Comment by palata

1 day ago

Reading the comments here, I see a lot of criticism along the line of "age verification doesn't work, it's completely stupid".

I believe it is counter-productive, because "not having age verification" is a lost battle. Unlike E2EE (where it is impossible to give access "only to the good guys"), it is possible to implement age verification in a privacy-preserving manner. And look at the ChatControl fight: even though it is not possible, we are still struggling to convince politicians of it. Good luck with age verification where it is actually possible to do something.

It should be a public service: just like the government issues IDs already, it should run the privacy-preserving system that allows citizens to prove their age. We should fight for that, otherwise we will get non-privacy-preserving systems managed by private companies (which is already starting).

That won't happen. Because the intent of the people pushing for "age verification" has nothing to do with the "think about the children" moral panic. It has to do with eliminating encryption and eliminating online anonymity. It is a dog whistle.

https://en.wikipedia.org/wiki/Dog_whistle_(politics)#

https://rationalwiki.org/wiki/Code_word

https://en.wikipedia.org/wiki/Moral_panic

  • I disagree. Maybe that's the intent of some people.

    Now go in the street and ask random people: "if there was a safe way to protect your children from accessing XYZ on the internet, do you think it would be a good thing?".

    Clearly one very real problem for parents right now is that if all the other kids do it, then it's hard to prevent your kid from doing it ("everybody is on TikTok, they make fun of me because I have no clue what's happening there"). If you can prevent most of them from accessing the service, then suddenly it becomes normal for kids not to use it.

I'm writing my (Canadian) MP to this effect.

There are a lot of issues with the UK approach. Privacy is a big one. But requiring this on every service is both a tax on the service and requires constantly authorizing stuff. That opens up the possibility for scams, data misuse, etc.

And no, saying we said to only use the data for verification clearly doesn't work. It didn't work for discord, or Persona, or Tea or AU10TIX or any others. Verification now means sharing that data with credit agencies and third party databases. Verification means keeping some data to resolve customer support disputes. There's data leakage for training and creating derived data products like biometric embeddings for future use.

Third party verification is a security nightmare.

I don't know why device based approvals abd controls aren't considered at all. Or really any privacy preserving technique.

And all this for ~54% efficacy?

  • There is no such thing as privacy protecting or anonymous age verification. If you tell Canadian that such a thing is possible, they are guaranteed to harm privacy with any legislation they proposal. Just tell them no.

    • > There is no such thing as privacy protecting or anonymous age verification

      There most definitely is privacy-protecting age verification. You go to a government office, you show your ID, they give you a piece of paper that officially says "over 18 years old". Now you have a piece of paper that says you're over 18 but doesn't say who you are, and the government won't know where you use it.

      On the Internet, the idea is the same, but with cryptography.

To date I haven't seen an implementation that preserves privacy and doesn't allow for easy bypass because person A generated infinite tokens and hands them out via a rest request.

  • I have seen implementations that preserve privacy. But fundamentally it means that an adult could give a token to a kid, as you say. But how bad is that? We don't need a perfect system, we just need it to be good enough that it prevents most kids from accessing stuff they shouldn't access. Some kids will always find a way anyway.

    A simple solution to "generate infinite token and hands them out via a rest request" could be one of:

    * Rate-limit the token generation. Nobody needs thousands per day, right?

    * Make it illegal to distribute tokens. The server sees if you request an abnormal amount of tokens, and... it knows who you are. Not too hard to investigate.

    * Make "honeypots" that scare the children when they try to access/buy the token.

    I don't think it makes the concept completely useless.