Section 512 roundtable, open mic

Open mic

Janis Pilch, Rutgers U: domestically it seems obvious that litigation on 512 can’t change the systemic problem of infringement and the impossibility for most rightsholders to litigate. 512 sets up a permanent conflict b/t service providers and rightsholders.  Illusion of balance. Need to amend to create a balance that favors rightsowners.  Second, internationally, the same conflict plays out.  EU Directive makes platforms more responsible versus rich tech companies fighting laws constraining their profits.  Those companies are using dominant economic position, made possible in part by 512, to distort public perception of law through misinformation and through academics & civil society organizations, paid or otherwise motivated.  Also those groups hacked … somebody?  USCO site was compromised in 2016 when written comments on 512 were due.

Q: do you contend that anyone was precluded from filing comments?

Pilch: no. But 92,000 comments were bots.  We see South African reform heavily influenced by US tech interests but presented as creator focused.  

Kupferschmidt: The problem isn’t that © owners shoulder most of the burden, it’s that they have very little to show for doing so, b/c the material goes back up online. Whackamole.  The result of the burden placed on the creative community is that we aren’t achieving the balance Congress intended.  On fraudulent notices: what we are doing to educate creators—Copyright Alliance website has FAQs and does presentations across the country.  [Unfortunately, and this really applies to all parts of this problem as well as many others, it’s really hard to deal with bad faith uses by trying to educate the ignorant; it tends to misdiagnose the problem.]  If you’re worried about paucity of challenges, support the small claims copyright act—it would be a lot less expensive in that context.  On red flag knowledge: no ISP could come up with one example that was red flag knowledge that wasn’t actual knowledge.  Proves our point.  Not sure it was intended to be narrow.  [Obviously I think that’s wrong—among other things, I think my example of getting told about a full copy creates a reason to investigate, but it’s not actual knowledge until I check and see if the report is accurate.]  Small businesses can’t afford to do this, Fourth Estate requires registration.

Q: opinion doesn’t mention 512. [And you don’t need a registration to send a notice!]

Rasenberger: one writer spends 50% of her time dealing with piracy; most give up. Mean author income is $20,000/year for fulltime authors; they don’t have the resources to fight.  There’s an absence of creators here b/c this was billed as an update on the case law and most creators don’t know the law well.  Step back and decide whether we want to protect © as a country; if we do, we need to amend 512. This is about who bears responsibility and risk; big ISPs have drained money out of the content industries.  The balance isn’t working.  We need no other proof than the transfer of wealth that’s already happened. The EU has the courage to take it on; we can too.  512(j) isn’t used b/c the relief is so narrow and b/c of the uncertainty over its application given what courts have done with other sections.

Google’s site demotion has been helpful to authors to do massive takedowns, but it doesn’t address the problem when the users know the name of the site and can just type it in.

Pariser: we may never get to STMs, but we should know factually what’s out there—that could help move the conversation so we won’t be discussing whether filtering tech exists in the abstract.  In the reply comment notice, CO asked are there any neutral principles? Our answer 2 ½ years ago was: look to notices. There’s a huge amount of notices, not dropping, and it can’t possibly be the case that the system is working, because if it were, piracy would be dropping, which would lead to fewer notices. Now notices are dropping, and we have changed our minds about what counts as evidence.  Notices are dropping not because piracy is dropping but because © owners have notice sending fatigue, and b/c Google’s demotion system has led © owners to focus on sending notices for top of search.  Piracy landscape is shifting from p2p to streaming and other forms of piracy, resulting in fewer notices. Torrents can generate 10s of 10,000s of noticeable links for a particular work, and you can send those notices if you have the resources. As piracy shifts to streaming, that will be 10, 20, 30 links b/c a site is doing all the aggregating for you. 

Q: does that make the notice system easier to enforce against streams?

A: of course not!  Titles still repopulate instantaneously. There are more streaming services and they proliferate easily. Finally, look at money: tech services are paying billions, and a rising tide is lifting all boats.  But the fact is that’s not true. Tech companies are making vast amounts of money, and becoming most profitable businesses, while content is shrinking relative to what it has been. The industry is worth $1 billion on an adjusted basis and it would have been worth $21 billion [if nothing else had changed but we extrapolated growth from the highest-growth period of the industry and people still bought records and didn’t play video games]. Without piracy, it would be different.  ISPs are spending a tiny fraction of revenue on takedowns, response to notices, etc. 

RIAA, Vicki Schekler [sp?]: Counternotices/notices to search engines—our experience is quite different as shown in comments submitted in the past.  96% takedown rate with Google, and 4% are those weren’t ever indexed. We send millions of notices annually.  Pariser mentioned the evolving nature of piracy, and our members experience streamripping.  Pirate sites circumvent DRM and then distribute the audio. Some of these sites don’t have a static URL so there’s no deeplink notice to send.  We’re happy to see recording revenues starting to rise again, but they’re nowhere near their peak in inflation adjusted dollars.  [I wish Glynn Lunney were here.]

Hatfield: Downward economic pressure that free access places on the entire ecosystem for creating music. No one can compete with free, especially not with our own music.  Eric Priest: what happens when © owners can’t monetize works at consumer value points—example of the music/movie industries in China—inability to monetize copies of works harmed monetization opportunities for smaller producers; market signals sent to producers are distorted; producers are disproportionately exposed to peculiarities of markets/exploitation by intermediaries.  The dystopian future is unchecked piracy + consolidated platforms, despite crocodile tears about startups. If & when a winning platform reaches monopsony standards, it will have little reason to maximize royalty payments.  If music is devalued anywhere, it’s devalued everywhere.  No musician now has more live gigs than they used to have. Famous acts now charge opening acts for exposure rather than paying them. We thought we were replacing greedy record company executives with the internet, but at least they invested in us.  [Where is that $6 billion Google pays going?]  Now they want to take a percentage of our tour money instead of giving us tour support.  Who wants a device devoid of access to interesting content?  We want a fair percentage of the revenues our works generate. 1998: music business was $15 billion; last year it’s $9.8. Now the internet is a trillion dollar industry.  Pay us less than 6/1000 of a cent per spin, and it’s less on YouTube.  [Really, read Glynn Lunney on this issue.]

Goldman: statement that no one can compete with free was contradicted by Polis who told us exactly what he does.  Has been confused about red flag discussion: 9^th Circ. has cleanly held that third party notices can constitute red flag knowledge: Shelter Capital, etc.  There’s been a lot of FUD today about red flag knowledge.  Google and FB are not the internet! There’s a whole lot of internet—the regulatory temptation is to think Google and FB need correction so we should regulate the whole internet. Please don’t do that.  Finally, 230 relationship: 230 excludes IP, but important to remember that it protects all kinds of curation and steps about what to publish, what to prioritize, what metadata to show. That’s helped sites understand what they can and can’t do.  512: if it starts as third party content, it should stay as third party content unless there’s evidence that the user no longer wants it up.  If the site makes the publication decision, not the user, then the site should be responsible—Batzel—but 230 is a good model for understanding what makes the transition from content submitted at the direction of the user to content not being at the direction of the user.

Levy: Representative list/red flag precedents may mean that notice has to ID the location. Lenz may also prevent the use of automated systems. This means publishers and songwriters are effectively prevented from protecting their works and many have given up. [Why hasn’t the Lenz liability risk deterred all those millions of automated notices to Google, including the 4% that weren’t ever on Google?  Interesting to hear from RIAA about what gives them such confidence to send so many notices that don’t identify an infringing link on Google even after Lenz.] We’ve heard that filtering works [and that it doesn’t]. We have a problem and the tools to fix it, and we need to rebalance the DMCA with the EU Directive as a positive roadmap to shift the burden of policing the internet from the © owner to the user.

Lavizzari: European countries don’t have anything to learn from the US on human rights, especially economic/social rights and healthcare. [I disagree in relevant part but appreciate his clear moral stance.] Our European report: what has been done in the Directive is relatively little. What’s not been done yet is implementation into national systems we’re trying to harmonize.  New art. 17: understands the concerns of tech companies & civil society organizations, but it’s also an issue of harmonizing secondary liability law, which we don’t have b/c we don’t have a common tort law. These are not perfect provisions, the result of very complex lawmaking. Harmonization is driven by cultural politics—France, which is not the most conservative gov’t in the union, negotiated the Directive through its Ministry of Culture, not Economic Growth or Development.  That’s a significant difference from the US. What motivated the Parliament majority was not the former art. 11/13 but the new exceptions, which get significant play in the new Directive—educational exceptions and text/data mining exceptions.  Read the final Directive—what motivated Parliament in the overall approval was the new rights that are being granted to authors and performers—the idea that the Parliament is run by socialists/democrats and it’s suspected that the majority will not be there at the next election. Platform regulation, data transparency, etc. are coming in the next iteration. 

Wolff: Notice & takedown isn’t adequate; professionals can’t spend their lives doing it. Need to stop infringement and encourage licensing that works.  Discourages real activity between content creators and ISPs. [Again, note who counts as “content creators”—in this version, not the people using the sites!] Filtering works; image recognition works. ISPs worry that they will lose protection if they do too much—perhaps should clarify that they won’t.  And should define STMs better b/c TMs aren’t done by broad consensus; they come out of different sectors that are familiar with their own type of content. What works for music might not work for visual arts.

Greenberg: does the statute allow industry specific STMs?

Wolff: unclear. Developed by broad consensus in multi-industry standard process. That’s not how it has worked.

Q: couldn’t multi-industry just mean ISPs + visual artists etc.?

Wolff: possible but it hasn’t happened in 21 years.  DMCA has discouraged platforms. There could be better content if there was more curation/working with content creators.  [Also I would like Google to give me a pony.]

Gellis: Q about the data from Sky is Rising: industry data; sources listed in report.  These aren’t hypothetical, idealized citizen creators; these are real people who need access to these platforms.  Mavrix: she thinks this was a wrong turn, given that the follow up decision moved away from it. Error was “at the direction of” the users. That’s creating the universe. The platform may then shrink the universe of what content is stored, but that doesn’t change whether what remains was “at the direction” of the user.  What we see is a conflict developed b/t 230 and 512: fear of liability leads platforms to crack down on speech—widespread damage from FOSTA/SESTA where large swathes of legal, lawful content were taken down.

Isbell: didn’t 230 already exempt criminal activity?

Gellis: why did they bother w/FOSTA when 230 already did the job? Good question!  Some ISPs thought that the unnecessary statutory change created enough uncertainty about the immunity that platforms reacted by sealing off areas, e.g., Craigslists adult personals.  We should be very reluctant: small changes have huge impacts on speech.  Finally, on Mavrix: moderation in general shouldn’t disable 512. The idea seems to be ‘you’ve seen the infringement’ but the same challenges that happen with any takedown notice appear—was there ownership, was there fair use, was there authorization—LJ moderators don’t have access to that information. For these protections to be useful & valid they have to be robust & reliable.

Band: agree w/ EU colleague: he supports Medicare for all.  (1) Don’t ignore the societal context of access to the internet. (2) note that we are living through an explosion of great content.  Netflix & Amazon & podcasts—we’re overwhelmed with content.  Rightsholders complain about piracy, but they also complain about how much competition there is from other musicians, photographers, etc. The barriers to entry have lowered, and that’s not a bad thing, especially from the CO’s perspective.  (3) Publisher’s right: they claimed not to want to regulate facts, free expression, quotation right, access to news. They did say that 4 words from a headline would be an infringement. That’s terrible.  [Note that they tried and failed on the same thing with databases; let’s not take our cues from that—or let’s take our cues from the US treatment of the sui generis database right, which is to say ignoring it out of embarrassment.]  EU Directive does have a couple of good things: preservation right for cultural heritage organizations, and contract overrides protecting exceptions from being removed by contract.

Gratz: (1) MP3tunes case: a Q about what a real world example of red flag knowledge; that case gives us one.  They knew that Beatles songs weren’t licensed anywhere; that was red flag.  (2) Long v. Facebook, a month ago: reflects the flexibility of courts taking into account the different facts. 

Greenberg: how unique was that case to the facts? They said 5 days was expeditious, but FB was receiving over 100 notices from one user.

Gratz: FB was receiving a lot of different communications from this user, and asked him which he wanted to deal w/first. That’s specific but there may be other situations, including ones where 5 days is too long and where that’s way too short, for example when there are layers of service providers. 

Castillo: definition of ISP is too broad.  It’s difficult to imagine any online service the definition wouldn’t encompass, per one court. AAP has proposed including an element of good faith into the definition of service provider for eligibility for the safe harbor.  [Yay litigation costs!]

Carver: Snoop Dogg says 360 deals involve the labels taking 360% of everything!  He works w/Google’s Content ID team & wants to give his experience. Across all Google products, we probably receive more abusive notices in a week than everybody else in a year. If we didn’t screen, Justin Bieber would be off YouTube.  One example from very recent history: Feb. 11 with “YouTube Extortion”—small game creators got fraudulent takedowns. We were fooled initially; we removed the videos/applied strikes. But once there was publicity we saw the fraud. When one provider decided to automate its process, over ½ of its notices were abusive.

Q: do you abide by them?

Carver: we try to detect and prevent them.  Another point about counternotices: 2% on YT is copyright removal request instead of Content ID—between 1-2% of removal requests get a counternotice. Small creators really are scared to counternotify even when they think they’re in the right. Polis is not particularly common among small creators in willingness to counternotify. But we also see some counternotification problems; we review them even though the law doesn’t require it.  If a takedown is from a composition copyright owner, then your objection that you sang the song yourself won’t work—we refuse to forward that counternotification, which is now more than half of the counternotifications we receive. To spare rightsholders from obvious misunderstandings of the process.

Vast majority of YT users never get a strike; the vast majority who get a strike only get one strike. Of those who do go on to have 3 strikes, the vast majority reach that point w/in 90 days of account creation. Two very different groups: people who don’t understand much but want to do the right thing v. those who are dedicated. Having one policy on repeat infringers doesn’t address that separating equilibrium.  Reasonable in the DMCA allows us flexibility.

Willen: The idea that we should redefine ISP to add good faith is inconsistent w/ the law—Shelter Capital, Fung etc already disable services that induce infringement from access to the safe harbor. 

Q: why not exclude all bad faith actors?

Willen: need to know what good faith is; opens the door to something impossible to implement. The Q of whether a service acts in good or bad faith can be answered through the way the courts are already applying the standards.

Troncoso: adversarial content v. tech tone is bad—not a zero sum game.  There’s a tremendous range of diversity on the 512(c) side alone. We’ve heard about a few providers, but a huge range of others could be threatened by sweeping changes. On the users: for particular users the DMCA works/doesn’t work for particular reasons.  Prof. Tushnet has talked a lot about the fan fiction community and how filters could be problematic for them; the same from open source software developers.  Bear in mind stakeholder diversity w/in categories.

RT: what counts as content stored at the direction of the user? When I’m on the bus home and I pull the cord, the bus driver stops at my direction even though she is the one hitting the brake and opening the door, and she may even decide not to stop immediately depending on the conditions; it's still at the user's direction when she does.