WIPIP: Copyright: Incentives and the Digital Age

Tang, Creative Labor in the Age of Platform Capitalism

Theories of expressive work and creativity: lead to idea that AI training itself is not a © problem b/c it doesn’t use work expressively. Past idea: digital creativity enhances autonomy by giving individuals greater roles in authoring their own lives. When people are creative, making new things out of old things, become producers, they exercise and perform freedom and become the sort of people who are free. Semiotic democracy.

This model is challenged by paradigm shift in digital creation. From early days of YouTube—hotbed of amateur creativity—to current situation of rightsholder synergy. “Collaborative model” in which “rights holders” and “online creators” are partners in sharing viral profits. YouTube’s ContentID, along with Meta and TikTok, allows users to take bits and pieces without themselves needing to pay. This changes how we think about digital creativity on the internet. Those autonomy arguments are actually no different than autonomy arguments made by Uber, Postmates, and other gig economy companies—greater flexibility for a bohemian lifestyle. But we know that’s not true. Creators relied on TikTok for their livelihoods—Uber drivers are subject to algorithmic black box whims; YouTube doesn’t make its monetization/demonetization policies public. At best, platforms are oligopolistic for creators.

Platforms have always gathered that meaning-making expression into data; made it clear when they changed TOS to make it explicit that they were using data to train generative AI. The analysis of the datafication of creative works can learn from privacy scholarship, which has asked: how does privacy law evolve from dignitarian individualistic notions focusing on noneconomic invasions for the era of mass privacy invasions through gathering data en masse?

Q: historically, the vast majority of artists have failed. Maybe the platforms make things better [or don’t make things worse] by allowing artists to find their audiences.

A: consider TV writers negotiating for uses of AI.

Rosenblatt: This is part of a story of precarity; sounds like freedom but isn’t—gig economy analogy is convincing. What does this do for ©? TV writers—that’s WFH. The one thing these creators have is ©, and what good does it do them? Very little. YouTubers have © but it doesn’t seem to be relevant. Is there something else we should be doing?

A: all these arguments—the payments would be minimal, it would be too hard to track all the uses—have been made in privacy, so we could look there. WFH: challenge to the idea of the inextricable link b/t work and the author.

Lunney, Incentives and Music Composition

Fundamental premise of ©: more rights means more $ for righsholders means more creative output. This paper is looking at music composition: does revenue increase output? Does revenue increase number of composers? RIAA shipment data in constant dollars—going back to 1962. Steady climb; 80s recession and recovery; Napster/filesharing decrease until 2015, back to the early 60s level, and then starts to rise again. ASCAP etc. payouts also have to be considered for composers, and there we see a steady increase rather than sharp drops, with a few blips. With both taken into account, you still see a filesharing effect, but not quite as drastic as for recording revenues (and not much recession effect).

What happened to composers? Hot 100: a relative measure of quality rather than absolute, since it’s competing with the current alternatives. More money means fewer top hits rather than more contenders. For absolute measures of quality, looked at decay over time. The 90s (high revenue period) songs did worse than expected. The low-revenue 2006-2015 period is where people are still listening to the songs today. Can control for various variables including teenage population and revenue is still uncorrelated or negatively correlated with revenue.

Superstar composers: only 2 from the 1990s, many more from other lower-revenue decades. So more money is associated with fewer and lower quality hit songs, fewer first appearance songwriters, and fewer superstar songwriters. Next steps: more data, looking for varied output based on quality of hit, and revisiting assumption of equal shares of songwriting credits.

Rosenblatt: what the trends are in publishing deals—are the deals better/worse/different/360? Industry practice has changed considerably over that period.

A: good question, but his core question is does more © yield more music? Why it does or doesn’t is an interesting Q but not his main one.

Fromer: Movies: Blockbusters are often perceived as less creative; gatekeeping around who is allowed to produce them—is there anything like that dynamic here?

A: certainly gatekeeping on the artist side. Harder to tell on the publishing/composing side. We hear about artists being taken advantage of. [We definitely hear about composers forced to share credit with performers in order to get the song recorded by them.]

Fromer: Artists wanting to work with a particular hitmaker may matter.

A: we’re increasing the number of composers on average in a hit song, for sure.

Q: some songs are on the Billboard 100 forever.

A: sure, it’s not a perfect measure of quality, but that’s why I also look at whether it’s still being streamed years later.

Q: another possibility: look at # of DMCA takedowns to measure popularity/quality.

Cathy Gellis: did the Copyright Act of 1976 coming into effect have any relation to what happened to spike revenue in 1978? But also note that the “filesharing collapse” was also correlated with the collapse of record stores, which wasn’t just about filesharing but was also about the prices record companies were charging to record stores that made them unsustainable.

A: hard to explain 1978. But you’re right, there’s a lot going on. And only looking at Hot 100, so there may be other things going on, although the music industry is highly skewed to the top.

Q: Number of composers being added: people are preemptively adding composers w/similar style b/c of fear of Blurred Lines type lawsuits.

Pager, Copyright's Extended Duration as Feature not Bug

Most valuable works may struggle before they are recognized as valuable—Van Gogh died penniless. Moby Dick, Citizen Kane were commercial failures. Art from marginalized communities may take time to be recognized.

Publishers will only invest in works that they think will pay off fast if © term is short. Does that mean © needs to last 100 years? Not arguing for any particular term. But could play with a variable term. Return to a renewal term—option. That would allow mainstream iconic works to renew, which would be bad; could couple it with some sort of revenue cutoff, so the works that have had returns on their investment can’t renew.

RT: Renewal restrictions: as far as I know most blockbuster movies have technically lost money; so how would that work?

A: could go by box office/gross.

RT: What about the naïve economic model here? The midlist is already gone from publishing with life + 70—publishers are already not investing in works that will pay off over 20 years, even with that very long term.

A: you might take more risks on an unknown if you think there’s more money around.

RT: what does that have to do with the term? Your time horizon is still going to be 1 year. [See also “I’ll be gone, you’ll be gone.”]

Jake Linford: you’re really telling a desert story not an incentive story: this person deserves a reward and we need to create some sort of system to give them a reward.

Tang: there’s tons of data that the industry is becoming more risk-averse and more winner-take all, not willing to tolerate failure, despite its concentration (because of its concentration?). [Cory Doctorow and Rebecca Giblin say that the way to deal with the bully taking your kid’s lunch money is not to give your kid more lunch money; this is a statement about © rights.]

WIPIP: Innovation policy

Michael Burstein, The Law of the Direction of Innovation

Foundational texts: NBER Rate and Direction of Inventive Activity, and revisited. But we need to ask: what innovation should we prioritize? What innovation, by whom, and for whom? These are contestable subjects of political choice. Health law scholars have started to ask why so many me too drugs instead of novel therapeutics; trying to generalize from these questions.

Positing two axioms representing conventional wisdom in law & business: (1) entrepreneurial activity, mediated by price signals and supplemented by exclusive rights, is the best way to establish the direction of innovation. (2) private ordering of innovation policy takes place largely in the absence of legal structure; law’s role is only to ensure that IP facilitates adequate price signaling.

This is wrong in a few ways. Startups direct innovative efforts towards funders’ interests—structured by VC incentives. Incumbent firms are influenced by corporate rules. Covenants not to compete matters. That’s all not just price. Even if was just price, IP doesn’t mediate price in a vacuum; IP makes choices. Market demand also doesn’t equal social demand/value of innovation. Gov’t funding remains an important tool of innovation policy.

How do we get an innovation ecosystem w/striking disparities in investment/development of tech that we see today? VC investment is mostly in IT and consumer-oriented stuff. Silicon Valley has become optimized to produce those kinds of innovations. Crypto, but not encryption from same underlying tech base. Crypto has low capital requirements, low barriers to entry. Encryption is deep tech, w/long and uncertain regulatory cycle/demand, but potentially much greater social value.

We’ve had different innovation systems over time—Bell Labs, corporate R&D.

One goal: challenge notion that innovation is part of market alone and not part of legal choice.

Gaps in literature to fill: economic literature has some useful modeling, but doesn’t explain mechanisms by which economic incentives actually arise. Innovation ecosystems may be described, but each part of ecosystem is molded by various laws, private law and public law. In law, we tend not to see the ecosystem as deserving of integrated study—we tend to atomize each piece.

Law influences direction by structuring relationships among firms, individuals, and markets (VC contracts are an example: generally applicable law rather than innovation specific); by delineating entitlements (creation/enforcement of legal rights or creating space for nonmarket innovation); creating or limiting markets—shaping demand (export controls that limit the addressable market for certain tech); subsidizing investments, generally through gov’t spending or the equivalent (gov’t grants).

Harder cases: corporate governance—that shapes both individual firms/people as well as markets as a whole. R&D tax credits; gov’t procurement—both a subsidy and market-making. Regulation—can prohibit and can structure what people can do, and direction of innovation responds in kind. IP may operate across modes. Delineating entitlements and subsidizing investments by overcoming public goods problems, but also shapes market, e.g. by identifying what’s protectable and unprotectable and by shaping licensing law.

Sapna Kumar: how does the model handle goods that more closely resemble public goods than private goods, like new antibiotics and vaccines for infectious diseases less common in the US now (but might be in the future). Infrastructural goods might be different.

A: the market paradigm tells us the underproduction problem is bigger with such goods, so it wouldn’t surprise him if things like regulation/gov’t procurement play a bigger role there.

Rosenblatt: people are uncomfortable with value judgments: which technologies are value-promoting and which aren’t.

Kumar, Scientific and Technical Expertise after Loper Bright

Her impression: majority fundamentally misunderstood intertwining of sci/tech expertise w/statutory interpretation. Lessons we can learn from patent litigation about how judges acquire needed knowledge about technical issues. Proposal: funding neutral experts for appellate courts for complex administrative law cases.

Loper Bright: Judges must independently interpret statutes. Judges must exercise independent judgment when they interpret ambiguous statutes; adopted very narrow version of Skidmore deference—can’t defer to agencies but can look for persuasiveness, doesn’t seem different from being persuaded by a party or amicus. Claimed holding based on APA, but framed decision w/in Art. III arguments—invoked Marbury twice.

Many prominent scholars claim lots of deference is left, but her own view is that lots has changed. Courts are forbidden from deferring. LB might exacerbate politicization for judges. Some judges may try to improperly defer. But what about judges who try to faithfully follow LB—what is a deference-free system of interpretation?

Patent has some lessons! District judges struggle to understand complex sci/tech, which contributes to patent cases being time-consuming. They cope by relying heavily on expert tech tutorials, technically trained clerks, tech-trained magistrate judges, and neutral experts for complex cases, sometimes special master or technical adviser. How does the CAFC cope? Expertise of the judges themselves, and technically trained law clerks.

We don’t have any of that for APA situations. No means for parties providing tech tutorials; appellate courts have few technically trained clerks; page limits on briefs and non-tech trained lawyers make it hard for judges to learn what they need; no means for hiring neutral experts at appellate level.

Neutral expert proposal has been floating around for a bit: Breyer & Leventhal, JJ, both proposed. Common in other countries, EU and England. Best of both worlds—independent judgment of judges & access to reputable info on sci/tech.

Implementation: Congress could fund it up front (unlikely); appellate courts could do this on their own and bill parties for cost (is this allowed?). Experts would explain background sci/tech, not opine in law. Give teachings in written form, subject to party response. Goal to prevent overreliance and opacity, while preserving adversarial system. Patent system suggests that this can work. This isn’t a substitute for Chevron, but it’s something.

RT: Could an appellate ct send to dct for this purpose?

A: would take amending the APA.

Burstein: can appellate courts appoint special masters?

A: seen conflicting things—need more research into their inherent authority.

Jacob Sherkow: there are original APA actions in the dct—those could have experts.

Q: generalist judges are always coming into situations they don’t know anything about—music, reinsurance markets, complex family situations, how do ski resorts work. That doesn’t mean anything you said is wrong. But why is science special?

A: dct often plays a filtering role.

Q: problem where you can’t find an expert who isn’t in one camp or another—an interest in advancing (or not) large hadron collider projects. Maybe you need to hire 2-3 with different ways of looking at the same thing.

A: that’s true.