Monday, April 10, 2017

LiveJournal's missteps threaten its DMCA protection

Mavrix Photographs, LLC v. LiveJournal Inc., No. 14-56596 (9th Cir. April 7, 2017)

Initial note: What the court here describes as LJ’s business model is in reality limited to its treatment of ONTD, the most popular community on LJ.  Most other LJ communities, not to mention its individual journals, operate very differently.  Query whether subsequent treatment will understand this important fact when dealing with the reversal of summary judgment in LJ’s favor on its 512 defense.

Mavrix sued LJ for infringement of twenty photos.  Users submitted the photos, but “a team of volunteer moderators led by a LiveJournal employee reviewed and approved them.”  Whether the acts of the moderators could be attributed to LJ was a disputed question of material fact under the common law of agency, which applies to DMCA analysis of whether material was “posted at the direction of the user.” [At the outset, I don’t get this—the analysis on knowledge etc. makes somewhat more sense, but they’re still user-submitted photos whether or not the moderators screen them.] The court of appeals also vacated the district court’s order denying discovery of the moderators’ identities.

LJ “allows users to create and run thematic ‘communities’ in which they post and comment on content related to the theme.”  There are three unpaid administrator roles: “moderators” review posts submitted by users to ensure compliance with the rules; “maintainers” review and delete posts and have the authority to remove moderators and users from the community; one “owner” per community can also remove maintainers.  Oh No They Didn’t! (ONTD) is a popular LJ community focused on celebrity news.

ONTD’s rules instructed users to “[i]nclude the article and picture(s) in your post, do not simply refer us off to another site for the goods.” Another rule: “Keep it recent. We don’t need a post in 2010 about Britney Spears shaving her head.” ONTD’s rules included a list of sources from which users should not copy, which were sources that informally requested that ONTD stop posting allegedly infringing material. ONTD also automatically blocked all material from one source that sent ONTD a C&D. Moderators reviewed proposed submissions and publicly posted about one-third of them. The substantive requirement for approval was new and exciting celebrity news, though they were also supposed to screen out copyright infringement, pornography, and harassment.

Like other LJ communities, ONTD used to be exclusively volunteer, without LJ involvement in day-to-day operation of the site. But it hit 52 million page views per month in 2010 and attracted LiveJournal’s attention. “By a significant margin, ONTD is LiveJournal’s most popular community and is the only community with a ‘household name.’” Thus, LJ determined to exercise more control over ONTD so that it could generate ad revenue from it. LJ hired a then active moderator, Brendan Delzer, to serve as the community’s full time “primary leader” with the intent to “take over” ONTD, grow the site, and run ads on it. Delzer instructed ONTD moderators on the content they should approve and selects and removes moderators on the basis of their performance, as well as performing moderator work of his own. Delzer was paid and expected to work full time, while the other moderators are “free to leave and go and volunteer their time in any way they see fit.”

ONTD posted the allegedly infringing photographs in seven separate posts between 2010 and 2014. Some of the photos contained either a generic watermark or a specific watermark featuring Mavrix’s website “” Delzer did not recall personally approving the seven posts, and LJ has no technological means to determine which moderator approved any given post. Mavrix didn’t send DMCA notices, but when it sued, LJ removed the posts.

The district court held that users’ submission of the posts was key to make them “at the direction of the user.”  The court of appeals disagreed, holding that §512(a) dealt with submission, while § 512(c) “focuses on the service provider’s role in publicly posting infringing material on its site.  Contrary to the district court’s view, posting rather than submission is the critical inquiry.”  This is … an interesting reading of the DMCA.  §512(a) is about transmission.  If §512(c)’s “hosting” means “publicly posting,” then what happens if an ISP enables private storage? 

Anyhow, the common law of agency could make LJ responsible for the moderators’ acts; the DMCA incorporates the common law of agency. To the extent that BWP Media USA, Inc. v. Clarity Dig. Grp., LLC, 820 F.3d 1175 (10th Cir. 2016), contradicted this holding by suggesting that ISP employees could be “users” under the DMCA, the court of appeals disagreed.

So, were the moderators LJ’s agents?  Agency requires actual or apparent authority to act on behalf of the principal as well as the principal’s right to control the actions of the agent.  There was evidence that LJ gave its ONTD moderators explicit and varying levels of authority to screen posts. [Note again that the court is writing as if moderators were this deeply embedded in LJ’s business model throughout LJ’s operations; in fact, I can make myself a moderator of a LJ community I create with no scrutiny/direction at all other than LJ’s general TOS.  This discussion is about ONTD, not LJ as a whole.]  Though they were “volunteers,” “the moderators performed a vital function in LiveJournal’s business model. There is evidence in the record that LiveJournal gave moderators express directions about their screening functions, including criteria for accepting or rejecting posts.”  There were genuine issues on actual authority. So too with apparent authority. LJ users “may have reasonably believed that the moderators had authority to act for LiveJournal”; for example, one user whose post was removed pursuant to a DMCA notice complained to LiveJournal “I’m sure my entry does not violate any sort of copyright law. . . . I followed [ONTD’s] formatting standards and the moderators checked and approved my post.”

Agency also depends on the level of control a principal has over the agent, and there was evidence that LJ “maintains significant control over ONTD and its moderators.”  Delzer supervised moderators and selected and removed moderators on the basis of their performance.  He also exercised control over the moderators’ work schedule. “For example, he added a moderator from Europe so that there would be a moderator who could work while other moderators slept.” Moderators’ screening criteria derived from rules ratified by LJ—LJ ratified them when one LJ employee discussed changing the rules with Delzer and declined to do so.

However, ONTD moderators “are free to leave and go and volunteer their time in any way they see fit.” The moderators can alos reject submissions for reasons other than those provided by the rules, which called into question LJ’s level of control.  Thus, reasonable jurors could find agency, but might not be compelled to do so.

If the moderators were LJ’s agents, the factfinder would still have to assess whether Mavrix’s photos were posted at the direction of the users in light of the moderators’ role in screening and posting. Activities “narrowly directed” towards enhancing the accessibility of the posts wouldn’t change the user-provided nature of the posts. Accessibility-enhancing activities “include automatic processes, for example, to reformat posts or perform some technological change” as well as “[s]ome manual service provider activities that screen for infringement or other harmful material like pornography.”  This follows from § 512(m) of the DMCA, which provides that no liability will arise from “a service provider monitoring its service or affirmatively seeking facts indicating infringing activity.”  What are the edges of accessibility-enhancing intervention?  When YouTube manually selected videos for front page syndication on the basis of substance, the district court on remand held that only those processes “without manual intervention” satisfied § 512(c).  The Fourth Circuit has approved a real estate website’s “cursory” manual screening to determine whether photographs indeed depicted real estate.  This would be an issue for the fact-finder.

The ONTD moderators posted only about one-third of submissions: only those posts relevant to new and exciting celebrity gossip. Betraying a bit of a prejudgment, the court of appeals said that the question was whether “their extensive, manual, and substantive activities went beyond the automatic and limited manual activities we have approved as accessibility-enhancing.”

The court also addressed LJ’s actual and red flag knowledge of the specific infringements alleged.  Failure to use a DMCA notice “is powerful, but not conclusive, towards showing that a service provider lacked actual knowledge.” Delzer didn’t remember approving the posts, but Mavrix didn’t have the opportunity to depose the moderators to determine their subjective knowledge [assuming they were agents, which I guess we are now]. “On remand, the fact finder should determine whether LiveJournal, through its agents, had actual knowledge of the infringing nature of the posts.”  Even without actual knowledge, red flag knowledge arises when a service provider is “aware of facts that would have made the specific infringement ‘objectively’ obvious to a reasonable person.” Watermarks were relevant even if Delzer didn’t know that Mavrix had a website: “The existence of a watermark, and particularly this watermark with a company name, is relevant to the knowledge inquiry…. [T]he fact finder should assess if it would be objectively obvious to a reasonable person that material bearing a generic watermark or a watermark referring to a service provider’s website was infringing.”

In addition, LJ would have to show that it did not financially benefit from infringements that it had the right and ability to control. LJ’s general practices would be relevant, not its conduct with respect to the specific infringements, since “right and ability to control” involves “something more than the ability to remove or block access to materials posted on a service provider’s website.” Something more can be present when the ISP exercises “high levels of control over activities of users,” such as when it “prescreens sites, gives them extensive advice, prohibits the proliferation of identical sites,” provides “detailed instructions regard[ing] issues of layout, appearance, and content,” and ensures “that celebrity images do not oversaturate the content.”

The court of appeals rejected the district court’s conclusion that LJ lacked “something more,” and sent it to the factfinder, because:

LiveJournal’s rules instruct users on the substance and infringement of their posts. The moderators screen for content and other guidelines such as infringement. Nearly two-thirds of submitted posts are rejected, including on substantive grounds. ONTD maintains a list of sources that have complained about infringement from which users should not submit posts. LiveJournal went so far as to use a tool to automatically block any posts from one source.

LJ also needed to show that it did not derive a financial benefit from infringement that it had the right and ability to control. “The financial benefit need not be substantial or a large proportion of the service provider’s revenue.”  The presence of a vast amount of infringing material supported an inference of such a financial benefit, where the service provider “promoted advertising by pointing to infringing activity” and “attracted primarily visitors who were seeking to engage in infringing activity, as that is mostly what occurred on [the service provider’s] sites.” Here, LJ derived ad revenue based on the number of views ONTD receives. Mavrix presented evidence showing that approximately 84% of posts on ONTD contain infringing material, although LJ disagreed; again, this was for the factfinder.

Also, on remand, whether the moderators were agents should inform the district court’s analysis of whether Mavrix’s need for discovery outweighed the moderators’ interest in anonymous internet speech. “Given the importance of the agency analysis to the ultimate outcome of the case, and the importance of discovering the moderators’ roles to that agency analysis, the district court should also consider alternative means by which Mavrix could formally notify or serve the moderators with process requesting that they appear for their deposition at a date and time certain.”

No comments:

Post a Comment