MACI

MACI (Minimal Anti-Collusion Infrastructure) is a cryptographic protocol designed to enable secure, private, and coercion-resistant voting on blockchain networks, primarily for applications like quadratic funding and decentralized governance. It ensures that a single, trusted coordinator can tally votes correctly while preventing voters from proving how they voted to third parties, which mitigates bribery and collusion. The system leverages zero-knowledge proofs (ZKPs) to allow the coordinator to prove the integrity of the vote tallying process without revealing individual votes.

The core mechanism of MACI involves a three-phase process: key submission, voting, and message processing. First, users generate a keypair and register their public key. When casting a vote, they encrypt their choice using the coordinator's public key. The coordinator then processes all encrypted votes in a batch, using their private key to decrypt them, apply the voting logic (e.g., quadratic funding calculations), and produce a final tally. Critically, the coordinator also publishes a zk-SNARK proof that verifies the entire process was executed correctly according to the protocol rules, without leaking any private voter data.

MACI's security model is based on the assumption of at least one honest participant—the coordinator. While the coordinator has the technical ability to censor votes, they cannot forge or alter votes without detection due to the public verifiability of the zk-SNARK proof. This creates a trust-minimized system where the coordinator's role is to provide compute resources and maintain liveness, while cryptographic guarantees ensure the outcome's correctness. The protocol's design makes it economically irrational to attempt large-scale collusion, as successful coercion would require compromising every individual voter's private key.

The primary application of MACI is in decentralized autonomous organization (DAO) governance and public goods funding mechanisms like Gitcoin Grants. In quadratic funding, MACI ensures that contributors cannot prove their donation patterns to projects, preventing sybil attacks and collusion where donors and projects might conspire to extract matching funds. By making votes private and un-provable, MACI aligns the economic incentives of the voting mechanism with the desired outcome of honest preference expression, moving beyond simple token-weighted voting.

Implementing MACI involves significant technical complexity, including managing key management for users, handling state updates, and generating the requisite zero-knowledge proofs, which can be computationally expensive. Projects like clr.fund and Ethereum's PSE (Privacy & Scaling Explorations) group have built production implementations. Ongoing research focuses on reducing the trust assumptions further, such as exploring decentralized coordinator networks or multi-party computation (MPC) to eliminate the single coordinator role entirely.

MACI stands for Minimal Anti-Collusion Infrastructure. The name was coined by Ethereum researcher Vitalik Buterin in a 2018 blog post, "An overview of quadratic voting and related mechanisms: collusion, bribery and the "antitragedy of the commons"". It was conceived as a cryptographic framework to enable applications like quadratic funding and decentralized governance while providing strong, on-chain guarantees against two primary threats: coercion (where a user is forced to vote a certain way) and collusion (where users are bribed to vote a specific way).

The "Minimal" in MACI refers to its design philosophy of achieving these anti-collusion properties with the fewest possible trust assumptions and cryptographic components. The core mechanism relies on a central, non-colluding coordinator—a role that is kept honest through cryptographic proofs (specifically, zk-SNARKs) that allow anyone to verify the coordinator processed all messages correctly. This creates a minimal, verifiable layer of infrastructure rather than a complex, multi-party system.

The term's etymology reflects its foundational purpose in blockchain governance. It is not merely a voting system but an infrastructure layer designed to be anti-collusion by construction. This distinguishes it from simpler voting smart contracts, which are transparent and thus vulnerable to bribery and voter retaliation. MACI's name has become synonymous with a class of privacy-preserving, coercion-resistant voting protocols, most famously implemented in projects like clr.fund for quadratic funding and various DAO governance experiments.

Minimum Anti-Collusion Infrastructure (MACI) is a cryptographic protocol that secures on-chain voting and decision-making by making votes private and coercion-resistant. It prevents vote buying and voter coercion by ensuring that even the voter themselves cannot cryptographically prove how they voted to a third party. This is achieved through a combination of zero-knowledge proofs (ZKPs), public-key cryptography, and a trusted coordinator who processes votes off-chain.

The process begins with a key ceremony, where each participant generates a public-private key pair, registering their public key with a smart contract. When casting a vote, a user encrypts their choice with the coordinator's public key and submits it on-chain, along with a ZKP that validates the vote's format and signature. Crucially, users can change their vote by submitting a new, valid message, with only the final vote being counted, which thwarts coercion attempts that demand proof of a specific vote.

The coordinator is a semi-trusted entity responsible for the protocol's privacy guarantees. After the voting period ends, the coordinator downloads all encrypted votes, decrypts them using its private key, and performs an off-chain tally. To prove the tally's correctness without revealing individual votes, the coordinator generates a zk-SNARK proof (a type of ZKP) that attests to the proper execution of the aggregation logic. This proof is then published on-chain, allowing anyone to verify the final result's integrity.

MACI's security model assumes the coordinator is honest but curious; it may try to learn individual votes but cannot alter the result without detection due to the public verifiability of the zk-SNARK proof. The system's resilience against collusion stems from the unforgeable signature scheme and the inability to prove vote content. Since a user's final message invalidates previous ones and the coordinator's decryption key is required to read votes, neither voters nor external adversaries can reliably demonstrate voting behavior.