Comment by rsynnott
15 days ago
It's kind of surprising that no-one has really come out with a proper privacy-preserving approach to this yet. It is clearly _possible_; there are reasonable-looking designs for this. But no-one's doing it; they're just collecting photos and IDs, and then leaking them all over the place.
Here's my solution: https://news.ycombinator.com/item?id=46447282
The problem is privacy activists and free speech activists (though there's some overlap between the two they aren't the same) oppose age verification by any means since it has the potential to infringe on both ever so slightly. Meanwhile age verification gates are being demanded and thrown up all over the Internet at a frightening pace. So we get only the maximal data collection solutions implemented by people who don't give a shit about privacy or free speech. And the mass surveillance cheerleaders egg them on.
If privacy and free speech activists understood that a proactive, privacy-preserving approach to age verification is the best outcome we'd be better off.
You need to process that other people disagree with that claim, and do not believe we'd be better off.
We should not accept the Overton window shifting here, and say "well, if we do it to ourselves, in a privacy-preserving way, that's less bad".
> You need to process that other people disagree with that claim
I think I already said that in my original post.
> We should not accept the Overton window shifting here
Great! Let's say you and I refuse to accept it. How do we keep Discord from demanding passports or selfies? How can we get France[1] or Finland[2] to roll back age restrictions on social media?
You'll never convince a majority of voters in democracies that nothing online should be age-restricted. These are the people that the enemies of anonymity and free speech are counting on to advance their agenda.
At the same time a majority of voters is currently quite content with the state of age verification for access to tobacco and alcohol. Both its strictness (or lack thereof) and privacy preservation (almost perfect).
I'm not saying my proposal is the one that should be adopted. I honestly don't care which idea gets picked and I don't want anything from it. But it's a virtual guarantee that in the absence of a competing good-enough, privacy-preserving implementation, only the most privacy-invasive idea will be implemented.
1. https://news.ycombinator.com/item?id=46838417
6 replies →
It really would be less bad though wouldn't it?
The more we resist turning this into a state-sided solution which provides a service to private companies with a YES/NO age verification, the more likely your data is going to be given to botton-of-the-barrel third party private companies.
I'm genuinely curious what the argument is against state-run privacy focused age verification is here. We already protect real life adult spaces with IDs. You hand your ID to a random store clerk who scans it with a random device when you want to buy alcohol or cigarettes.
What makes these social media platforms special that they have entirely different rules?
I will say, if they came for small privately-hosted communities, I can understand the cause for alarm. But so far it appears to be limited to massive misinformation machines.
1 reply →
Much like DRM, there is no good option. Its a fundamentally bad thing. If parents want to abdicate their parental responsibilities, their children should bare the cost of that, not millions of strangers.
> If parents want to abdicate their parental responsibilities, their children should bare the cost of that, not millions of strangers.
Oh but we all will. One way or another. There is a least bad option.
1 reply →
The issue with your solution still comes down to yet another centralizing middleman with no real incentive to be efficient. And all the incentive to lobby governments and extract more wealth from the people.
This can of course be done government by government, but that isn't scalable for a global company.
It doesn't have to be 1 middleman. Multiple companies can issue the cards, just like there are multiple beer and cigarette and lottery companies.
I wish I could edit my post because a lot of people had the same misconception when I first wrote it.
3 replies →
There is no 'half-pregnant' option. Compromise is synonymous with 'bring into danger' for a reason. They are right to be dogmatic about rights - believing that is like believing it will really be 'just the tip'.
> ever so slightly
It’s not “slightly”. They’ll start with claiming to protect people under 18 from obviously problematic content — porn, grooming, etc.
It won’t stop there. The scope creep will extend to expressing or reading “incorrect” or “dangerous” views.
They’ll probably call some of it “hate speech”, but hate speech is whatever the people in power say it is; on X, “cisgender” is designated as a slur and gets your post censored.
The slippery slope fallacy is only a fallacy if the slope isn’t slippery — “think of the children” is a wedge bad actors are once again trying to use to open the floodgates of censorship.
They don’t even need to target adults; if you control what children can see and express, you have enormous control over all future generations of voters.
I agree, but the powers that be loathe the phrase "hate speech". I'm betting the next encroachment will be on "violence", "terrorism" or even Russian-style "promotion of nontraditional values".
It's already happening. What's your alternative? Not VPNs because every jurisdiction and website will eventually have equivalent laws or terms of service.
4 replies →
It is only a matter of time before ID verification means the camera is always on watching the face of the person looking at the screen.
That's why I put black tape over my front-facing camera.
Expect a visit feom ho.eland security 10 years from now
They do not want to solve the problem, they want to collect our IDs. If they would have wanted to actually solve it they would not have done this on legislations where it is not a requirement.
> It is clearly _possible_
Is it?
I don't think it is.
I truly don't believe that there's any possible way to verify someone's age without collecting ID from them.
It's possible to (cryptpgraphically verifiably) split up the age verification and the knowledge of what the verification is for.
It would seem like a naive solution would be some arrangement where Discord would ask for a proof-of-age from an official service ran by the State (which issues your ID)
Well you could have government-run cryptographically signed tokens. They're already in the business of holding ID data (i.e. they don't need to collect it and this wouldn't increase the attack surface).
But assuming it has to be a private solution, you could do the same thing but make it a non-profit. Then at least _new_ services you wish to use don't need to collect your ID.
many countries already have a working system mostly integrated, so yes, i would say it is possible.
the government should issue physical tokens that are sold wherever you can buy booze or smokes. when you login to a service that needs age verification, you type in the code from your age token.
its pretty cheap, its low-tech, we are already accepting of showing id to a store clerk privacy-wise, we generally trust the enforcement mechanisms around smoking/drinking already, it would be easy to expand existing laws to accommodate selling them/punishing misuse.
https://www.apple.com/newsroom/2025/06/apple-expands-tools-t...
What are your thoughts on Apple's approach? You still have to provide your birthdate to apple. But after that, it only only ever shares your age range with other companies that request it, not your birthdate.
This is great, but if and only if it remains an opt-in choice that enables parents.
There is a stark difference between enabling choice or compelling it.
Somehow in the last 15 years, we have completely lost sight of agency-based ethics as a founding and fundamental principle of western liberalism.
This has been replaced with harm-based ethics. Harm has no fixed definition. There is no stopping rule — when will we have eradicated enough harm? It’s declared by fiat by whoever has the means to compel and coerce — and harm inherent in that enforcement are ignored.
As others have said, it’s obvious that no real attempts have been made by anyone to create a privacy-focused solution because the end goal is to collect photo IDs.
Occasionally in my free time I have been tinkering with a certificate-based solution that could fulfill this sort of need for age verification. It’s not the most robust idea but it’s simple enough using most of what we already have. Creating a minimal protocol which doesn’t share actual identifying information nor metadata of the site you’re accessing is trivial. If I can make an 80% solution in less than 100 hours of my free time then some groups with more money and intelligence could propose a dead-simple and easy-to-adopt solution just as easily.
No privacy is simpler and the simpler solution is cheaper. If there's no real incentive to go with another option, companies will go with the cheaper option.
They dont want it to be private. They want to profile you and sell your data
mDLs support selective disclosure of age.
[dead]