amm-fairness-impossibility

## Context This methodology comes from arXiv:2606.04959 "Fairness and Strategy-Proofness in Automated Market Makers" by Frank M. V. Feys (June 2026, 52 pages). The paper establishes a fundamental impossibility result for Automated Market Maker (AMM) design in decentralized finance (DeFi). ## Core Finding: The AMM Impossibility Theorem **No deployed AMM lets liquidity providers (LPs) vote on the trading function. This is structural, not an oversight.** On the weighted-product family with n assets: - **No aggregation rule is simultaneously fair AND strategy-proof** when n > 2 - **Arrovian fairness** forces a unique form: the weighted Aitchison centroid (weighted geometric mean of LPs' preferred pools) - **Fairness forces mean-type aggregation** while **strategy-proofness forces median-type aggregation** - The only rule satisfying both is **single-provider dictatorship** - **The obstruction is sharp**: it vanishes at n=2, where a fair strategy-proof rule exists ## Key Mathematical Results ### 1. Fairness → Weighted Aitchison Centroid - Under Arrovian fairness (unanimity, scale-invariance, independence of irrelevant alternatives), the unique aggregation rule is the weighted geometric mean of providers' preferred pools - This is the **weighted Aitchison centroid** in compositional data analysis - Under the Frongillo–Papireddygari–Waggoner equivalence, the centroid corresponds to **Genest's logarithmic opinion pool** ### 2. Strategy-Proofness → Median-Type Aggregation - Strategy-proofness (no LP can benefit from misreporting preferences) forces median-type aggregation - Mean-type and median-type aggregation are fundamentally incompatible for n > 2 ### 3. The Impossibility Transfer - The impossibility transfers to **externally Bayesian pooling** via the Frongillo–Papireddygari–Waggoner equivalence theorem - This connects AMM design to the broader literature on opinion aggregation and belief pooling ### 4. The n=2 Exception - When there are exactly 2 LPs (n=2), the obstruction vanishes - A fair, strategy-proof rule exists for 2-provider AMMs ## Implementation Steps ### Step 1: Identify the AMM Family - Determine if the AMM uses weighted-product trading functions (e.g., Uniswap v3, Balancer) - Weighted-product family: ∏(x_i^w_i) = k, where x_i are reserves and w_i are weights ### Step 2: Count Liquidity Providers - If n = 2 LPs: fair strategy-proof rules exist → proceed with design - If n > 2 LPs: impossibility holds → must choose between fairness and strategy-proofness ### Step 3: Choose Trade-off Strategy **Option A: Prioritize Fairness (recommended for DeFi protocols)** - Implement weighted Aitchison centroid aggregation - Accept that LPs may have incentives to misreport preferences - Use reputation systems or stake-weighted voting to mitigate manipulation **Option B: Prioritize Strategy-Proofness** - Implement median-type aggregation - Accept that some LPs' preferences may be systematically underrepresented - Use rotation mechanisms to ensure long-term representation **Option C: Restrict to n=2** - Design AMMs with exactly 2 LPs or 2 LP groups - Achieve both fairness and strategy-proofness - Practical for bilateral trading venues ### Step 4: Design Governance Mechanism - If using centroid aggregation (Option A), implement: - Stake-weighted preference submission - Time-weighted preference aggregation (prevent flash preference attacks) - Preference smoothing to reduce manipulation incentives ### Step 5: Verify Properties - Check: Does the aggregation rule satisfy unanimity? - Check: Is the rule scale-invariant? - Check: Does the rule satisfy IIA (independence of irrelevant alternatives)? - Check: Is the rule strategy-proof (or accept it isn't)? ## Pitfalls - **Assuming fairness and strategy-proofness are compatible**: They are not for n > 2 LPs on weighted-product AMMs - **Ignoring the n=2 exception**: Two-provider AMMs CAN be both fair and strategy-proof - **Confusing weighted-product with other AMM families**: This impossibility applies specifically to weighted-product family (Constant Product, Constant Sum, etc.) - **Not considering the Frongillo–Papireddygari–Waggoner equivalence**: This equivalence transfers impossibility to externally Bayesian pooling, affecting broader mechanism design - **Overlooking Genest's logarithmic opinion pool connection**: The centroid is mathematically equivalent to this pooling method, with known properties and limitations ## Verification 1. For any proposed AMM aggregation rule on weighted-product family with n > 2 LPs: - Verify it satisfies Arrovian fairness axioms → it must be the weighted Aitchison centroid - Verify it is strategy-proof → it must be median-type aggregation - Since centroid ≠ median-type for n > 2, both cannot hold simultaneously 2. For n = 2 LPs: - Verify existence of fair, strategy-proof rule - The obstruction vanishes at exactly 2 providers ## Activation AMM design, automated market maker, DeFi protocol, liquidity provider voting, fairness in DeFi, strategy-proof mechanism, Arrovian impossibility, Aitchison centroid, opinion pooling, mechanism design, game theory, weighted-product AMM, Uniswap governance, Balancer governance, prediction market design, externally Bayesian pooling