4 ms·

Yes, but at the same time it’s similar to the issues of the Texan abortion changes and “3rd party enforcement” as a side step around the problem. Google aren’t

Yes, but at the same time it’s similar to the issues of the Texan abortion changes and “3rd party enforcement” as a side step around the problem. Google aren’t allowed to discriminate, but are they required to build their systems to prevent the discrimination of 3rd parties who come along, flag videos and by proxy train the algorithms that automatically flag other similar content? Is it legal to compel them to build their systems this way? Do user moderation/flagging count as “content” for the purposes of safe harbour legislation and thus protect Google from consequences regarding users “uploading” it while simultaneously not displaying it to anyone but the user that flagged it also preventing any possible takedown claim or other legal action against inappropriate/malicious flagging/moderation data? This is Google, they have “fight Larry Ellison in court for a decade” levels of Legal team money. Taking the issue to court would basically be a decade long fight that would need help from someone like the ACLU that knows ahead of time the case will drag on to the Supreme Court and knows what that really means and how to handle that.

BugsJustFindMe 5y ago
I think you make interesting points that all became moot the second that a human dispute reviewer on behalf of youtube decided to uphold the penalty. That's an official action by agents of the company, not unaffiliated 3rd parties.
- techdragon 5y ago
  Oh absolutely, it rapidly falls apart when people employed by Google make choices about any of this. But Google is famously insistent on how humans don’t scale and for trying to automate everything with all their ai and algorithms. It feels like a real possibility that this entire process was only “supervised” by humans without any of them having an “actionable level of culpability” or whatever wording of choice the Google lawyers use to say “look people are involved but not involved enough you can sue us over this”. The level of stupidity we see with the ContentID system and it’s insane copyright claims on things like silence, white noises static, birds chirping, cats meowing, etc… it’s very very easy to imagine that an entire pathway of their flagging and moderation system exists to throw away false reports on obviously ok content or automatically accept flagging on content they have trained machine learning models on to cross check borderline cases, or even automatically flag and demonetise content matching previously flagged and moderated content even without a human ever seeing it. The process may be “match past flagged content -> auto flag -> user requests review -> AI checks it against a model trained on other data -> still rejected -> user re-requests -> Google algorithms check some other thing before deciding if it’s even worth showing to a real person -> final rejection without even showing it to a person. All this kind of stuff is private info Google doesn’t share and doesn’t want to share about internal processes and would have to be compelled legal disclosure involving an entire round of “sub-lawsuits”, blocking meaningful progress of the main lawsuit before it can even be established just how culpable Google is either by a specific human employee deciding, or some complex web of obligation on them being required to design to fulfill legal obligations such as not discriminating against protected classes.

BugsJustFindMe 5y ago

I think you make interesting points that all became moot the second that a human dispute reviewer on behalf of youtube decided to uphold the penalty. That's an official action by agents of the company, not unaffiliated 3rd parties.

techdragon 5y ago
Oh absolutely, it rapidly falls apart when people employed by Google make choices about any of this. But Google is famously insistent on how humans don’t scale and for trying to automate everything with all their ai and algorithms. It feels like a real possibility that this entire process was only “supervised” by humans without any of them having an “actionable level of culpability” or whatever wording of choice the Google lawyers use to say “look people are involved but not involved enough you can sue us over this”. The level of stupidity we see with the ContentID system and it’s insane copyright claims on things like silence, white noises static, birds chirping, cats meowing, etc… it’s very very easy to imagine that an entire pathway of their flagging and moderation system exists to throw away false reports on obviously ok content or automatically accept flagging on content they have trained machine learning models on to cross check borderline cases, or even automatically flag and demonetise content matching previously flagged and moderated content even without a human ever seeing it. The process may be “match past flagged content -> auto flag -> user requests review -> AI checks it against a model trained on other data -> still rejected -> user re-requests -> Google algorithms check some other thing before deciding if it’s even worth showing to a real person -> final rejection without even showing it to a person. All this kind of stuff is private info Google doesn’t share and doesn’t want to share about internal processes and would have to be compelled legal disclosure involving an entire round of “sub-lawsuits”, blocking meaningful progress of the main lawsuit before it can even be established just how culpable Google is either by a specific human employee deciding, or some complex web of obligation on them being required to design to fulfill legal obligations such as not discriminating against protected classes.