[Nursie]

> Age verification obliviates anonymity on the internet.

How so?

Please explain in detail, because there are already schemes such as "verifiable credentials" which allow people to prove they are of age without handing over ID to online services.

[shakna]

Last time my government tried that, they failed. [0]

You need to 100% trust those verification services. And considering their success rate [1], you shouldn't.

[0] https://thinkingcybersecurity.com/DigitalID/

[1] https://discord.com/press-releases/update-on-security-incide...

[Nursie]

> You need to 100% trust those verification services.

First link - mitigation: use a well supported standard like OIDC, not a home-cooked scheme. Duh.

Second link - this is part of the problem such schemes as verifiable credentials are designed to address, random third parties collecting ID they don't need.

Yes, any system needs to be executed well. Neither of these really display that.

[shakna]

If _the government_ can't be trusted not to use a dumbass scheme, then no, it isn't a duh moment. You don't exactly get to dictate how the government implements it!

The point is that systems today, aren't really well executed. So it is unreasonable to expect them to be well executed.

If you can't trust people not to build the bomb well - then don't let them build a bomb.

[Nursie]

> You don't exactly get to dictate how the government implements it!

Who was talking about the government implementing it? I wasn't.

And also "This has been done poorly in the past so we should never attempt to do it again, better" seems an odd way to go about things. There are well put together schemes by international standards bodies in this area now. Neither of the above links followed them.

[shakna]

If neither follow them, why do you have such faith that anybody would...?

[Nursie]

I mean, your example of the ATO there isn't even an age verification thing, it's a defective clone of OIDC, so by that logic we should ban all SSO or identity delegation solutions?

Because we don't believe anyone will ever use the standards in this area, despite loads of companies and government bodies actually using OIDC already?

I'm not really sure what you're driving at.

[afiori]

because most implementations are not going to be like that.

[Nursie]

In the context of "Age verification should be banned" though, we're already talking about legislative intervention. If there's no particular problem with schemes that are like that then we don't necessarily need a blanket ban on age verification.

Perhaps what we're really saying is "Ban age verification that collects lots of personal information".

Or perhaps we could distil it down further to "Ban unnecessary collection and storage of PII". In which case, Congrats! You've arrived back at the GDPR :)

Which I think is a good thing, and should be strengthened further.

(Also the other response to "because most implementations are not going to be like that" is "why not?". People are already building such ecosystems.)

[AnthonyMouse]

> If there's no particular problem with schemes that are like that then we don't necessarily need a blanket ban on age verification.

There is a problem with schemes like that.

The way computer security works is, attacks always get better, they never get worse. A scheme that nobody has found any privacy holes in when it's enacted will have one found a week after.

The way governments work is, the compromise bill passes if the people who care about privacy support it because then it has the votes of the people who care about privacy and the people who want to ID everyone. But then when the vulnerability is found, the people who care about privacy can't get it fixed because they can't pass a new bill without also having the votes of the people who want to ID everyone, and those people already have what they want. More specifically, many of them then have what they really want, which is to invade everyone's privacy, as they were hoping to do once the vulnerability was found.

Which means you need it to be perfect the first time or it's already ossified and can't be fixed. But the chances of that happening in practice are zero, which means it needs to not happen at all.

[Nursie]

> There is a problem with schemes like that.

/goes on to discuss how government legislation of specific schemes is the issue, not the schemes themselves.

Then we don't legislate specific schemes? The GDPR doesn't do that, for instance, it spells out responsibilities and penalties but doesn't say "Though shalt use this specific algorithm".

Remember, this discussion started with a call to ban all age checks, which itself is a government action and restriction on the agency of private business.

There are ways that private entities can implement age checks both securely and without leaking much other information, so it seems very heavy-handed to ban them. Private entities are building such systems between themselves already, without government mandates on the specifics.

[AnthonyMouse]

> Then we don't legislate specific schemes?

Except that you have to in this case because IDs are issued by the government and then it's the government having to provide some privacy-protecting means of using them, which is the thing they're incapable of in practice.

> There are ways that private entities can implement age checks both securely and without leaking much other information

I have yet to see a single one implemented in real life. People point to attempts and then you look at the implementation and it's full of dubious choices and unforced errors, before you even start looking for bugs.

Moreover, private entities have the perverse incentive to do the opposite of implementing it securely, because they find it profitable to track people, or find it unprofitable to spend the resources necessary to prevent themselves from being infiltrated by foreign governments when their business is the sort which is useful to them as these are.

[Nursie]

> it's the government having to provide some privacy-protecting means of using them

Nope, not necessarily.

> I have yet to see a single one implemented in real life.

There are likely to be a lot more coming as the newer standards in this area were finalised last year. Online identity is a continually evolving space.

> Moreover, private entities have the perverse incentive to do the opposite of implementing it securely, because they find it profitable to track people

Some do in some circumstances, but far from all. Others (often financial institutions) have wised up to PII being a liability rather than an opportunity and some are working on frameworks and capabilites in this space that don't involve any more storage or transfer of anyone's ID than already happens in banks.