> No, ZK schemes have absolutely nothing to do with sybil resistance. That's a threat modeling concern anyways.

Any effective form of sybil resistance itself compromises the privacy that ZK proofs allegedly provide.

Example: You create an account which is then unintentionally linked to your identity in some way, or someone doxxes you. That account is now irrevocably compromised and you need to create a new one, but any effective anti-sybil mechanism prevents you from doing so.

Example: Services use "sign in with Google" or similar to link your activity across services. To prevent this you need a separate Google account for each service, the exact thing any effective anti-sybil mechanism would prevent.

Example: A journalist is doing a series on some company's shady practices. After the first installment, the company uses the published description of the account activity required in order to write the story to identify the small subset of accounts that could belong to the journalist and then bans them or makes their accounts behave differently than ordinary accounts to prevent them from investigating further. An anti-sybil mechanism prevents the journalist from creating a new account not in the identified subset.

Example: A new company wants to create a search engine. Shady websites want to present different content to the search index than they do to users who visit. To detect this the company needs its systems to sometimes request content in a way the site can't distinguish from an ordinary user, to make sure it matches what will go on the search results page. Provide the scammers with an anti-sybil mechanism and once they identify which clients are validating their responses, they give the identified clients the same responses as they give the indexing bot, the search engine can't create new unidentified clients (even while attackers can via identity theft) and you get more scams in the search results page.

Example: A researcher is trying to determine if a service is using redlining, i.e. discriminating on the basis of location. To do this they need to create separate uncorrelated accounts that differ only in their address. An anti-sybil mechanism prevents this.

Example: There is a service the use of which intrinsically reveals some data point about you, e.g. narrows you down from one in 8 billion to one in a billion, each time you use it. Use it twice and it's one in 125 million, then one in 16 million and so on. If your current use isn't correlated with your past use then this is fine. If it is, which is the thing anti-sybil measures require, then repeated use of the service forces you to become uniquely identified.

That's the motte and bailey of zero-knowledge proofs. You present something that would protect privacy as long as it isn't combined with something else that would compromise it, but when it is then the proposed system doesn't actually protect privacy, and if it isn't then it serves no access restricting function and makes the system ineffective and pointless.

> What you describe is (or at least should be) a non-goal for any scheme in this context, because it is fundamentally unsolveable.

That's the point. Identifying someone without identifying them is fundamentally unsolvable. Zero-knowledge proofs cannot make the impossible happen, there is no cryptography that can give you that. But that is the claimed goal of the system -- to prove that someone has characteristic X without having any other information about them.

When you have no other information about them, you cannot establish that they are the person with characteristic X. Anyone can let anyone else use their ID. When you do, the use of zero-knowledge proofs does nothing to prevent the use of that other information to identify them.

> This is (loaded) political speculation, not an argument.

Reasonable political predictions are a valid form of argument. Authoritarians using mass surveillance for oppression and worse has an enormous amount of precedent and the consequences are orders of magnitude more severe than any possible advantages of any form of digital ID.