privacy-technology-comparison
Use when the agent must choose a privacy technology — eERC (zk-SNARK + ElGamal) vs FHE (Zama/Fhenix) vs shielded pools/mixers vs a private L1 — and justify the pick for a given use case.
SKILL.mdskills/privacy-technology-comparison/SKILL.md
GitHubPrivacy Technology Comparison
Overview
eERC is one of several ways to get privacy in web3. Picking correctly depends on what you need private (token amounts? arbitrary computation? the whole chain?) and your constraints (EVM compat, gas, trust).
The landscape
| Technology | What it hides | How | Best for | Trade-offs |
|---|---|---|---|---|
| eERC | token balances + amounts (+ metadata) | zk-SNARK (Groth16) + ElGamal homomorphic, client-side proofs | confidential tokens/payments on EVM, compliant | token-specific; addresses/graph still public; proof gas |
| FHE (Zama fhEVM, Fhenix CoFHE) | arbitrary encrypted state & computation | fully homomorphic encryption on encrypted types | confidential general-purpose contracts (auctions, voting) | heavier compute; different programming model |
| Shielded pools / mixers | sender↔receiver link, amounts | zk shielded UTXO / anonymity sets | anonymity / unlinkability of transfers | anonymity-set dependent; compliance friction |
| Private / permissioned L1 | the whole chain (access) | restricted RPC + allowlists | enterprise/consortium walled gardens | insiders see plaintext; not data privacy |
Decision guide
- Confidential ERC-20 amounts/balances, EVM-compatible, compliance-ready → eERC ([[eerc-introduction]]).
- Arbitrary confidential logic (private auctions, sealed-bid, encrypted voting, confidential DeFi math) → FHE. See sibling collections (
fhenix-skills,zama-skills). - Unlinkability of who-paid-whom → shielded pool / mixer (note compliance considerations).
- Hide the entire network from outsiders → private L1 ([[private-permissioned-l1s]]).
- Maximum coverage → combine eERC + private L1 ([[combining-eerc-with-private-l1]]).
eERC vs FHE (the common question)
- eERC: purpose-built for token value privacy; partially homomorphic (ElGamal add) + zk proofs; lighter, EVM-native, auditor built in.
- FHE: general computation over encrypted data; more powerful but heavier and a new mental model. Use FHE when you need encrypted logic beyond balances.
Common Pitfalls
- Forcing eERC to do general FHE. It encrypts token amounts, not arbitrary contract state.
- Treating a private L1 as data privacy. It's access privacy ([[private-l1-metadata-leakage]]).
- Ignoring compliance when choosing mixers.
AI Agent Prompt
"Act as a privacy-tech advisor. Given my requirement (what must be private + constraints), recommend eERC, FHE, a shielded pool, a private L1, or a combination, with justification and trade-offs."
References
See references/README.md (eERC comparison academy page; sibling fhenix-skills/zama-skills for FHE).