The Ninth Circuit has decided Mavrix Photographs v. LiveJournal, and the outcome is in every respect bad news for LiveJournal. In some respects, it’s also bad for the safe harbors themselves, as I’ll explain below and in subsequent posts. The district court in the case granted summary judgment for LiveJournal on grounds that there were no material factual disputes concerning LiveJournal’s eligibility for safe harbor under Section 512(c) of the DMCA. Mavrix alleged that LiveJournal infringed copyrights in its watermarked photographs. Users submitted the photos to LiveJournal along with celebrity gossip news items, and the site's moderators posted them following a fairly intensive screening process (including screening for copyright infringement). There was no question in the case that LiveJournal complied with the DMCA’s notice and takedown requirements when it received notices from right holders. However, Mavrix did not send notices for any of the photos in suit. LiveJournal removed the photos when Mavrix filed its complaint.
Depending on how you count, there were six issues on appeal, all related to LiveJournal’s safe-harbor eligibility: (1) whether LiveJournal’s unpaid moderators were its agents, so that any knowledge they may have had of specific infringements could be attributed to LiveJournal; (2) whether the allegedly infringing photos appeared on LiveJournal’s website “at the direction of users” (vs. on LiveJournal’s own initiative), which is a requirement for safe harbor under Section 512(c); (3) whether, in the absence of takedown notices, LiveJournal had actual or red flag knowledge that the photos were infringing; (4) whether LiveJournal had the right and ability to control the infringements, as evidenced by the required “something more” than the right to remove or block access to infringing material; (5) whether Mavrix financially benefited from infringements that it had the right and ability to control; and, finally, (6) whether the district court erred in denying Mavrix’s motion to compel production of the identity of the moderators who screened the posts at issue in the case.
This case is important to the growing body of DMCA case law for the large number of discrete issues it addresses relating to the application and scope of the safe harbors. I’ll focus on issue (2) in this post and on issues (3) and (4) in two subsequent posts. These three issues are the ones that I believe will be most relevant in future cases. I’ll say in passing that I think the court’s analysis of agency—issue (1)—is correct. As the court describes LiveJournal’s operations, the platform’s moderators play a significant enough role in its curation process, and LiveJournal’s relationship to its moderators is sufficiently “thick,” to justify further inquiry on that issue. With respect to the court’s vacatur of the district court’s denial of the motion to compel—issue (6)—I think the court has gotten that right, too. If (but only if) the moderators are found on remand to be LiveJournal’s agents, discovering what they subjectively knew about the alleged infringements will be necessary to determine LiveJournal’s knowledge or lack thereof. If the moderators who screened the posts containing Mavrix’s photos are LiveJournal’s agents, I don’t see how LiveJournal's knowledge could be determined without their testimony.
Issue 2: Were the posts “at the direction" of users?
The court begins its discussion of this issue with an incorrect description of the different activities covered by the different safe harbors. It then goes on to misapply its own precedent—UMG Recordings v. Shelter Capital Partners. This part of the opinion is highly problematic, and it’s a good candidate for rehearing en banc considering how it might reverberate in future cases. The court’s reasoning on this issue undermines the safe harbors for all sites that use moderators to screen and curate user-submitted content before posting it, including the world’s largest platforms—Facebook, Twitter, and YouTube. Such curation is precisely what Congress encouraged when it enacted Section 230 of the Communications Decency Act. In this era of rampant online harassment, “fake news,” and revenge porn, it seems especially misguided for a court to interpret the DMCA in a way that could discourage human moderation of user-submitted content.
According to the court, the section 512(a) safe harbor covers users’ submission of material to providers, and section 512(c) covers the providers’ subsequent posting of that material to their sites. There is no such submission-posting distinction in section 512. On the face of the statute and in the legislative history, it’s quite clear that section 512(a) is meant to cover user-initiated, end-to-end routing of information across a provider’s network. A residential broadband access provider is the paradigmatic section 512(a) provider. Section 512(c) covers hosting providers like LiveJournal that receive, store, and provide public access to stored user-generated content. To characterize LiveJournal as a hybrid 512(a)/512(c) provider misapplies the statute and introduces into the case law a wrongheaded distinction between submitting and posting material.
Putting aside the peculiar submission-posting dyad, the dispositive question concerning LiveJournal’s eligibility for the section 512(c) safe harbor is whether the site’s moderator-curated, user-submitted posts occur “at the direction of users,” taking into consideration the nature of moderators’ review and the fact that only about one-third of user submissions are ultimately posted. That question can be answered entirely within the ambit of section 512(c) and the existing case law interpreting it, including the Ninth Circuit’s own decision in Shelter Capital. There was simply no need for the court to invoke section 512(a) in this case.
Shelter Capital held that a provider can undertake “accessibility-enhancing” activities that go beyond storage without surrendering the storage safe harbor in section 512(c). Included among these permissible, “storage-plus” activities are active behaviors like transcoding and displaying user-uploaded files. The nub of the “storage-plus” inquiry is when it is no longer fair to say in light of a provider’s post-upload actions that material appearing on the provider’s site appears there “at the direction of users” (versus on the initiative of the provider itself).
To my mind, this is a much simpler question to answer in LiveJournal’s case than the court makes it out to be. If a user submits material to a website with the intention that it be posted, and the material is ultimately posted, that posting should be held to have occurred “at the direction” of the user, regardless of what the provider did between the time the material was submitted and the time it was posted. The fact that a provider culls user-submitted material does not alter the fact that all users who submit material to the provider, including those whose material is ultimately posted, are in effect directing the provider to post their material. For a curated site, sometimes the provider heeds the user’s direction, and sometimes it doesn’t. But the existence of discarded user-submitted material does not undermine the proposition that posted user-submitted material was posted “at the direction” of the users who submitted it. It was therefore correct for the district court to conclude as a matter of law that LiveJournal was not disqualified from the storage safe harbor by virtue of its moderators’ review and curation of user-submitted posts. The Ninth Circuit was wrong to reverse on this issue and has created uncertainty where none existed before.
Concluding that the court erred on this point is not to say, however, that LiveJournal’s activities in relation to user-submitted material are irrelevant to the safe harbor analysis. On the contrary, what a provider does between the time a user submits material and the time that material is posted—assuming it’s posted—may give rise to knowledge of specific infringements that disqualifies the provider from safe harbor. The court’s analysis of actual and red flag knowledge under the DMCA will be the subject of my next post in this series.