Monday, April 08, 2019

Section 512 roundtable, part 4: International

SESSION 4: International Developments     

Carlo Scollo Lavizzari International STM Association: (Scientific Technical Medical publishers, also arts & humanities publications.) Internet hasn’t been static. Dynamic developments in Europe from platform liability and safe harbor to one of responsibility, not just from legislation but case law in cases from Germany, Italy, France, Spain, Britain, ECJ. The Q is not how to fix a broken whackamole system but how do platforms discharge their duties based on the risk they introduce, not one size fits all [just two sizes, I guess].

Stan Adams Center for Democracy & Technology: Directive provisions are fundamentally problematic and unbalanced v. 512. The internet is an incredible tool to market content, but it’s also the default option for sharing and expressing between people. 512 is foundational; the EU removed that foundational stability.

Eric Cady Independent Film & Television Alliance: Global problem, massive online infringement w/no way to stop the proliferation of illegal copies.  Encouraged by the new Directive: need to rebalance framework for sharing service providers, which haven’t had incentive to discourage users from uploading infringing content.

Danielle Coffey News Media Alliance: Yay Directive.

Alec French Thorsen French Advocacy: 512 was the price we paid for the rest of the DMCA [which we would now like to not pay any more, though we’d definitely like to keep 1201, thanks!]  Directive distinguishes small and big UGC sites; innovation by startups wouldn’t be impacted by requiring companies w/over $500 million in market cap to secure licenses.  Limiting 512 to startups would not prevent innovation but would stop allowing multibilliondollar companies to ignore and profit from infringement w/impunity. 

Ashley Friedman Information Technology Industry Council: innovation.

Joshua Lamel Re:Create: Coalition members are very concerned: impacts not just on European consumers & innovators but American investment. Many smaller startups can’t meet the startup test; US creators will have a lot more trouble reaching the American market. Two things have changed: (1) profits are up in the creative industries, and (2) piracy is down—those are worth noting. Exponential increase in creators foregoing traditional intermediaries and choosing Amazon, YouTube, TikTok, etc. Approx. 17 million Americans are creating & distributing content online w/o traditional intermediaries, some of which are small businesses; Europe threatens them all.

Stan McCoy Motion Picture Association EMEA: Supports the EU approach to no fault injunctive relief from the 2001 directive. Liability: look to ECJ case law, not new EU directive as the model. The original clarification of communication to the public was good, but we dislike the emphasis on licensing over enforcement and dislike the UGC language contradicting the Commission’s own language.

Corynne McSherry Electronic Frontier Foundation: © policy is speech policy and innovation policy. If we see upload filters across Europe, they’re inevitably going to flag lawful content—we have a decade of experience w/Content ID, in which YT has invested millions of dollars. Still routinely misidentifies birdsong, public domain works, and other types of content. At 300,000 videos a day, 1/10 of a percent is 300 lawful posts blocked per day. 

Katherine Oyama Google: We do agree that the DMCA has allowed an explosion of creativity and economic growth. Enables more than 27.7 trillion in global ecommerce. Balanced US legal approach has driven billions to the entertainment industries that could have been lost to piracy. Global music revenue and box office revenue is up.  Directive will set Europe backwards.  Unlike recently passed Music Modernization Act, which was win-win-win, the Directive poses the potential for massive and dramatic overblocking of content. Details will matter as will implementation. Will work with Member States.

[I really want to hear more about who gets this Directive-mandated license money. It sounds a lot like French photographers are going to get money from online use of photography in France, but won’t have to distribute it outside Europe and in particular won’t have to send it to US photographers whose works might well make up a big chunk of the works used under “license,” if the collective licensing works the way it has to date.]

Christopher Randle Facebook: Reiterate strong support for the US framework. Our measures are enabled by the strong & balanced US approach.  We’re excited by our new tools/partnerships, such as video matching/Rights Manager, alongside with use of Audible Magic.  Important illustrations of how collaboration can work, but only if it’s voluntary, adaptable and flexible. Developing strong partnerships with rightsholders in all segments. 

Steven Rosenthal McGraw-Hill Education: WHOIS has conflicted with GDPR restrictions on access to information.  We’ve seen a number of instances where identifying data previously available was suddenly redacted. We’ve seen a proliferation of content delivery networks like Cloudflare anonymizing providers under the pretense of security. DMCA subpoena provides no alternative solution b/c they lead to useless, inaccurate ID info self-reported by the infringer. Negatively affect rightsowners’ ability to enforce rights.

Matthew Schruers Computer and Communications Industry Association: EU and US approaches starting to diverge.  Sky is Rising shows numerous 3d party industry organizations reporting growth across sectors, showing that notice & takedown is working. The new Directive provisions, by contrast, are out of step with the US/increasing international norm, creating great uncertainty, which we should regard w/skepticism.  Deters investment/creates risks to free speech.

Lui Simpson Association of American Publishers: Encourages CO to take into account effects of website blocking on pirate sites. In Europe alone, 1800 sites have been blocked and the internet hasn’t been broken.  US should have additional meaningful tools to block piracy as a mere takedown is not enough.

Sherwin Siy Wikimedia Foundation: Lots of uncertainty about the new Directive both in its provisions and in its implementations in Member states. There’s tension b/t the recitations and provisions, and tensions w/in the provisions themselves that recapitulate what we’ve been discussing today.  Wikipedia/Wikimedia commons occupy an interesting space in this discussion: large, prominent websites with a small staff and a large userbase/contributor base; exists for specific purposes that aren’t often discussed. Need to talk about sites that don’t fit the model of a general purpose sharing sites.

Abby Vollmer GitHub: Puts the internet at risk; we were able to secure a carveout for open source, but people are building a lot of stuff w/open source that isn’t exempt.  This demonstrates the difficulty and ham handedness of what they’re trying.

Rachel Wolbers          Engine: Startup exemption isn’t workable. Will have to rethink startups worldwide/ restrict UGC.

Strong: how will this affect doing business in Europe, and then how will it affect doing business here?

Lavizzari: (1) to what extent will it codify the ECJ case law on active platforms/structurally infringing platforms that can’t hide behind the user and are carrying out communication to the public. YouTube and Elsevier cases are pending before the ECJ.  (2) different standards for platforms’ responsibilities: will that change?  ECJ has different standards depending on whether you choose to have unidentified/anonymous users—higher standard.  We hope that case law isn’t impaired by the promises of licensing that Article 17 also created.

Simpson: we don’t think Art. 17 is a problem b/c it’s intended to clarify existing EU law. German cases: Rapidshare case was a clear enunciation of the principle that if you’ve set up a platform to facilitate infringement, you have responsibilities. Implementation uncertainties, but fundamentally EU caselaw is sound on platform responsibility and not just mere liability.

Vollmer: Number one problem from our perspective is filtering.  Our carveout for github itself is not enough for software development/innovation. Reality is that the requirements will make platforms filter to avoid crushing liability. Why is filtering bad?  Github is the home of open source software. Software developers who use © by using open source licenses embrace four freedoms: to study, use, modify, and redistribute. Rightsholders sharing code want it to be shared. If you’re going to legislate on this level, think about how content varies and whether your requirements help all rightsholders. If open source software disappears, that harms rightsholders.  And code has dependencies.  If a filter mistakenly detects a block of code, a whole set of programs/functions can collapse.

Siy: Online encyclopedias are only parts of our projects; we’d want to argue that our other projects should also be excluded, but we are concerned about whether that’s a risk. There’s an unresolved tension b/t what it means to make best efforts to obtain authorization for media we don’t intend to have on our platform. Wikimedia commons is devoted to hosting content that is public domain or licensed for free use (not even all forms of CC license qualify). We have no intention of hosting even fair use works.  The Q of what it thus means to seek permission for those uses is an open question—and to make best efforts to ensure their unavailability while also not resulting in prevention of lawful uses.

Cady: Art. 17 isn’t perfect, but it’s good b/c it’s premised on getting authorization, and b/c larger platforms have to prevent future uploads of notified works. 

Q: given worldwide nature of your members, say a little more about what “shall obtain” a license means?

Cady: we license exclusively, so the premise of these platforms obtaining authorization may not work out.  It does have the potential to impact the way members finance their productions.

Lavizzari: one of the beauties of the emerging case law and art. 17 is creates an incentive for rightsholders and platforms to cooperate lacking in 512. Rightsholders do want works to be available [not Cady’s members, he just said], so there are policies that won’t stifle free expression or make works unavailable. We use artificial intelligence to deal w/plagiarism, but there are more sophisticated options/identifiers and we’re eager to work w/incentivized platforms to devise a system that will work for everyone.

McCoy: Berne-inconsistent notification requirements: structure says that OSPs shall be liable unless … they act expeditiously upon receiving sufficiently substantiated notice.  That requirement is a formality under Berne. [Wow.]  We would have preferred to wait until the ECJ ruled.  Second, emphasis on licensing over enforcement: for many rightsholders, the idea of licensing UGC platforms isn’t something they’re interested in b/c they have exclusive distribution models. Enforcement is the key for them. Licensing leaves us concerned about how it’s going to be implemented for the benefit of those rightsholders who want enforcement, though filtering along the lines of Content ID is one promising step [except for all the shit rightsowners shovel on Content ID for not being good enough].

Oyama: Our primary concern is conflict b/t the two frameworks, EU and US. Direct liability for any type of conduct uploaded by anyone imposes lots of fear and risk. One place we’d like to focus on in implementation is making more clear what is sufficient notice for platforms.  Final version had some positive steps beyond Parliament—platforms making a good faith effort shouldn’t face direct liability based on best efforts, but there’s a need for clarity about what that would be.  UGC is hugely beneficial. Even 5 years ago, the concept of beneficial UGC was more controversial; on YT, the vast majority now choose to monetize instead of block, and more than 50% of the revenue we send to the music industry comes from claims against UGC. The risk is harm to US creators as well—68% of US users’ views come from outside of the US. Significant risk of overblocking.

Schruers: Obligations in Directive conflict—ensure the availability of parody (but not satire) and ensure the unavailability of infringing conduct. Unmanageable filtering obligation, as well as an obligation to prevent upload of future infringing works, when the only technologies that even imperfectly do that deal with AV works, but the Directive is a mandate for all works.  Tech that doesn’t exist.

Simpson: The EU is already about proportionality and reasonableness. There’s no general obligation to filter in Germany, but once you’re notified about specific works you have to take measures to prevent reupload of infringing content. There isn’t yet an effective filter for all types, but surely legislation can get us to that point. [!]  We have a framework to move us to find a workable, reasonable, and proportionate solution.  Let’s not yet look to the Directive as a problem; we have 24 months to see if they’ll get it right or mess it up. The case law is already there; this clarifies/codifies it. Extremely problematic, though, is that it seems to say that rightsholders have to put up with whatever is being done w/their content; contrary to the fundamental right to control. When you have no choice but to monetize or take it down, that’s not what © is for. But given that’s our world, we need the tools.

McSherry: copyright owners have the right in many instances to control how their works are used, but those rights aren’t unlimited—subject to limitations and exceptions.  Moving to a licensing regime ignores the lots of uses that don’t need permission. Robots are very bad at telling the difference.  May work for some kinds of content, but it’s not a good answer overall. Art. 13’s negotiation involved a lot of uncertainty about whether filters were required; lack of clarity is a feature though now we are hearing filters were definitely going to be required.  Third, we have a competition problem.  Google and FB will be fine. But the platforms that could emerge & compete in the social media space are not protected by the size exemptions.  An investor will ask how they plan to comply w/Art. 13: the business plan will always have to include the ability to filter, and that’s expensive and unaffordable for many [and again may be completely irrelevant to what the service actually does!].

Wolbers: the concept of having to build different platforms for each country isn’t feasible, plus the cost of implementing filters. We’ve seen in Art. 17 a number of exceptions, and it’s politically popular to say we’ll carveout the small guys, but that actually doesn’t help startups.  3 years, $10 million in turnover, 5 million users—those are tiny & create perverse incentives to stay under the threshold. And when you create carveouts like this for open source/wikis, you are not futureproofing the legislation but are instead creating anticompetitive situations—the legislation now protects incumbents like Google and FB against new entrants. 

Isbell: the purpose of the Digital Single Market was to create a single platform, not 27 different platforms. But don’t we already have that issue? Germany requires you to monitor hate speech. Thailand requires anything derogatory to the king to be taken down. Isn’t that just the cost of doing business? In the analog world, you have to comply w/each country’s safety laws.  [We have a whole lot of law of jurisdiction governing this; unless you actively screen, you can’t stay out of Germany with your website the way you can decide not to sell your widget to a German address.]

Wolbers: we’ve worked with Kickstarter, and Bandcamp, and Soundcloud: a lot of the content doesn’t involve the German hate speech law. It’s not a fundamental shift in the way UGC is uploaded in the same way as Article 17.

Lamel: Not all creators are the same or want the same things.  EU didn’t hear enough from actual creators, and we haven’t heard from enough creators here today.  Europe has very different view of issues like fair use than the US. In over 50% of European countries, there is no educational exception so that showing a YouTube video to the class is technically infringing. Thai king mention: the US should stand up for free speech around the world, not assist in that suppression.

Coffey: Springer, one of their members, gets a © over its news that create efficiencies similar to music. More complicated w/US publishers—will they be able to assert the right in the EU?  They want to!

Adams: This will cause harm in the US. For some subset of startups, the decision will be if I stay in business I build to the most stringent standard, with filtering and licensing, both of which conflict with fair use here.

Lavizzari: you shouldn’t build a startup that will be a victim of its own success. If your business is from works being shared on your platform then you should build compliance in early on. If you have a bakery that doesn’t principally attract users from works of others, you’ll be safe.

[This point deserves more emphasis: these descriptions of social media/similar sites are quite elitist, despite the rhetoric.  Almost every post on my FB feed is a “work,” and those that aren’t, aren’t because they’re too small; every picture on Instagram is a “work.”  He equivocates on “works shared on your platform,” because there are lots of sites that involve lots of “works shared on the platform” of which very few are unauthorized, much less infringing.  So it is probably right to say a bakery that doesn’t allow public comments probably doesn’t have too much to worry about, but that is very much not the point about the burdens on sites that aren’t even causing problems by the copyright industries’ own lights.]

Siy: wikipedias of all languages are blocked in Turkey b/c of a dispute over government’s characterizations. Sometimes we make that call. Distinction b/t © and these discussion of lese-majeste, hate speech, etc. is that those aren’t ex ante.  Anonymity/privacy: wikis are available all through the world and we take privacy very seriously, including in restrictive regimes. The considerations today do exist in a larger sphere in which privacy is relevant.

Schruers: there are sites where content that might violate lese-majeste laws appears even in nations that have such laws. There’s a huge difference b/t that and having to install filters in year 3.

Oyama: we are interested in giving publishers maximum control. Where rights are not waivable, that leads to unintended, regrettable consequences across the board.

Q: injunctions?

McCoy: we support the EU model.  [site blocking] Injunctions plus takedowns can give you flexible tools to address piracy. Not sufficiently implemented in some countries, but still good. By no means cutting off access to legitimate sources of film & TV.

Q: have you noticed a difference in use of injunctions across systems?

Simpson: publishers have taken advantage of European remedy in 6 countries. Main goal is to disrupt the availability of that service in that country. There are limits to the effectiveness of this remedy. Sometimes the operator moves to a different server, but now we don’t have to redo the entire process; can just amend the order to cover the masking sites as well.  Notice and takedown doesn’t need to be accounted for.

Oyama: hard to say we never encounter instances of abuse, where legit sites are targeted. Australian implementation is recently passed; not aware of any orders actually issued. One approach we’ve taken in search: if there is a site blocking order that an ISP receives, even if a link showed up in search they couldn’t access the site, but we also have search ranking demotion that works with the DMCA.

McCoy: MPAA likes the Google demotions. Doesn’t know if there’s any correlation b/t site blocking and take downs. Panels this morning drew out experience on expense of notice & takedown, so rightsholders are selective about what they target. In some cases availability of injunctive relief might provide a way of addressing worst of the worst in markets that aren’t a high priority for notice and takedown.

Q: has heard that Mexico without legally required notice & takedown still sees informal use of that practice.

Schwartz: for a long time, before the Canadian system was implemented, an informal industry agreement enabled informal notice forwarding. We’ve seen that in other markets, where intermediaries don’t want their services to be perceived as places for misuse. Allows more capable services to invest more. Important takeaway: absence of statutory mandate doesn’t mean that services aren’t implementing misuse and misconduct policies.

Simpson: when we were successful in enjoining SciHub in SDNY, Chinese operator actually recognized the judgment and did block the site in China for its subscribers. If no mandate is in place, though, an ISP will choose not to act if there’s no obligation.  [Which is contradicted by …] We sent the copy of the judgment to the Chinese operator, and it was unavailable w/in 2 days. Totally informal—just sent notice. 

Lanza: CRTC denied website blocking and said there were alternatives. Agree/disagree? 

Strong: or thoughts about Canada’s notice and notice regime from 2015?

Simpson: We obviously think that’s notice and nothing. In the past there were some private cooperation agreements that were favorable, but notice and notice alone won’t do anything.  [Stats?]

[I got distracted by a side issue.  I missed some discussion of various countries’ systems.]  McCoy: Belgium ISPs will do voluntary takedowns. 

Lamel: US ISPs are less competitive than other countries’ broadband service providers.

Simpson: One size solutions don’t fit all. We don’t have an adequate definition of an OSP or an ISP. If you just manage the pipes your responsibilities will be very different from an online service provider. Need to be a parsing out of what these types of intermediaries are.

Q: for those of you who want 512 outside the US, do you support an obligation to have a repeat infringer policy? In some countries, they tend to take it to mean adjudicated infringer.

Simpson: Adjudication can take years—strips out what makes notice & takedown workable (though it’s not workable now) [the food is so bad and the portions so small].

Isbell: couldn’t you do both, go to court and notify?

Simpson: yeah but why.

Schruers: statute says repeat infringer, not repeat alleged infringer. That being said, has seen many online services operate a far more strict process that functionally encompasses accusations. That’s reasonable in an arms length relationship in the private sector, a user who causes a lot of problems might lead the intermediary to discontinue services. Many online TOS terminate users long before the statutory definition comes into play. But that doesn’t change the words of the statute.

McSherry: world has changed; people are reliant on internet service in the way they weren’t 2 decades ago. A household can rely on the internet; cutting it off b/c of one person’s behavior has serious consequences that have to change how we think about this. And in the US we don’t have a lot of choices, particularly for high speed services. Our approach to repeat infringers needs a rethink. Bad idea to go in the direction of making it easier to cut people off—far beyond speech, to work and education and other interests.

French: strongest justifications for safe harbor involve need for access to the internet. Promoting startup innovation. But those two justifications apply to online access providers, not to UGC sites, not to digital media services, not to anyone who doesn’t get you online. And once a provider has $500 million market cap and $100 million in the bank, they don’t deserve that protection any more.  [Interesting variation on what the actual carveout in the new Directive is; really makes the point that we should be talking about antitrust law, not copyright law.] Directive is really narrow: only applies to UGC sites that are consumer-oriented.  That would be useful to US law. 

Lamel: these © policy conversations are happening around a larger conversation about internet policy generally. Just as disconnection destroys your ability to participate in the economy/politics, far beyond the jurisdiction of the CO—also privacy and cybersecurity policy. There are other important issues in play. 

Vollmer: it’s really hard to find a working example of a startup that would actually be protected, so be careful when you talk about those carveouts being real.

Amer: do you see internet access concern mitigated by voluntariness of the system?  ISPs have the choice whether or not to participate, but if they do participate they get a limitation against monetary liability. In exchange for that benefit, we ask them to do something otherwise against their economic interest: terminate repeat infringers.

Vollmer: access is not fair collateral damage. The goal of notice & takedown is to prevent infringement. We’ll voluntarily take steps but the cost is great.  Counternotice exists, but it’s such a small fraction for a lot of reasons; accept the cost of things coming down/becoming inaccessible.

Siy: the idea that it’s a voluntary system isn’t practically true, if the alternative is strict liability/statutory damages.  ISP v. OSP: we don’t think that distinction is good. Access to knowledge is important even though we aren’t a conduit.

Schruers: need to distinguish b/t (a) and (b)-(d) services—the calculus for those constituencies is different. We shouldn’t necessarily ask (b)-(d) businesses about the incentives for (a) businesses, who aren’t represented on this panel.

Lamel: Consumers also need to be part of the conversation. Second, we’re seeing integration b/t ISP and content providers, and that has to be taken into account.  Comcast & AT&T have huge shareholders that are content providers. This integration changes the economic incentives.

McCoy: the whole point about not taking away access speaks to the larger balancing of interests that has to take place. Jurisprudence in Europe/ECJ have taken very seriously the obligation to weigh the different rights at stake [though not fair use] and concluding that site blocking that meets certain basic criteria is consistent w/fundamental rights.

No comments: