In principle, it is possible to make a privacy-preserving age check.

Site/app asks "≥18?", device generates a UUID specifically for that [app/domain + device], the device passes that UUID to government database along with the "≥18?" question and the ID card info but not the app/domain, government database gives the answer and signs just the UUID and the answer and doesn't include any ID card info.

Government doesn't know what you're looking at, website doesn't know your ID.

(There's probably also better ways to do all this, I'm not a cryptographer and I expect that will be obvious from this comment to people who are).

▲

Xelbair 10 hours ago | parent | next [-]

In reality, the ones making such check will want extra capabilities 'just in case' as we observe now, or just go for simpler solution because it's cheaper.

How does it prevent fingerprinting if you get access to both logs and try to time it? Even more so if you include already present tracking infrastructure, which as a government you can just request data from.

It is possible to de-anonymize people based on aggregate data already, and this proposed solution just adds extra data points.

Even in countries in which this requires a subpoena, agencies break the law frequently and don't get punished.

Legal systems aren't computer systems. This isn't a technical problem but a social/political one.

▲

ben_w 9 hours ago | parent [-]

> How does it prevent fingerprinting if you get access to both logs and try to time it? Even more so if you include already present tracking infrastructure, which as a government you can just request data from.

True, but as this problem exists without any ID at all*, I don't see how the addition of the ID cards makes any difference?

> It is possible to de-anonymize people based on aggregate data already, and this proposed solution just adds extra data points.

This is why I specified a "UUID specifically for that [app/domain + device]". Can't aggregate when each app/domain gets a different signed UUID.

> Legal systems aren't computer systems. This isn't a technical problem but a social/political one.

While true, the reverse also applies: computer systems are not legal systems.

I think many lawmakers' ignorance of this is to the detriment of everyone.

In this case, the social/political problem is: we want to stop kids accessing age-inappropriate material.

The options-space for doing this appears to be:

(0) give up

(1) require parents to limit their kids' behaviour (to which I say: "Have you met a kid? Do you remember being one?")

(2) require websites to age-rank appropriately (to which I say in a sarcastic tone of voice: "Gee, that worked soooooo well for GDPR")

(3) require operating systems to intermediate. Will this suck? Yes. Will it be buggy? Also yes. Will there be false positives and false negatives? Yes to both. Will it be a constant fight as kids keep finding loopholes? Indeed.

But you know what else? All that psych testing Facebook have used for evil, can also catch bugs and loopholes faster than kids can figure them out. A school full of kids can beat their parents at the security game because they have more time to spend on finding exploits than the parents have to spend keeping up, the reverse is true of the difference between kids and Google LLC etc.

It doesn't need to be perfect, the kids aren't a computer program.

What does need to be held to a high standard is making sure the OSes don't leak all over the place.

* to be specific, the Investigatory Powers Act 2016 is one of two reasons I left the UK, just look at how over-broad the "Internet connection records without a warrant" list is https://en.wikipedia.org/wiki/Investigatory_Powers_Act_2016#...

▲

baby_souffle 8 hours ago | parent | next [-]

> All that psych testing Facebook have used for evil, can also catch bugs and loopholes faster than kids can figure them out

What would be the incentive for meta to deploy that against their own self interests?

> Will it be a constant fight as kids keep finding loopholes? Indeed.

Good thing there are no other issues we as a species face. Let’s burn resources on a sysphian task because what else were we going to do with them…

▲

ben_w 8 hours ago | parent [-]

> What would be the incentive for meta to deploy that against their own self interests?

The normal method is passing laws. That's kinda the point of laws.

Zuckerberg may be arrogant as hell, think he can bully governments into bending over backwards for him, governments have guns and get to arrest (and have in the past arrested) anyone local who does what Zuckerberg says instead of what the laws say when they are in conflict.

> Good thing there are no other issues we as a species face. Let’s burn resources on a sysphian task because what else were we going to do with them…

Hardly. This is more like gardening. It matters much less if kids get a few days of messing around where they're not ment to be, than if it's continuous.

And ultimately, if you want to run a business without spending money on legally required actions, you're in the wrong business, nobody should shed tears for you.

	▲	handedness 4 hours ago \| parent [-]
		> Zuckerberg may be arrogant as hell, think he can bully governments into bending over backwards for him, governments have guns and get to arrest (and have in the past arrested) anyone local who does what Zuckerberg says instead of what the laws say when they are in conflict. In an ideal world. Personal experience has shown that isn't the case with him.

▲

Xelbair 5 hours ago | parent | prev [-]

>The options-space for doing this appears to be:

(0) give up

...

(n) waste resources

yeah, i think current state is good enough, might as well slap regulation and ban any form of targetted ads.

	▲	ben_w 5 hours ago \| parent [-]
		I'd be on board with banning micro-targeted ads, and more broadly any use of psychology to induce similar states as regulated exogenous drugs equivalent to the simple language descriptions of US Schedule I, II, and III. However, the issues are rather broader than showing ads to kids who have no money to spend on whatever is being advertised.

▲

baby_souffle 8 hours ago | parent | prev | next [-]

This can’t work because it creates a market for people that have an ID that cleared gates to lease/borrow to those that can’t.

	▲	ben_w 4 hours ago \| parent \| next [-]
		It cannot work perfectly, but unlike most crypto stuff, it doesn't have to (at least, that part doesn't need to). It's a social/political problem, not a technical one. The goal doesn't even need to be to make a completely perfect, impossible to circumvent, barrier for all minors; all this needs to do is just make it equally difficult for a minor to get adult (for whatever definition thereof) content online as it is to get tobacco or alcohol or gambling in real life, the hard part being to do so without compromising privacy, privacy being the bit which has to be done perfectly.
	▲	undersuit 4 hours ago \| parent \| prev [-]
		And the verification systems rushing at us are immune to this?

▲

fsflover 9 hours ago | parent | prev [-]

> Government doesn't know what you're looking at, website doesn't know your ID.

But then the government knows your location at any point of time and how often you use websites requiring the age check. Also, if your device is configured to do it automatically, a child can also follow this verification.

▲

ben_w 9 hours ago | parent | next [-]

> But then the government knows your location at any point of time and how often you use websites requiring the age check.

Not as described. There's no location info in that path, and the signed statement of that UUID passing the age check does not need to be re-signed because I've not given any consideration to expiry.

(Should I consider expiry? It's not like people age backwards?)

> Also, if your device is configured to do it automatically, a child can also follow this verification.

Yes, if that device has been associated with a government ID and also the government ID signing process fails to make use the things we've already got on-device like how my phone reads my fingerprint to know I'm me and can store copies of some forms of government ID (in some places but not where I live, Apple Wallet apparently only supports some US states, Japan, Greece, and UAE).

▲

baby_souffle 8 hours ago | parent [-]

> Not as described. There's no location info in that path

Ip address and general time of day will tell you a lot about location. Not street level, but country or state level.

> UUID passing the age check does not need to be re-signed because I've not given any consideration to expiry.

So as soon as any one uuid is leaked, it becomes plausible for any child to bypass the gates until the uuid is manually revoked?

	▲	ben_w 8 hours ago \| parent [-]
		> So as soon as any one uuid is leaked, it becomes plausible for any child to bypass the gates until the uuid is manually revoked? How? Phones are already locked down pretty hard. Anything like this would need to be in something secured at the OS level just to stop signatures getting leaked between apps. If you're thinking "kid roots device, replaces OS entirely", that's not the problem of the manufacturer of the OS that just got deleted.

▲

mrguyorama 6 hours ago | parent | prev [-]

Google and Verizon sell that data to the government right now. The US's official position is that buying info they aren't legally allowed to collect is perfectly fine, as if you were given a privacy right in the constitution only to enable an info broker economy, and not because of the obvious and understood harms of the government having whatever info about you they want.

If you want the US government to not know something about you, unfortunately there's a lot of changes that need to happen, including entirely new political parties and making changes to the Supreme Court, and popular support for taking the privacy rights you already have much more seriously.