(2 eps) with reports bounded in (eps, 1-eps) | M^k contract function · arbitrage freeness | arbitrage possible | Improved |
| allow extremal reports (0.98) instead of bounding reports away from 0 and 1 | M^k with k=2 · Example 1 beliefs (0,0,0,0.98,0.98) | per-agent payment under collusion: 0.8208 -> 0.8568 if X=0; 0.2808 -> 0.2968 if X=1 | Hurt |
| use large k to guarantee no arbitrage for eps = 0.1, n = 5 | M^k · payment magnitude | principal's worst-case payment: k > 35, payments up to about 10^13 | Hurt |
| pay all agents by one strictly proper scoring rule (more than 2 choices) | fixed scoring-rule contract function · arbitrage freeness | arbitrage possible | Hurt |
| restrict to 2-choice scoring rule such as the 0-1 score | fixed scoring-rule contract function · weak distinguishability | weak distinguishability | Improved |
| restrict to 1-choice (constant) scoring rule | fixed scoring-rule contract function · weak distinguishability | weak distinguishability | Hurt |
| pay every agent the score of the median report | strictly proper scoring rule · arbitrage and dominant properness | arbitrage freeness and weak dominant properness | Improved |
| median payment rule used in place of individual scoring | median rule · weak distinguishability | weak distinguishability | Hurt |
| add a scaled median-rule term c/(n+1) * r(med) to a 2-choice scoring rule | 2-choice scoring rule · arbitrage, distinguishability, dominant properness | combined properties | Improved |
| add the median-rule term without the small scaling c/(n+1) (ablation) | 0-1 score plus quadratic median rule · p=(1,1,0) vs reports (0.5,0.5,0.5) | total payment: $1 -> $2.25 if X=0; $5 -> $5.25 if X=1 | Hurt |
Closest papers by meaning
- Domination-Avoiding Learning Agents Cannot Collude (NeurIPS 2026 · similarity 0.85)
- Hedging and Approximate Truthfulness in Traditional Forecasting Competitions (AAAI 2025 · similarity 0.84)
- A Unified Framework for Uniform-Price Resource Allocation Mechanisms (NeurIPS 2026 · similarity 0.83)
- Plant-and-Steal: Truthful Fair Allocations via Predictions (NeurIPS 2024 · similarity 0.83)
- Carrot and Stick: Eliciting Comparison Data and Beyond (NeurIPS 2024 · similarity 0.83)
- Network Intervention by Polling Strategic Agents (NeurIPS 2026 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
