- cross-posted to:
- technology
- cross-posted to:
- technology
You must log in or register to comment.
From the article…
But while many think that YouTube’s system isn’t great, Trendacosta also said that she “can’t think of a way to build the match technology” to improve it, because “machines cannot tell context.” Perhaps if YouTube’s matching technology triggered a human review each time, “that might be tenable,” but “they would have to hire so many more people to do it.”
That’s what it comes down to, right there.
Google needs to spend money on people, and not just rely on the AI automation, because it’s obviously getting things wrong, its not judging context correctly.
US Corporations: But we can’t start paying people to do work! That would completely wreck our business model!
Workers: So you would actually be bankrupt? Your corporation is that much of an empty shell?
US Corporations: Well, we really just don’t want to have to spend less time golfing, and having to pay people might eventually cut into golf funds and time.
YouTube is already a giant cost sink lmfao. It’s basically the one decent thing they’re keeping up still which is why they’ve been monetizing it as much as possible lately.
And I just canceled my YouTube premium family in favor of SmartTube and Spotify.
Somehow I’m yet to encounter a single ad in Spotify Free and I have no idea how or why.
But the downside is that I want to subscribe to CuriosityStream/Nebula and I can’t find a referer link for the channels I like because they are all being skipped.
Google is absolutely allergic to hiring humans for manual review. They view it as an existential issue because they have billions of users which means they’d need to hire millions of people to do the review work.
This isn’t unique to google but if the system continues to be designed to allow companies to mask the true cost of doing business we will never move ahead past it.
We undervalue ourselves repeatedly at the sake of cheap products.
I’m not sure what you mean by “true cost of business.” The biggest cost here is the issue of copyright claims and takedowns which were created by law in the first place, not by a natural phenomenon.
No matter what system we design, you’ll find that people adapt to take advantage of it. Well-meaning laws frequently have large and nasty unintended consequences. One of the biggest examples I can think of is the copyright system — originally intended to reward artists — which has led to big publishers monopolizing our culture.
That seems a bit excessive, say all 8 billion people were using Google products, 8 million reviews would be 1 per thousand users which seems like many more than are needed since almost all users of Google are passive and don’t create content.
There are an estimated 720,000 hours of video uploaded to YouTube per day. At 8 hours per day it would take 90,000 people just to watch all those videos, working 7 days per week with no breaks and no time spent doing anything else apart from watching.
Now take into account that YouTube users watch over a billion hours of video per day and consider that even one controversial video might get millions of different reports. Who is going to read through all of those and verify whether the video actually depicts what is being claimed?
A Hollywood studio, on the other hand, produces maybe a few hundred to a few thousand hours of video per year (unless they’re Disney or some other major TV producer). They can afford to have a legal team of literally dozens of lawyers and technology consultants who just spend all their time scanning YouTube for videos to take down and issuing thousands to millions of copyright notices. Now YouTube has made it easy for them by giving them a tool to take down videos directly without any review. How long do you think it would take for YouTube employees to manually review all those cases?
And then what happens when the Hollywood studio disagrees with YouTube’s review decision and decides to file a lawsuit instead? This whole takedown process began after Viacom filed a $1 billion lawsuit against YouTube!
But they don’t have to review every video, just the ones that are flagged by the AI then contested, which is probably a fraction of a percentage of all of them.
Just go to a public library, get on a computer and search for transparent undergarments. Or better yet, “the black tape project”.
This will ensure the computer is going to be tainted forever with soft YouTube porn for everyone to enjoy.
They could also punish false claims. Currently the copyright holders (and not even that, just something that might vaguely sound like your stuff) can automatically send out strikes for any match in the system. The burden to prove it’s fair use goes to YouTube channel, and if it’s found to not be copyright infringement nothing happens to the fraudulent claimer.
A big step would be to discourage the copyright holders from shooting from the hip.
Just because a claim doesn’t stand in court, doesn’t make it fraudulent. Actual fraudulent claims have landed people in prison.
ETA: Once again, I have no idea why I am being down-voted. The copyright fanatics here are really something else.
You are missing the point.
Well, what is the point?
The burden of proof is on the wrong party and convictions for fraudulent dmca claims are extremely rare.
Most false claims are not fraudulent. The burden of proof is where US law puts it.
Thanks for explaining how people see this.
You’re talking about the court system. They are talking about Content ID. YouTube makes it easy to submit faulty copyright claims with little repercussions if they fail, so there are more fraudulent claims than you’d see in the actual court system. They want YouTube to penalize the abuse of their system more strongly so people that upload videos don’t have to deal with so much shit.
I understand the insanity. They want a private company to prosecute “fraud”. Yikes. Less Ayn Rand and more civics lessons, please.
The ask that YouTube manage their system better. Currently, they assume that a copyright claim is valid unless proven otherwise, and it is difficult for content creators to actually get them to review a claim to determine if it is invalid. So, a lot of legitimate users that post videos without actually violating anybody’s copyright end up being permanently punished for somebody illegitimate claim. What we want is for YouTube to, one, make it more difficult or consequential to file a bad claim, and two, make it easier to dispute a bad claim.
However, that’s not going to happen because the YouTube itself is legally responsible for copyrighted material that is posted to their platform. Because of that, they are incentivised to assume a claim is valid lest they end up in court for violating somebody’s legitimate copyright. Meaning that the current system entails a private company adjudicating legal questions where they are not an impartial actor in the dispute.
So your concern is legitimate, but it’s ignoring the fact that we already are in a situation where a private company is prosecuting fraud. People want it to change so that it is more in favor of the content creators (or at least, in the spirit of innocent until proven guilty), but it would ultimately be better if they were not involved in it whatsoever. However, major copyright holders pushed for laws that put the onus on YouTube because it makes it easier for them, and it’s unlikely for those laws to change anytime soon. That’s what I’d say we should be pushing for, but it’s also fair to say that the Content ID system is flawed and allows too much fraud to go unpunished.
Thanks for the explanation. You are certainly more polite and productive than most people here.
The DMCA gives explicit rules on takedowns in section c here. Complying with DMCA notices is not adjudicating the law, nor prosecuting anyone. It is simply taking the necessary steps to avoid liability. If youtube were to prosecute fraudulent DMCA notices, then it would be engaging in (probably) criminal vigilantism.
Courts have ruled that merely reacting to DMCA notices is not sufficient to avoid liability. Youtube was taken to court over this, and Content ID is the result. (EU law is considerably harsher and positively demands something like it,)
It was a predicted consequence of these laws that they would favor major rights-holders. Mind that the same people here, who want youtube to adjudicate the law, also are against fair use. They would have cheered these lawsuits against youtube/Big Tech, just as cheer now cheer lawsuits against fair use. They want more capitalism. Maybe they delude themselves into thinking that more of the same will have a different outcome.
I have no idea why I am being down-voted.
Just FYI. I downvoted you for the whiny edit.
Removed by mod
Really mod? Saying “The fact that you think a little copyright notice is going to do shit to prevent AI companies from utilizing comments you leave on the internet is laughable” got my comment removed? Lol. Honestly, just ban me, if that’s how you moderate here, I don’t really care to participate.
Typically no, that wouldn’t get it removed.
Only removed because the user is getting harassed elsewhere for the signature, and the comment was border-line rule 3.
Ah, I see. I understand how it could be border-line rule 3. I wasn’t intending to harass or personally attack, but that makes sense. I appreciate the response and rescind my prior douchebaggery lol.
Thanks for being understanding!
Linkspam
From the article…
But while many think that YouTube’s system isn’t great, Trendacosta also said that she “can’t think of a way to build the match technology” to improve it, because “machines cannot tell context.” Perhaps if YouTube’s matching technology triggered a human review each time, “that might be tenable,” but “they would have to hire so many more people to do it.”
That’s what it comes down to, right there.
Google needs to spend money on people, and not just rely on the AI automation, because it’s obviously getting things wrong, its not judging context correctly.
I hereby grant approval for anybody to change, alter, and or use my comment for AI and commercial means.
I hereby grant approval for anybody to change, alter, and or use my comment for AI and commercial means.
I’m guessing this is what gets you down-voted. The “information wants to be owned” brigades are out in full force today.
Oh yes, let’s make a private company adjudicate the law. That’ll teach em.
That’s already what they’re doing essentially. This person is just advocating for an actual human to review these rather than some black-box algorithm.
Not really. They have to do something, or they become liable. If youtube decides that something is fair use, and a court disagrees, then they are on the hook for damages. They’d have to pay a lot of money to copyright lawyers, only for the chance of having to pay damages.
And, you know…, The same libertarians, who are now attacking youtube for not going full feudal, would be absolutely outraged if they did fight for fair use. It’s stealing property, as far as they are concerned.
I’m not even sure what you’re arguing for since you seemed to have done a complete 180 on your stance. You earlier said you don’t want YouTube adjudicating the law (by choosing sides in a copyright claim), but now you’re arguing that they have to do this in order to avoid liability.
The issue here is copyright trolls claiming copyright over things that don’t belong to them. In many cases, YouTube sides with these trolls and steals revenue from the actual content creators simply by virtue of them having made a claim in the first place, which seems to lend a lot of legitimacy to the trolls even if it’s complete fraud (similar to police testimony in court being treated like gospel). Currently, these cases are reviewed by bots, and people here are asking for them to be reviewed by actual people with real brains instead because the system is completely broken as there are no consequences for these trolls making false claims.
I’m not even sure what you’re arguing for since you seemed to have done a complete 180 on your stance. You earlier said you don’t want YouTube adjudicating the law (by choosing sides in a copyright claim), but now you’re arguing that they have to do this in order to avoid liability. I see the problem.
EG Young people may not buy alcohol. When a cashier asks for ID, they are not adjudicating the law but following it. Right?
When you personally copy something, you must follow the law. EG When you re-upload some image for use on Lemmy, you must “judge” if you can legally do so. Maybe it’s fair use, but that’s not as straight as age. When you make the call, that does not mean that you adjudicate the law.
Under US law, someone can send a DMCA notice to the server. If the server owner ignores the take-down request, then they become liable to pay damages for the copyright infringement. Maybe the owner decided that it was a case of fair use, but that does not mean they adjudicate the law.
I hope that helped.
The issue here is copyright trolls claiming copyright over things that don’t belong to them.
That is criminal fraud. A copyright troll usually means someone on the legal side.
Currently, these cases are reviewed by bots,
That is wrong. But thank you for helping me understand the problems of the people here.
God I fucking hate YouTube.
I copy&paste the one unavoidable youtube video per week to mpv, so i’m not bothered with whatever they do (as long as yt-dlp works).
For playlist followers, there’s a bunch of cli/gui tools.
Okay.
Don’t use it then.
You’re not using it, right?
I’m allowed to use it and be critical of it at the same time. But I use it way less these days because adblocker or not, it’s become a user-hostile and censored place and every video is following the same formula in order to get seen the most and the whole thing feels gross.
Ya I’m so sick of hearing “can’t mention that, YouTube will demonetize me” on videos (or what amounts to that with censored language, topics, blurred guns now, etc.) Just makes it very clear we’re living in a corporate echo chamber where everything must align with what advertisers want. How about what the viewers want??? Fuck the advertisers
How about what the viewers want
As long as the viewers refuse to pay for content, they get what the customers (the advertisers) want.
YouTube Premium actually pays out to “demonetized” channels. What people call “demonetized” is actually called “limited ads”.
Pardon me Sir/Madam,
This is Lemmy.
All everyone does is shit on YouTube because they can no longer easily grab content without paying for it and they expect YouTube to constantly serve them content forever for free because “I’m SuRe GoOgLe CaN aFfOrD iT”
Even beeping and blurring is punished by the algorithm. A YouTuber called RTGame decided to start censoring swearwords and content by using the word YouTube and their logo. Either YouTube has to start flagging their own name or it won’t detect it. Nothing has been flagged so far.
RTGame always comes up with some of the best workarounds. Assuming he was the one to come up with it
This is the best summary I could come up with:
Albino, who is also a popular Twitch streamer, complained that his YouTube video playing through Fallout was demonetized because a Samsung washing machine randomly chimed to signal a laundry cycle had finished while he was streaming.
To Albino it was obvious that Audego didn’t have any rights to the jingle, which Dexerto reported actually comes from the song “Die Forelle” (“The Trout”) from Austrian composer Franz Schubert.
Albino suggested that YouTube had potentially allowed Audego to make invalid copyright claims for years without detecting the seemingly obvious abuse.
"Ah okay, yes, I’m sure they did this in good faith and will make the correct call, though it would be a shame if they simply clicked ‘reject dispute,’ took all the ad revenue money and forced me to risk having my channel terminated to appeal it!!
YouTube also acknowledged in 2021 that “just one invalid reference file in Content ID can impact thousands of videos and users, stripping them of monetization or blocking them altogether.”
“That rings hollow,” EFF reported in 2021, noting that “huge conglomerates have consistently pushed for more and more restrictions on the use of copyrighted material, at the expense of fair use and, as a result, free expression.”
The original article contains 981 words, the summary contains 200 words. Saved 80%. I’m a bot and I’m open source!
Machines can’t tell context
They could back when everyone was using pre-AI context engines that were actually capable of it. Autocorrect is in the same boat. It used to change things correctly to match the context, and now a days it will change words to other words that entirely don’t work within the rest of the context.
Though I am doubtful whatever detects music and sounds in the video literally ever had any kind of context seeking in the first place.
tldr: “Soon after, YouTube confirmed on X that Audego’s copyright claim was indeed invalid. The social platform ultimately released the claim and told Albino to expect the changes to be reflected on his channel within two business days.”
After the first wash cycle I turned the tune off. I mean, the guys putting the damn thing in place, your machine’s legend, as well as the manual all tell you how.
There’s kind of a scary face in the middle of the washing machine in the thumbnail.
Totally missed that lol
