Harvard Law School
Opening
Keynote
Yochai Benkler, Harvard University
How do we understand what we do in our field against the background
of profoundly increased inequality and stagnation for all but the top 5%, including
an unprecedented increase in death rate among US whites from diseases of
dispair? The economy functions in relation to power. Prior stories that have
failed us: Mainstream economics treated technology as exogenous; it has become
skill-biased. Monotonic story: the more skilled people are, the more they can
adapt to tech, so college degree gets a premium. Fit the 1980s US data
beautifully. Institutions played a relatively limited role in this theory and
we were supposed to educate workers. But that story didn’t work in the
1990s. Early 2000s: “tasks framework”:
technology still exogenous, deterministic (some things are easier to automate
than others). Middle class jobs—people in the back office of the CPA—were now
being done by Excel. People on assembly line replaced by robots. Nonroutine jobs—the
janitor dealing with irregular floors, the CPA using Excel—still had demand. But that still didn’t fit the overall
differences b/t different countries, all at the same tech frontier—substantial empirical
threat.
Then the story became: robots will take jobs, cause
sustained levels of structural unemployment; all work will be Uber. Insitutionalists fought back: politics
led to sustained shift of bargaining power b/t labor and capital, b/t financial
sector and others; free movements of goods/capital but not people across borders
created new methods of offshoring, outsourcing, etc. Tech plays no independent
role in the institutional story.
So, where do we stand, who do law/institutions but also
believe that tech has an important role?
Innovation economics/IP scholarship has some familiar forms. E.g., Classic/neoclassical
incentives/access model; Schumpeterian: market structure is more important, antitrust
is the key area of law; industrial policy intended to have exogenous effect. Institution
design can seek better innovation w/o attention to distributional effects. Or
you can have distribution sensitive innovation policy—primarily about access;
accepting a market society and looking for redistribution through rules. Market
systems will predictably not deliver access to certain goods, e.g. treatments
for tropical diseases. Central to Reichman’s work. How to structure markets and
more importantly innovation systems to deliver outcomes that are not maldistributive.
Next to this: people looking at tech rather than at IP,
sounding more in STS than in economics or law (Winner, Do Artifacts Have
Politics?, Nissenbaum, Bias in Computer Systems/Values in Design, Lessig, Code),
focusing on the fact that technologies settle politics, whether in the form of
bridges that block access to public beaches or machines that were less
productive than workers but still managed to break unions, tech is used to
resolve social conflict in favor of the party implementing the tech.
Look at ©, patent, TRIPs, WIPO treaty wars: the experience
is of active, knowing, strategic action both by firms and individuals, alone
and collectively, to shape the contest they understand will give them greater
or lesser market power. Rent seeking, particularly on the side of firms seeking
to increase the value of first mover advantages. (For Schumpeterians, rents
aren’t inherently bad—they drive innovation—but it matters how long they will
last before competition comes in.) Individuals come in with: freedom oriented
and other social values. Trying to nullify state capture. Strategic lobbying on
DRM, WIPO, etc. to create technologies that allow incumbents to control
entrants. Today: surveillance capitalism, dark patterns/manipulating consumer
demand.
Markets are arenas where power is negotiated in context—actors
spend some of their rents to create the context that will make it easier to
extract rents next time. Tech is endogenous and plastic. There’s enough play in the joints that how it’s
designed makes a huge difference including for bargaining and distributions.
Markets are usually not efficient and power therefore matters. Institutions,
tech, and ideology each shape policy.
Hired scholarship on one side justifying extractive practices; academics
trying to do something else—like Reichman on the relationship b/t academic
science and innovation. Shapes knowledge framework to shape the institutional
framework.
New institutionalism in sociology: models of practice are imitated/normalized
across groups.
Microfoundational component, just looking at the firm: every
firm faces a choice of action in any given context. It can increase
productivity (becoming more competitive) or decrease competitiveness (make products
for which demand is less elastic). Tech and institutions are part of the
investment pathways. Each pathway has an ideology too. They’re constantly
trading off options to shape future rents. Options: increase entry barriers, prevent
entry by innovators, increase competition among workers in the labor market and
decrease competition among employers, increase monitoring of employees, disrupt
collective action among employees/consumers, manipulate demand among consumers.
DRMs, terminator gene, rent maximizing price discrimination are all mechanisms
to control demand and disable competition. Older work on ways in which tech is
used to deskill and homogenize, to monitor labor, to decrease labor’s
bargaining power. “Ghost Work” by Mary Gray—platform organization makes it
harder for workers to organize, and that’s part of the point. It’s not just
about extraction; it’s in part about productivity and it’s also about ability
to strategically grab rents. In each
relation, the objective is to decrease competition/entry, increase the size and
lifetime of the strategies and the rate of rents; larger rents give you more
bargaining power to capture more.
All this explains dramatic tech innovation/growth in one
sector along with higher rents, more markups, more concentration of firms and
wealth and increasing economic insecurity for everyone else—once the state is
removed as counterbalance through regulation, once unions are weakened through
labor law, then firms can dedicate more resources to rent seeking and rent
extraction over productivity.
Q from Reichman: What is to be done?
A: (1) Knowledge production is part of the process and
having a different story about how tech works opens up different avenues; many
regulators today are highly constrained by fear of upsetting the applecart by
regulating/pressuring tech firms. That’s changing a bit. Having a better story that understands
structural/distributive things at play in tech policy, not just more is better,
is one part of moving forward. Invite distribution sensitive innovation
theorists to look more at extraction and competition. (2) Invite a multisystem
approach: look in the mirror and worry. He’s spent plenty of time being
skeptical about the state, trying to find self-organizing approaches like free
software. That’s not good enough. That’s not to be nostalgic for the 1950s and
60s—there are excellent reasons that model of regulation lost out in the
developing world. We need to find new ways to make regulation more effective—democratically
accountable but flexible sources of counterpower. Consider: who can actually do
this? What kind of actors can serve as counterpower? What innovations will
undermine the financial power of certain actors?
Maggie Chon: building on rent extraction being not
inherently bad—is it reinvested in R&D and innovation generation that might
lead to more growth/inclusion of labor?
Q: rate of innovation v. direction of innovation—are we even
capable of directional?
A: one of the outcomes of understanding things this way is
to get comfortable nudging tech development into different structures. There’s
no reason to think we should focus on rate; if we focus only on rate based on
letting market actors choose whatever they want we aren’t doing our jobs.
Social/political consequences can be vastly different and disruptive. Power of
labor, 15 years after you go to robot nurses versus PAs with computers in their
pockets, is completely different. Understanding how that relates to the broader
framework may make you demand to see the interactions b/t “rate” and the power
of incumbents.
Reichman: but where does the counterpower come from?
A: Democratic party in the US has changed a bunch in 20
years. Green New Deal. EC’s willingness to regulate tech now compared
to its willingness to hand out monopolies in database rights 15 years ago is
different.
Reichman: so institutional tools can evolve into
counterpower.
Do antitrust and competition law trust intellectual property
law too much?
Rebecca Tushnet, Harvard University, session chair
Graham Dutfield, The evolving role of branding in
pharmaceutical management: how should competition law respond?
Are there things other than patents that raise issues with pricing/competition?
Looking at TMs. Norlutin: on the market in UK since 1936. Pfizer sold marketing authorization to Flynn;
Flynn debranded the product and escaped all price controls. Pfizer became the
sole supplier of the active ingredient to Flynn; Pfizer raised the price and
Flynn did as well (and then some). Then Flynn sought to block importation of Norlutin
from Italy based on its rights. Competition authority imposed a fine, but they
appealed and won. Trademarks do matter.
In historical work, patent based rights are often subordinate
to brand based rights. Dyestuff/chemical companies onward.
What about the science?
Plain packaging of pills conveys little info. Text/visuals affect how consumers perceive
and even react to the products. Bayer puts its cross on its aspirin. When TM litigated, Learned Hand says “aspirin”
has two meanings. Nowadays pharmacos use colors as well. When it’s transferred
to a me-too product that has a completely different price, what happens? Zantac product expansion.
Looked at old cases on visual aspects of pills, UK case in
1972: court recognizes rights in “getup” or coloring. Some problems: AstraZeneca, which has gotten
in trouble w/competition authorities over Nexium, has used color to protect
rights.
How does this matter?
Tentative conclusions: Look further than patent system. Competition law
should reflect this broader focus. Branding
can actually generate confusion about the product provenance/uniqueness, and
that’s a problem.
Maybe shape/color shouldn’t be protected/shouldn’t be
enforceable against authorized generics for the same conditions. Placebo &
nocebo effects are relevant here; if the color matters, it should be
transferred to the generic drug to avoid artificial creation of nonsubstitutability.
The consumer may reject/experience less effect from the generic.
Sean Flynn, Enforcing Fair Following Rights Through
Competition Law
Confronted an argument in assisting South Africa: you can’t
use competition law to force licensing of patent rights b/c they are completely
separate fields. Reichman was
foundational in response to that. Also relevant to ©: rights to education and
research w/in the © system can also be seen as competition issues.
How do we craft the interface b/t IP and competition? Louis
Kaplow has done work on finding a realist/realistic proposal. US courts have
looked at the “scope of the patent” but that’s what Felix Cohen would call
transcendental nonsense. If competition law cabins the power of the patent
holder, you can’t use the scope of the patent to determine when competition law
operates: it has to have the ability to operate on the core of the rights,
including the right to exclude. Kaplow said it’s basically a ratio. The more
you have a small patent reward and a large monopoly loss, the more competition law
should bear on the situation and vice versa.
Can we identify scenarios where one or the other will typically be the
case?
Where monopoly power is on essential goods/services w/
highly inelastic demand curve, and a country has extremely high income inequality,
it will almost always be the case that full exercise of monopoly power will
create small monopoly reward and huge deadweight loss. Edmund Kitsch was wrong:
we didn’t see, after TRIPs, proportional reductions in prices in countries with
less income. It turns out that charging
high prices in India is still what happens. In the US if you charge a price
that only the top 10% can afford you make less than if you charge a price to
allow 60% to afford it. In South Africa, though, there’s a small # of people
who earn global high incomes and a large # who have very low incomes, and the
profit maximizing price turns out to be to serve only the small #. You might even make more profit having a
higher price in a poorer country because you’re not serving much of it.
We also have textbook examples: books, like drugs, have
essentially zero marginal cost to distribute. But b/c you’re not trying to
serve the entire market, a higher price earns more profit. Exclusionary pricing:
more profit through massive deadweight loss. 2-3x more than monopoly profit. We
should be restraining that through competition law (or other means).
How to do it?
Competition, fair use, international law. In South Africa, they forced the licensing of
AIDS drugs, and after that the advice IP lawyers began giving to drug suppliers
was that any pharmaco has to give at least 4-5 licenses to local producers to enter
the market. So we shouldn’t look at the
scope of the patent, but at whether there’s massive exclusion in the market.
For ©, in South Africa there are affordable locally produced
textbooks, but not for niche topics. They can be 10-20x higher priced than normal.
Current SAfrican bill, on the President’s desk now, adds a competition standard
to fair use, allowing use of entire work where authorized copies can’t be
obtained at a price reasonably related to the normal price for textbooks in
South Africa. Used language from the Berne
convention that was highlighted by Reichman.
Duncan Matthews, A Pro-Competitive Strategy for Developing
Countries: Do we still need this and where do we go from here?
Reichman’s work on flexibilities in TRIPs. Not just the use of competition law per se,
but invoked Art. 7 & 8 of TRIPs. Art. 8 para. 2 creates potential to
prevent abuse of IP rights. UN panel on access to medicines referred to this as
underutilized. OECD: competition
committee, where national enforcement agencies are looking closely not only at
excessive pricing and pay for delay and vexatious litigation but also at
patenting strategies, patent thickets, defensive patenting, divisional
applications causing noise & confusion to increase uncertainties for
generics. TRIPs council: South Africa
has taken the lead, supported by China, India and Brazil, acknowledging that competition
law is one of the least discussed flexibilities and asking for information
gathering.
European perspective: the EC is led at WTO/TRIPs council by
DG Trade, which has pushed back against discussion of this topic, which is somewhat
surprising given what DG Competition has been doing in Brussels. Pharma inquiry by DG Competition raised patent
strategy issues as recently as Jan. 2019.
Policy incoherence in the EU prevents the issues from being discussed in
progressive way at TRIPs council; can be harmful to regional/EU discussions as
well. For developing countries, simple information gathering is being blocked.
Reichman’s ideas remain as topical and crucial as they did when TRIPs was new.
Pam Samuelson, The Infusion of Competition Policy in
Copyright Law
When they were starting out, the idea of considering
competition policy in © was relatively new. Reichman’s work on computer
software as applied knowhow was significant. Reichman argued against
overprotection via patent/©. At the time, Whelan v. Jaslow had suggested that
only the general purpose or function of a computer program was unprotected by ©
and everything else was protectable look and feel. Nimmer’s treatise tried to rewrite the
statute, which is not the way it’s supposed to go. Reichman & Samuelson
started writing amicus briefs and getting more active—Lotus v. Borland, where
the dct protected the command menu hierarchy of a program even though the dct
also said that was the fundamental part of the functionality of the system. The SCt blew it by splitting 4-4 when it got
to them, and we wouldn’t be in this Oracle mess if it hadn’t, but at least the
1st Circuit got it right. Likewise, worked on Sega v. Accolade to
allow decompilation to get access to un©able subject matter. This was an uphill battle! The 9th Circuit tracked their
amicus, yay. The court said that decompilation
was important to innovation and competition—likewise the Altai case. The cycle of overprotection seemed to have
come to an end.
But then there’s a threat of underprotection; recent
decisions in CAFC in Oracle v. Google overreacts to give us Whelan v. Jaslow
all over again. So it’s still amicus
work for Samuelson et al. Wrote a paper:
if we’re worried about cycles of over and underprotection, we should think
about what kind of protection programs really need—the industrial compilation of
applied knowhow is what’s valuable about them.
She’s still convinced that’s right. They’re not suitable for ©; they
should be protected, if at all, through a sui generis right. She’s not expecting
that, but we should think it through regardless.
Peter Yu, Revisiting the Historical Lines of Demarcation:
Competition Law, Intellectual Property Rights, and International Trade When
TRIPS Hits 25
EU proposal on data producer’s right for anonymous data.
Japan actually did change its law in 2018 with respect to big data. “Data is the new oil”: misleading but
persuasive description. Unasked important question: should data be owned in the
first place? Reichman & Samuelson have addressed this in an important
article dealing w/ the modality of protection. There are methods of regulation
between nothing and property rights, including unfair competition law.
Complementary operation of competition law: US filed an IP complaint
against China, its second under TRIPs, 2 years ago, that could have shed light
on these important issues. National treatment (art. 3) and about
license/contract in patent law. Some of the potential defenses come from art.
40, abuse of IP rights. Technology transfer and joint venture issues have been
important for decades, before TRIPs. In
June, however, they suspended the complaint so we won’t have a ruling.
Traditional knowledge: do people overstate the importance of
protection? Does competition law have
the same relation to TK as to other IP—do we want to encourage more
competition in this space?
Reichman’s 1983 article on design and the law: foundational.
Justin Hughes: is Altai about as good as sui generis would’ve
been?
Samuelson: yes, given that there would’ve been issues w/sui
generis protection too. Cases mandating
filtering algorithms out as unprotectable procedures; interoperability as
unprotectable procedures improved the Altai standard. As long as there’s
meaningful effort to filter out the functional, it’s not so bad an outcome. The
software industry is doing well. Microsoft endorsed flexible fair use in its
amicus supporting cert in Oracle, which is not where Microsoft was in the
1990s. Fed Cir threw out 102(b), merger,
and fair use, and you can get to the Fed Cir by alleging a patent claim; there
are no defenses left in (c) (they would get rid of the scenes a faire doctrine too
if given the chance).
Q: PTO asked for comments on protection of AI. Asked if we
need data protection. Seems quite concerning: comment period extended to Nov.
8.
Samuelson: she’s heard noises that extensive protection for
structure, sequence and organization is necessary because patents are being
invalidated under Alice; she thinks it’s healthy and that the industry doesn’t
need patents as much as all that. There
should be a layer that is unprotectable except for trade secret; that’s the
most sensible.
Reichman: was puzzled by Alice, which seemed like a move to
the European approach (technical step/solution) but stopped short of it, which
makes it conceptually worse.
Samuelson: Alice involved 10 judges with 7 opinions, which
was part of the problem. Fed Cir reacts
by saying that SCt threw out patent, so © must step into the void.