Shyamkrishna Balganesh, The Uneasy Case Against Copyright Trolls
Three things: unpack the idea of copyright trolling, an undertheorized idea compared to patent trolling. What trolling is, how it’s recent, and how it’s a bit surprising it took so long after the 1976 Act to emerge. Second, show that it’s bad even though what the troll does in in strict compliance with the letter of the law. Third, offer a solution for the troll’s behavior.
Patent: troll is a NPE (nonpracticing entity) that only wants to monetize, not work or commercialize, the patent. Copyright troll: interest in copyright entitlement is limited almost entirely in enforcement, not distribution, use or “working” of underlying work. Interest in monetizing infringement. First copyright troll: England, Harry Wall, used power of attorney privileges to collect royalties after artists/composers died; didn’t remit royalties to them. Described as “not caring for the work” or the creative process.
Structural incentives from the 1976 Act. Independent standing for exclusive licensee; copyright bundle can be broken down into an infinite number of idiosyncratic subrights called exclusive; statutory damages.
Righthaven: locates infringers, obtains exclusive assignment from owner, commences action for statutory damages. Brought about 285 lawsuits and settled most. Everyone dislikes this, including courts, and they find ways to shut it down.
Righthaven’s mistake: strategic alliance agreement. Exception to infinite divisibility: in 9th Circuit, right to sue cannot be assigned in itself. Turns out, the Stevens-Righthaven agreement has language that can be construed this way; a number of courts find that Righthaven lacks standing. Righthaven loses, is sanctioned, go under.
In trying to fix the problem of Righthaven, courts lacked a long-term solution to what it was doing. A smarter troll with a better structured agreement would have gotten away with it.
Why is copyright trolling bad?
It doesn’t normally implicate fair use. The kind of actions RH was going after were hard to classify as clear-cut fair use; otherwise RH might have lost fairly early on. They involved verbatim copying, often without copying. RH wasn’t just over-enforcement; we’ve seen that before, with for example the RIAA’s action, which were considered [by whom?] legitimate over-enforcement. The problem seemed to be the identity of the plaintiff.
Why should it matter who sues?
In copyright theory there is a presumptive alignment of incentives to create and enforce. When that alignment breaks down, there are problems. Because copyright is structured as private law, there’s a hidden equilibrium that can be broken by trolls. The equilibrium provides a norm for coordination between owners and users, which the troll disrupts.
Copyright’s primary purpose is incentives to create. Copyright’s ability to protect incentives depends on enforcement and its costs. Incentive to litigate arises when payoff is positive. If infringement is widespread and litigation is too expensive and the creator knows that, the incentive to create diminishes. Well-known in tort law.
As private law, enforcement decisions are delegated to rights holders. This creates sorting. Type I: actionable as a matter of law and enforced. Type II: actionable but tolerated. Type III: Non-actionable as a matter of law (paradigmatically fair use). The ratio of I and II to III is determined by Congress entirely. [This seems quite wrong to me; among other things, the ratio depends on what people are actually doing both in terms of infringing and making fair uses, and they do those things based on various considerations including but not limited to likelihood of enforcement. Either that or the definition of “ratio” is not mathematical at all.] The principle variable for I and II is litigation costs.
Breathing space is injected by Type II, determined by copyright holders themselves. It’s not about optimality but rather a reflection of what copyright owners see as their incentives moderated by litigation costs. Thus, this can be changed when copyright owners determine they need more incentives, for example in the RIAA litigation.
Has no normative significance as a matter of law but is a focal point for coordination between copyright owners and users. When users can determine an equilibrium at a particular type, certain kinds of de jure activities nonetheless continue to thrive because there’s no enforcement: a credible expectation users develop based on concerted inaction.
This is what the troll disrupts. Troll’s expertise lies in ability to lower litigation costs/raise probability of positive outcome. Result is change in Type I/Type II equilibrium. The problem isn’t just in the change in equilibrium. The problem is the fact that the troll’s entry into the system disrupts the function of the focal point in a coordination game. For a focal point to work, actors need sufficient information about each other and need to be able to assess each other’s motivations. Troll’s motivations are fundamentally different from those of copyright owners, making it hard for users to figure out their motivations.
Why “uneasy” case? Because what the troll did wasn’t regulated by formal law. Solution: tinkering with statutory damages.
Extensive reactions follow:
Well, this paper made me think! Balganesh argues that the wrong of trolling is that it encourages trolls to sue people whose uses are harmless to the true author/owner and therefore, in the absence of trolling, tolerated though infringing. Among other things, Balgenesh suggests that it will be easy to solve the Righthaven standing problem by transferring enough of an exclusive right to confer standing. But how would you word that grant? An online newspaper won’t, for example, want or be able to grant the next Righthaven “the exclusive right to distribute in [defendant’s home city, or home block],” since the online version of the paper will also be available there. I’m not saying it can’t be done, but I am interested in how exactly.
I also think there’s a lack of clarity in the concepts at work in the paper of actionable/not actionable claims—they are categorized as Type I (infringing/actionable and likely to be brought), Type II (infringing/actionable but unlikely to generate actual litigation), and Type III (noninfringing/fair use etc., apparently without regard to whether they are likely to generate litigation). Actionable could mean “one could sue with almost no worries about paying the other side’s attorneys’ fees,” or it could mean “one could expect to win if litigated out.” Those are very different things; there’s almost no claim that falls outside of the first definition, including plenty of actual fair uses, but many claims fall outside the second definition. I see a lot of Type II cases (what Tim Wu calls tolerated uses, which Balganesh suggests populate his Type II category) as at least arguable defendant victories. If actionable means “could expect to win,” then I’m going to classify fact patterns very differently than Balganesh apparently does. (If actionable means instead “one could expect to win because the defendant would be fearful/unable to bear the costs of litigation,” then Type III also changes substantially in its contours; I don’t think that’s what he means.)
And this leads me to some serious disagreements with the paper’s analysis. E.g., “The fair use doctrine, in short, was both useless to Righthaven’s defendants, and structurally facilitative of Righthaven’s strategy.” Except that several defendants won fair use rulings (and, not for nothing, it’s not accidental that they did: Righthaven’s acts seemed unreasonable even if the law developed for non-troll situations didn’t necessarily predict this result). Balganesh thinks fair use shouldn’t have helped these defendants, but to call it “useless” seems to represent a triumph of hope over experience. This is related to the fact that the paper deems fan fiction infringing (apparently with no real hope of fair use, which I think is a serious misdescription of the law). “Gray areas” in this paper are just where enforcement hasn’t happened (yet), not where the uses might actually be legal; compare, for example, the recent approach of Canada’s Supreme Court. In this sense, the paper offers a direct challenge to arguments made by Michael Madison, Peter Jaszi & Patricia Aufderheide, Pam Samuelson and others that norms can solidify particular practices as noninfringing.
This idea that there is an abstract answer to the question of whether a use is infringing that is both static and known leads Balganesh to suggest in his dynamic analysis that a particular set of facts can only move between Type I and Type II (actionable but likely to be sued or unlikely), at least in the absence of congressional action. By contrast, I would say that types of uses can move out of Type II (not likely to be sued, maybe because no one is really sure what would happen and potential plaintiffs don’t want to risk the various expenses of a loss) to Type III (fair use and generally accepted as fair). Even Type I (likely to be sued) can move to Type III, as Jeff Koons and Google might have noticed. The same unwillingness to admit that categories of fair use are shaped by practice, including litigation, comes in the claim that “given that the decision whether something is in Type II or instead in Type I (and therefore enforced) is entirely a private decision delegated to copyright owners, potential defendants have little ability to contribute to the scope of Type II claims on their own.” Tell this to the Righthaven defendants who fought, or any of the recipients of C&Ds who’ve done the same. (This somewhat depends on what actionable means, but a defendant’s known willingness to scrap can certainly change a potential plaintiff’s mind about actually suing.)
I see a similar issue arising from the claim, “Individual home downloading is today viewed as likely to trigger an infringement action.” By whom? Certainly not by individual home downloaders. Illegal, maybe—but not likely to get you caught. That’s just a wording problem, but it reflects a lack of a sense of contestedness of facts and law. So, then the paper calls Sony a Type II case because many copyright owners were ok with home taping—but now I’m even more confused: the Supreme Court told us it was a Type III case because time-shifting is fair use! Perhaps Balganesh thinks Sony is wrongly decided. But here again the paper seems to exclude the possibility that norms, some of them created by non-copyright owners, can/do in fact help determine whether something is a fair use, even though in Sony the idea that “many owners tolerate it” apparently contributed to the finding of fair use.
All these disagreements don’t necessarily go to the paper’s core: assuming we don’t like copyright trolls, what is to be done? The paper’s proposed solution is to require trolls, though not authors, to show harm in order to be allowed to sue for statutory damages. I’m not clear on the statutory basis, but ok.
It’s not clear to me that this principle would have disallowed Righthaven’s claims: if the argument for harm is “I’d like to charge you a $750 licensing fee,” as many online newspapers would, unless the use is fair that seems like a pretty standard harm argument. We still have to figure out the baseline. And this perhaps connects my discomfort with the paper’s typology with the core question about trolling: there doesn’t seem to be any basis to condemn trolling if it’s just about the cost curve. Current owners have better things to do than sue ordinary noncommercial copiers, but that has no deontological or utilitarian force if someone shows up to say “I’ll specialize and do it cheaper!” You need a defense of Type II behavior in itself, as Tim Wu has offered.
Side note: I also wondered about the difference between authors and owners in the proposed fix, since “authors” and “trolls” don’t seem to fill the entire set of owners. Arguing that we can infer harm from infringement when an actual author brings a claim, Balganesh points out that we give authors special privileges, such as termination rights. But again there’s equivocation about meaning: Warner Bros. is legally the author of a large number of works, but has no termination rights because it is the author of works for hire.
Industries Without IP
Kate Darling, What Drives IP without IP? A Study of the Online Adult Entertainment Industry
Adult entertainment online: big advantages, but easy copying and sharing.
Copyright enforcement is difficult because of DMCA/whack-a-mole strategies, and P2P end user litigation problems (a wave of litigation against AE downloaders); the lawyers involved are not representative of the industry as a whole. Even those who litigate only litigate non-standard-het content: gay porn is easier to get settlements for. Courts have begun to see that these cases are about settling to avoid embarrassment, which courts don’t like. (RT: There’s an interesting side note here about the history of gay blackmail, or the history of the heartbalm torts and dislike of use of the court system for that.)
Also, standard content is a commodity: others are giving it away for free. Undermines model of trying to sell. So, effectively operating in copyright’s negative space.
Shift towards goods that are difficult to copy: convenience goods and services and experience goods. Similar to other fields: Sprigman & Oliar find that negative space in comedy shifted production.
People will pay for convenience of not sifting through free stuff; cloud and cross-platform services (despite Apple’s restrictions, apps are thriving). Increased interactivity: live chat/cam shows; choose your own adventure; games.
There is still an incentive to produce traditional content, basis for these other goods and a very powerful marketing tool as long as it’s branded. Works because production costs are low.
A lot of consolidation, vertical integration; weaker players are being weeded out. But stronger players are surviving and even thriving and that seems it may continue.
Industry-specific factors: consumers highly value privacy and convenience; high deamnd for content = viral marketing; low production costs. Non specific factors: experience goods, free content as marketing.
There may well be market failure. But the fact that production incentives remain at all challenges basic assumption of simplistic copyright theory.
Ramsey: any product placement?
A: not seen that. Companies struggle with the fact that they don’t have mainstream deals. As they professionalize, some companies are trying to make deals with men’s cologne etc. Not clear that will work.
Goldman: the list of industry-specific factors describes almost all content except big Hollywood blockbusters and other niches. Books, music, some movies may have low production costs. Consumers want convenience and privacy elsewhere too. Not everything goes viral, but everyone has local communities. Are these really different?
A: There are parallels but there are relevant differences. Production costs: theory of copyright assumes that investment in quality of content makes it better for consumers. But at least for some of the industry there’s preference for cheap to produce amateur content: optimally cheap. You don’t see much actual amateur content because of recordkeeping requirements, but bad lighting/shakycam/few people involved is both cheap and very popular. Also privacy is way more important.
RT: The promise of feminist porn/porn for women was that it would have nice production values and would be something women would want to see, or women would want to watch it with men. What happens to that in an optimally cheap world? Is this optimally cheap guy porn?
A: niche markets can still sell because of convenience: it’s still hard to find that stuff in the slush free. If you want to watch something with your partner, you’ll buy a DVD & create an experience at home, so that still works. (I still think there’s an interesting question here about the distribution of kinds of content.)
Heymann: say more about branding. What is the brand signalling? Niche, quality?
A: basically the name.
Kristellia Garcia: in music, the argument is that experience etc. doesn’t make up the difference lost to piracy. True here?
A: the producers say that the difference is made up: people are willing to pay.
Sprigman: why isn’t there a Viacom case against these sites? You could also flood the site with bad clips, or with short clips as ads, or what.
A: premium Tube sites are now owned by content production companies, and making deals with content producers to police for unauthorized content but in return the companies provide 5-minute clips of good quality and have revenue-sharing. Producers prefer good quality content; also prefer hosting on Tube sites, since bandwidth costs are much higher than actual production costs.
Matt Sag: are other industries being harmed by the reluctance to allow John Doe joinder of torrent swarms because the plaintiffs come from porn?
A: a lot of the reasons judges use do translate, even though the underlying reasons may be reluctance to support the extortion scheme—although another reason they don’t like these cases is the huge amount of paperwork.
Erez Reuveni, Copyright, Neuroscience, and Creativity
Disclaimer: only his opinions. The neuroscience of creativity allows us to design legal instructure that completments the processes within our brains responsible for creativity.
Models: utilitarian/economic, moral rights/personhood. Also romantic authorship. Creativity is, in fact, a function of many variables and interactions between people within systems: cultural environmentalism. Copyright is one element in designing an optimal environment for creativity; the neurobiology of creativity can serve as a baseline within that environment.
Simple rule: the more (connections) the better. Convergent and divergent thinking: when you encounter a problem, look for low hanging fruit: what you consciously know. If that doesn’t work, divergent thinking: pattern-sifting, search algorithm, and then a convergent/executive process managing things to see how well what you find works. Result: idea.
Creativity is a function of the richness of brain’s internal environment, immediacy of communication, and efficiency of interface with external environment.
Internet effects: Transactive memory: actual information v. locational information. When you don’t remember where your car is, but remember that your spouse knows, that’s locational information. Not a bad thing! Notebooks, bookshelves, etc. Increasingly, new information in longterm memory is locational.
So: let’s complement these realities of transactive memory and externalized memory storage. Law should facilitate an efficient interface between neural networks and external networks housing information referenced by transactive memory. Things that matter: internet infrastructure, including net neutrality and network management (data caps, graduated response, metered billing); internet content and copyright enforcement. DMCA, SOPA/PIPA, domain seizures. Criticisms: over-incentivize content removal, under-incentivize challenges to removal; reverse traditional burdens of proof in prior restraint stetting; overbroad targets: whole domains v. specific pages; ignore the real party in interest. Who is the real party in interest? People relying on the stored information?
Domain seizure: this is a prior restraint, which in analog world requires notice and adversarial proceedings. This default has been flipped in domain-wide seizures/takedowns. Neuroscience should move the needle away from them. Minimizing false positives: the people using Megaupload legitimately despite the massive infringement. Online storage lockers are not illegal Put the gov’t to its burden: if it wants to take down an entire domain, it should take account of individuals who are legit. We don’t use missiles to take down a whole block for one person with drugs in one house (RT: though overseas, one Al Qaeda member probably qualifies for that treatment). Minimize false positives by requiring a bond, swearing under penalty of perjury, etc.—sword over seizers’ head. Smaller domains/claims should lead to smaller bonds.
Immunity: claims of infringement should bear some risk. Complete v. partial infringement: claiming wholesale infringement could be tilted to remove complete copies more readily, whereas partial/potential fair use should stay innocent til proven guilty.
Q: how do we get legislators to consider this?
A: SOPA/PIPA shows the possibilities: the tech community woke up. In the next go-round we can put out more of the evidence.
Gilden: are you actually proposing a more radical First Amendment argument? To have the basic tools of our brains/express ourselves, we need this information.
A: maybe the cognitive domain is protected by the First Amendment. Now that we know better how the brain works, we can add that to the argument.
Brauneis: are there any particular kinds of content/services that deserve more protection than others?
A: Google is the connector—you know you’re looking for something but not exactly. Not sure how to make that argument, but maybe the tools have a heightened obligation to behave as connectors.
Eva Subotnik, Subconscious Fair Use
Recent Georgia case analyzing fair use on a case by case basis for use in course assignments. Compare that to Cariou v. Prince, where there were a variety of pictures with greater and lesser alterations: the district court analyzed the pictures in one fell swoop. Cariou: affected by the court’s reliance on statements by the defendant. Intent of the putative user: not just binary in terms of bad faith, but more substantively whether there is or should be a fair use intent component to the fair use analysis. District court: “Prince did not intend to comment on any aspects of the original work or on the broader culture.” Multiple references to intent.
Campbell: whether a parodic character may reasonably be perceived. Suggestion of objective standard. Kennedy’s concurrence: we shouldn’t make it easy for musicians to exploit existing works and then claim their later rendition was valuable commentary—take care to ensure against post hoc parody rationalizations. Is there an expectation of rationalization pre hoc?
Authors’ Guild v. Google: Google didn’t conduct a case by case inquiry into copyright ownership or conduct individualized evaluation as to whether posting snippets would constitute fair use.
Of the factors, factor one is the one over which the defendant has the most knowledge/control. Can think about affirmative defense in this light. Not arguing for a new intent-based component, but rather trying to distill descriptively what’s going on, and characterizing intent in fair use compared to other parts of copyright: easy to become an author (don’t have to intend it in any sense—but see joint authorship); intent not required in ordinary direct liability, which is strict; secondary liability often considers intent/knowledge.
Intent to do what? Perhaps whether the putative fair user had authorial intents and purposes. Blanch v. Koons: court says we don’t need to depend on our own poorly honed artistic sensibilities, because Koons explained why he used Blanch’s image.
Some suggest we should look at responses of interpretive communities to the second work. (I agree!) Argues for a role for subconscious fair use: unknowing fair use serves the goals of copyright, and should be possible under our law. Objective purposes should trump subjective intent. (RT: is objective/subjective the only binary? Communities of practice/interpretive communities may not have an objective view either.) Interested in how this translates from appropriation art to corporations like Google.
Katz: Abraham Drassinower has written about relevant issues.
RT: asked my question/best practices as interpretive communities deciding what’s fair use.
A: Sure, but will that lead to increased documentation requirements that will backfire if there’s no documentation in similar circumstances.
Zahr Said: how do we distinguish legal from artistic intent? We are skeptical of post hoc intention description that matches transformativeness. But artistic statements of intention are apparently more common in the art world than elsewhere, where curatorial practices encourage them. If you disaggregate legal and artistic intention, you can unpack the effects of the theory of the death of the author. Even if the author says she meant to do something, can we trust that? Can we trust her memory? Should we still find it fair if the author refuses to interpret meaning?
Brauneis: interpretive communities often (necessarily?) impute authorial intent. If you find out that a blotch on a painting was added not by the author but by a pigeon, that may well change the community’s view of its meaning. (Though perhaps the person who sees it could create their own authorial intent in appropriating that randomly modified piece.) The interplay between subjective and objective is difficult because the thing you impute may be modified with further information.
Heymann: post hoc rationalizations are also possible strategically in interpretive communities. One question for the paper: How would you deal with those in a litigation setting?