Hacker News new | past | comments | ask | show | jobs | submit login
[dupe] Ten Hours of Static Gets Five Copyright Notices (eff.org)
336 points by DiabloD3 on Mar 9, 2018 | hide | past | web | favorite | 71 comments




Since you have to attest in a legal document that you are the copyright holder in order to file a DMCA, then you are guilty of a crime (would it be perjury?) if you provide such an attestation under false pretenses.

If we don't punish these false claims when they happen, then they will continue.

An alternative approach based on civil disobedience would be in the form of a large group of individuals filing claims against large swaths of content (basically using the same broken rules of DMCA against the people who advocate them).


There's a bit of conflicting verbiage around this, but the 'copyright notices' that are mentioned in the linked article aren't actually DMCA takedown requests; they're Content ID matches.

The Content ID system is a private arrangement between Youtube and various rightsholders, and it allows them to assert control without the legal repercussions involved in DMCA claims.

The EFF has another article[0] which goes into more detail on the two systems, but I found that to be a bit confusing as well - it seems to conflate the two systems at times.

[0]: https://www.eff.org/issues/intellectual-property/guide-to-yo...


Somewhat tangential, I think it was also Cory Doctorow or Larry Lessig arguing that DRM is bad because it goes beyond what the law requires (especially in relation to blocking all fair use).

This is similar in that Google goes beyond what the law requires to appease certain content providers. People should start boycotting YouTube in the same way they have started boycotting Facebook. We won't see a major shift overnight, but eventually enough of us will do it and Google, like most other companies, will only notice when it's too late to do anything about it.

Vote with your feet/clicks and plead with indie creators to switch to a different platform (such as D.tube, for instance, but any would work).


Youtube was in perpetual high-cost lawsuits until they implemented Content ID to media conglomerates' satisfaction.

The problem is the law.


At least in the US, they tended to win those lawsuits, thanks to the fact that online content hosts receive a liability safe harbor if they comply with the DMCA provisions around takedowns and terminating repeat infringers.

This was one of the good details of the DMCA, although other countries implementing the WIPO Internet Treaty may have done a less hoster-friendly job.

Unless the problem was lawsuits outside the US or they wanted to make the US lawsuit nuisance go away (both entirely possible!), I imagine they started playing nice with the media conglomerates so as to be able to offer Google Play Music, Movies, and Books, for all of which they need licenses.

Note for anyone who looks at my profile and sees Google among my former employers: I never worked for YouTube, Google Play, or Google legal and had no involvement with these decisions. I'm not speaking based on inside information or otherwise speaking for Google here.


Yeah, they won, but it still cost many millions ... I guess I mean the flaw is with the judicial system.

But you have a great point, the media conglomerates could with-hold cooperation which Google needed for Google Play, so Google needed to make them happy anyway ...


And this enabling of nonsense is why I refuse to use YouTube to host anything.

Those provisions in the DMCA were about the only good thing in it, providing a way to sidestep the consequences of this kind of action has been immeasurably toxic and the fault lies squarely with YouTube and Google


I recently read about a decentralized alternative to YouTube called DTube.

https://d.tube/

https://www.polygon.com/platform/amp/2018/3/7/17087668/steem...

It’s built on STEEM and IPFS.

> DTube is an application fully written in javascript, that runs in the browser, that allows you to upload and watch videos on top of the IPFS Network. Moreover, it uses STEEM as a database and enables earning [...]

https://steemkr.com/video/@heimindanger/introducing-dtube-a-...


Honestly, that sounds a lot like the slow and tremendously unreliable P2P services that used to be popular before BitTorrent came out. Gnutella for web 2.0.


I thought so too originally but if you go to the site it’s actually acceptably fast IMO.


There is also lbry, a decentralized media hosting platform based on cryptocurrency.

Although I'm still not convinced that the crypto is necessary here...


Depends what "cryptocurrency" means. If it's proof-of-work, it's only meaningful if there's an actual double-spend problem in the problem domain, otherwise it's just waste and buzzword. If it doesn't solve proof-of-work, it could make sense—there are lots of production systems that use technologies also used in cryptocurrencies, like Merkle trees for Certificate Transparency and for Chrome OS image signing—but "cryptocurrency" is misleading.


A lot of places use blockchain as some kind of "all you have is a hammer" solution to hashing.

Which... I mean in some (rare) cases is a good thing but is usually - as far as I can see - just a way to get VC money or buzz by saying "we use crypto".


Right, and it's also usually significantly more inefficient than just using normal cryptography, so it seems like it's a long-term negative.

For lbry itself it looks like they're using coins for actual payments to creators, which seems like it's a worthwhile project (and definitely has the double-spend problem), but it's not clear to me that it needs to be connected with the content distribution problem.


A blockchain is necessary because there is no other method of having a shared, distributed database of digital content that is not owned or controlled by a single party.


It’s rarely necessary


The only thing you have to attest under penalty of perjury in a DMCA claim is that you (the person filing the notice) are authorized to act on behalf of the copyright owner you are claiming has had their copyright violated.

The attestation that the use is unauthorized is merely a "good faith belief", which is a standard of evidence so low as to be meaningless. I don't believe there has ever been a successful case going after someone for filing frivolous DMCA notices.

(As others have said, YouTube has their own inhouse process that does not actually use the DMCA in most cases, but I wanted to elucidate this point, because lots of people seem to think the DMCA takedowns have some repercussions for fraud. In my opinion, they don't.)


That's why the article says "Copyright Notices" and not DMCA takedown requests. Youtube has a not-legally-binding made-it-up-ourselves copyright notice system where you're allowed to file as many false claims as you want without breaking any laws. No need for an actual DMCA request.


Has this changed at some point? Many years ago I had a takedown request on one of my videos. It was me showing a musical composition I made in some sequencing software and the claim appeared to be from the published of that software. I don't know why that company would have done that but I wondered if just anyone could have filed it claiming to be that company.

They told me that to contest I had to provide loads of legal documentation and essentially escalate it to a serious claim. It wasn't worth my time or worth risking actual legal trouble. I decided to just never upload a video to Youtube again and felt like it was a case of guilty until proven innocent. I'm sure it mentioned the DMCA in the request, though.


They still respond to real takedown requests (it'd be illegal not to), so you may have gotten an actual DMCA takedown.

But the term "copyright notice" as I understand it applies to less serious flagging, mostly done automatically by some kind of content fingerprinting system.


this is a really helpful distinction; thank you for pointing it out.


These are not DMCA complaints.

They are a Youtube copyright notices. The difference is that a DMCA will tell Youtube to remove content from Youtube, while a copyright notice is Youtubes way of telling you that they are giving partial control of your content to someone else, because Youtube believes that someone else have the rights.


Only partial control if you don't file a counter. I've had just around ten or so of these notices for videos I've uploaded, all related to iTunes bits I've used or in one case music playing in the background at a sporting event. In every case I easily "won" the counter and was never bothered with a notice again.


For the sake of argument, assume that we want to protect copyright on sites like YouTube. What is a good, fair solution here?

Given the scale, I think that automated solutions are required to have any reasonable enforcement. However automated (and non-automated) solutions will have false positives. Right now it seems like there is no cost to a false positive (claiming that a video is copyrighted when it is not), and so systems are happy to be aggressive with these notices. Likewise, the cost of a false negative (missing a copyrighted video) is high to the rights holder, as they are losing revenue.

Since these systems use machine learning, it seems that it’s easy to add noise and fool them. I’ve seen videos that are flipped, slightly skewed, and have audio sped up or slowed down, presumably to get around the system. So I would assume that the system would have to be a bit aggressive to catch added noise.

I think it’s ultimately an economics problem though. Since there’s basically no cost to the person making the claim when there’s a false positive, there’s less incentive to avoid false positives. However, it seems hard to charge fees here - if you’re charging people to take down content that they hold the rights to, that’s basically extortion.

Any thoughts on what could be done to improve it? I think long term, we have to evolve our understanding of copyright and licensing, but given the current regime, is there even a solution?


Why should it be YouTube’s responsibility to enforce copyright? Shouldn’t the beneficiaries of copyright be the ones to pay for its enforcement? If that were the case we might see many fewer claims, simply because it’s not worth the time/effort.

This doesn’t directly address false positives, or the problem of fair use (which seems to be a real issue on YouTube), but it could be a start.


Are YouTube beneficiaries of copyright? I think they'd make more money if users were allowed to copy and remix video.

Edit: And Viacom sued them for a billion dollars because YouTube executives knew most of their money was coming from straight-up movie rips anyway. Viacom only lost the lawsuit because the memos didn't name specific videos.


Without YouTube's system, you'd have automatic takedowns based on nothing but keywords in the title: https://arstechnica.com/tech-policy/2011/11/warner-admits-it...

There's no cost when you have bots sending emails. You'd have to fix the law first.


But as mentioned elsewhere in this discussion there are legal repercussions for filing an incorrect DMCA takedown request. The problem is Youtube have their own version of the law and just automatically believe the claimant.


there are legal repercussions for filing an incorrect DMCA takedown request

Can you find a single entity that has ever been hit with these "legal repercussions"?


The responses to you will mostly be people who haven't read the details of the Diebold case and assume that it set a precedent for false claims of infringement; it didn't, and it's still basically impossible to get someone penalized for that.

The issue, as other commenters noted, is that the only claim made under penalty of perjury is that you are the copyright holder, or are authorized to act on behalf of the copyright holder, of a copyrighted work.

The claim that someone has infringed your rights (or the copyright holder's rights) is not made under penalty of perjury. The only way to get someone penalized for it is, essentially, to get them to admit in court "Yeah, I knew it wasn't infringing, but I sent a notice anyway out of malice". Which is, well, pretty much what happened in the Diebold case. Other issuers of mass automated takedown notices are dumb, but not quite that dumb.


> For the sake of argument, assume that we want to protect copyright on sites like YouTube. What is a good, fair solution here?

Well, it may not be a “good, fair” solution, but the process provided a safe harbor by the DMCA is a better, fairer process than the actual process used by YouTube.

The problem is that, as it stands, YouTube likely has no liability to users for taking content down without any good justification, and do has no incentive to follow the part of the safe harbor process that provides protection against liability to the accused user.


The problem with the current system is that it‘s the opposite of due process. Instead of letting some copyright violations slip as the cost of freedom of expression, the current system restricts freedom as the cost for minimizing copyright violations.


The solution is to review claims and punish abusers on both sides of copyright. There is essentially no review process. All claims are automated, which is pretty bogus, and false claims are not punished.


There doesn’t need to be a charge to remove legit content violations but if the removal is contested and found to be false the original claimant should pay up. I’d like the penalty to be steep enough that if you’re using machine learning to flag content a human should still be incentivized to take a look before the claim is submitted.


This certainly does not deal with gray areas, but quite well copes with blatant cases: monetize false claims on both sides. The protocol:

  * Copyright holder fills a copyright claim, video is not taken down or demonetized, ad revenue is held in escrow
  *  * If content uploader agrees with claim - video is removed/demonetized/ad revenue directed to copyright holder
  *  * Content uploader fills a dispute *claiming fully original content*, under potential monetary penalty of false claims
  *  *  * Copyright holder does not dispute the dispute, ad revenue is released to video uploader
  *  *  * Copyright holder fills counter-dispute, under potential monetary penalty of false claims
  *  *  *  * Content is reviewed by human, decision awarded, false claim party penalized, human reviewer expenses paid from penalty
In cases where the content is clearly not what copyright holder claims to be or content is clearly original this system disincentivizes automated, "what we have to lose" counter-claims. Of course, in gray area cases like "does this still fall under fair use" cases this is probably even worse than current setup, therefore I stress fully original content in the protocol, which should be different for fair use disputes.


One entirely fanciful solution would be to DDoS YouTube's copyright system with bogus takedown requests. Place multiple takedowns on EVERY YouTube video. Then they would have no choice but to come up with a system that relies on more than just the say-so of a so-called "rights holder."


They'll just shut down/limit the easily accessible ones and let the ones for big customers running as is.


As I understand it, appeals against these Content ID strikes are assessed by the (alleged) rightsholder.

A good start would be having appeals handled by an impartial party, though I'm sure everyone involved would be loathe to pay for that.


Didn't people discover sound of rain is copyrighted as well?

Along with bird song, <s>children smiles, and joy?</s>



If you watch static long enough, then at some point you will be able to read the collected works of Shakespeare.

So this probably also holds for copyrighted works.

https://en.wikipedia.org/wiki/Infinite_monkey_theorem


There have been similar experiments & results posted before. Seems like a consistent issue with the platform: https://news.ycombinator.com/item?id=16075325


Surprised Google don't detect white noise, and stop if being copyrighted, if for no other reason than the bad publicity this generates.


From their perspective, the purpose of the platform is not to host white noise. I could see them seeing white noise itself as such an extreme corner case as to be not defendable from their content identification. One could imagine a ddos of white noise in an attempt to overwhelm their system. (Devils advocate.)


They let people upload video/audio, I'm sure their definition of corner case[0] is very narrow and I'm not sure this would qualify. It's really no different than computer generated music.

[0] I don't think corner case is appropriate terminology in this discussion. I think it should be edge case. But even still, I don't think Google would think this is an edge case.


Sure, but from a PR perspective, it should be easy enough to check for, and then this never happens again.


My favorite recent copyright story is the Happy Birthday song.

It may be common knowledge by now, but Warner copyrighted this song from the 1800s and profited for years until someone challenged it in court.

I've been waiting for someone to post one of those "this was way more complicated than that and this is why it wasn't as outrageous as people make it seem", but nothing yet.

This in a addition to the Micky Mouse laws really makes the system feel flawed, even though it has its clear benefits.

https://www.nbcnews.com/business/business-news/happy-birthda...


Bogus claims are an ongoing issue and this same basic article (about static videos in particular!) has been posted before -- which is not to say it shouldn't be posted, but rather that Youtube hasn't fixed this known, albeit difficult, problem. Is it considered WONTFIX?

It's one of the reasons that makes me self-host videos, which is actually very easy to do with HTML5. Check the examples section here:

https://en.wikipedia.org/wiki/HTML5_video

It's just a few lines of HTML and you're done. Try it with a test video -- it's surprisingly easy.


What if one of your videos goes viral? Would that not get very expensive?


The videos are only of interest to family and friends.

That said, I've seen a few stats of top-page HN links generating 10,000 or more unexpected visitors, and a 1TB cap with a 20MB video would easily handle that.

But if that weren't the case, if I were making professional videos intended for a mass audience, then I would just add more servers with appropriate bandwidth. Scaling for your own content is vastly different and easier than scaling for everyone's a la Youtube or imgur.


This seems to be one of the less surprising false-positives of Youtube's copyright detection algorithm. You match millions of hours of copyrighted material vs. noise, seems like you have a multiple testing problem and at least one thing will inadvertently match somewhere in the 10 hours noise video

Wonder what Youtube can do here, except speed up dispute resolution. With datasets of their size, false-positives seem inevitable.

Do copyright-owners just set the system to auto-monetization or -takedown? If so, maybe Youtube should change their policy to require more manual action from copyright owners.


So original creator posts videos of static. Then another company says, wait, we have a copyright on that and so they start getting revenue from the videos. The original creator now looses revenue while they go through a process to counter the notice.

This is fair, how?

Why should the other party automatically start getting revenue? Why can't that money go into escrow and whichever party prevails gets that money?

The problem is Google has no incentive to make this better.


Do the authors mean that the claimant had uploaded their static noise video before 2015? Or were the videos matched after the fact?


It doesn’t matter. You can’t copyright white noise. Unless you’re John Cage, or possibly Donald Trump.


I can't see any reason that a work comprising primarily of white noise couldn't be copyrighted.

What you can't do is stop others copying your idea (as long as they don't copy your implementation), nor stop others implementing the same thing who don't copy you.

Now, a work needs in most jurisdictions to be distinctive, but IMO the length of the work would inter alia be enough to make it distinctive, or perhaps use of a specially designed noise generator. But when you release "10 hours of noise" and I copy your idea and make "9 hours of noise" there will be no tort there.


In the US, the threshold of originality seems to be a bit higher than that. E.g. this logo was denied copyright registration: https://en.wikipedia.org/wiki/Threshold_of_originality#/medi...

Refusal letter: https://web.archive.org/web/20160412165054/http://ipmall.inf...


It's not that simple though, Cage's piece -- Silence 4'44" -- I don't think was tested, but taken as a whole I'd grant it copyright. It was art that caused reflection on the nature of a "work", it was designed to be "played" by a musician with any instrument, it was 273s long presumably to reflect on whether silence could ever be absolute. I suspect it was also intended to encourage the listener to observe ambient sounds as if they had meaning. That makes it distinct as a performance piece.

That logo is, within the field of logos, indistinct, boring if you will, if you sat down 100 logo designers I bet you'd get a few that look very like it. I'm not sure I entirely agree with it but having copyright but for a logo RTM is probably a stronger right anyway.

In short I don't think just being white noise is sufficient, but I don't think it excluded a piece of itself.


Importantly, if I sit down at a piano for a few minutes, stare at the keys, and don't play, I have not performed Cage's work. There is a subtle nuance here.


Exactly, you have only performed Cage's work if you set out with the intent to play Cage's work. This obviously makes it impossible to determine if a particular performance of silence is a performance of Cage's work without reading someone's mind, but that same issue of determining intent comes up a lot in legal matters.


We have the best noise.


I sleep with a powered air filter running (HEPA and charcoal) every night and its a perfect noise (maybe not exactly white but minor variance) to sleep by. I should upload a recording and see if anything complains.


They don't complain, they just insert ads and gain money from your recording.


Probably by 65daysofstatic, I mean it's part of their work!


When the "copyright holder" decides to monetize a video, google gets a cut, right? Doesn't this create an incentive for google to prefer false positives?


Google already takes a cut of ad revenue regardless.


How do you copyright leftover radiation from the big bang (since that is how white noise is generated) when the "static" is 13.8-billion years old?


Sounds like an opportunity for copyright "trolling": upload videos, wait for the strikes and removals, sue.


Right, but what's wrong with that? you can't win the suit if the people who issued the request are acting in good faith, so you can't trick them into issuing a takedown if they are not acting maliciously.

in the early days of the do not call list, you could make quite a lot of money off of bad actors who ignored it. there was incentive to get your number on both the do not call list, and cold calling registries, and very quickly made malicious telemarketing untenable.


On what grounds would you sue?


I wonder if this can be considered the precursor to adversarial attacks on neural networks?




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: