Was reading through Dusk Network's Citadel protocol during the task. The citadel repo on GitHub picked up commits on August 8 — active work, not a static spec — and I ended up stuck on the three-party model longer than intended.
The phrase "selective disclosure" gets used a lot in the $DUSK @Dusk #dusk materials. It's accurate. But there's a structural thing the architecture reveals that the pitch doesn't foreground.
Citadel has three parties: User, License Provider (LP), Service Provider (SP). The ZK proofs protect you from the SP — they verify you meet a compliance threshold without seeing your actual data. That part works as described. But the LP does full KYC. They hold your data. They issue the on-chain license. Every subsequent service provider only gets a proof, which is elegant. The friction is earlier, at onboarding, not at every gate.
So Dusk's privacy model isn't "hidden from authority." It's "hidden from counterparties, visible to your chosen authority." The LP knows everything. The SPs know nothing. For regulated finance that's probably the correct design — someone has to be the responsible data custodian for regulators. But it reads differently from how most people interpret "blockchain privacy," which tends to mean hidden from everyone by default.
Hmm… the interesting open question is who actually plays LP in practice. If it's a regulated custodian or a licensed KYC provider, that's outsourced identity infrastructure with better UX, not decentralized identity. I'm not sure those two framings ever fully resolve into each other.
The phrase "selective disclosure" gets used a lot in the $DUSK @Dusk #dusk materials. It's accurate. But there's a structural thing the architecture reveals that the pitch doesn't foreground.
Citadel has three parties: User, License Provider (LP), Service Provider (SP). The ZK proofs protect you from the SP — they verify you meet a compliance threshold without seeing your actual data. That part works as described. But the LP does full KYC. They hold your data. They issue the on-chain license. Every subsequent service provider only gets a proof, which is elegant. The friction is earlier, at onboarding, not at every gate.
So Dusk's privacy model isn't "hidden from authority." It's "hidden from counterparties, visible to your chosen authority." The LP knows everything. The SPs know nothing. For regulated finance that's probably the correct design — someone has to be the responsible data custodian for regulators. But it reads differently from how most people interpret "blockchain privacy," which tends to mean hidden from everyone by default.
Hmm… the interesting open question is who actually plays LP in practice. If it's a regulated custodian or a licensed KYC provider, that's outsourced identity infrastructure with better UX, not decentralized identity. I'm not sure those two framings ever fully resolve into each other.