Session 3: Licensing I
Moderator: Yonathan Arbel
Jonathan M. Barnett, “Why is Everyone Afraid of IP Licensing?”: Conventional view—be wary of encroachment on public domain from licensing transactions. Medley of limitations lay minefield for construction of transactions w/minimum legal risk. Preemption, misuse, naked licensing/assignments in gross in TM; first sale, exhaustions. Legal scholarship thinks these limits are good and courts should be more vigorous about enforcing them. IP markets haven’t agreed, nor have lower courts until recently—attenuated or abolished by courts. 1976 Copyright Act abolished the doctrine of indivisibility. 1988: Congress amended Patent Act to say that some misuse claims needed a showing of market power. Exhaustion/first sale commonly detoured around by recharacterizing sale as a license. Antitrust has treated non-price-vertical restraints under rule of reason, including IP licenses; now includes price restraints. Freely divisible and tradable: a core element of every robust content and tech market. Lawyers have mostly figured out ways to detour around these limits.
But out of 6 license-related decisions since 2006, 1 was split, Monsanto favored patentee, but 4 decisions strengthened these limits, even specifically rejecting suggestion to reconcile IP law with antitrust’s rule of reason. Thesis: both recent decisions and dominant legal scholarship misappreciate/underappreciate critical role played by licensing in content and tech markets and that function is predominantly efficient. Licensing is critical as an enabler of commercial transactions that would otherwise be infeasible, given risks of expropriation from unrelated third parties. IP scholarship in general focuses on incentives to upstream innovator, but real world markets only generate value by embedding that innovation in a product or service for the end user, and that requires a multitude of costly commercialization steps that require expertise, almost always carried out by actors who care only about delivering value to shareholders, not the things that matter to artists and scientists.
Three main categories of efficiency gains through licensing. Thesis of the paper: secure IP rights + secure licensing deliver these sources of value. Efficient supply chains; efficient risk diversification; efficient fractionalization. Most important: supply chains. Traditional story of using license to encroach on public domain always assumes that the licensor has market power, dictating prices and nonprice terms to market, but that’s the exceptional case, not the typical case. Most patents and © don’t have value. Even a powerful standard-holder is often subject to competition that exists or could be introduced; the history of IT markets shows that dominant standard-holder often gives away at zero or below market royalties in order to grow the market/sell related services. Antitrust would require market power to even begin thinking about liability. Can’t presume market power in a patent claim; IP should do the same.
Next step: revisit these limitations and ask the added value of an IP-law specific limit on this practice given that antitrust law already regulates it?
Licensing transactions can be evaluated: (1) is it one-way or reciprocal exchange of knowledge? (2) Is it horizontal or vertical? Anticompetitive risks are higher in horizontal. Any tech/content market, to extract value from innovation, has to go through different levels of the supply chain. Classic decision: make or buy. You will choose the cheaper one. But when you choose buy, you have expropriation risks identified by Arrow’s information paradox. License + IP right is modularization, but must be combined with contract that allows you, not necessarily to expand the grant you got, but to fine-tune the relationship in customized way to regulate info flow between third parties who otherwise lack reliable instrument to do that. Combination of secure IP + secure licensing is good because it maximizes the total universe of transactions because a priori we have no idea what the efficient supply structure is in any given market. Licensor has that info and will adjust that structure as long as it’s subject to competition (or even not; it will do so to pocket more gains for itself).
Semiconductor industry: 25 years ago they’d always make; 25-30% of manufacturers today are fabless. How did they enter? Didn’t get $1 billion to set up new fab, but stayed in design market and monetized by licensing—dependent on the licensing structure, not encroaching on the public domain. Licensing obliterates an entry barrier.
Risk diversification: tech markets & content markets have extreme skew problems. Most movies are losers; hits cover the losers. How can you spread that risk? Create a portfolio, whether internal (studio system) or external. Biotech market now looks like Hollywood: risk spread by external portfolio; innovators shoulder the risk and contract w/large incumbents who have scale to carry out the rest of the supply chain.
Fractionalization: Combine divisibility with full alienability: you can slice and dice among multiple users; field of use restrictions; time delay. Lower entry barriers into industry by allowing IP owner to sell off among universe of licensees who can fund distribution costs up front (movies). Field of use restrictions/windows: this is just price discrimination, whose effects are ambiguous in theory but are efficient here, eliminating deadweight loss.
Commentator: Brett Frischmann: There’s some description here; factual claims may be in need of empirical support. How are you defining innovation? Hard to evaluate assertions; also doesn’t think that traditional scholarship ignores claims about the benefits of licensing. Us/them, black/white framing may not be helpful in discourse, even as to private law/public law.
What’s the theory? Is it refutable? Can it be evaluated/tested and does it have boundaries? Is it a normative claim? Yes—suggests that we should adopt a new view supportive of IP licensing. But most scholars understand that IP enables licensing in various segments of the supply chain. As a result of framing, the if/then statements in the paper collapse. If we understood the facts, then we won’t necessarily adopt this new view of licensing: I need more proof. Different baselines might give different answers. IP isn’t fundamentally about transactional efficiency or market-based metrics of efficiency—that’s important, but other things are important too. Even understanding the facts Barnett offers, judges might disagree w/the baseline.
Even w/the same baseline, we might disagree—not clear that these examples generalize. Hollywood, big pharma—but does that reach a wide variety of other industries, markets, and nonmarkets that shape/rely on IP? Not clear that examples prove the point about efficiency—do we believe that Hollywood and big pharma are efficient structures w/which to produce movies. Laughed when I read “the market evidently prefers hub & spoke.” Did you interview the market? The idea that the market speaks about what it prefers as if it’s not shaped by the law is confusing to him.
No discussion of spillovers: how can you talk about supply chains w/o spillovers? Even w/efficiency—depends on free flow of spillovers; there’s lots of empirical/theoretical work on this. IP is a little bit about internalizing externalities, but also about promoting participation in activities. Some of the limits on licensing freedom might be welfare-enhancing and even efficient if they promote spillovers.
Barnett: I think the literature does focus on the expansion of the monopoly by licensors—SCt decisions do that—I want to shift the focus. Distributive effects: price discrimination; semiconductors—when you don’t allow free licensing choice you limit transaction structures, forcing them to take place in house which can raise capital costs/favor incumbents. On the market’s preference: we’re agnostic about market structures, and as long as there’s free entry, whatever structures we observe are efficient, so I don’t need to interview Sony or “the market.”
Arbel: Market power (lack thereof) is key to your analysis. Circularity: if you allow that, won’t market power increase/the empirical world may change. Comparative institutionalism: compare other jurisdictions, where there are different licensing rules; could show inefficiencies in those industries perhaps. Do you see a problem of the anticommons, fragmentation?
RT: In Barnett’s model, what are TMs for? What do they incentivize the creation of, because the model of efficiency offered in the paper depends on incentive structures and not on consumer understanding? More generally, why class patent, TM, and © together here with antitrust law (and notable that most papers here don’t cover TM, which suggests some uncertainty about the overlap between private and public law here—larger question for the group, is TM already private law so we aren’t worrying about it here (I have thoughts about why that shouldn’t be true)?), and what about trade secret, misappropriation, or advertising law? Relatedly: Different kinds of wrongs to consumers and competitors exist—antitrust law doesn’t recognize most false advertising as actionable, or most product liability (I think), but that doesn’t mean those torts shouldn’t exist.
John Golden: You argue we shouldn’t do more than antitrust, but do antitrust scholars disagree? Herbert Hovenkamp wants to shift the problems of regulation from antitrust to IP b/c he thinks antitrust isn’t designed well to deal with the problems of IP.
Van Houweling: paper sets up troubling mismatch b/t critiques of licensing and benefits of licensing. That’s a problem if there’s a complete overlap. But we aren’t skeptical of all these licensing practices; the skepticism isn’t that thoroughgoing, but covers overreaching rather than all licensing. Would be more convincing with some examples of beneficial licenses that the critics undermine.
Barnett: Mismatch is worth addressing, yes. Division of labor b/t antitrust and IP—do they have the same normative objectives, maybe we shouldn’t have overlap, but otherwise it might be ok. Major antitrust case of the last 3 decades is Microsoft, all about IP. Market power is always important b/c you’re taking terms from the market rather than dictating terms.
Greg R. Vetter, “Opportunistic FOSS Development Pathways”: Modular, in H. Smith’s sense, not technical sense. Software licenses deploy permissions against a set of rights—©, trade secret, patent, but mostly ©. Whatever license you have targets some type of opportunism. License might work with or against the rights basis. FOSS/GPL works against the rights of © in that it uses © to enable copying and to defeat trade secrecy in the source code. One person’s opportunism is another person’s business model. Some modes of opportunism are illegal. Licenses operate as quasi public instruments.
Forking is allowed; GPL is hard to get out of—locks usable value of software to the license b/c there are so many contributors and it would be hard to get them all to agree to a different licensing scheme. Property as a shortcut over many contracts that would otherwise have to take place. Small startup can shift to permissive license, but hard to shift back; could start a permissive fork.
Commentator: Christina Mulligan: Are incompatible licenses a problem of property at all? Potentially separate issue—could be a problem of modularity or a problem of anticommons. The worry is that public licenses create a bad interface b/t pieces of property blocking us from bringing 2 pieces of software together. The public license didn’t cause the problem; companies can license a work under a public license such as the GPL, and then if there’s some other reason they might want another deal, can license it to specific people in a proprietary/specific way. Conceivably, you could imagine a license that tried to be an exclusive GPL/public license, but she doesn’t think any license purports to do this. Would create the same problem as the fee tail, where no combination of people could make a separate deal; our commitments make us not want have attempts to control property like that work. The software: can conceptualize it as the software as it currently exists and as its possible derivatives, but she thinks that’s wrong. The property right can include the work and the right to make derivative works, but not the derivative works that don’t yet exist. Breeding racehorses: you want to blend them together—we wouldn’t about whether it’s A or B; we can see it’s a third thing and the owners can agree about how to share rights in the new thing. Pre-commiting to waiving your right to exclude in a certain circumstances doesn’t redraw the boundaries of property; it’s not a problem of defining the property interest correctly b/c you can always make side deals to license, which means that the fundamental problem is one of the anticommons.
Vetter: one point to remember is that GPL uses © to defeat trade secrecy—this makes me think there’s more instability here than there otherwise might be.
Smith: One way of reconciling this idea of anticommons v. modularity: maybe here your identification of the difficulty of modularity arises not up front but after the fact. You have an intersection b/t the running covenants problem w/confusion (you have grain, I have grain, they intermingle) or conceivably accession (if you don’t have a deal and there’s a new calf, it belongs to the owner of the mother)—the reason we don’t worry about that in property is that even ex post we can modularize really easily—to the mother; divide pro rata; punish the person who caused the problem. Not as difficult as intertwined software with conflicting licenses. There is an anticommons aspect but it’s a lack of modularity, b/c even after the fact we can’t draw lines about contributions very easily.
Michael Abramowicz: Biggest concern justifying viral license was risk of proprietary fork that would add so much it would be hard to avoid; GPL is attempt to deal with that risk.
Vetter: thought that forking was usually a result of disagreement, not a purposeful deviation.