> According to YouTube, about 6 percent of removals from July to December 2023 were abusive, along with 10 times more attempted abusive removals
I uploaded the recording of my father's funeral to youtube, to share as an unlisted video. It attracted two copyright claims - for hymns written in the 18th century.
These were claims - not takedown requests - but I didn't think it was appropriate to share a link to the family that was unable to attend in person that would serve them ads. So I instead made the video private and shared it a different way.
But if I don't contest the claims, how can youtube possibly count them as abusive?
YouTube matches a lot of classical music even if it was played by a different person. I mean, I uploaded a recording of myself playing something and it was flagged. But you can just contest it and most of the time it just goes away.
It's not valid. But some punters try to make such a claim, in the hopes of a few dollars. It's fraud, but youtube/google cannot police it properly, because the bigger media companies _want_ to be able to make claims easier than a true DMCA claim. I do not believe this can be fixed, as they're diametrically opposed.
>But if I don't contest the claims, how can youtube possibly count them as abusive?
I'm not saying Alphabet/Google/Youtube did this, but they do things on a massive scale and they could "look into" a reasonable sample of claims all on their own and generate very good statistics. (always good to remember, statistical significance is independent of population size, only dependent on sample size, and rather small samples have a lot of significance, see birthday paradox)
Scale is part of the problem. It actually hurts small businesses.
There is a probably on Google Maps where a marketer will setup a website, a lot of business listings, and get a Google Business Profile for a "local business" which doesn't actually exist. This is all setup to attract unsuspecting customers to share their information which is then sold to local contractors often without informing the customers and/or the local contractors.
It's a practice known as lead gen and/or rank-and-rent.
For an example, you can look up a "business" called Seattle Grease Trap Services on Google Maps. It's part of a network of fake businesses run by marketer in Hawaii. Check the profiles of the people who leave reviews to find some of the others.
Another example is found by searching for deck builders in Chicago. There are several variations of Koval, TOD, CPC, and similar companies. This is all operated by a local marketing company that I managed to identify in less than an hour.
I report these frauds, nothing happens. I escalate. Nothing happens. So I've given up, but I know that these listings cost local businesses actually work. And even though they're against Google guidelines, nothing happens.
I don't think it's right that statistical significance is independent of population. 1 sample out of a population size 1 is plainly more significant than 1 in 100.
I'm not good enough at this to be explaining it, but statistical significance is based on probabilities.
if humans have a 10% chance of having cancer at any moment, and you have one human on an entire planet, the probability for that planet is still 1/10 and not 100% or 0 based on whether that person has or does not have cancer, and having one sample from that population of 1 does not increase your confidence in the answer as to whether his cancer represents a cancer cluster.
or something like that. If you need a sample of 30 people to achieve a certain degree of confidence, whether that sample comes from a population of 30 people or a population of 100,000 people, same confidence. We aren't trying to establish what percentage of people have cancer at any given moment, we are trying to establish whether this population matches what we know about the percentage who "should" have cancer.
I'm really bad at explaining this.
the more times you flip a coin, the more confidence you'll have in understanding the probabilities of that coin. for a given confidence level, you don't need to flip that coin more by 100x, you already know. There might be a population of a million coin flips, but you just need a sample of a certain size to obtain confidence that your sample is representative.
It gets a little complicated around old pieces like that.
As usual, I am not a lawyer, consult a legal professional before making any decisions based on this information.
Even though an old work like that is in the public domain, if it was played as a recording, whoever owns the rights to the recording still has a claim to copyright.
But even if it was played live and sung by the attendees, that doesn't necessarily mean the sheet music is public domain. Pianos from hundreds of years ago didn't play exactly like a modern piano, so there may have been an intermediary step where the original piece was modified to work on a modern piano when the sheet music went to print. The modifications to the sheet music could grant the person who sold the book a copyright.
Doesn't necessarily mean the person has a valid claim, but it's not impossible.
The piano thing is basically nonsense, but you're in the right general area.
The copyright may have elapsed on the original composition, but a specific arrangement of a public domain composition may be entitled to copyright protection as a derivative work. For example, the song "Scarborough Fair" pre-dates the modern concept of copyright, but Simon & Garfunkel's performance of the song contains novel elements that are subject to copyright.
> Even though an old work like that is in the public domain, if it was played as a recording, whoever owns the rights to the recording still has a claim to copyright.
I am aware of that. In this case it was played from sheet music by a professional organist.
> The modifications to the sheet music could grant the person who sold the book a copyright.
Here's some discussion on the matter that indicates that a transcription can be, but a lot of what I've seen is much like the rest of copyright, and it's a hot mess and very specific to what the circumstances are.
The "might have a claim" about that topic in my original post is pulling a lot of weight for me.
> Pianos from hundreds of years ago didn't play exactly like a modern piano, so there may have been an intermediary step where the original piece was modified to work on a modern piano when the sheet music went to print.
This, i argue, has nothing to do with the amount of work, but everything to do with the simplicity of proving a copy.
A specific arrangement of words, music, or light (in the case of film) are well protected. Painting & sculpture, perhaps less so; fashion, dance, and architecture, sweet fuck all. Which all have as much if not more creative input than an arrangement.
How I think they should be handled? Same as fashion or architecture, only direct copies in the same format are protected, everything else is derivative.
You don't own "the piece", you have a copyright claim to the specific parts you contributed of the specific arrangement/transcription you produced for the duration of copyright. Someone else can take that original work and create their own transcription/arrangement that you would have no claim over. ContentID is unable to reliably distinguish these two.
I don't see the difference. It means that if you transcribe a piece, I can't play the piece you transcribed based on your transcription for well over a hundred years. That's the bullshit.
The fact of copyright is different from the term of copyright. If we changed the term to 1 year tomorrow, there would still be arrangement and transcription copyrights. I also dislike the term, but it's a separate discussion.
I think it would be reasonable for someone to own future performances of a piece based on a transcription they made for a year, or a couple of years. I don't think it's okay for them to own it until 70 years after their death. Given that the term is what it is, it is unreasonable for transcriptions to be covered by copyright. The two things are inseparable.
No; if you create your own transcription, and it just happens to be identical, you own the copyright on your transcription. You can't steal someone else's, but you aren't forbidden from making your own, even if the end result is the same. (Better be able to prove that you made it yourself though.)
Transcription takes time and effort. Time and effort equals a copyright.
This is so confusing. What stops people from copying music verbatim (with little effort) and obtaining copyright? And once someone has done so, if I try to do it too how can I claim I was transcribing the original and not the copy, since it’s identical? And if two identical transcriptions can exist, how to adjudicate which one to credit?
Copyright isn't exclusive, so the original rights holder you derived from would still have claim. You only have copyright on the derived work over the creative input and changes you provided, if any. If two people independently make the same creative changes, they both have copyrights on distinct works that happen to be identical. There's a well-known blog post on this subject called "What colour are your bits" (https://ansuz.sooke.bc.ca/entry/23) that goes into more detail, but the gist is that the copyright is not solely determined by a work's representation in the world, but rather by the process to create it (of which the representation is simply a product). If this needs to be adjucated, it's done by the courts.
This is true, but I believe exceedingly uncommon. There's a requirement for a creative element, and if there's only one way to adapt the music to a modern piano it shouldn't fulfill that requirement and be eligible for copyright at all. I would be unsurprised if copyrights like that existed, though.
If it were at all common, copyright would cease to function. E.g. someone could, in theory, write Romeo and Juliet word-for-word from their own imagination and then apply for a copyright. How on earth would enforcement work? There's no distinction between the copyright-free Shakespeare version and the newly copyrighted version, so do performers get to choose which version they're acting under?
Music has, through various court decisions through the years, gained a particularly arcane and arbitrary set of rules on what parts of a piece are copyrightable, and to what extent. It's a mess, as some parts are overly-protected, like the layout of the music on the page (unlike the typesetting of a book, or the font used!), and some under-protected in comparison, like the rhythm of a song compared to it's melody (which has a particularly arbitrary rule of 7 notes long). Though, to answer the question of the last part, if that version was copyrightable, it depends what book the performers are using! Though again, it could be that the essence of the performance is not considered to be a violation, even if photocopying the book would be, depending on what form any differences take.
If you're concluding copyright is impossibly complicated, vague, and confusing and the only reason the system works at all is because hashing out the details in court is either a nuclear option between two large companies, or enforcement on a large scale against small actors is basically impossible (but will unpredictably come down like a ton of bricks on some unlucky individuals), then you'd be right!
But there'd be a finite number of ways to transcribe a piece. This someone would be able to claim the only ways to faithfully play the original public domain piece.
In that case, the law is not incompatible with you both owning your own transcriptions, even if they are identical. You might need some proof that you created your own if sued, but that's the copyright system.
They don't. ContentID makes the claimant determine if a counterclaim is valid or not. YouTube forwards the messages, diverts revenue into an escrow bucket for the duration, and washes their hands of any decision-making in the matter after the automated notice.
The same applies to books - even if you bought a public domain book like Alice in Wonderland, you could be sued if you make a scan. The typesetting and layout is copyrightable.
Even if the claimant's sheet music is used, how do they know that it is their sheet music specifically, and not someone else's transcription of the same music?
They could include "paper towns" in the music, but that's not what these automated systems are looking for.
And rightfully so, back in the day I watched more than a few episodes through secret delisted playlists. They still exist, presumably hidden so that one of those AI copyright bots ("companies") can't take down every video in one fell swoop.
Back in the day I found them on forums that required accounts. I've long since switched to other mechanisms for pirating content not available in my region, so I wouldn't know where to find them these days.
Presumably, there are Discord servers for that kind of thing, seeing as Discord has mostly replaced the forums of yore.
Not really, considering the origins of ContentID. It's just the same story over again. Google/YouTube didn't really want to build ContentID. It was a concession as a result of lawsuits from Viacom (and others) from the early days of YouTube, where they were about to be sued out of existence for having a platform, which at the time, was filled with blatantly pirated content.
I can confirm from my experience that Google also automatically scans private videos.
My YouTube account was recently mistakenly suspended; luckily, after a few minutes of panic, Google reinstated it.
I discovered the suspension because my kid wanted to see a video on my TV's YouTube app (LG webOS), which showed an error without further explanation. Trying to find out what happened, I saw an email from YT about my account breaking the ToS with a video. "WTF!? I got pwned!" was my first reaction (followed by panic as YT uses my Google account). But, no, the culprit was a screen recording in 2015 that I did at work to showcase a UI prototype about a key-based VPC setup workflow I was working on. The video was private and unlisted, and I didn't remember it as I did it as a quick capture to share an idea with the team. I hypothesize that the video got flagged because the algorithm confused it with a "crack/hack tutorial."
I was happy to see my account reinstated the same day. However, the whole experience didn't leave good taste: a small algorithm mistake can make your life miserable, and you don't have any human or support to correct it. Also, the "private" moniker in YT videos is an illusion.
I have refused to create any content on YouTube for this reason. My Google account is too valuable to my digital identity.
When Google banned a user because their Google Drive was used to share a picture of their child’s skin with the doctor for diagnosis reasons and Google flagged the account as CSAM, I felt vindicated that I don’t create content for YouTube. Not because I believe my content would get banned, but because their automated systems catch irrelevant content and the user experience of those instances is like a DoS on your digital identity.
Probably YouTube scanned it, found something potentially actionable, automatically informed the representatives. The video was unlisted, which is not private, technically totally accessible to anyone with the link.
Only if the 74 quntillion possibilities exist. There's a pretty good percentage of 404s.
Plus, they don't have to do it every day, and clearly don't, since some people get dinged for videos they uploaded months or years earlier.
It's not like these organizations don't have the cash for the system to do it. It's not going to be one guy in the basement of the building overclocking a 386.
You are making zero sense. The point is the ids are random, so yes there are 74 quintillion possibilities. And roughly 74 quintillion 404’s. You need to check all of them. If you want to keep up with them you need to check them often. No one has the cash to check 74 quintillion anything even with a million 386s. Even the 386sx.
You won’t know if you’ll get a 404 until you make a network request. How many of those can you make per second? Hundreds? Thousands? Let’s be very generous and say millions.
That means you just need millions of years to go through the entire range.
Last year I uploaded a recording of the mass for the wedding of my sister's sister in law (the husband had terminal cancer) My homebrew recording was expected days before the official recording.
Youtube decided it was agaisnt community guidelines or something. Religion? Music? Child abuse (the ones spreading flowers)? Who knows. My appeal was successful like 6 months later.
For the time being, upload it to somewhere else than youtube, like pcloud, dailymotion, etc. Youtube is the best techwise, but the legal situation is not stable. Best left for professionals (influencers)
Recordings are copyrightable even if the underlying music is public domain. If you record a performance of Mozart's 40th symphony on your phone, you can be sued for copyright violations by the concert hall. (Don't be surprised that concert halls perform for more than just the attendees.)
To me it seems he didn’t win, he just lived to fight another day. Winning would be if YouTube would treat him as if or better than if this never happened, or if the guy got in trouble, or if it would be harder to do this in the future.
> No one knows how much copyright abuse occurs on YouTube. According to YouTube, about 6 percent of removals from July to December 2023 were abusive, along with 10 times more attempted abusive removals. But if a significant number of users never flag abuse—out of fear they could be sued for contributing to copyright infringement—then the true figure could be higher.
Sounds like the YouTuber won a DMCA fight with YouTube, as well.
If a provider does nothing and they're wrong, they become legally liable for lawsuits. So they default to accepting the request because it is legally safer to do so.
Unfortunately, improving this will probably require changes to the DMCA, and given the general state of politics I would honestly prefer to let sleeping dogs lie on this one.
Dmca is under prejury laws. However content id has no legal protection. You can sue google for false accusation if you want (good luck winning since their terms of service allow this - you would have to also argue the terms of service should not apply)
Some comments are wondering how DCMA claims work and where ads revenue go to. So here is the gist.
The current YouTube DCMA system is broken. Here's how it (mal)functions:
- Creator uploads video, gets ad revenue.
- First valid claimant gets ad revenue.
- Subsequent claimants are ignored, first claimant keeps all revenue.
This leads to a first-come, first-served system where the first claimant wins all, regardless of the actual IP ownership.
Would need to implement proportional revenue allocation and a more transparent with arbitrage claiming process, e.g moderators with subject expertise chiming in.
That's if YouTube wants to fix this. But they are rather worried about a/ being accused of copyright infringement for not taking "some" action and b/ aggravating their profit making distribution agreements with major content owners.
To note there is also an automated id system which applies primarily on audio fingerprints, to a degree visual frames as well.
- Creator uploads video, id system finds matching fingerprint against a large DB of copyrighted content and send ads revenue to whoever owns the match.
Also to note, that applies to largely most audio content at this point. Not much to big films (studios are still onto trying to keep distribution via their own streaming platform, to be platform or preferred partner platform).
We are in this mess, not because YouTube is dragging its feet, but because the IP laws simply aren't reconcilable with the way we create and consume content these days.
A brick needed to build a more fair revenue distribution system could be to revise legislations, e.g banning distribution exclusivity. That would give incentive for innovation, a startup could build an engine that redistribute somewhat fairly.
Ultimately, creators own their arts and can pretty much dictate how their works can be used, and call out their licensing fees. But when intermediaries step in and figure out exclusivity would milk more cash from consumers, it goes all the way to effectively lobby for the laws to remain crooked to their advantage.
> Subsequent claimants are ignored, first claimant keeps all revenue.
Are you sure? I remember Jim Sterling used to show content from multiple companies because multiple claims locked them all out from monetisation. Maybe it's changed since, though, that was years ago.
If the first-come first-serve is true, what prevents a creator from self DMCA-ing all his videos so as to get all the revenue? Possibly by embedding a copyrighted sound that belongs to a company they create just for that purpose.
Perjury charges. They're almost never pursued because of the difficulty of proving what someone knew or didn't know, but DMCA'ing yourself is perjury per se because you cannot infringe your own copyright. The RIAA/MPAA would absolutely lobby to make pursuing those perjury charges a priority.
I DMCAed GitHub and Stack Overflow for not letting me delete my own content (both of them ignored it and then I sent GDPR notices to both; GitHub ignored that one a d SO told me to fuck off. Next step is a lawsuit, but that's expensive). The perjury declaration you have to make is that you believe everything in the notice is accurate and you believe you have the right to issue this notice. Actually, the less well you know the law, the safer you are from perjury.
Plus, they'd have to actively sue you for perjury which costs them a lot of money for zero gain, unless you're spamming bad requests or something.
How does detecting the emails were fake address the DMCA claim on his YouTube channel? It’s my understanding that YouTube won’t intervene in DMCA claims even if the claims were false.
the most important thing here seems to be that the fraudulent claims were retracted by the person who sent them, and the statement from nintendo that the claims were not legitimate.
but it's not true that youtube won't do anything about false claims. both the DMCA and youtube's pseudo-dmca takedown process care that about the identity of the person making the claim. if you claim "i am an authorized agent for Nintendo of America" and that's not true, then they'll throw out your claim. as long as you are who you say you are and are an authorized representative of the company you claim to represent, then google tends to trust you even if you're claiming ownership of content you don't actually own.
When a claim is filed through content-ID, the claiming company basically gets two options: disable all ads (taking out any revenue for the uploader), or run ads and get paid out. They also get to pick how many ads to run, which means a three second segment can insert minutes of ads into a video. They can also mute audio during infringing sections, or block videos in certain countries.
Stephanie Sterling famously came up with a trick to make their videos run without ads, even when including fair use content: by intentionally including content from multiple companies, each with different preferred monetisation preferences, introducing a conflict in the Youtube content-ID system that didn't get resolved automatically. Their videos were intentionally run without ads, being supported by Patreon and all, and companies kept claiming sections for themselves and inserting ads into videos.
Bizarre that to shield oneself from content ID bullshit (like including a three second snippet of a Ninendo trailer blocking an entire video), one needs to intentionally infringe on more copyright, but that was the Youtube system of yesteryear.
I don't know if the system still works like that, but based on the seemingly random snippets included in Sterling's videos I'm guessing they're still applying this trick.
In theory one could take this further, uploading parts of one's own video to content-ID so the system hands over some partial control over the content, but I feel like Youtube wouldn't take too kindly to that method.
What on earth could be the motivation for this not-lawyer? A malicious thrill?
"The goal is to take the monetization of videos. You get enough 200k+ videos under your belt for a form letter and bot writing in requests. For little work you can get in the 100K per year income, providing you don't get shutdown."
Hopefully this type of blatant abuse of the law will motivate some changes. If not, expect whole offices to pop up in developing countries with fake lawyers pulling off this scam en-masse.
I uploaded the recording of my father's funeral to youtube, to share as an unlisted video. It attracted two copyright claims - for hymns written in the 18th century.
These were claims - not takedown requests - but I didn't think it was appropriate to share a link to the family that was unable to attend in person that would serve them ads. So I instead made the video private and shared it a different way.
But if I don't contest the claims, how can youtube possibly count them as abusive?
reply