Comment by teravor
1 day ago
> use of a persistent identifier
at the terminus, yes. there is no other way to avoid the homeless problem you listed. by terminus I am referring to where a central authority vouches for unforgability. this does not mean advertisers will have a token they can use (see remote attestation infrastructure).
> tied to a person
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.
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.
> at the terminus, yes. there is no other way to avoid the homeless problem you listed. by terminus I am referring to where a central authority vouches for unforgability. this does not mean advertisers will have a token they can use (see remote attestation infrastructure).
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.
it's actually clear that you are the one who isn't familiar with this, I referenced remote attestation which you appear to know little about as it addresses the problem of identifying information (the service has no way to link tokens across without help from the CA).
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.
> it's actually clear that you are the one who isn't familiar with this, I referenced remote attestation which you appear to know little about as it addresses the problem of identifying information (the service has no way to link tokens across without help from the CA).
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.