The observable (and controversial) effect of this is an outcry by content creators, I suspect the measures they take around accounts identified as children are much less obvious and entirely orthogonal.
I guess if you are speaking of "what is best for society", ok, but that's a completely different thing. It sure seemed like you were debating the post above, but I guess not.
It was my own post and it had nothing to do with legal liability. As I said (clearly, I thought), I think Google is reluctant to show the world just how unreliable their demographic data is.
I think that's a bridge too far. Unfortunately, I don't think anyone would actually want a world where ad-targetting data could be used as a form of ID to prove your age...
If you want the data to be as good as an ID... then everyone would need to offer some kind of government issued ID in order to create an online account, in order to make sure they weren't a child. Anything short of that, and you're going to have accounts that porport to be 21 when the 'real' user is only 12.
Or maybe you think they should use Machine Learning? So you then monitor every youtube video the 21 year old user watches, and determine that they are watching My Little Pony content over and over... so surely they must be a child. And then lock down the account and prevent them from making comments/purchases/etc?
The child protection laws are very strict. I don't think any of us would like a world where you get treated like a child (e.g. no right to view or participate on most of the internet) because of some algorithm. Civil suits against content creators seems like one of the only viable ways forward to me, as stupid as it is.
Theses whole changes are happening because they were tracking childrens (well they were tracking everyone, which included childrens). They can't store any data from a child, none at all. That means that they can't detect if it's a child, and even if they had a magical way to do it, storing the "isChildren" variable set as true would be illegal in itself.
COPPA decided that a video targeting child was a proof that they knew child were watching, and thus tracking them. So instead they move the responsibility to the content creator to say whether the content is for kids or not (thus not storing data on the child, while still being able to not track the user which is now assumed to be a child).
The only way they can store data about a child require parent permission.
Targeting content aimed at children is... I'm not going to pretend it's good, or even long-term viable (malicious videos are still going to slip through). But it passes the extremely low bar of being better than age-verification gates.
If their demographic detection was bad, they would have no revenue and no customers since that's the flagship product that they are selling.
Course the data is all there, but they just really promise they never put it together and knowingly went after kids.
Will this help?
To start with, I'm not a huge fan of the idea that the Internet as whole could be legally obligated to assume I'm a child unless I provide every site I visit with a drivers license, for example.
But let's take your assumption to the next level: let's assume that there's some privacy-respecting way of doing this (I don't believe there is, but we'll temporarily pretend it exists).
I'm also not a huge fan of the idea that the Internet as a whole could be legally obligated to assume I'm a child based on criteria that are outside of my control. I personally think most of our experience with filters online have taught us that they're best implemented close to the consumer, not close to the source.
To be clear, both Youtube and the FTC are turning this into a dumpster fire. But the idea of filtering content that's designed for children is better than the idea of filtering consumers. It's just that this implementation of that idea is garbage and unnecessarily hurts creators.
(What content/functionality is available to what user is a different question, and, I think we both agree, should be left for the provider to decide.)
(And I always find internet-as-whole arguments as a straw man. We are taking about YouTube and Kids. Nothing more.)
> Filtering? It’s a discussion about privacy
The point being, there's no way to determine whether someone is an adult without violating their privacy, and even if there was, the false positives would be (imo) an unacceptable tradeoff. If you want to have a version of COPPA that works without violating privacy, you have to focus on either labeling or filtering content that's uploaded to a service. Trying to filter users based on their age just doesn't work.
Will this help?
(There’s an elephant in the room. All-of-internet wants it straw elephant out (I think). The lawyers just got confused.)
Even in that scenario, content filtering is more feasible and practical than detecting and banning children.
But, pushing that problem aside as well, sure. Assuming that you didn't have false positives, and assuming that you didn't have to violate anyone's privacy, and assuming that the solution wasn't obscenely expensive, that theoretical solution would be very helpful. It would basically solve the entire problem, for pretty much everyone.
But that theoretical solution doesn't exist.
If we're willing to make those assumptions, I would also like a special gun that only shoots bad people and doesn't work in schools or churches. And it would be nice to give law enforcement with warrants access to lawful information without requiring encryption backdoors. Of course, we'll also need to implement the new training regiment that gets rid of any corrupt prosecutors or officers.
I'll happily posit things for the sake of discussion, but there's a point where I'm suspending so much reality that the discussion just breaks down. I can't stop looking at the proposed solution and saying, "yeah but this thing you're talking about doesn't work." I guess I'm just not sure what you're getting at.
Asking for a friend.
Note that this scenario isn't entirely theoretical. It's rare, but there are already subsets of Youtube (trailers for horror movies are where I see this most often) that are locked behind a login screen because the uploader has indicated they're intended for mature audiences. You either log into Youtube (sacrificing some privacy) or you just don't get access to those videos.
This ends up being not a huge problem today, because the amount of content falling into that category is very narrow. Imagine a world where you fail Youtube's theoretical privacy-preserving test, and suddenly you can't watch anything, regardless of the content, that isn't designated as being specifically directed at children.
Of course, on some level that's just Youtube's choice. But now imagine if a government agency like the FTC dictates that every large Youtube-scale platform has to implement this test. Suddenly you find yourself not only unable to watch videos on Youtube, but also unable to make a Facebook/Twitter account. Imagine if you can't search for certain things on Google.
Note that this final scenario also isn't entirely theoretical with the FTC. My reading of the comments that the FTC commissioner has made is that the value of targeting Youtube is specifically that it's a centralized choke-point for this kind of content. I don't see any reason to assume that if creators migrate off of Youtube, that the FTC won't follow them and impose similar rules wherever they end up.
(I have a lot to say about content, including suggesting specific solutions to specific problems, but I’ve decided to not air this in a public forum. Further discussion is pay gated. Sponsor me on github for more !!! :))
Or
Google figures it out the same way they infer any number of things about you and use it to resell advertising. If they can figure out someone is a woman between 19 and 25, I'm pretty sure they can figure out with a good degree of accuracy that a user is under 15.
There is pretty much 0 way that you could say that purchasing a universal communication device is carte blanche for a child to view any content online.
Not always. Minors are allowed to have cellular with parental consent in some cases.
Content is likely the strongest indicator determining if someone is a child or not.