Monday, April 08, 2019

Section 512 Roundtable part two

Sofia Castillo Association of American Publishers: still bad.

Stephen Carlisle Nova Southeastern University: has one client, a small music publisher, and can’t keep stuff down.  Red flag knowledge should come from having a full song + album cover on YouTube.

Caleb Donaldson Google: DMCA is balanced and supports $6 billion of YT ad revenue to music industry alone, along with 17 million Americans selling their creativity online.

Kenneth L. Doroshow Recording Industry Association of America: Not good enough. Motherless: evident problems w/takedown practice didn’t preclude safe harbor.

Douglas T. Hudson Etsy: 2 million microbusinesses & creators not fully represented in these discussions.  Have seen dramatic increase in false takedowns, phishing and scamming, and other abuses.  Have seen false counternotices.  Need more teeth in antifraud, including 512(f). Should simplify DMCA for microbusinesses w/small library of materials. Prefilters don’t work for everyone. When you deal in physical goods, creative services that don’t match digital content, it’s not a viable solution; 512’s flexibility allows all creative endeavors, not just digital audio/video

Keith Kupferschmid Copyright Alliance: mass © infringement is regular and ISPs are routinely shielded from liability. 2 years: more of the same; red flag has been written out of the statute. Repeat infringer decisions are no panacea. Fourth Estate compounds the problem by adding a registration requirement to the DMCA notice requirements.  New limits on WHOIS database = we are worse off than two years ago.

Arthur Levy Association of Independent Music Publishers: Whack a mole is still a problem.  Lack resources to enforce rights=DMCA offers no remedy. Should seek legislative rebalancing.

Peter Midgley Brigham Young University: we’re content owners and a big ISP for our students/faculty; administrative burdens and uncertainty following Cox and Grande are problematic.

Sasha Moss R Street Institute: DMCA is better than the alternatives.

Mary Rasenberger      Authors Guild: Ebook piracy is big.  512 isn’t incentivizing cooperation and notice and takedown is ludicrous. Authors’ main issue is that we can’t address ebook piracy sites—hide behind 512.  Switch burden to ISPs as EU Directive does.

Amer: repeat infringer policies: general Q: to what extent have recent decisions affected or clarified the state of the law?

Midgley: we want clarity on what that policy should be. We do our best to forward the notice to the user, but we can’t always do it given the dynamic nature of our network. Students get a scary legal notice & show up in my office denying any knowledge. Now the question is: what’s my obligation?  I could start having hearings but I don’t know what burden I bear as an ISP.

Amer: elaborate more on the notices. P2P?

Midgley: We get notices styled as (c) but they’re really (a)—this is content that flowed through our network but we have no way of verifying whether it was actually there since we don’t store the content, as per 512(a)’s requirements. We can refer it to our Honor Code department and allow a student to dispute.

Smith: you take them as data indicative of infringement.

Midgley: we have, but we don’t know what’s an adequate repeat infringer policy. We’re doing what we think is reasonable and hoping we’re eligible for the safe harbor.

Amer: would you favor more specificity?  One size fits all has come in for criticism.

Midgley: 512(e) for educational institutions is basically worthless; he’d like clarifications.

Isbell: does your university either post a policy or get back to rightholders after a complaint?

Midgley: it’s posted online and visible to students and public. We have Higher Education Opportunity Act, which has © specific provisions and we send out an annual notice to every member of the community making them aware of the policy etc. 

Band: libraries are another kind of ISP—the place where many people get internet access is the library. It’s very important to recognize that the standards for Verizon & Cox are not the right standards necessarily for policy across the board.  We don’t see a need for statutory amendment; we think the language as is provides enough flexibility with “appropriate circumstances.”  Internet access has a constitutional dimension but also a practical concern—30-40% of the population has broadband only at the public library because there is a lot of poverty/lack of coverage in America. Access to the internet goes beyond the 1A to the ability to function in society. You can’t apply to Medicaid/meet work requirements unless you file routinely online. That assumes you have access!

Smith: should libraries educate about copyright?  You have to repeatedly infringe to risk termination.

Band: libraries take education seriously, particularly in higher ed.  When we’re balancing issues on terminating internet access, it goes beyond “the First Amendment” to life, liberty and the pursuit of happiness.

Isbell: Does LCA view the fact that libraries provide the physical facilities to access the internet as making them 512(a) ISPs?

Band: yes.

Greenberg: do you still think ISPs are applying a higher standard than the law requires? Last time, ISPs were saying that repeat infringer means adjudicated infringer and Cox said otherwise.

Band: there’s lots of different kinds of ISPs and they have different opinions. It seems to him that an infringer is an infringer, not an alleged infringer, according to the plain language of the statute, but courts do seem to be going in a different direction and they have Article III authority.

Amer: What do you want from us?

Band: Not congressional intervention, as long as courts don’t start restricting “appropriate circumstances” to allow libraries and universities to craft specific policies.

Amer: statute contemplates that at some point repeat infringers will be terminated.  Take your point about the need for that to vary depending on nature of ISP, especially given the importance of internet access, but statute does seem to contemplate. Do any ISPs not need to comply?

Band: No, but what’s appropriate for a public library/university may be different from what’s appropriate from a large commercial provider.  [So very tempted to try to get a notice sent to the CO’s public internet to make that point, except that I don’t actually know where the kids are getting their downloads these days.]

Castillo: Disagree w/notion that it was difficult for Cox & Grande to implement a repeat infringer policy. Cox had a policy & decided not to implement it; Grande just decided to ignore notices.  It’s not about difficulty levels.  ISP should meaningfully & consistently enforce its own policy, whatever that policy is.  (Cox.)  Grande: ISP should be keeping a log of repeat infringers to reasonably implement a policy.  ISPs should prevent terminated users from opening a new account using a different email address or username. [This is one of the things where operating a website gives you a very different perspective. It’s not that damn easy to prevent a terminated user from resurfacing.  In fact it can be impossible.]  512(i)(1) requires ISPs to inform users/subscribers of a repeat infringer policy. The policy “anything legal stays” doesn’t convey to users that there’s a potential for termination if they repeatedly submit infringing content. Motherless was also wrong in allowing personal judgment of operator, w/o a log, to satisfy the statute.  The Cox decision requiring meaningful and consistent enforcement is more in line w/Congress’s intent to share responsibility.

Midgley: Unlike Cox or Grande, we do receive notices from subscriber that there is no infringement. What do we do with conflicting info? Whose word are we supposed to take?  512 refers to system or network; we provide a network and also provide a system, which is how the students access their education.  Do we have to terminate both of those? That’s an important distinction for us.  Nonprofits are notoriously risk averse. Uncertainty makes it v. difficult for nonprofits to provide a robust environment that we all depend on.

Donaldson: Cox & Motherless show judicial involvement in appropriateness tailored to size of platform. This shows how it would be a bad idea to write a single regulation for all. The size/resources available dictate that repeat infringer policies have variation—even w/in Google’s 512(c) products, of which there are many, we tailor policies to the purpose of the platform.

Doroshow: Importance of repeat infringer policy: very important, but just to make the point that there was a failure here, the rights owners had to send millions of notice. Unreasonable up front burden on © owner.

Amer: how do you provide this information?

Doroshow: variable: we send DMCA notices to 512(a) and 512(c) providers.

Rasenberger: best practices could help for repeat infringer policies and for red flag notice.

Moss: some possibility for CO to offer assistance to rightsholders.

Smith: our public information office does answer 100,000s of queries.

Kupferschmidt: Doesn’t disagree w/one size doesn’t fit all for ISPs, but that’s also true for the creative community. Doesn’t work for the notice system either. 

Q re human moderation.

Band: muddied the law—Mavrix went in a bad direction, Motherless improved it. Treading in very dangerous area—the issue of moderation and what’s appropriate goes way beyond copyright to what we want the internet to look like.

Smith: 512(c) says ability to control.

Band: but you can’t condition eligibility on monitoring under 512(m): Congress in 1996 and 1998 wanted no monitoring requirement; encouraging people to moderate content is good.

Amer: But you think Mavrix muddied things.

Band: yeah, there’s a spectrum. If you let most stuff through, that doesn’t seem appropriate to call the ISP the publisher. If you screen out 90%, then it starts to look more like a publisher.  Motherwell to some extent corrected the broad suggestions of Mavrix but we don’t want to put platforms in the impossible place where if they try to look at whether the stuff is appropriate they lose the safe harbor.

Castillo: Motherless was “anything legal stays” screening. The court found this still user-directed storage. The screening in LJ was different – it was substantive.  [This is a non sequitur.  Whether the content is legal is substantive, it’s just smaller.]  If the ISP is screening for substance but not for infringement, it may lose the safe harbor.  We also disagree w/512(m) interpretation—the intent was to protect privacy by preventing ISPs from violating privacy laws when they were pursuing efforts to address infringement, not to keep them from having any obligation to monitor.

Greenberg: Does that mean if you’re screening for child porn and snuff films, do they lose safe harbor? If not, what do you mean?

Castillo: screening for illegal content was what Congress could not have meant to discourage. If you’re screening for illegal content including © infringement they shouldn’t lose their safe harbor.

Greenberg: but what if you’re only screening for porn/snuff [not all of which is illegal]?

Castillo: that’s a closer question. 

Carlisle: to get songs heard, I put my songs on Reverb Nation. I had to warrant that my songs were noninfringing.  A lot of problems w/red flag knowledge could be solved w/looking at whether user claims to be the owner. [I wonder how many websites Carlisle thinks don’t have this in their TOS already.]

Amer: doesn’t Google require people to affirm they have the right to upload?

Donaldson: yes, it does.

Strong: what happens when people use ContentID—how do you connect copyright owner and alleged infringer to take their dispute offline? 

Donaldson: ContentID resolves 98% of disputes on YT; Copyright Match allows smaller creators to find matches and file takedown notices.  400,000 smaller creators; continuing to expand eligibility. We’ve seen good results. Beyonce songs: a demonstration that the label wants the songs on the platforms. They’re licensed under ContentID. If Beyonce monetizes a fan upload, we’re happy to help with that.

Smith: is it clear Beyonce opted to leave that up?

Donaldson: not easy for public to find out, but we have 1000 deals w/music groups, and the vast majority is licensed. There is a huge music industry problem w/incomplete data: labels, collecting societies, etc. can’t or won’t tell you the list of © they represent.

Levy: Content ID and Match rely on representative lists, which is fine for publishers w/direct arrangements w/YT, which our independent publishers/songwriters don’t have, so they can’t submit a representative list.

Donaldson: Content Match/ID doesn’t rely on a representative list, but on ingesting the music itself.

Amer: Individual creators complained that Content ID wasn’t available to them. Has that changed? [As I recall, G’s position was that it wasn’t true then.]

Donaldson: growth of 3d party aggregators; Copyright Match as better tailored to small creators.

Amer: why not Content ID?

Donaldson: it’s inordinately powerful & complicated. Allows partners to specify threshold amounts they’re willing to allow use.  Even from our partners, we’ve seen a user who isn’t experienced take down or wrongly monetize a broad swath.

Q: could you take it all down through Content ID instead of monetizing?

Donaldson: yes.

Doroshow: if you screen for illegal content but not © infringement, what is the rule? Our position is that if the ability to screen is there, then you have the obligation to do so. [Sigh.]  The availability of these tools exists—there are other solutions than Content ID.

Greenberg: last time we did the Roundtable, you said $60 million investment in Content ID; now $100 million.  Is that right?

Donaldson: it was more than 60 million; he thinks that’s accurate. 

Greenberg: he’s sure more has been developed [though we have nothing specific on that, but ok], but what else has changed?

Donaldson: Content ID is not static; subject of major ongoing investment.  $40 million over 3 years seems like a reasonable number to him. Are they STMs?  No, not under the statute. They’re not in widespread use. 

Hudson: Dealing w/long tail—small creators, nondigital content—filtering is just not going to be comprehensive. So now how much filtering will be enough? We’re moving the question but the uncertainty still remains. That’s why flexibility of the current regime needs to be taken into account. Changing to add a filtering requirement won’t solve the problem.

Amer: how do you respond to the argument that you could filter entire works?  Why couldn’t filtering tech capture full works?

Hudson: what if the full work is a quilt?  You’re thinking about digital content, but a lot of the content shared/discussed isn’t digital. The picture may be digital, but not the content. Inordinately complex once you get beyond a full copy of a movie or an audio work, where tech work has been done. The point is that there’s a huge long tail that is significant in volume and that the tech doesn’t work on.

Greenberg: full image of a movie poster on a t-shirt—could it be screened out?

Hudson: it depends on the tech. There are also issues with things that are old and things that are new.  A vintage T-shirt or poster; we’re not in a position to know what’s correct.  [Also see 113(c) protecting use of images incorporated into useful articles.]  We don’t handle goods, don’t do drop shipping, don’t handle content—don’t think Zazzle is relevant. We do have repeat infringer/counterfeiting policies.

Kupferschmidt: filtering: perfect shouldn’t be enemy of good. There’s a middle ground. There can be monitoring that can be done in a way that takes into account different concerns & types of examples.  Full movie that isn’t licensed to anyone: if you notify a platform that shouldn’t be up, that should never require a takedown again [wants notice and staydown, but this time without any messy congressional intervention].  Photo ID: why don’t we ask the user whether they consider this fair use if they’re trying to upload a photo with a watermark.  Sites use CAPTCHA so they could ask you this.  [I wonder how awesome he’d find this if applied to every post he made on FB.]

Band: Repositories may want some degree of moderation to make sure they’re getting the right content. Should not lose your 512 safe harbor for checking whether the content is appropriate for the site—SSRN for example checks to see if it’s basically an academic paper. 

Amer: that sounds like volitional conduct.  [I don’t think he is familiar with SSRN.]  If I were just to email you some materials, and you post 100% of them, there’s an argument that you’re the one who says yes or no. 

Band: 512(c) purposes: that’s at the direction of the user.  [If I pull the string on the bus for my stop and the bus driver opens the doors, I would say that’s still done at my direction, even if the bus driver needed to push the button to open the doors.]  Very different from traditional publishing model.  They’re not choosing 1 out of 100 or 1 out of 1000. 

Rasenberger: Terms of service aren’t enough for affirming that you have the right to post. Whenever you upload to any site, you should have to say you own it, you licensed it, or you believe it’s fair use.  [Again, I encourage her to apply this to her own internet use—and by the way, since we’re talking all 512 here, we’re talking about her emails too.]

512 is really ineffective against bad actors.  [So we should break it for the good actors.]  ebookbike: founded Pirate Party in Canada, principal members of Copyism religion—the sacrament is copying is a sacred duty. Hides behind 512.  To upload content you have to become a member, and most members are part of/related to the Piracy Party.  Instructed on how to buy ebook, strip out DRM, upload it, then return it. We’ve sent notices to the site, the server provider [it switched servers], and to Google.  Can’t do anything but litigate, and litigation costs too much. [It really sounds like 512 is not the problem.  Suppose there were an EU style law—you would still have to litigate against his defiance.]

Isbell: do we really think Congress intended to cover those sites?

Rasenberger: of course not, but Viacom/Youtube and Veoh make it possible for the bad actors to be protected. It’s possible that we can win a lawsuit, but that’s expensive.

Isbell: but is the answer to get rid of 512 for everyone?

Rasenberger: Best practices; Congress should clarify red flag is not just knowledge of a specific infringing item at a specific location—knowledge that your site is a place for piracy should take you out of 512. You should be able to win that on summary judgment.

Amer: 512(f)?

Moss: Filters aren’t working the way people say: EU parliamentarian had her own content taken down as infringing even though it was a public speech. 

Levy: Lenz requires us to consider fair use before sending a takedown but doesn’t say what that means.  A timebomb for small publishers/ songwriters who may have massive amounts of infringing examples out on the internet, to engage in a 4 point analysis before sending each notice is very expensive.

Smith: do you interpret that case as imposing one size fits all standard or does it matter who the © owner is?

Levy: the ruling doesn’t help us.

Greenberg: automation in making assessment? We heard some sense that there’s room for automation.

Hudson: as intermediaries, we don’t have access to the information that the users or the senders have.  That’s why 512(f) enforceability is important for platforms to enable users to express their own creative content while protecting other © owners.

Carlisle: for small creators, independent musicians, Lenz is good news/bad news. Fair use is incredibly complex, and it’s burdensome to do that analysis before sending a takedown when sophisticated companies profess they have no idea whether something is red flag.  It’s easier to figure out something is infringing than it is fair use.  [§107 says a fair use is not an infringement of copyright.]

Amer: isn’t that the premise of the statute, that the burden is on the © owner to send the notice?

Carlisle: yes, and that’s unfair to make them monitor the entire internet.

Smith: 512(f): if it’s complicated, then you shouldn’t have a problem with an honest mistake.

Carlisle: material misrepresentation/ultimate standard is not clear.  Lenz court struggled w/whether Universal, a very sophisticated actor, was misrepresenting.

Smith: but for the little guy, isn’t “knowingly” very helpful?

Carlisle: yes, but the creative person may know something about copyright—musicians can get very aggressive about asserting © and sometimes they’re right and sometimes they’re wrong. A lot of musicians hear any similarity as infringement.

Castillo: takedowns of legal content by filtering: for those cases we have the counternotice system that is working. [Citation needed.] That’s not a reason to avoid filtering. [Also, filtering doesn’t actually give you access to counternotice as currently set up.]

Band: Courts are very good at figuring out who’s a good guy and who’s a bad guy; they tend to find ways to hold bad guys liable. Rightsholders aren’t always as careful as they should be in selecting defendants or misperceive who’s the bad guy (Google, HathiTrust).

Rasenbergers: good actors [with lots of money] can keep pirate books off their sites. Amazon is pretty successful at keeping pirate copies off their sites.

No comments: