Last night, YouTube posted a note on the official YouTube Blog concerning the recent firestorm over Content-ID takedowns like the one I wrote about earlier in the week regarding my Downfall DMCA Parody.
In the post, they are kind enough to link to my video (now back up on YouTube thanks to my disputing the Content-ID takedown) as an example of a fair use parody, and to a talk by (former) fellow EFF director Larry Lessig which incorporated some copyrighted music.
However, some of the statements in the post deserve a response. Let me start first that I hope I do understand a bit of YouTube’s motivations in creating the Content-ID system. YouTube certainly has a lot of copyright violations on it, and it’s staring down the barrel of a billion dollar lawsuit from Viacom and other legal burdens. I can understand why it wants to show the content owners that it wants to help them and wants to be their partner. It is a business and is free to host what it wants. However, it is also part of Google, whose mission is “to organize the world’s information and make it universally accessible and useful,” and of course to not “be evil” in the process of doing so. On the same blog, YouTube declares its dedication to free speech very eloquently.
As such YouTube does want to avoid the blocking of non-infringing videos while trying to help content owners get rid of actual infringements on the site. These recommendations apply on what to do for partial Content-ID matches where the upload is not simply a verbatim audio/video copy of the content owner’s work, but is possibly transformed into something else which may be non-infringing.
Find only, don’t takedown.
The simplest plan would be to simply offer the Content-ID system as a means to help content owners find violations but not a means to have them command DMCA-free takedowns, particularly ones that take place when no human being has reviewed the material to assure it is not a protected non-infringement. And it already does this, but it offers more to the content partner, namely the ability to auto-takedown various matches. We at the EFF don’t think it is necessary to offer the auto-takedown at all, except on verbatim, un-transformed uploads, but I have heard the complaint that in many cases, infringements are uploaded at a rate faster than a company can individually examine. This surprises me, but if need be, auto-takedown could be offered as a service only to content owners who do indeed get really large numbers of matches for their properties. Those who get only a few might be expected to do a personal inspection to assure takedowns are not done on non-infringing works. There is no need to give them the bulk service. You could set the bar at a workload that it would be hard to deem unreasonable.
Next, auto-takedown could be made less immediate, if not for new uploads then at least for older uploads which are well established on the site. Is there truly a need for an immediate takedown of videos that have been there for weeks or even years? What is the emergency? Even the more powerful DMCA takedown is not instantaneous. However, if a content owner needs an urgent takedown, let them do it, but only after direct examination of the video. Surely there is not a need for a blanket emergency takedown?
With a non-urgent takedown, the uploader would be notified of the requested takedown, and have some period of time, perhaps just a couple of days, to file a dispute. If they don’t, down it goes until they do. This could harm uploaders who are out of touch for a while, but the majority would not be hurt. The content owner would suffer a few more infringements but the cost would be minor. Instead of timing it in days, it could be timed in downloads. Perhaps a very frequently downloaded file would only get a minimum time (one day) to respond. This would inherently limit the damage to the content owner from an actual infringement. Uploaders of files which get really high download rates might expect to be more available to receive notice.
Of course for new uploads, the uploader is present and can file a dispute right there. The video need not go up at all until the dispute is filed on a new upload that matches a fingerprint.
Another alternative would be to allow uploaders to check a box that says, “I made use of copyrighted materials in the creation of this video, but warrant that I did so without infringing copyrights.” This would in effect be an “automatic dispute.” Checking it could come with the same warnings that come when you try to file a dispute today, about how serious the matter is, and how you are opening yourself up to liability if the claim is false. If you fear that every infringer would routinely check the box, you can be harsh on those who do this falsely, and ban them from YouTube, as you already do when you discover a user is routinely infringing.
With this automatic dispute, your content owner partner can decide then and there to file a real DMCA takedown, the rules for which are already laid out.
As noted, YouTube punishes repeat offenders who infringe. YouTube should also consider restricting the actions of content partners who misuse the Content-ID system to do takedowns on non-infringing works. Particularly, if I may make a personal point, if they apply them to works that are critical of the content owner. In your blog post, you write “Rights holders are the only ones in a position to know what is and is not an authorized use of their content, and we require them to enforce their policies in a manner that complies with the law.” I fear that the first part of this is quite at odds with one of the core doctrines of fair use — that it is permitted because criticism of works and their creators is just the sort of thing that the rights owners can’t be trusted to judge. (I also think you might not want to cite the dancing baby as an example of infringement. Lenz vs. Universal is an EFF case, and I think you’ll find things are going well for the mother of the dancing baby.)
But the second part of the sentence speaks to a solution. The DMCA takedown process, for all its flaws, provides some disincentives to false takedowns. The EFF has helped get judgements against parties who misused the DMCA takedown process to censor speech. The Content-ID takedown process seems to have no disincentive for misuse.
I propose that you fix that. If a content partner misuses the Content-ID system to takedown non-infringing videos multiple times, you should treat them as you treat users who do wrong multiple times. This partner should be barred from using the Content-ID takedown system any further. They can still use it to find infringement, if you wish, but they should be denied the privilege of automatic removal at the very least. I feel that Constantin Films has already gone way over that line. They knew — it’s well documented — that the majority of the Downfall videos on YouTube were parodies that substantially transformed the underlying work, only used a few minutes from the film and did not harm the market for the film. (These are among the key fair use tests.) Perhaps a few out of hundreds were just clips from the actual film, un-doctored. Another handful at most were of the type which Constantin says were generating “sick, offensive humour” complaints. Not that these are any more infringing than the others; they were simply quite few in number. In spite of that Constantin Films used what could have been a scalpel and used it like a sledgehammer. If this is not an example of misuse, then what is? If YouTube is able to agree that any tool should have checks and balances against misuse, and this doesn’t count, then what does?
Thanks for responding
It’s good to see you responding. I hope you will find these suggestions constructive. I hope you will continue your dedication to free speech and the company mission of making information available. Google has shown some great leadership in this area of late, and there’s no reason the YouTube arm can’t carry that torch well.
Fix some bugs
And on a minor technical note, there are a few bugs where the system is not operating as you hoped. I checked my mail logs, I got no notification of the Content-ID takedown. I learned of it first from readers of my blog saying the video was not showing. Once I learned about it, it was not at all easy and clear, even after reading the help system, how to file the dispute. After I filed the dispute, the status page still says, “the video is blocked worldwide” even though it has been re-enabled. When it was re-enabled, it was enabled with embedding disabled, even though it had been embedded beforehand.