byon November 13, 2020
Just two weeks ago, my colleague Denver posted a criticism of Microsoft's GitHub for its capitulation to the RIAA takedown notice — which alleged, with specious evidence, that youtube-dl violated 17 USC § 1201. Frankly, this is the kind of behavior we'd expect from the RIAA — an organization controlled in recent decades by two dudes who seem to have helped write § 1201. No one is surprised when the RIAA attacks FOSS projects.
Last night, though, I was shocked to learn that a company that generally has a much better track record on DMCA matters (and with FOSS projects) joined the recent onslaught of DMCA takedown notices against FOSS projects. Namely, GitHub announced yesterday that Google sent a § 1201 DMCA takedown notice for a FOSS project called widevine-l3-decryptor.
Google is the primary provider of browser-based DRM technology for nearly all of the well-known entertainment streaming services, through a product called Widevine. If you've watched a streaming video from a major provider (such as Netflix, Prime Video, and Hulu), then you've probably used Google's DRM. For the past two years, researchers have been in a DRM arms race with Google in cracking the lowest level (and lowest video quality level) of Widevine (called “L3”). The most recent crack inspired creation of a FOSS project, called widevine-l3-decryptor. If successfully integrated into browsers and other platforms, this new freely licensed code may well allow a 100% FOSS solution for viewing videos at this lowest level of DRM0. As always, though, DRM and software freedom remain on an irreconcilable collision course; the function of one always precludes the other.
If you just had déjà vu, it's likely because the narrative here resembles the story of DeCSS from about 20 years ago. The big differences are: (a) cracking L3 isn't as big of a threat to DRM technology (since it yields video output at very low quality), and (b) more importantly, and strangely, Google takes on the role of the MPAA in this repeat dance of DMCA history.
What Should Activists Do?
I admit that this situation kept me up half the night. My first thought was to come out blogging today that we needed to immediately institute a full boycott of all DRM, and the companies that produce it. Yes, a boycott would surely be effective, but it is effectively impossible. Seth Schoen, who has spent nearly a lifetime working to fight DRM, told me once that the talking point inside the industry circles in the early 2000s was:
DRM is inevitable. Indeed, the media companies succeeded in inserting that phrase into culture, research circles, and everywhere else — so much so that it eventually became a running joke for activists.
We erred in our arrogant belief that DRM would remain clunky and rare. Media companies and their technology providers have laughed at us all the way to the proverbial bank. Then, the W3C and Mozilla Foundation capitulated with EME. Simply put, the boycott won't work now because DRM, along with the ubiquity of proprietary software in (at least as some component of) every popular platform means that DRM is seamless, easy-to-use, and rarely gets in the way of paying customers. We in fact tried, and mostly succeeded, in boycotting DRM when it was cumbersome, full of bugs, and annoyed users in the early 2000s. Today, most users of DRM don't even know it's there, or who provides it. While I'm not a user of browser-based streaming, I am still embarrassed to say that until yesterday I didn't even know Widevine was a Google product. I thought others were fighting the good fight against DRM and I mostly ignored it. But no one really is. DRM may not rule the technological world for software freedom activists who shun proprietary platforms. But it silently rules everyone else's tech world, and boycotting DRM effectively means boycotting most technology.
This led me to think of political polarization and a failure to compromise, and how it puts policy issues into gridlock. So, for a moment, let's step aside from our visceral negative reaction to 17 USC § 1201. Of course, we should never forget that § 1201 frustrates many Free Speech rights with respect to software in the USA; however, we also must admit that strategies until now have failed to repeal that abhorrent law. So, let's attempt for a few minutes to see the other side's position, as it might help us find other ideas to try while we wait indefinitely (22 years and counting ☹) for restoration of technological Free Speech rights.
Toward that mindset, consider that the copyright statute is ultimately a tool. We in the FOSS community created copyleft as a method to use that tool for good rather than ill. Meanwhile, the media and big tech companies lack any moral motivations on this issue, so they see this tool as merely a method to keep paying customers paying over and over again for the same content. It seems on the surface that there's no zone of agreement, but perhaps there is. Maybe we can agree, especially when we look back to the era of DeCSS, that §1201 is a tool so sharp that it instead became a clumsy weapon. Herein, I propose a compromise that slightly blunts §1201's sharpness. That compromise can be found by focusing on the consensus we already have regarding what parts of copyright that copyleft enforcement should avoid.
Short Term: Copyright Enforcement Parity
Two years ago, Google along with many other companies signed on to the IBM's Red Hat initiative and agreed to the RHCC. The RHCC is a pledge (similar to the KESAP, the latter of which we endorsed) whereby all copyright holders in GPL'd software agree to allow infringers 30 days to repair any copyright infringement consequence-free. During that 30 days, the infringer can continue acts of copyright infringement (e.g., violating the GPL) with impunity. On the thirtieth day, if the infringer achieves full compliance with the GPL, all is forgiven and no penalties are imposed.
While 30 days of unabated GPL violations are quite problematic and often result in thousands of customers remaining uninformed forever that their products contain copylefted software (and thus are possibly never informed that software freedom exists), GPL enforcers have always understood that it takes time for folks to coordinate a response and fix the situation. We have remained steadfast in our focus on beginning with friendly and respectful conversation with any infringer. We never demand immediate injunction; in fact, we usually don't even request one for about 120 days or more. By contrast, DMCA takedown is the exact opposite approach. Takedowns are unwarranted for FOSS projects that develop and operate transparently in the open and facilitate important work for the public good.
Thus, I hereby call on all these companies who signed onto the RHCC to agree immediately to the following pledge:
We agree that regarding any and all alleged copyright infringement committed by any software project that is licensed under (or intends to license under) an OSI-Approved License, that we will give notice to the project, and take no action of copyright enforcement (including but not limited to DMCA takedown) for at least 30 days from the date of notice of our concerns made to the project.
I will be writing to my contacts at all the companies who signed onto the RHCC to ask them to sign onto this provision as well.
Medium and Long Term Solutions
An agreement to slow the rising DMCA takedown onslaught won't actually solve the scourge of §1201. But it can raise awareness. We live every day under unreasonable restrictions from the DMCA and similar laws worldwide. We sometimes forget the urgency of this problem, but this fall's reemergence of the DRM wars should remind us that we must regularly discuss concrete ideas for response. Over the coming weeks, I'll be blogging more on the topic with more ideas to address this problem. In the meantime, I hope I've inspired some of you to propose ideas on how to respond in this struggle, and please do share this post and the request above on social media and any other fora you frequent!
0Google does claim, without evidence, that some of the files in the project were copyrighted by Google and therefore not licensable as FOSS. If true, Google would also have a more mundane copyright infringement allegation unrelated to §1201. However, recall that it's typical for pro-DRM organizations to (incorrectly) claim that databases of keys or even keys themselves are independently copyrightable, and we strongly suspect that's occurring here.