Harvard Law Review Harvard Law Review Harvard Law Review

Harvard Law Review Forum

Free to Be You and Me? Copyright and Constraint

Joseph P. Fishman’s Creating Around Copyright advances a provocative thesis: some restrictions on creativity can spur the development of additional creative solutions, and (some level of) copyright might be one of those restrictions.1×1. Joseph P. Fishman, Creating Around Copyright, 128 Harv. L. Rev. 1333 (2015). If Picasso was right that “forcing yourself to use restricted means is the sort of restraint that liberates invention,”2×2. Françoise Gilot & Carlton Lake, Life with Picasso 57 (1964) (emphasis added). then being forced by law to use restricted means might do the same thing, ultimately leading to more varied and thus more valuable works.3×3. Fishman does distinguish between invented or chosen constraints and imposed constraints like law: he concedes that, all things being equal, invented and chosen will be better, but argues that externally imposed constraints can work, too. Fishman, supra note 1, at 1373–76.

At the outset, it’s important to know the baseline against which we ought to evaluate Fishman’s claims. Most copyright restrictionists, of whom I count myself one, don’t want to eliminate all copyright law.4×4. Cf. id. at 1335, 1369–70 (suggesting that Fishman is evaluating “copyright” versus “no copyright”). Fishman’s argument is directed at creators who want to take an existing work and do something with it — incorporate parts of it into a new creative work or make a derivative work based on it. Because the question is the proper scope of copyright as applied to these works, the comparison should not be to a world without copyright, but should instead focus on the marginal effects of expanding or contracting copyright’s definitions of substantial similarity and derivative works. Once the question is properly framed, I have concerns about the major analogies Fishman uses — patent law and experimental evidence about other types of constraints on creativity — as well as his model of the rational creator.

Patents, Copyrights, and Progress

Fishman analogizes to patents5×5. Except when he concludes that patents aren’t a good model. See id. at 1399–400 (rejecting patent’s treatment of independent creation, while relying on creativity literature that doesn’t distinguish between copyrightable and patentable subject matter); id. at 1394–96 (rejecting blocking copyrights as too risky for creativity despite the existence of blocking patents). and applies to copyright a rationale often given for exclusive patent rights — that they encourage innovation not only by rewarding patentees but also by giving people who can’t get a license from the patentee, or don’t want to pay for that license, incentives to design around patents. However, while innovation in copyright shares features with innovation in patent, it has marked differences as well. Doctrinally, unlike patent, copyright protects expression and not ideas. Why allow free riding on ideas but not on expression? Constraint theory can’t tell us.

Perhaps even more significantly, “designing around” requires a deliberateness about the law that isn’t common in many copyright sectors, though it may occur with large copyright exploiters. Relatedly, Jeanne Fromer has drawn on the empirical creativity literature to identify “problem finding” as a vital driver of copyrightable creativity (how do I express the alienation of my generation?), whereas “problem solving,” where the contours of the problem are already known (how do I make a fixed-wing aircraft more fuel efficient?), is more representative of patentable creativity.6×6. See Jeanne C. Fromer, A Psychology of Intellectual Property, 104 Nw. U. L. Rev. 1441, 1444 (2010). Where problem finding is more important, constraints outside the artistic process may be especially detrimental.

In addition, in both of the alternatives to inventing around — licensing or infringing — the person who isn’t inventing around may just be copying the patented device or process, not generating patentable subject matter. By contrast, creating a derivative work — authorized or not — also requires the addition of creativity. Tom Stoppard’s Rosencrantz and Guildenstern Are Dead (based on Hamlet) involves creative effort, as does Stoppard’s Arcadia (based on history). It’s not evident that the nonderivative work requires “more” creativity or is “better” at giving us the new expression copyright aims to incentivize, because “derivative” and “lots of variation” are not really opposites.

Consider also how we think about performances of cover songs; although the cover song is copied, it is also altered, and often the alteration gains meaning from being put in dialogue with other versions. It is hard to see how to analogize that kind of cultural conversation with patent improvements. Meanwhile, six of the top country songs of 2014 sound almost exactly alike, apparently not as a result of infringement but based on multiple singers’ understanding of what sells.7×7. Michelle Geslani, This Video Proves Every Hit Country Song Sounds the Same, Consequence of Sound (Jan. 9, 2015, 11:20 AM), http://consequenceofsound.net/2015/01/this-video-proves-every-hit-country-song-sounds-the-same [ http://perma.cc/X8ZK-YNTA”>http://perma.cc/X8ZK-YNTA]. In copyright, there can be a lot of sameness without infringement and a lot of copying without sameness.

The role of iterative creativity in copyright is related to an important difference between “progress” in patentable subject matter and “progress” in copyright. As Barton Beebe and others have noted, what constitutes progress in utilitarian fields is often relatively clear compared to progress in art.8×8. Barton Beebe, Intellectual Property Law and the Problem of Aesthetic Progress, Inaugural Lecture of the John M. Desmarais Professorship of Intellectual Property Law, New York University School of Law, (Jan. 28, 2014), http://www.law.nyu.edu/news/barton-beebe-inaugural-desmarais-lecture [http://perma.cc/5HTJ-9VQ4]. Very few of us would choose to ride in a Model T for anything other than novelty value. We prefer cars with seatbelts, airbags, and powerful acceleration. What constitutes “progress” in copyright is much more up for grabs, and any consensus is likely to be fragile and generational. As a result, anyone who doesn’t want to have to design around the derivative works right can easily go to another deep well of inspiration, one in the public domain (this is also relevant to the experimental evidence on which Fishman relies, discussed below). If you can’t write the next Iron Man movie without a license, you can always write another Sherlock Holmes movie without a license.9×9. As long as you leave out details from the ten stories still in copyright in the United States. See Klinger v. Conan Doyle Estate, Ltd., 755 F.3d 496, 497 (7th Cir. 2014). By contrast, riffing off the Model T engine instead of a currently patented engine would be a trickier, and likely commercially unsuccessful, prospect.

But what if you really want to write a superhero movie, not a Robert Downey Jr. movie? In that case, copyright does indeed pose unique constraints. If we accept that some artists are creatively committed to particular genres or tropes, though, it is hard to deny that some artists are also creatively committed to particular characters or works and won’t do their best work without them. As Wendy Gordon insightfully observed, there are cases in which a copyrighted work has such a powerful influence on us that being barred from creatively reacting to it imposes an injury on us.10×10. See Wendy J. Gordon, A Property Right in Self-Expression: Equality and Individualism in the Natural Law of Intellectual Property, 102 Yale L.J. 1533 (1993); see also Rebecca Tushnet, Economies of Desire: Fair Use and Marketplace Assumptions, 51 Wm. & Mary L. Rev. 513, 532–33 (2009) (describing this artistic response).

Moreover, even if a creator doesn’t feel compelled to work with specific subject matter because of her exposure to it, working with existing characters and situations is working with specific constraints, just as writing a sonnet is.11×11. See, e.g., Henry Jenkins, Why Heather Can Write, MIT Tech. Rev. (Feb. 6, 2004), http://www.technologyreview.com/news/402471/why-heather-can-write [http://perma.cc/9PZQ-JWUQ] (discussing the benefits for young writers of starting with familiar characters and situations). Indeed, classic derivative works such as translations or movie versions of books inherently pose important creative constraints, because of their goal of moving from one medium or language to another. Consider an account of sampling from electronic musician Kid606 that differs markedly from Fishman’s characterization: “Sampling is like Legos. If you give someone a bunch of Lego blocks and tell them to put something together, then they have something to work with — as opposed to saying, ‘Here’s a bunch of plastic, go mold it and then build it.’”12×12. Kembrew McLeod, An Oral History of Sampling: From Turntables to Mashups, in The Routledge Companion to Remix Studies 83, 85 (Eduardo Navas et al. eds., 2014).

Similarly, as children’s literature scholar Catherine Tosenberger notes, the “limitations of writing preexisting characters within an established universe” themselves “offer opportunities to play with constraint” — “in order to be recognizable as such, a recursive text [based on an existing text] must agree to be limited by some element of the source material.”13×13. Catherine Tosenberger, Mature Poets Steal: Children’s Literature and the Unpublishability of Fanfiction, 39 Child. Literature Ass’n Q. 4, 13, 22 (2014).

It’s unclear why further restraints would be useful or important for such works, though Fishman’s argument may have more relevance for the bloated reproduction right, which allows copyright owners to suppress works in the same market niche as their own. Moreover, legal constraints are likely to vary in ways unrelated to artistic aims or processes: a rapper who signs with a major label, for example, will have much more freedom to sample than an independent artist, based on the label’s cross-licensing practices within its own properties.

Precisely because the aims of art are up for grabs in a way that the aims of technological invention are generally not, constraints that come from law are regularly mismatched with the constraints an artist would otherwise choose. With their much-vaunted reluctance to judge the quality of creative works, courts are not in a good position to determine which constraints are useful to creators. Even more important: courts are not in a better position to identify proper constraints than the artists themselves (or even the funders, whether nonprofit or for-profit, who bankroll artists with the hope of finding something artistically or economically successful). Artists are indeed regularly overoptimistic about the value of their creations,14×14. See Christopher Buccafusco & Christopher Jon Sprigman, The Creativity Effect, 78 U. Chi. L. Rev. 31 (2011). which may help explain why experimental subjects overestimate their own creativity,15×15. See Fishman, supra note 1, at 1365. but that doesn’t validate copyright’s limits. Creativity is too varied to conclude that copyright’s constraints are useful ones, even if we believe that patent’s are.

Experiments with Constraints on Creativity

Along with the analogy to patent, Fishman also draws on experimental evidence about the effects of constraints on creativity. While the research indicates that the nature of constraint matters,16×16. See, e.g., id. at 1364–65 (discussing studies in which subjects were given modular components from a presorted set of noncopyrightable elements — shapes or yarn). Constraints on choice within a set of noncopyrightable elements, or limits on the size of the set, are very different from the constraints imposed by copyright’s substantial similarity and derivative works rules. that very context-specificity limits the applicability of evidence about nonlegal constraints to legal regimes. Research suggests, for example, that being given an exemplar of a solution to a problem limits subjects’ imaginations with respect to design problems, causing them to stick close to the exemplar even when it exhibits obvious flaws.17×17. Id. at 1366–67 (citing David G. Jansson & Steven M. Smith, Design Fixation, 12 Design Stud. 3 (1991)) (reporting research on the effects of pictorial examples of solutions to design problems; explicit instructions to students to avoid using the specific problematic features of the exemplars did not work, where problematic meant undesirable from the perspective of the functional design goal).

Two points are worth noting. First, further studies showed negative effects of pictorial exemplars for mechanical engineering students, but only marginally significant effects for students in industrial and interior design.18×18. See Evangelia G. Chrysikou & Robert W. Weisberg, Following the Wrong Footsteps: Fixation Effects of Pictorial Examples in a Design Problem-Solving Task, 31 J. Experimental Psychol.: Learning Memory & Cognition 1134, 1135 (2005) (citing A. Terry Purcell & John S. Gero, Design and Other Types of Fixation, 17 Design Stud. 363 (1996)). This result is another indication that creativity is highly sensitive to contextual factors, like genre and training, and that extrapolations to copyright law from this evidence are risky. Second, and more significantly, in the experimental condition, exemplars can be withheld, and thus people can be directed to create without a model of possible results.19×19. See, e.g., Fishman, supra note 1, at 1396 (discussing a study in which coders were denied access to others’ solutions and did better than coders who were free to build off of others’ solutions). In the copyright condition, they cannot. Only the legal regime can be tweaked.20×20. The studies differ on whether instructing people to stay away from the exemplar helps when designing functional devices to solve specified problems. Id. at 1367 & n.201 (citing Chrysikou & Weisberg, supra note 18). Chrysikou and Weisberg used the exact same examples, instructions, and tasks as Jansson and Smith, but they got conflicting results. See Chrysikou & Weisberg, supra note 18, at 1144. This variation isn’t surprising as replication is difficult, but Chrysikou and Weisberg attribute their differing results to steps taken to ensure that participants carefully read the instructions and didn’t report on perceived ambiguities. It’s not clear what the analogy in copyright would be — could we make sure that the people exposed to existing works are carefully educated about the derivative works right and instructed that they’re not allowed to license? Even if such instructions could work, instructions to avoid copying functional features are still not very much like copyright’s definitions of infringing similarity or derivative works. Relatedly, the research Fishman discusses involves specific tasks set by experimenters (design a rack for a car), not tasks chosen for intrinsic or market-based reasons.

Moreover, Fishman’s theory of copyright constraint requires a legally educated, risk-averse creator who wishes to commercialize her work and thus knows of and desires to respond to copyright’s constraints.21×21. This is probably a somewhat different set of individuals than the set of people who are in fact incentivized by copyright, though a good enough fit for that set. But that’s not the full set of creators who are affected by copyright law, or by beliefs about that law, and if we’re judging copyright’s fitness for purpose we should think about those people as well. As long as copyright only protects expression, then, this type of creator can copy a lot. Unlike experimental subjects, the lazy creator has a whole menu of possible existing works both in the public domain (the King Arthur mythos) and not (Outlander) that she can use or license if Game of Thrones is unavailable. The make-or-buy decision, in other words, is quite different from the make-or-copy decision faced, at least subconsciously, in the creativity studies. Someone will license a sword-and-sorcery fantasy to the profit maximizer at a price that at least matches the cost of creating around.22×22. Cf. Ty Inc. v. Perryman, 306 F.3d 509, 512 (7th Cir. 2002) (discussing, in the trademark context, purely rational licensor’s disinterest in the content of licensed material if profit is the only concern). Because progress in copyright doesn’t work like it does in patent, and yesterday’s stories can be made fresh and profitable in a way that buggy whips can’t, copyright’s constraints have much less effect on “inventing around.”

Indeed, if Fishman is correct, it’s not clear why we should protect derivative works at all — why not send them immediately into the public domain? That would provide even more economic incentive to design around existing works.23×23. Fishman’s reasoning could even be read to imply that derivative works should be banned, preventing anyone, even the initial author, from relying too heavily on an existing work. After all, as author Don Marquis reportedly said, “a sequel is an admission that you’ve been reduced to imitating yourself.” Edward Anthony, O Rare Don Marquis 487 (1962). But if we want (some) derivative works to be created because they’re creative and valuable, then we’ve arrived at the position that designing around is sometimes beneficial and not designing around is sometimes beneficial — a result which seems innocuous but unhelpful.

Thus, even accepting the relevance of the creativity studies, it’s too simple to say that “[w]ithout a derivative work right, we may get more homogenization.”24×24. Fishman, supra note 1, at 1395. The derivative work right doesn’t protect us against the creation of lots of derivative works, as current experience with serialization and merchandising demonstrates.25×25. See, e.g., Cecilia Kang, Hollywood’s Addiction to Franchises Is Reaching New Extremes, Wash. Post (Jan. 16, 2015), http://www.washingtonpost.com/business/economy/hollywoods-addiction-to-franchises-is-reaching-new-extremes/2015/01/16/4b8bc934-96a9-11e4-aabd-d0b93ff613d5_story.html [http://perma.cc/AD3E-MKT6] (“The result, say many observers, is a slate of movies that endlessly recycle the same characters and franchises, to the exclusion of practically anything else. . . . And with elaborate plans for trilogies and spinoffs going into the next decade, the industry is counting on the fact that audiences won’t get tired of it all.”). Robust licensing already means that new works are less different from old works than they would be if derivative works were banned. Licensing also delivers more profit for already-wealthy content providers, which has its own effects on the creative ecosystem.26×26. See, e.g., Neil Weinstock Netanel, Market Hierarchy and Copyright in Our System of Free Expression, 53 Vand. L. Rev. 1879, 1911–15 (2000). In all of Fishman’s anecdotes related to copyright, being denied a license is what spurs creativity. How often that happens, versus how often a Marvel property spawns another Marvel property, seems at least on first inspection to favor the granted license. And moves by major copyright owners to license almost every use — and monetize it afterwards — are only increasing.

Quite separately, many creators aren’t legal experts subject to careful calibration of the substantial similarity standard and the derivative works right. Professor Jessica Silbey’s qualitative empirical work with creators, lawyers, and businesspeople involved in commercializing intellectual property reveals that artists are usually uninformed about and uninterested in the details of intellectual property law. They just don’t think about copyright at the point of creation. As she explains,

[I]ntellectual property law is largely unfelt and unseen by these interviewees either as a guide or as a constraint in the early development of works of art and science. . . . [I]ntellectual property law does not productively structure the beginnings of creative or inventive experience. Instead, the creative or innovative impulses described by the interviewees arise from diverse and serendipitous personal experiences, from doing what pleases them, from pursuing what appears necessary and important, from the pursuit of personal and professional freedom, and from within communities of influence.27×27. Jessica Silbey, The Eureka Myth: Creators, Innovators, and Everyday Intellectual Property 53 (2015); see also id. at 96, 100 (finding that intellectual property owners underenforce, are ignorant of, or misunderstand the law, particularly copyright).

Other recent empirical work finds similar results across a number of artistic and innovative communities. For most creators, rights talk and economics are at best loosely linked to formal law, which is “invoked or imagined occasionally, opportunistically, or instrumentally,”28×28. Laura J. Murray, S. Tina Piper & Kirsty Robertson, Putting Intellectual Property in Its Place 1 (2014). more as fantasy, rumor, threat, or “a symbol of corporate power rather than as a specific set of rules.”29×29. Id. at 2. Even when statutory and case law is readily available, “people actually choose to understand the law through information and opinion gathered from friends, strangers, coworkers, and the media.”30×30. Id.

Silbey finds that law and legal advice often only come in at the last minute, after the work has been created.31×31. See Silbey, supra note 27, at 188; see also id. at 189–90 (finding that creators feel that lawyers make the questions too complex and offer unhelpful advice, treating everything as too risky); Murray, Piper & Robertson, supra note 28, at 2 (finding that intellectual property claims are “adopted to cross or police boundaries long after a work has been created or an innovation has taken place”). As Fishman notes, this is a likely unhelpful “middle” or “late constraint,”32×32. Fishman, supra note 1, at 1388. and one that the empirical evidence suggests can’t routinely be moved forward. Lawyers do help with designing around patents, but not with other parts of the creative process.33×33. Silbey, supra note 27, at 206–07. Instead, Silbey concludes, creators value autonomy: the freedom to choose for themselves what their aims and constraints will be.34×34. See, e.g., id. at 248. This leads to my next major concern: variation in who benefits from legal constraints, and who doesn’t.

The Appropriate Model of the Creator

What about creators who care more about the content of what they create than an indifferent profit maximizer does? Some people enjoy creating alternatives to sampled music and feel their work has been improved by finding workarounds.35×35. Fishman, supra note 1, at 1370–71. Interestingly, however, several of Fishman’s examples suggest a different reaction to copyright law from the compliance he proposes as generative: for example, a producer says, “[Y]ou tend to take less obvious bits of records and obviously you hunt for more obscure records, or you chop something within an inch of its life so even you’ve forgotten what you sampled . . . The new cautious approach in itself becomes a limitation, but not necessarily a bad one.”36×36. Id. at 1371–72 (quoting Justin Morey, Copyright Management and Its Effect on the Sampling Practice of UK Dance Music Publishers, 3 J. Int’l Ass’n for Study Popular Music, no. 1, 2012, at 48, 59 (omission in original)); see also id. at 1371 (citing other examples of musicians who hide but do not cease their copying). But a sample’s obscurity doesn’t make it any less copied; it just means you’re less likely to get caught.37×37. The courts are, unfortunately, split on whether a completely unrecognizable sample is actionable — but all would agree that a recognizable sample, even an obscure one, is actionable unless the use is de minimis or fair use. See, e.g., Bridgeport Music, Inc. v. Dimension Films, 410 F.3d 792 (6th Cir. 2005); VMG Salsoul, LLC v. Ciccone, No. CV 12-05967, 2013 WL 8600435 (C.D. Cal. Nov. 18, 2013). This producer has taken an idea of the law and used it to impose a different kind of constraint, which is certainly noteworthy but not very supportive of the idea that restraints on unlicensed sampling encourage creativity.

Nonetheless, I readily accept that people with the resources to re-record performance tracks in a studio, such as Kanye West,38×38. Fishman, supra note 1, at 1370. can conclude that their workarounds are better artistically than their original ideas. However, as Fishman notes, additional constraints don’t work as well for people without other resources and support.39×39. See also Marcus Boon, In Praise of Copying 147 (2010) (discussing sampling’s relation to lack of access to equipment and musical instruments); Sheila Rule, Record Companies Are Challenging ‘Sampling’ in Rap, N.Y. Times, Apr. 21, 1992, at C13. Plenty of other creators find that the inability to sample makes their works worse.40×40. See Kembrew McLeod & Peter DiCola, Creative License 188 (2011). Moreover, this isn’t an apples-to-apples comparison, since the problem is not just that some works are “worse” (by whatever standard one cares to apply) but that some potential producers and messages leave the field. Rap, for example, has depoliticized sampling, and thus the very specific historical references and allusions made possible by sampling have declined:

[S]ampling provided an important engagement with musical and political history, a connection that was interrupted by Grand Upright and the cases after it [requiring sampling to be authorized], coinciding with a growing disconnect between rap music and a sense of social responsibility.

. . . But as Hank Shocklee, pioneering member of Public Enemy’s production team The Bomb Squad, told me, having open access to samples often did significantly impact artists’ lyrical content: “A lot of the records that were being sampled were socially conscious, socially relevant records, and that has a way of shaping the lyrics that you’re going to write in conjunction with them.” When you take sampling out of the equation, Shocklee said, much of the social consciousness disappears because, as he put it, “artists’ lyrical reference point only lies within themselves.”41×41. Erik Nielson, Did the Decline of Sampling Cause the Decline of Political Hip Hop?, The Atlantic, Sept. 2013, http://www.theatlantic.com/entertainment/archive/2013/09/did-the-decline-of-sampling-cause-the-decline-of-political-hip-hop/279791 [http://perma.cc/C3M4-NZFC].

Likewise, works of history now “design around” copyright by only using government artists to represent post-1923 U.S. art, thus overrepresenting their role to students of history.42×42. See Brief for College Art Ass’n et al. as Amici Curiae in Support of Petitioners at 7, 18, Eldred v. Ashcroft, 537 U.S. 186 (2003) (No. 01-618). The recent film Selma rewrote Dr. Martin Luther King Jr.’s speeches because the King family sold the rights to King’s speeches to a different filmmaker, forcing distortion into a creative work of historical fiction.43×43. See Jonathan Band, Can You Copyright a Dream?: How the Martin Luther King Estate Controls the National Hero’s Image, Politico Mag., Jan. 12, 2015, http://www.politico.com/magazine/story/2015/01/selma-martin-luther-king-can-you-copyright-a-dream-114187.html [http://perma.cc/44ZN-HKCT].

Fishman alludes to this lumpiness in the effects of copyright constraints, suggesting that “we lack a fine-grained understanding of who is benefiting, who isn’t, and what explains the difference.”44×44. Fishman, supra note 1, at 1403. I disagree. Along with the copyright restrictionists who recount the kinds of creativity lost to overreaching copyright, especially creativity from younger and less wealthy classes of artists with lower tolerance for risk,45×45. See, e.g., Lawrence Lessig, Remix (2008). scholars focusing on race and gender have elaborated who is benefiting and who isn’t, and suggested reasons that aren’t founded in optimizing creativity.46×46. See, e.g., Carys J. Craig, Reconstructing the Author-Self: Some Feminist Lessons for Copyright Law, 15 Am. U. J. Gender Soc. Pol’y & L. 207 (2007); K.J. Greene, Intellectual Property at the Intersection of Race and Gender: Lady Sings the Blues, 16 Am. U. J. Gender Soc. Pol’y & L. 365 (2008).

Relatedly, if the goal of copyright law is to optimize differentiation, then we are wildly overprotecting. The constraint/incentive model covers only situations where people are trying to generate “new” works. But copyright protects a huge number of works that aren’t generated because people are trying to make money being creative, or even trying to make money by using expressive works as such. Protecting email, homework, product manuals, and many other kinds of copyrighted works creates pure deadweight loss from this perspective.

The usual copyright theory answer is that it’s not worth sorting out these incentive-indifferent works, especially since there might be a little bit of copyright-relevant motivation even in some of these categories. That response may well be true; regardless, the nondiscrimination principle demonstrates that our calibration of rights will almost inevitably, for many reasons outside optimizing the number or quality of creative works, be chunky. Where we can only move in large steps, appeals to precision of the sort that requires categorizing seven types of constraint may come at the expense of accuracy and fairness.47×47. Cf. Oren Bracha & Talha Syed, Beyond the Incentive–Access Paradigm? Product Differentiation & Copyright Revisited, 92 Tex. L. Rev. 1841, 1882 (2014) (“Calibrating copyright on the basis of the trade-offs involved along the various supramarginal and inframarginal parameters requires predicting a host of complex effects in multiple markets and then attempting to fine-tune doctrines that are not always well-suited to the task. . . . [S]uch complexity should give pause when it comes to embracing the theory’s prescriptive relevance, and should motivate further reflection on second-best, comparative considerations regarding what kinds of necessarily rough judgments or imprecise proxies are most plausible to distill and implement as the theory’s takehome lessons.”).

What Should We Do Next?

I welcome Fishman’s call for more empirical research.48×48. See Fishman, supra note 1, at 1403. Nonetheless, uncertainty about calibration means that “[h]ow broad or narrow constraint scope should be to promote creativity” should not “color what kinds of adaptations the fair use doctrine should permit.”49×49. Id. at 1385. Even if we get more empirical evidence, it won’t correlate with results in litigated cases — at most, constraint theory offers a general reason that fair use defenses should be rejected in transformative content cases. That would be a bad result.50×50. Cf. Bracha & Syed, supra note 47, at 1883 (“The more complex and information-demanding the model upon which decisions about shaping and applying the law are based, the more vulnerable such decisions may be to manipulation by private rent-seeking efforts to tilt such decisions in their favor.”).

I’m suspicious of claims about improving fair use doctrine through the use of creativity research, not just on empirical grounds but because fair use supports other values. For example, freedom of speech means that you need a good reason to tell someone that they should express themselves in some other way, because of that person’s own interests in communicating. Communicating easily, via shortcuts instead of “designing around,” is often to be valued rather than disparaged because it helps a speaker reach her audience. Moreover, Fishman’s account of fair use is neither descriptively accurate nor normatively attractive: he suggests that transformativeness is a good standard because it encourages designing around and altering existing content.51×51. Fishman, supra note 1, at 1397. But in the current fair use system, exact copying coupled with transformative purpose is arguably even more likely to be fair than transforming content. As a result, books can accurately illustrate history52×52. See Bill Graham Archives v. Dorling Kindersley Ltd., 448 F.3d 605 (2d Cir. 2006). and libraries can digitize their collections for analysis and access for those who are print disabled.53×53. See Authors Guild, Inc. v. HathiTrust, 755 F.3d 87 (2d Cir. 2014).

Other relevant areas of copyright law are equally complicated. While constraint theory could be used to justify expanding the derivative works right or limiting compulsory licenses to record cover songs in the hope that we’d encourage more singers to become songwriters as well, its marginal effects are unlikely to overwhelm other considerations.

More generally, even those of us who dislike how the broad derivative works right and the substantial similarity test currently operate commonly think that authors should be able to license movie versions of their works. We question whether law is the appropriate source of constraints that ought to (and do) loom large in artists’ minds, instead of artistic genres, audience demands, artistic integrity, economic incentives, and so on. And we also doubt that the law presently has the right balance, primarily because, even if some people respond well to legal constraints, others — especially those without resources to access lawyers or pay large licensing fees — respond by shutting down. At a minimum, those who wish to constrain expression by pointing to the creativity that often emerges under censorship54×54. Fishman, supra note 1, at 1378. surely bear a heavy burden of proof. The artists (not to mention the art) lost to the gulag, the blacklist, the closet, and the asylum deserve a presumption in favor of freedom.

In today’s politicized context, the rhetoric of freedom — of speech, of thought, of artistic choice of which constraints to follow — is vitally important to check the automatic deference to “property” that too often dominates government treatment of copyright issues. Neither information nor creativity wants to be free (or chained), because neither of those concepts wants anything. People do. And when copyright restrictionists speak of “freedom,” it’s not because we want to make up our own languages or breathe on the moon, awesome as that might be. It’s because broad copyright produces specific winners and losers, and the winners are gaining too much at the expense of the losers. If copyright law is about incentives for individual creators, then caring about how copyright doctrines distribute the benefits of copyright law between the haves and have nots is fundamental.

* Professor of Law, Georgetown University Law Center. Thanks to Glynn Lunney and Jessica Silbey for comments.