HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

51
Dark Markets, Bad Patents, No Data Institutional and Information Failure in the Digital Economy Brian Kahin Fellow, MIT Sloan School Center for Digital Business Senior Fellow, Computer & Communications Industry Association Digital Initiative Harvard Business School March 26, 2014 This version enhanced for those not present at the event.

description

This presentation for the Digital Initiative at Harvard Business School looks at how digitization interacts with the patent system and the systemic dysfunction that results. This version is annotated for the benefit of the reader.

Transcript of HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

Page 1: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

Dark Markets, Bad Patents, No Data

Institutional and Information Failure

in the Digital Economy

Brian Kahin

Fellow, MIT Sloan School Center for Digital Business

Senior Fellow, Computer & Communications Industry Association

Digital Initiative

Harvard Business School

March 26, 2014

This version enhanced for those not present at the event.

Page 2: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

digital technology effects

• in most economic policy domains: financial regulation, tax arbitrage, regulation of services, telecom regulation, competition policy, intellectual property….– an enabler (usually taken for granted)– and source of disruption, opportunity, arbitrage,

disequilibriuum….

• effects mediated/amplified/extended by firms– but institutions and policies typically slow to

change

Page 3: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

• examples: financial system, patent system, US healthcare• mix of public agencies, institutions, markets, instruments,

and relationships– less “system” and more “ecosystem” :– many points for private ordering and state intervention

• mix of different languages, different logic – especially in patents– technical explanation directed at “person having ordinary skill in

the art”– claims: legally operative terms that require interpretation (“claims

construction”) – algorithmic language of code– globalization

complex public-private systems

Page 4: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

need for multidisciplinary approachto public-private systems

• law and economics

• political economy

• applied transaction cost analysis

• institutional/policy design

• complex systems

place in the curriculum?

Page 5: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

digital economy fundamentals

• exponential– increasing processing power, bandwidth, storage– rapid but unpredictable change

• digital– content, communications, algorithms/logic

• combinatorial– layers, modules, systems, platforms, functional

integration, things

Brynjolfsson and McAfee, The Second Machine Age 2014

Page 6: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

combinatorial(can be viewed from two directions)

• bottoms-up functional aggregation and integration

• top-down (architectural)– platforms

• degrees of openness• DoD, IBM, Microsoft

– modules and interfaces– standards

• compatibility vs substitute• function vs implementation• public vs private

Page 7: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

digital economy fundamentals

• exponential– increasing processing power, bandwidth, storage– in general, rapid change over many years

• digital– content, communications, algorithms/logic

• combinatorial– layers, modules, systems, platforms, functional

integration, things

• connected (a critical 4th fundamental)– global reach, addressing, search; boundary-spanning,

complements

Page 8: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

primary/secondary/tertiary effects(a quick sense of the complex interplay)

• boundary spanning >> category blurring• perfect sharing and global scaling• competition for the market• cheap transacting >> open innovation

– efficient markets– new business models, transformed markets

• modularity and complementarity – relations across boundaries

• algorithm-based management of data, information, knowledge

• documenting, monitoring, measuring, and modeling • asset sharing/partitioning

Page 9: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

first-order policy implications

• exponential -- schizophrenic response to rapid change– rapid reaction: copyright, obscenity – inertia: patent law, taxation, antitrust

• digital – radical change for content and communications (confronts copyright, free expression, telecom)

• combinatorial – too complex for intervention

• connected – discourages local intervention, promotes harmonization, creates transborder tensions/arbitrage

Page 10: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

• copyright – radical enablement of intermediaries and end users (owner-

enabler-user); reuse and piracy– policy response: “notice and takedown”

• trademark– conflict with domain names (network admin)

• domain names: global, generic, unlimited…• Internet governance

• trade secret– rapid dissemination by insiders, third parties… (similar to

primary copyright problem)– hacking, industrial espionage, cyberconflict– tool for discouraging executive mobility

impact on intellectual property regimes

Page 11: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

implications for patent system

• exponential – unlimited capacity for software functionality– accelerated time-to-market and product cycles

• digital– algorithms as “basic tools” (at what point do “basic tools”

become applied and patentable?)– enables content, process, and communications covergence

• limited by patent portfolios (smartphone wars)

• combinatorial (complex products/systems)– vast functionality at multiple overlapping levels drives patent

proliferation– content, communication, GPS, sensors, other technology….

Page 12: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

Colleen Chien, Functional Claiming, Presentation to the 2/12/13 USPTO Software RoundtableColleen Chien, Functional Claiming, Presentation to the 2/12/13 USPTO Software Roundtable

problem of structured complexity in “combinatorial”

Page 13: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

implications for patent system

• exponential – unlimited capacity for software functionality– accelerated time-to-market and product cycles

• digital– Benson: algorithms as “basic tools” (at what point do “basic tools”

become applied and patentable?)– enables content, process, and communications covergence

• limited by patent portfolios (smartphone wars)

• combinatorial (complex products/systems)– vast functionality at multiple overlapping levels drives patent proliferation– content, communication, GPS, sensors, other technology….

• connected– extreme vulnerability of networks ($612M Blackberry settlement)

– global markets, global “patent warming”, global prior art

Page 14: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

up 66% in last 4 years

Page 15: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

updated from Bessen, A Generation of Software patents (2011)

up 100% in last four years!

Page 16: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

rapidly rising litigation for software-related patents

Page 17: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

first presidential initiative on patents in 47 years – and only two years after the America Invents Act

Page 18: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

White House report:• PAE suits are concentrated in the IT sphere;

according to one estimate, 82% of PAE defendants were sued on the basis of a software patent (in contrast to only 30% of those sued by non-PAEs) (Chien and Karkhanis, 2013).

• Software patents are nearly five times as likely to be in a lawsuit as chemical patents; business method patents are nearly fourteen times as likely (Bessen 2011).

Risch 2011:

• Average PAE suit is filed 8.3 years after patent issues

Page 19: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data
Page 20: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

percent of patents allowed.JPG

Cotropia, Quillen, and Webster 2013

difficult to get at it because of lag and continuations (in the U.S. an application cannot be conclusively rejected); authors had to get underlying data via FOIA request

Page 21: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

“Best places to work in federal government”

PTO moves from 105th to 1st place from 2009 to 2013

as allowance rate rose, patent office became a more pleasant place to work

Page 22: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

“The U.S. has had a 230-plus year love affair with innovation. It started with our Constitution, in which our Founding Fathers made patents an affirmative right the government is required to grant to anyone who meets the legal requirements.”

former Under Secretary of Commerce for Intellectual Property David Kappos, Stanford Technology Law Review, May 2013

But the patents clause is in Article I (enumerated powers), not the Bill of Rights; Congress is empowered but not required to grant patents. And patents are a negative right, not an affirmative right. They do not give owner the right to practice invention – only the right to exclude others from doing so.

Page 23: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

… [T]he defense establishment insisted on an intellectual property rights regime that was scandalously – and productively – loose relative to what has evolved over the three decades since then.

William Janeway, Doing Capitalism in the Innovation Economy, 2012

The landscape was once very different, especially in semiconductors and software.

DoD’s insistence on second sourcing and technology sharing was a major force behind the culture of Silicon Valley.

Page 24: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

• 1952 Patent Act, written “by patent lawyers drawn from the Patent Office, from industry, from private practice, and from some government departments.”

• (1966 President’s Commission – no patents on computer programs)

• (1972 Benson – patent on algorithm denied by SC) • 1982 Federal Circuit, created to hear all patent appeals and

make patent law more uniform– made patents easier to get, easier to assert, more powerful,

harder to invalidate, and more abstract….• 1990 USPTO becomes fee-funded

– 1996-2002 mission: “To help customers get patents.”– subsidy for examination creates incentive to grant patents

• 1998 State Street decision– business methods patentable, as well as software– activist opinion by judge who drafted 1952 Act

institutional background

Page 25: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

market developments in IT

• defensive portfolios, cross-licensing• portfolio monetization• secondary markets• assertion/monetization specialists• aggregators

– catch and release, sale (and buy back)

• privateering (surrogates raise rivals’ costs)• sovereign/state-supported patent funds• transaction costs as weapons

Page 26: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

historical average cost of processing, search, and examination for USPTO late 2000s: $3713

subsidized fee: $1600; $400 for micro-entitiesFrakes and Wasserman, 2013: Grant rate responds

to fiscal constraintsBUTcompare to costs facing accused infringer before

suit (AIPLA 2013)• validity opinion: $12759 • infringement opinion $11626

knowledge about patents wants to appear cheap

Page 27: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

average costs of full litigation (AIPLA 2013)

amount in controversy

legal costs per side

x2 (total) minimum legal costs

relative to

amount in controversy

>$25M $5.9M $11.8M n/a

$10-25M $3.2M $6.3M >25%

$1-10M $2.1M $4.2 >42%

<$1M $.97M $1.9 >194%

and ultimately

The patent system does not scale down efficiently!

Page 28: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

reaction to expansion

• 1999-2004 NAS study and report• 2002: 30 days of FTC/DOJ hearings followed by FTC report

(2003)• 2005: Intensive Supreme Court review of Federal Circuit begins• 2005-2011 legislation introduced, eventually passed as America

Invents Act– new forms of administrative review: post-grant, covered business method

• 2011 FTC report on evolving IP marketplace– competition agency focus on standard-essential patents (private

ordering): ambush, licensing commitments, injunctive relief, damages

• new round of legislation aimed at trolls• White House initiative• FTC launches study of patent assertion entities

Page 29: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

private ordering (standards)

• informal – FRAND commitments (“fair, reasonable and

nondiscriminatory” – but no guidance on interpretation)• royalty-free (RF) alternative, esp for software

– background arrangements: cross-licenses, nonassertion agreements, comity, MAD

– licensing overlooked for small players

• formal – patent pools: precise and costly to set up– ex ante licensing (disclosure of maximum terms in

advance – legal but rarely done)

• useful point of insight into patent markets

Page 30: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

“TI has something like 8000 patents in the United States that are active patents, and for us to know what's in that portfolio, we think, is just a mind-boggling, budget-busting exercise to try to figure that out with any degree of accuracy at all.”

Frederick J. Telecky, Jr., Senior Vice President and General Patent Counsel, Texas Instruments, FTC/DOJ hearings Feb 2002

widely cited comment made in the context of patent disclosures in SSOs but reveals much about the costs of patent-related knowledge [What about the hundreds of thousands of patents that belong to somebody else?]

Page 31: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

IT patents routinely ignored

“[B]oth researchers and companies in component industries simply ignore patents. Virtually everyone does it. They do it at all stages of endeavor. From the perspective of an outsider to the patent system, this is a remarkable fact. And yet it may be what prevents the patent system from crushing innovation in component industries like IT.”

Mark Lemley, Ignoring Patents (2008)

… costs leading to this widely cited comment by the dean of patent scholarship

Page 32: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

Fontana et al, Reassessing Patent Propensity, 2013

Out of 2802 winners of R&D 100 awards, less than 10% were patented.

Is patent rhetoric wildly overblown?

Page 33: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

“IP is fast becoming the world’s currency of innovation.”NBER, Innovation Policy and the Economy, April 20, 2010

“IP rights are the global currency for creating value for products and services, for all innovators, in all markets.” Center for American Progress, Nov 20, 2012

while Under Secretary Kappos championed the “financialization” of the patent system…

Page 34: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

multiple patent markets

• markets for technology [commonly assumed model]

– patents as marketable products (pharma) – patents as complements?– ex ante (“carrot”) licensing

• markets for enforcement [actually dominates in IT]

– based on evidence of use (ex post “stick” licensing); patent = option to litigate

• markets for freedom of action– mitigation of risk/liability

• profit-shifting among international affiliates– tax benefits

Page 35: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

“[T]here has been increasing value in capturing patents that have demonstrated value, that is, there are issued claims that you can show actually are infringed by folks.”

“Yes, unpredictability is a bad thing. But unpredictability is the only thing that’s allowing these patent owners to get the access to capital which allows them to actually try and get a return on the patents.”

-- Ron Epstein, CEO of IPotential, FTC hearing at Berkeley, May 2009

IP intermediary explains that secondary market is based “evidence of use” [opportunity for holdup]

and speculation:

Page 36: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

After Mansfield, then Levin (Yale survey), no large-scale survey since Carnegie-Mellon survey in 1994 (Cohen et al.)

Golden Era of Survey Data

surveys do not review well at NSF:

expensive and they’ve been done before

Page 37: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

NSF CSESBusiness R&D and Innovation Survey (BRDIS)asked in 2009 about patents and 5 other IPRs

• single question: "very important," "somewhat important," or "not important." [3-point Likert scale!]

• “important” – why? volume, value of individual patents,

defense, exclusivity, salvage value

• according to whom?– radical differences between lawyers and software

designers [two surveys in the 1990s]

Page 38: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

• “order of magnitude estimate” for industrial process licensing (Robbins 2009) – $50B/year revenue from domestic sources in 2002 pieced together from BEA,

Census, IRS data– compare $7.8B worldwide receipts by U.S. patent licensing establishments (Census

2002) includes trolls, spinoffs, etc.

• problems in data– patents vs trade secrets– assignment vs licensing (and exclusive vs. non-exclusive)– cross-licensing: net vs gross (barter reporting)

• non-assertion agreements

– expenditure data needed to balance receipts– copyright and patent overlap for software– in-house vs spinoff vs specialists– primary vs “secondary” markets– ex ante vs ex post (“carrot” vs “stick”)

• value creation vs rent seeking

data on licensing is dated, inconsistent, and problematic

Page 39: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

validity

• PTO study (Graham, Journal of Economic Perspectives, Jan 2013): Of 21 smartphone software patents evaluated by district court 17 held valid or likely valid.

• Compare Shawn Miller study (2013): – 210 cases with holdings on innovation-related validity: 49% of

software patents invalidated in whole or in part, compared to only 33% of non-software patents

• Strong presumption of validity can only be overcome by “clear and convincing evidence” so court decision does not correspond to PTO grant

• Selection bias? (not discussed in Graham article)– discovery of prior art leads to dropping assertions

Page 40: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

“In the last 5 years, Nokia has defended against over 150 patents in Europe. Many were dropped at an early stage. 71 were pursued to judgment. Of those 71, the courts have determined that only one was valid ‐ and even that was later revoked by the EPO.” -- Tim Frain, Nokia, April 2013

public comments by Nokia in context of industry concerns about Europe’s pending unified patent system (which promises to enhance the market for enforcement)

and these are European patents, which are reputedly of higher quality

Page 41: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

competing definitions of quality

• legal– whether the patent is upheld in court

• technological (inventiveness)• market value • indicators (patent text – “where the light is”)

– “Wacky patents” meet economic indicators (Czarnitzki et al, 2011)

• bureaucratic– “Patent quality isn’t broken at all. In fact, our decisions on

both allowances and rejections correctly comply with all laws and regulations over 96 percent of the time.” [Under Secretary Kappos]

contingent on liquidity, evidence of use

(at what level?)

Page 42: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

anomalous characteristics

• right to block, tax, regulate….

• one size fits all

• legal system to produce economic result

• territorial rights in global commerce– tax arbitrage from profit shifting

Page 43: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

• uncertainty, not just risk– uncertainty is socialized, weaponized– fee-earning professionals/ intermediaries/ agents bearing no risk

• positive feedback loops– portfolio racing– profession, agency, and court

• widespread information asymmetries• contingent and context-dependent value

– goes up with “evidence of use”– goes down with discovery of prior art

• widely distributed risk/liability– app developers, Main Street Coalition

• asymmetries of cost, risk, and liability– start-ups vs contingency-fee lawyers

• aggregation for predictability– cf RMBS (failure of quality and information)

• multiple marginalization (patents in standards), anti-commons• legal process misaligned with speed of technological advance and business practice• “highest and best” use is holdup

systemic problems in the patent

system

Page 44: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

• original “patent paradox” -- most patents where least valuable. (Hall and Ziedonis 2001)

• public grants premised on disclosure– but not of legal and market behaviors– failure of disclosure function

• value diluted by complexity• strategic withholding• legal constraints on reading patents

• entrusting governance of innovation to legal system >> cannot adapt to out-of-the-box insights; precedent oriented

– innovation and governance (patent institutions very conservative, not innovative)

• despite lowering of transaction costs (thanks to digital technology), institutional arrangements bury information/ avoid transaction costs (cross-licensing, non-assertion covenants, MAD, FRAND)

• territorial vs global; the more widespread innovation, the greater the patent costs, uncertainty, opacity, asymmetries….and temptation to intervention .

– harmonization (TRIPS) addresses minimum standards – not excesses– IP mercantilism vs cash mercantilism?

paradoxes

Page 45: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

Do nanotechnologists read patents?

0102030405060708090100

had read a patent looked forTECHNICALinformation

found informationuseful

believed technology

reproducible

Lisa Larrimore Ouellette, “Do Patents Disclose Useful Information?”, 2011

%failure of disclosure even in a physical technology

Page 46: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

• original “patent paradox” -- most patents where least valuable. (Hall and Ziedonis 2001)

• public grants premised on disclosure– but not of legal and market behaviors– failure of disclosure function

• value diluted by complexity• strategic withholding• legal constraints on reading patents

• entrusting governance of innovation to legal system >> cannot adapt to out-of-the-box insights; precedent oriented

– innovation and governance (patent institutions very conservative, not innovative)

• despite lowering of transaction costs (thanks to digital technology), institutional arrangements bury information/ avoid transaction costs (cross-licensing, non-assertion covenants, MAD, FRAND)

• territorial vs global; the more widespread innovation, the greater the patent costs, uncertainty, opacity, asymmetries….and temptation to intervention .

– harmonization (TRIPS) addresses minimum standards – not maximums or excesses– IP mercantilism vs cash mercantilism?

paradoxes

Page 47: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data
Page 48: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data
Page 49: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

motives for sovereign/state-supported funds

• commercialize technology– assemble and exploit “sleeping” patents

• defense against foreign aggression– in high-value markets (US, EU soon, China)

• return on investment– no different from SWFs– private aggregators (Intellectual Ventures, Acacia

Research, etc.) directly or via hedge funds and private equity

>>relationship to SOEs/SSEs and industrial policyOnce patents are acquired by governments, they can be used

for any purpose – or sold to those who will do so.

Page 50: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

“The land of Intellectual Property Rights is manipulative, predatory and vicious. And here is the best part: it’s managed at the national level.”

Jim Basillie, ex-RIM (Blackberry)

a public-private system

Page 51: HBS seminar 3/26/14: Dark Markets, Bad Patents, No Data

long-term recommendations

• raise standard of invention– peer review standard

• economic oversight – international dialog needed

• explicitly address diversity (Burk & Lemley, 2009)– tailor administrative processes to technology and business environment

• develop administrative alternatives to litigation– tie presumption of validity to level of administrative review

• eliminate or tailor subsidies– ad valorem service fees/taxes (in lieu of property tax)

• explore independent invention defense at least for software• open to experimentation

– government should not have to buy back (via eminent domain) what it grants