Friday, February 14, 2020

DMCA at 22: my notes from the Senate hearing

Senate Judiciary Committee, Intellectual Property Subcommittee, The Digital Millennium Copyright Act at 22: What is it, why was it enacted, and where are we now (archived video; apparently you have to wait 13 minutes before the hearing actually starts, though)

Sen. Tillis: DMCA was passed when Chumbawumba and Myspace were popular. Time to hear about modernization. Protection was given to new platforms in exchange for quick removal. Almost everything has changed in 22 years. Goal: introduce reform bill w/bicameral, bipartisan consensus. Top priority for him. Today: background and level-setitng.

Panel I
The Honorable Edward J. Damich, Senior Judge, United States Court of Federal Claims
Damich was chief IP counsel for Senate Judiciary at the time; not providing views of judiciary but recollections as staffer. Internet just coming into its own: 36 million web users in 1996. Highlights: (1) ISP liability for users’ infringement, (2) TPMs were key issues. Sen. Hatch didn’t want to disturb common law of secondary infringement; wanted to focus on practical reality of internet at the time: email, system caching. Hosting required more care: notice & takedown or actual knowledge requried action. Middle ground b/t duty to monitor and actual knowledge: red flag provision. ISPs should not ignore clear indications of infringement however discovered & should cooperate w/copyright owners. Anticircumvention: controlling access was becoming more important than making copies. Exceptions included periodic rulemaking for temporary exemptions.

Mr. Jonathan Band, Owner, Jonathan Band PLLC
TPM control appeared in EU Software Directive. Problem: legitimate reasons to unlock TPMs, including for interoperability; he worked on exceptions. Safe harbors for basic online functions: Yahoo provided the then-leading directory, but arguably fell outside the info location tool safe harbor, so negotiated legislative report language to make clear that it was covered. Database protection was also proposed by the DMCA House version, fortunately failed. DRM and safe harbors were a compromise: you can’t consider the effectiveness and fairness of either in isolation. Each industry believes its own section was successful; each title includes internal compromise, but that was a grand bargain between each one. Congress made policy choices w/open eyes and courts have generally complied with Congressional intent. Note also term extension at the same time: windfall to © owners.

Mr. Robert S. Schwartz, Partner, Constantine Cannon
Road began in 1981 with 9th Circuit’s ruling against VCRs. Even after the SCt overturned that, content industry proposed levies on tech. Consumer groups formed home recording rights organization. Negotiated serial copy management system designed to prevent impact on general computers—AHRA—which became a dead letter quickly. Computers have to be covered, so there was a copy protection working group.  Bad experience with mandatesàTPM provisions. Should have tied circumvention to copyright infringement; should be limited to expressive uses and allow for uses like vehicle repair. Trafficking provisions inhibit fair use and vehicle repair. Even with an exemption, user can’t get lawfully produced software to take advantage of exemption b/c of trafficking provisions.

Mr. Steve Metalitz, Partner, Mitchell, Silberberg & Knupp LLP
Speaking in personal capacity though represented © industries in negotiations. Implementing WIPO digital treaties: 1201 was for that. On the whole, remarkably successful: lots of access to content, due in large part to TPMs [now wait until he tells you about what the creative environment is like due to 512]—allowed creation of marketplace, not thieves’ bazaar. Congress got this right.  512 was a different story: politically necessary, not necessary for treaty compliance. Telcos wanted payback for the IP carveout in 230. Copyright owners didn’t need or want 512. Goal: provide incentives for © owners and services to work together. Infringement is pervasive and so are infringement-based business models. Close to blanket immunity for OSPs if they take relatively minimal steps out of scale with the infringement problem; can base their business models on infringement. Dominant online businesses have to shoulder responsibility for bad behavior, including IP theft. Increased volume of infringement, increased velocity, need for voluntary arrangements that haven’t yet materialized.

Sen. Coons: This is the most active subcommittee of Judiciary. Core © industries add $1 trillion to GDP. [And core OSP industries?]  Google didn’t exist in 1992. Digital piracy has exploded. Unauthorized copies are just a click away. [Yep, 1201 totally worked.] Digital video piracy costs $29 billion/year and 250,000 jobs.

Tillis: what areas are most contentious? Given where we are now, is there opportunity to bridge the gap with modernization?

Damich: Ethically can’t give opinions on future legislation.

Tillis: root causes of differences?

Damich: did try to compromise on 512 & 1201. Hatch broke impasse: made sure there was a core group of stakeholders then a larger group of interested parties like libraries, reverse engineering researchers, law enforcement were allowed to comment.

Band: 1201, 512, databases all happening together. Most of the focus was probably on safe harbors b/c players were bigger on both sides: telcos v. big content providers. Hard to get focus on 1201/databases. Hard to get researchers to believe 1201 would happen: encryption researchers insisted Congress would never do anything so stupid. More people are familiar now with 1201 and auto repair type problems.

Schwartz: Home Recording Rights Coalition was included up to a point, then excluded from negotiations. We were considered too tough to make a final deal with. Sen. Ashcroft worked with us to work on legislative history. Didn’t anticipate that at most recent triennial rulemaking we’d be there on behalf of farming bureaus. With short growing seasons + distant repair facilities run by manufacturers, a software-dependent equipment breakdown means a waiting list for authorized repair, while independents would do it if they could. It’s not trafficking to obtain and use software; the Copyright Office has gone as far as it thinks it can go in allowing repairers to act on behalf of owners but still needs guidance on trafficking.

Metalitz: also surprised to represent auto manufacturers at the triennial rulemaking. 1201 speaks in very general terms of all TPMs. Stood the test of time better than 512, which depended on the tech of the time. Storage at the direction of the user was straightforward at the time. Other functions were added: indexing, highlighting content for users. [This is ahistorical, albeit algorithmic discovery was in its infancy; indexing and highlighting were definitely known at the time.] Unfortunately courts said this was all ok but that wasn’t intended. Legislation in general terms (except for the exceptions in 1201, which are good) last longer v. tech-specific 512.

Panel II
Professor Sandra Aistars, Clinical Professor Senior Scholar and Director Copyright Research & Policy, Center for the Protection of IP, Antonin Scalia Law School, George Mason University

Informed by clients’ epxerience but views are her own. 512 hasn’t helped creators; the bargain Congress expected was not achieved. Twin goals of quick removal and support for innovation. No blind eye to infringement—red flag provisions have been misinterpreted and eviscerated, especially in 2d and 9th Circuits [those hotbeds of hostility to the © industries]. Perfect 10 v. CCBill—“illegal/stolen” label wasn’t a red flag [to credit card processors].  Viacom v. YouTube limited obligation to knowledge of specific infringing activity, and UMG v. Shelter Capital agreed. Capitol Records v. Vimeo: red flag not triggered by full length copies of © works, even when leaders winkingly encouraged infringement. Notice and takedown alone isn’t enough. Representative list requirement isn’t enforced. Works with artists and small businesses; even trained and coached, students found removal frustrating and time consuming; searches led to phishing sites [note that this is already illegal; these sites aren’t likely to be complying with the DMCA and it has yet to be explained to me why they’d comply with DMCA-Plus.] CASE Act would allow recipients of bad notices to sue in small claims court.  Cox is a good case: they failed to enforce even a toothless repeat infringer policy.

Professor Rebecca Tushnet, Frank Stanton Professor of the First Amendment, Harvard Law School
I would like to start with some perhaps surprising numbers that illustrate what §512 has meant for creativity and innovation online.  I work with a nonprofit, the Organization for Transformative Works (OTW), that doesn’t have a single paid employee. With a budget of under $400,000 per year, it operates a website for user-generated content that currently has over four million creative works from over a million registered users. Those millions of works attract 1.12 billion page views per month. That’s more than double what it was only three years ago, and rapid growth continues. Last year we were honored by the World Science Fiction Society with a Hugo Award for our service to the global community of fans.

Despite the large numbers of works posted on our site, we get less than one DMCA notice per month. Most of those notices are invalid, such as attempts to claim rights in a title or name. At the cost of significant amounts of volunteer lawyer time—something very few services can afford—we carefully review each notice and explain to the senders of the invalid notices why their claim is mistaken. Unfortunately, our experiences with mistaken and abusive notices designed to suppress noninfringing and critical speech are common among service providers.

Empirical research reveals that most of the internet service providers who rely on §512 are like us: Most service providers receive relatively few notices and handle them individually. Only a few entities receive millions of notices. While market pressures and business decisions have led a few large sites like YouTube to rely on automated systems and sometimes on filters, it is important not to treat YouTube as a model for the internet at large. If we did, the only online service to survive would be YouTube.

If there is one message I would ask the members of the Committee to take away today, it is that most beneficiaries of §512 are not like Google or Facebook. Big or small, most sites that use §512 don’t need and couldn’t survive a requirement to use technologically complex mechanisms to filter out the relatively rare infringements.

There are serious problems of market concentration in the content and telecommunications industries, and §512 has been vital to preserving the competition that exists. If Congress changes §512 to target Google and Facebook, or because of rogue overseas sites that already ignore the law, it will ensure that only Facebook, Google and pirate sites survive, making the problem of market concentration even worse without protecting creators. A mandate for filtering, whether called “staydown” or something else, would destroy the small and medium entities that are vital to innovation, creativity, and competition on the internet.

The system is by no means perfect—there remain persistent problems with invalid takedown notices used to extort creators or suppress political speech—but, like democracy, it’s better than the alternatives that have been tried.  The numbers of independent creators and the amount of money spent on legitimate content are growing every year. Changes to §512 would be likely to make things much worse.

My written testimony addresses some other specific issues, including section 1201. Section 1201 is broken: it is mostly used to suppress competition rather than protect copyrighted works from infringement. It would benefit from requiring a nexus between circumvention and copyright infringement. Otherwise it will continue to be used to prevent diabetics from getting information from their own medical devices and researchers from investigating security vulnerabilities in voting machines.

Professor Jessica Litman, John F. Nickoll Professor of Law, University of Michigan Law School
Three points: (1) In 22 years since DMCA, anticircumvention hasn’t lived up to its promise. Immense symbolic value, not effective v. piracy. No additional deterrent to people who are already infringing. DRM is not impregnable and it is buggy—prevents licensed uses. Americans believe that if they buy a copy, they should get to use it. If they need to repair it, they often try to find fixes and usually succeed. Circumvention tools are widely available. Meanwhile, businesses that embed software have harmed aftermarket parts and independent repair. It’s unreasonable to tell a farmer who wants to repair a tractor to petition the Librarian of Congress for permission to do so. That has nothing to do with © piracy.

(2) 512 safe harbor tradeoff worked well for longer than we had any right to expect. Courts have been good at interpreting it. Principal weakness: imagined human judgment applied in good faith to send and receive notices. That didn’t scale for large © owners and [some] large services.  Automated solutions are buggy with false positives/wrongness. Larger © owners and services have negotiated private deals, but smaller services without the ability to strike private deals still find 512 essential.

(3) © owners complain about “value gap” based on market power of ISPs leading to lower license fees. I get why © owners are upset, but this is a bargaining power/competition problem, not a © problem. Antitrust would be a better solution.

Professor Mark F. Schultz, Goodyear Tire & Rubber Company Chair in Intellectual Property Law Director, Intellectual Property and Technology Law Program, University of Akron School of Law
Good intentions—sought to shield infant industry, which is grown up now into world’s most wealthy and powerful businesses; outdated assumptions. Courts narrowed service providers’ responsibility too much. Congress assumed ISPs would be indifferent to infringement, not supportive of it. Infringement notices in billions, files spread fast. Not just one upload, nonstop uploading by many users. Finding one copy is worthless. Many online services profit from availability of infringing material. [I always wish they’d name names. Lots of services that did profit in this way have been sued out of existence, and also Veoh, which didn’t; who exactly do they think, other than Google, is an illegitimate business that survives only because the DMCA doesn’t have a filtering mandate?]  Courts have excused businesses from many duties intended by the DMCA even where they knew of and profited from infringement—not what laypeople would call knowledge. Red flag now meaningless. [This is the specific knowledge/general knowledge distinction: general knowledge that infringement is most likely occurring on your platform is not red flag knowledge.  I wonder whether he thinks the University of Akron has red flag knowledge, defined as he would like to redefine it.]

Solutions: (1) amend 512 to require notice and styadown. (2) Revise to define willful blindness; not encouraged to ignore infringement.

Sen. Tillis: What did you learn from panel 1?

Schultz: Metalitz is right: red flags.

Litman: 512(m) explicitly relieved sites of a duty to monitor; that was a good idea; courts have been really careful to limit knowledge so as not to interfere with 512(m).

RT: These red flag claims are not right. Viacom seeded YT with apparently leaked footage; what counts as red flag knowledge is appropriately determined case by case.  On the volume of infringement: Almost all the sites you’ve heard of, including Wikipedia and Amazon Kindle/digital, have very few problems with actual infringement. Most can’t afford and don’t need filtering.

Aistars: 1201 engendered cross industry cooperation; worked w/entertainment and computer industry to not burden consumers and build new businesses. 512 didn’t engender affordable tech to filter works before they were uploaded. Court decisions around 1201 have hewed to parties’ expectations. We didn’t expect printer cartridges and courts didn’t apply 1201 to them. Lacked same luck for 512.

Sen. Leahy: Success reflected a years long bipartisan process. To RT: is it working?

RT: Yes: Most sites don’t have problems.  Google receives billions of notices, but it is doing a lot/used to get other sites not searched.

Coons: Does 512 discourage sites from gaining knowledge?

RT: No, it encourages them to monitor for other bad things like terrorist content without fear that if they see something else they’ll be liable for infringement.

Litman: also encourages monitoring for ©: if they catch one infringement, 512 means they can’t be held liable for the ones they missed.

Coons for Schultz: is burden on rightsholders to police fair?

A: should be shared burden. YT had scroll showing one Comedy Central clip after another on homepage. Knows content owner objects; that’s wrong.

For Aistars: repeat infringers/whack a mole?

A: (1) isn’t a shared burden to find infringement; (2) there has been some success against ISPs that don’t implement their own extremely lenient policies, but that’s not much of a victory—14 notifications before action is taken, and customer can just start again; sets us up for bad actors trampling small businesses. Have to decide on creating v. sitting at home searching for infringement.

Coons: working together on voluntary measures?

Aistars: STMs are supposed to be accommodated, but there’s no process to ID tech though many could qualify, like Audible Magic. For a # of years, relatively inexpensive [$10,000/month for small websites, yay]. 1201 worked b/c there were incentives for cross industry cooperation. 512=no incentives for anyone. No requirement to work together.

Coons: are there easy modernizations to improve balance?

Aistars: red flag.

RT: (1) apply antitrust to © and telecom. (2) 1201 infringement nexus.

Litman: narrowing 1201 would help. 512: these are computers, they aren’t as smart as us. They don’t recognize what humans would. Can’t tell who uploaded content. Can’t simply assume that online services perceive as humans do.

Schultz: red flag. Willful blindness isn’t just computers; Vimeo says employees can interact w/famous content, but b/c they’re not © experts, can’t hold them to objective knowledge. Every layperson knows where to go to get infringing content. [Is it Vimeo?]

Sen. Tillis: What are decisions that are problematic?

Aistars: See her testimony: Perfect 10 v. CCBill, Viacom, UMG v. Shelter Capital, Capital Records v. Vimeo.

RT: those cases were rightly decided. Perfect 10 was specific to sexual content, Viacom had those “leaked” uploads.  The alternative, filtering via staydown is terrible. Google spent $60 million [correction: now $100 million] on a system that is 60% effective according to the major labels who use it, and we’re better at ID’ing music than at other types of works.

Litman: dislikes some results. In Cox, they lost the safe harbor for not terminating users, which could be right, but Cox was receiving automated takedown notices with thousands of dollar-demanding settlement letters attached. They wrote back refusing to forward the settlement letters. The court considered these to be notices that Cox ignored.

Schultz: dislikes same cases as Aistars. The understanding of what constitutes knowledge defies common sense/most other areas of law.

Staydown should only occur after notice. [I don’t know why people think this is an argument in favor of filtering. The primary problem with filtering is not that there’s no list, it’s the expensive and error-prone apparatus of filtering that most sites can’t afford and don’t need. It’s as if the police came to every business on Main Street and said, we’ll be sending you a list of our 10 most wanted, and if you don’t employ facial recognition at each entrance to catch them you’ll be breaking the law, but don’t worry, it won’t be burdensome because we’ll send you the pictures.]  Affordable filters exist but there’s no market demand; if there were a duty we’d see more systems [magic, or Audible Magic?]. Congress can make exceptions.

Tillis: infringement online is big; number of notices is difficult for new entrants [again, this is not true in the way he thinks—most entrants don’t get big numbers of notices, but quality screening can be a big deal]. Neither is wrong but the DMCA is not working in every case.

No comments: