the truth is that the two extremes you listed can be titrated.
if you use nullifiers you can trade some privacy for some security. basically you convert your true identity into a private token which you can use to authenticate aspects of yourself, the price being that the token can be tracked with some effort across services. better than just using your identity at least. if a token/nullifier is abused it can be revoked and then you have to jump through a bunch of hoops to get another.
there are some other trade offs that can be made.
What combination of details can you validate on that is meaningfully privacy-preserving and couldn't result in wide-spread re-use of tokens?
Additionally - what would prevent some kids from getting a homeless man in the city to hand them his ID, get a facial scan, and everything else you can think of to generate a token and then pass that token around?
ZKP are a cryptography-nerd's joy but are are categorically unsuitable for the purpose of age verification. I stand by this without the slightest reservation.
You can do this: when you want to log into a service, the service provider gives you a fresh challenge C, bound to that service/session. You sign the challenge, and then generate a zkp of the fact that:
1. you have the signed challenge C with a certain public key P 2. you have a state-signed credential/certificate that binds P with a person with birth date BD 3. current date - BD > 18 years 4. optionally, you derived a per-service nullifier, e.g. from the card/credential secret, the service origin and a time bucket, so the service can rate-limit abuse without getting a global cross-site identifier
You send the proof to the service provider, that verifies it, and learns nothing about you (except for the fact that you're of age).
An adult can of course give away the card/PIN, but you need to have it physically to sign fresh challenges, so it cannot be passed around as easily as a bearer token. Moreover he loses access to his actual ID, which is required for other services.
tying multiple accounts and services together isn't ideal but its inarguably better than tying your real world identity to every single service.
To clarify - it's not cryptographically necessary to present the same token for each and every transaction and serves to categorically defeat the entire privacy guarantee of ZKP.
It also makes it trivial to associate your ZKP token with your real identity.
> use of a persistent identifier
> tied to a person
contrast this with the situation as it currently is (under ideal assumptions) where a central authority verifies your real identity and issues temporary rate limited tokens which are then saved by each service and can at any time be linked to you whenever the central authority can get the service to disclose the database entry. the nullifier will force the central authority to do an investigation about who the nullifier actually belongs to which may actually fail.
realistically I expect VPNs and Tor to just become more popular in response to such nonsense. I wouldn't be using government issued tokens for anything that isn't trivial to tie to your identity already: such as a personal bank access.
Where to even begin here....
To generate the token, it needs to be based on specific data. How do you prevent people from generating tokens based on fake data and submitting that to the "terminus" that you mention? We already have cases of people bypassing facial scan liveliness checks for banks using AI-generated footage.
What about validating tokens during the token enrollment process based on your government ID? Though that makes sure that poor or undereducated people who don't have such an ID are locked out of large swaths of Internet services.
Though there's also the matter of it being trivial to generate fake IDs using AI.
If you have no gatekeeping for the token enrollment process, anyone can submit an arbitrary number of new tokens.
And if you do have gatekeeping, you're right back to square one of needing to validate against more than just your age.
After all - the cryptography algorithms will be publicly known. If the only thing ZKP is validating against is age, it won't take long to figure out how to generate identifiers based on fabricated information.
> whether or not the terminus can tie a token to a real world identity will depend on how careless the user was and how much collusion there is between the terminus and the services. at the very least it will impose an investigation cost.
No it won't. A user submits a token to a server. The user also logs in with their e-mail address or phone number. Their email and/or phone number is hashed and it, along with the ZKP token and any additional information the website has on you, will be sent to data brokers.
This is the same as any other bit of information out there that data brokers collect on the internet. They just associate your new info with other info you are required to provide in order to use various services.
This will be automated and will cost next to nothing for data brokers to take advantage of.
> contrast this with the situation as it currently is (under ideal assumptions) where a central authority verifies your real identity and issues temporary rate limited tokens which are then saved by each service and can at any time be linked to you whenever the central authority can get the service to disclose the database entry. the nullifier will force the central authority to do an investigation about who the nullifier actually belongs to which may actually fail.
....what? What investigation by central authorities? You are talking of a system that would constantly mediate permissions for billions upon billions upon billions of devices across dozens of services and accounts per device.
You couldn't hire an army of people large enough to handle this and AI is infamously awful at detecting when a given image has been generated with AI.
> realistically I expect VPNs and Tor to just become more popular in response to such nonsense. I wouldn't be using government issued tokens for anything that isn't trivial to tie to your identity already: such as a personal bank access.
Their popularity would only rise in order to VPN into jurisdictions that don't enforce this. Assuming major websites don't just mandate age/identity verification for all new users regardless of jurisdiction just because it's easier and cheaper to apply one system to everyone.
Look - I know you mean well, but it is clear from this discussion you aren't familiar with cryptography, system security guarantees, Internet infrastructure scaling, or what would be needed to introduce new descriptive information about a person on the Internet and not have it become a new privacy risk.
This is an issue that has no tech-only solution. The specifics aren't just something to just figure out at a later date - the specifics are everything. And it's something that is enormously difficult to get right and extremely easy to get very, very wrong.
> Look - I know you mean well, but it is clear from this discussion you aren't familiar with cryptography, system security guarantees, Internet infrastructure scaling, or what would be needed to introduce new descriptive information about a person on the Internet and not have it become a new privacy risk.
you also don't appear to know what a nullifier is, in a ZKP system you submit identifying information and a hash of a secret string. the CA adds the hash to a public database and in the future you prove you one of the members of the database with a nullifier - the anonymity-set is everyone in the database who entered it prior to your submission. this can also be done with a blind signature to the same effect.
there is no further point to this discussion.
You've promoted mutually exclusive concepts with regards to cryptography which is why I said you don't seem to understand it. And again - and again and again and again and again and again - what is the additional information you are authenticating based off of beyond age? Remote attestation provides absolutely zero privacy utility here whatsoever on its own! So you've remotely attested this ZKP key represents a person who is an adult. Creating another key based on that information alone is trivial to spoof - for it not to be trivial, it would require validating additional information!
What is your root of trust? What is the basis by which age is verified in a way that can't readily be spoofed?
> you also don't appear to know what a nullifier is, in a ZKP system you submit identifying information and a hash of a secret string. the CA adds the hash to a public database and in the future you prove you one of the members of the database with a nullifier - the anonymity-set is everyone in the database who entered it prior to your submission. this can also be done with a blind signature to the same effect.
That's nice and all for trivia on ZKP but how does that touch upon the problem being discussed?
The mechanics of ZKP are not relevant to the problem of ZKP being categorically worthless for the problem at hand. I don't say ZKP is worthless out of ignorance - more discussions about it won't change that.
The specifics of ZKP do not change the fact that you are validating either too little information to be useful for preventing fraud or too much to have privacy-preserving value.
> there is no further point to this discussion.
Evidently not.
We can't solve private age verification with blockchain tech. I'm happy you're so passionate about it, but it isn't a silver bullet.