Tuesday, October 13, 2015

CFP: Intermediary liability

Intermediary Liability: How to Kill Content on the Internet. Thanks to statutes like Section 230 of the Communications Decency Act and even the Digital Millennium Copyright Act, intermediaries in the United States are at least theoretically shielded from liability for how actions by users on their systems. As a result, the internet has become a diverse and vibrant platform for speech of all sorts. Some, however, seek routes around these limitations on liability. One vector of attack is through the DMCA, but new tactics are also being tried, including the pursuit of domainwide injunctions, site blocking, and threats to payment providers, as well as attempts to limit Section 230. Many of these attacks on intermediaries are launched by entities focused on controlling intellectual property rights . In this talk we’ll consider those, and also how these tactics are being used by those with objections to other types of speech, such as adult content, including “revenge porn” and online advertisements, or content governed by the “Right to Be Forgotten.” And we’ll look at what can be done to better protect all online speech from censorship by strengthening intermediaries’ legal and political positions.
Cathy Gellis, Attorney (moderator): §230 and DMCA for intermediaries.  §230 is very strong for intermediaries.  Plaintiffs hate that.  Can’t be compelled to remove content.  But see Barnes v. Yahoo! (contract claims are allowed).  Carveout for federal crimes; IP.  No state carveouts, but they try.  DMCA for ©—much more complicated and conditional than §230. 
Emma Llanso, Director, Free Expression Project, Center for Democracy & Technology: §230 is incredibly strong legal protection for intermediaries—really important.  Strength has drawn the attention of people who’d really like to be able to sue an intermediary for a harm caused by a third party. Various attempts to amend/work around.  Victim of child trafficking who has judgment against the person who trafficked them, but that person is judgment-proof.  Seeking to go after website that posted ads related to trafficking: very sympathetic P; law enforcement learns that §230 bars making the website liable. 
53 State/Territorial AGs sent a letter to Congress proposing an amendment to allow state laws for intermediary liability to override §230.  Defeated due to outcry; since then, would-be imposers of liability have gotten more clever. Big one, especially this year: looking at the exemption for federal criminal law.  Efforts to create new federal criminal laws with some kind of liability for hosting third-party content, such as the SAVE Act: Stop Advertising Victims of Exploitation Act.  Additional liability beyond existing trafficking liability—the crime of advertising a person for trafficking purposes.  Passed earlier this year, without formal hearing.  Ultimately added to Justice for Victims of Trafficking Act, which got tied up in Lynch confirmation so rational debate ended.  Not clear what “advertising” means.  Not defined in statute.  Point of law according to advocates was to be able to go after platforms that host advertising.  But the person who creates the ad is arguably the advertiser—still unsettled.
Rebecca Tushnet, Professor of Law, Georgetown University Law Center: Compared to §230, §512 notice and takedown doesn’t make anybody happy.  That doesn’t mean it’s good, but it does mean it’s a compromise that may be extremely hard to change, despite the unanticipated volume of notices—hundreds of millions overall.
Takedown problems: (1) algorithmic overreach—recently dealt with a Harry Potter takedown notice based on the fact that the name of the file is Harry Potter and the Deathly Hallows; have had similar issues with fan fiction stories that happened to use the same name as a popular song—insufficient use of metadata—also a problem of figuring out that some online instances are in fact noninfringing because authorized by the rightsholder—for example, NBC Universal recently took down its Canadian affiliate’s stream of Mr. Robot; (2) claim overreach—attempts to use DMCA for privacy, trademark/getting competitors’ links removed, suppressing criticism (claims made on behalf of the Argentinean gov’t; Apple trying to hide its contract terms)—Google’s Transparency Report, WordPress’s hall of shame.  Need for better deterrent to claim arbitrage; some hope that Lenz will improve matters, at least for problem (1).  Big differences in behavior of the small number of senders who send millions of notices—algorithms are the problem—versus the large number of senders who send a few notices a year—for whom misuse or at best misunderstanding is the problem.  For fascinating empirical data: Daniel Seng, The State of the Discordant Union: An Empirical Analysis of DMCA Takedown Notices, 18 Va. J. L. & Tech 369 (2014)
Other than judicial decisions interpreting §512, what’s going on?  Experience from PTO/NTIA DMCA best practices group: weak tea.  Didn’t even go as far as Lenz in its recommendations for submitters of DMCA notices: “it is a good practice to take measures that are reasonable under the circumstances (e.g. taking into account the information visible to the notifier and the apparent volume of infringement at the location, etc.) to … appropriately consider whether use of the material identified in the notice in the manner complained of is not authorized by the copyright owner, its agent or the law.” Didn’t specify much of anything for senders or recipients of notices b/c the DMCA already gives plenty of detail about what you’re supposed to do. 
Participants insisted on having best practices for counternotification even though we all understood that counternotices don’t come from people who read PTO/NTIA best practices.  Compare the bad practice for notice senders: Bad practices for notice senders include “Falsely asserting that the notice submitter has a good faith belief that use of the material in the manner complained of is not authorized by the copyright owner, its agent or the law.” Bad practices for counternotice senders include, “Failing to take reasonable efforts to form a good faith belief that the material was removed or disabled as a result of a mistake or misidentification of the identified material.”  Again, compare Lenz: the courts are doing a better job.
Larger lesson: this was a meaningless exercise, and we still couldn’t agree on much of anything. Stalemate that will leave current law in place, making judicial interpretations the most important source of law unless/until raw political power produces a different result for copyright owners.
Speaking of which: testimony on §512 before Congress.  Content owners have come up with a phrase that is Frank Luntzian in its brilliance: “notice and takedown” is insufficient, so we should have “notice and staydown.”  Just a little change!  Essentially indifferent to the reality that this is filtering with a rhyme.  Attitude from representatives at the hearings earlier this year was essentially “get the nerds on it” and “Google can do this with Content ID for video and sound on the YouTube servers it controls, so it must be possible for everyone to do this with everything for the entire internet.”  Risks: It is very difficult to get a representative to understand something when her campaign donations depend on not understanding it.
Final random note: design patent—none of the exclusions we’re used to, including fair use; people are getting design patents on logos and graphic interfaces, potentially creating huge new sources of infringement litigation.  Newly discovered by plaintiffs’ lawyers and ordinary businesses, which have less interest in balancing free speech interests than media plaintiffs do.  So, like the ITC, this is an issue for intermediaries to keep an eye on.
Gellis: People want content down from the internet.  How will they use/abuse/change/route around laws to get the content taken down?
What other content is being seen as bad and worthy of takedown?
Llanso: We are increasingly hearing “Google can do it” from regulators around the world—drawn from implementation of DMCA notice and takedown. A lot of policymakers think of notice from private individuals to intermediaries resulting in content takedowns, without court involvement, as standard/comfortable.  Proposals around “extremist” content—but one big problem is that people don’t offer much of a definition of extremist content.  Concerning to look at a category that runs the gamut from political/religious speech to incitement to commit imminent violence.  France: wants public-private partnerships w/intermediaries. UK has internet referral unit, official gov’t unit to use content policies defined by major platforms like YouTube and content flagging mechanisms, as a way for gov’t to get rid of content that may or may not be inconsistent w/ the law.  No court involved.  No obvious angle for users whose content has been taken down/opportunity for appeal/oversight from independent third party.
(RT: May be another reason to endorse algorithmic approach approved by the court in Lenz: makes more clear that © notice and takedown is a poor model for other types of content, which can’t be determined in an algorithmic way.)
Gellis: right to be forgotten.
Llanso: Right.  Even if it’s private action, speakers feel censored.  Right to be forgotten has created a circumstance for search engines that they have to respond to assertions from private parties or they’ll violate European data protection law—have to be responsive to private parties, not court orders.
Gellis: DMCA is about not using the courts—lowers transaction costs but also creates them.
RT: if you get sued as an intermediary, hard to get out on a motion to dismiss.  Means that to get big you need an agreement with the industry. 
Gellis: out of the courts: validity of the claim is never tested, which means there’s vulnerability to abuse.  Small intermediaries: not as bullish as RT—many in the content industry regret allowing Google to grow big.  Small intermediaries can’t take the risk of ignoring the takedown notice—it may be crap, but the safer course is just to delete/just censor.
Llanso: threats to credit card companies led them to refuse to process transactions for Backpage—not just for adult content but for everything. Backpage sued sheriff for pressuring the credit card companies as a prior restraint on speech.  Important case to watch—TRO granted against sheriff.  But denied PI, so the case is being appealed to the 7th Circuit.  One of the most concerning elements of the dct opinion was characterizing the letters as containing threats but also containing the Sheriff’s own 1A-protected advocacy—a stunning result.  Adding advocacy shouldn’t be able to avoid liability for censorship—a roadmap for any official to threaten.
Gellis: nothing on the internet happens w/o the help of someone else.  Intermediaries are everywhere—some issues are restricted to a particular sector, but that’s why people try to recharacterize issues.  Protecting intermediaries is important b/c it’s difficult to get the good stuff if we pressure intermediaries to take down everything bad.
Q: View of speech as an absolute and internet as exceptional medium.  Financial intermediaries are also intermediaries, but we restrict them offline, for a variety of different transactions, including restrictions on arms trade, other blockades.  When this is electronically, and we can’t get jurisdiction over people, how do we get at those kinds of transactions w/out targeting the intermediary?  What’s the minimal harm way of targeting the intermediaries?
Llanso: there are laws proportionate to the harm.  When we don’t have gov’t acting under a clearly defined law, with balances worked out, we have a big problem.  If the rule is “don’t make the sheriff angry” and the content isn’t adjudicated unlawful, that’s a big problem.

No comments: