Equilibrium Infra Bulletin #47: Privacy Cocktails, State of Onchain Futarchy, and more...
Equilibrium designs, builds, and invests in core infrastructure for the decentralized web. We are a global team of ~30 people who tackle challenges around security, privacy, and scaling.
🔍 Is ZK-MPC-FHE-TEE a real creature?
⚡️ Topic Summary
Individual PETs, such as ZKP, FHE, or MPC, can work well for specific use cases but are limited in their expressiveness and/or hardness of guarantees. However, the different PET combinations, their usefulness, and limitations can be challenging to reason about. This post by Lisa breaks down the different combinations in detail, including:
ZK-MPC: Facilitates collaborative proof generation among multiple parties, ensuring correctness without revealing individual inputs. Can be used for both private proof delegation (outsourcing client-side proving to a proving network while retaining some privacy guarantees) and private shared state (enabling multiple parties to compute over a private state).
MPC-FHE: While FHE enables computing over encrypted data, the key question is “who controls the decryption key?”. By leveraging MPC, the decryption key can be distributed among multiple parties, improving security and robustness.
ZK-FHE: Integrating ZKPs with FHE enables verification that the encryption and subsequent computation over that encrypted data were done correctly. However, a key challenge with ZK-FHE is the performance overhead, with current implementations facing challenges such as prolonged proof generation times.
ZK-MPC-FHE: Combining the two above, to get verifiable FHE while using MPC to distribute the decryption key. While this is more robust and reduces trust assumptions, it also adds significant overhead (at least for now).
TEE integration: Running computation within a trusted execution environment (TEE) keeps data isolated during execution, reducing the attack vector. TEEs can be used to construct ZK proofs and to participate in MPC and FHE protocols, increasing the robustness of the overall protocol.
🤔 Our Thoughts
Much of the debate around PETs (often facilitated by companies championing a specific solution) sounds like it’s an either-or solution, where you can only choose one PET. In reality, it’s more of a question of how these different PETs will work together (and where the tradeoffs make sense). However, as Lisa shows in the blog, the different combinations are not straightforward to reason about and introduce new tradeoffs and challenges to consider.
The ultimate privacy cocktail is combining all four of these PETs (ZK-MPC-FHE-TEE). While this is feasible in theory and may sound tempting, it introduces a lot of uncertainty around implementation and potential attack surface. We’ve argued before that all solutions that want to avoid single points of failure and enable shared private state, the trust assumptions ultimately boil down to MPC.
💡 Research, Articles & Other Things of Interest
🤓 FHE-SNARK vs. SNARK-FHE: From Analysis to Practical Verifiable Computation: This paper explores whether these two are identical in terms of security and their respective practical efficiency. While SNARK-FHE has a slight security edge, FHE-SNARK is far more practical in terms of performance and can be made more secure through knowledge soundness.
📚 The State of Onchain Futarchy: Futarchy is a market-driven governance system that allows anyone to express their views. While still nascent, futarchy protocols such as MetaDAO have been gaining traction recently with initial adoption from DAOs.
📚 The Next Billion Users Won't Know They're Using Crypto: Most users don’t think about TCP/IP when browsing the web, database architecture when using Instagram, or payment rails when buying something online. The most successful technologies are invisible. Similarly, crypto will fade into the background as it becomes more widely used - something we’re already starting to see.
🤌 Personal Recommendations From Our Team
📚 Reading: Meltdown: Scandal, Sleaze and the Collapse of Credit Suisse
🎧 Listening: Most recognizable song each year of the past 100 years (2024 version)