Janet Freilich

Patents’ New Salience

The vast majority of patents do not matter. They are almost never enforced or licensed and, in consequence, are almost always ignored. This is a well-accepted feature of the patent system and has a tremendous impact on patent policy. In particular, while there are many aspects of patent law that are potentially troubling—including grants of unmerited patents, high transaction costs in obtaining necessary patent licenses, and patents’ potential to block innovation and hinder economic growth—these problems may be insignificant in practice because patents are under-enforced and routinely infringed without consequence.

This Article argues that technological developments are greatly increasing the salience of patents by making patents easier and cheaper to find and enforce. These developments—including private platforms’ adjudication systems and AI-driven patent analytics—profoundly impact how the patent system functions and upend the system’s present dependence on under-enforcement and ignorance. Where most patents could previously be safely disregarded, formerly forgotten patents now matter.

This Article makes four contributions to the literature. First, this Article explores the technology that is rendering patents newly salient and explains how this alters basic assumptions underlying the patent system. Second, this Article demonstrates that although new technology is increasing the number of patents that can be reviewed and enforced, this transformation sometimes decreases the depth of patent analysis. Because it is difficult to draw conclusions about patent scope or validity without in-depth analysis, this omission means that technological review of patents may give patents unmerited influence.

Third, this Article shows a sharp divergence between public policy goals and private use of patents. For several decades, the courts and Congress have been reforming patent policy to decrease the impact of patents to alleviate concerns that patent owners hinder innovation by others. This Article demonstrates, in clear contrast to this goal, an increase in patent salience that is due exclusively to the use of private platforms and technologies. Further, the use of private platforms to find, analyze, and enforce patents creates the risk that choices made by companies and software developers will displace substantive patent law. Finally, this Article suggests policy reform, including ways to improve technology and patents and adjusted approaches to patent doctrine and theory.

Introduction

It is quite likely that you, the reader, have infringed a patent today. There are millions of in-force U.S. patents, and many cover routine, everyday behaviors. Perhaps you used a smartphone, which are covered by thousands of patents, and liability for infringement extends not just to the phone manufacturer but also to the consumer.1 1.Colleen Chien, Predicting Patent Litigation, 90 Tex. L. Rev. 283, 289 (2011); Gaia Bernstein, The Rise of the End User in Patent Litigation, 55 B.C. L. Rev. 1443, 1452–53 (2014).Show More Or you used Wi-Fi, also covered by many patents.2 2.Mark A. Lemley & Carl Shapiro, Patent Holdup and Royalty Stacking, 85 Tex. L. Rev. 1991, 2027 (2007).Show More Alternatively, your infringing act may have been low-tech—playing on a swing3 3.U.S. Patent No. 6,368,227 (filed Nov. 17, 2000).Show More or throwing a stick,4 4.U.S. Patent No. 6,360,693 (filed Dec. 2, 1999).Show More for example. You were probably not aware that you took an action covered by a patent, but this is no defense to patent infringement, which is a strict liability tort and does not take intent into account.5 5.In re Seagate Tech., LLC, 497 F.3d 1360, 1368 (Fed. Cir. 2007).Show More

Fortunately, the vast majority of patents are never enforced so the likelihood that you will be sued for infringement is infinitesimally small.6 6.Mark A. Lemley, Rational Ignorance at the Patent Office, 95 Nw. U. L. Rev. 1495, 1497 (2001).Show More The patent system relies heavily on under-enforcement: if most patents were enforced, day-to-day activities would be impossible because the transaction costs required to find and license all relevant patents would be prohibitively high.7 7.Mark A. Lemley, Ignoring Patents, 2008 Mich. St. L. Rev. 19, 25. This is analogous to many other areas of law—torts, criminal law—where the system is characterized by pervasive under-enforcement. See Richard Abel, The Real Tort Crisis—Too Few Claims, 48 Ohio St. L.J. 443, 447 (1987); Richard Frase, The Decision to File Federal Criminal Charges: A Quantitative Study of Prosecutorial Discretion, 47 U. Chi. L. Rev. 246, 246 (1980).Show More Patent scholars, policy makers, and the U.S. Patent and Trademark Office (“USPTO” or “Patent Office”) all recognize that many potential problems with the patent system are avoided because patentees rarely enforce patents and infringers generally ignore patents.8 8.See, e.g., Jonathan M. Barnett, Property as Process: How Innovation Markets Select Innovation Regimes, 119 Yale L.J. 384, 392 (2009) (noting that criticisms of the subject-matter expansion of patents as excessive propertization are overblown because most patents are ignored); Tun-Jen Chiang, Fixing Patent Boundaries, 108 Mich. L. Rev. 523, 542 (2010) (suggesting that the notice functions of patent claims work poorly in part because competitors ignore patents); Lemley, supra note 6, at 1510–11 (arguing that low-cost, error-prone patent examination is rational because most patents are ignored).Show More

This Article argues that we are at the beginning of a technological shift that is changing this pattern of under-enforcement and ignorance.9 9.See infra Part II.Show More Because patent policy relies so heavily on ignorance and under-enforcement, the shift towards patent salience has important implications for both doctrinal and theoretical reform.10 10.See infra Part III.Show More

This shift from ignorance and under-enforcement to salience is caused by new technologies that make patents easier to find and use. This Article illustrates the shift with three case studies: First, automated freedom-to-operate algorithms, which are computer programs that take a desired endpoint and design around any relevant patents.11 11.See infra Section II.A.Show More Such a program was used, for instance, to suggest ways to avoid patents on remdesivir (VEKLURY®) in order to increase production during the COVID-19 pandemic.12 12.Sara Szymkuc et al., Computer-Generated “Synthetic Contingency” Plans at Times of Logistics and Supply Problems: Scenarios for Hydroxychloroquine and Remdesivir, 11 Chem. Sci. 6736, 6736 (2020).Show More Second, Amazon’s Utility Patent Neutral Evaluation program, a company-run system to adjudicate claims of patent infringement and remove infringing products from Amazon’s platform.13 13.Ganda Suthivarakom, Welcome to the Era of Fake Products, N.Y. Times: Wirecutter (Feb. 11, 2020), https://www.nytimes.com/wirecutter/blog/amazon-counterfeit-fake-produc‌ts/ [https://perma.cc/B3LJ-UACW].Show More The program provides fast and cheap ($4,000) opportunities for arbitration.14 14.Tammy Terry & Lisa Margonis, Unpacking Amazon’s Patent Infringement Evaluation Process, Law360 (Mar. 19, 2021), https://www.law360.com/articles/1366714/unpacking-amazon-s-patent-infringement-evaluation-process [https://perma.cc/TQ48-KTBY].Show More Third, analytics software that uses machine learning and artificial intelligence to produce patent landscape reports.15 15.Leonidas Aristodemou, Frank Tietze, Nikoletta Athanassopoulou & Tim Minshall, Exploring the Future of Patent Analytics: A Technology Roadmapping Approach, at Abstract (Univ. of Cambridge Ctr. for Tech. Mgmt., Working Paper No. 5, 2017).Show More These reports are detailed accounts of trends in patenting across a field that inform a varied set of decision makers—for example, a report on hydrogen fuel patents designed to help companies find collaborators and make investment decisions.16 16.Chem. Abstracts Serv., Am. Chem. Soc’y, Hydrogen Fuel: Insights into a Growing Market 12 (2019).Show More

With each of these new technologies, patents that would previously never have been enforced, licensed, or likely even read now impact behavioral choices. Because automated freedom-to-operate analyses show users how to avoid all patents in a field, a patent need simply exist to cause a response, even though many such patents would not—indeed could not—be enforced.17 17.See infra Subsection II.A.1.Show More In the case of Amazon’s program, the low cost of the program compared to litigation incentivizes additional enforcement, as does Amazon’s ability to reach beyond traditional jurisdictional limits.18 18.See infra Subsection II.B.1.Show More Further, by providing an easy way to search for products, Amazon’s platform makes it considerably simpler for patentees to find infringers.19 19.See infra Subsection II.B.1.Show More Patent landscape analyses provide information on all patents in a field so that decisions can be made based on a great breadth of patents.20 20.See infra Subsection II.C.1.Show More Patents that were formerly overlooked are now found and integrated into decision-making. Previously ignored, these patents are now impactful.

The technologically driven shift from under-enforcement to salience has created a second fundamental change in how patents are used: the greater impact of patents is accompanied by a move away from deep legal analysis. This shift is most stark with respect to patent validity. Granted patents can be found invalid, and indeed many are.21 21.35 U.S.C. § 282.Show More The mere presence of a patent therefore means little without some evaluation of its validity.22 22.Lemley, supra note 7, at 27.Show More But not all of the case studies highlighted in this Article evaluate validity.23 23.See infra Part II.Show More Amazon’s adjudication system explicitly excludes a validity analysis—a significant difference from litigation, where validity is an issue in almost every case.24 24.Terry & Margonis, supra note 14; Lemley, supra note 6, at 1502 (“Virtually every patent infringement lawsuit includes a claim that the patent is either invalid or unenforceable due to inequitable conduct (or commonly both).”).Show More Some algorithms that run automated freedom-to-operate analyses and create patent landscapes do not account for the possibility of invalidity nor do they discount patents of dubious validity.25 25.See infra Sections II.A, II.C.Show More Rather, each patent is given equal weight in the analysis.26 26.See infra Sections II.A, II.C.Show More Though technology allows analysis of more patents, the analysis can be cursory and blurs the major quality differences between patents.

The trend toward greater patent salience and the changes in how patents are analyzed have substantial implications for patent theory and policy. One notable example is the Patent Office’s “rational ignorance” approach to patent examination.27 27.Lemley, supra note 6, at 1497.Show More Examiners spend relatively little time reviewing each patent and make many mistakes, meaning that many invalid patents are granted.28 28.Id. at 1500.Show More This is justified because more careful examination would be expensive and, if most patents are ignored, these errors have little practical effect.29 29.See infra Section III.B.Show More If, however, more patents impact behavioral choices, the rational ignorance approach breaks down, particularly if new technology does little or no analysis of validity.30 30.See infra Section IV.C.Show More This Article highlights several additional policies and doctrines that are central to the patent system—including the lack of a research exception, methods by which remedies are determined, and the potential for a patent anticommons to block follow-on research—where potentially disastrous consequences are brushed aside on the grounds that patents are ignored.31 31.See infra Section IV.C.Show More

Another key consequence of patents’ new salience is that choices about patent impact are increasingly privatized, which creates concerns about the influence of private platforms and their divergence from public goals. First, the technological shift highlighted in this Article predominantly involves private platforms.32 32.See infra Part II.Show More When private platforms design algorithms and choose training data for patent analysis, they inevitably make choices about how to interpret and prioritize substantive law.33 33.See infra Section III.D.Show More To the extent that algorithmic output influences decisions and is not subject to judicial review, it raises the risk that private choices about enforcement mechanisms or platform design will displace substantive law.34 34.See infra Section III.D.Show More While these privatization concerns have been well-aired in the context of copyright law and other fields, the concerns apply with equal force to patent law.35 35.E.g., Matthew Sag, Internet Safe Harbors and the Transformation of Copyright Law, 93 Notre Dame L. Rev. 499, 499 (2017).Show More Moreover, to the extent that substantive patent law is woven into private designs, it is often in a black box without transparency about how and when patent law is incorporated into the analysis.36 36.See infra Section III.D. More specifically, technologies that rely on AI do not always disclose the data used to train the AI, making it difficult to predict bias in output. See, e.g., Shlomit Yanisky-Ravid & Sean K. Hallisey, “Equality and Privacy by Design”: A New Model of Artificial Intelligence Data Transparency via Auditing, Certification, and Safe Harbor Regimes, 46 Fordham Urb. L.J. 428, 474 (2019) (recommending increased disclosure of data inputs in order to prevent discrimination).Show More

Further, the increasing patent impact documented herein is in striking contrast to a countervailing trend in congressional and judicial action which is towards making patents less impactful.37 37.Paul Gugliuzza, Quick Decisions in Patent Cases, 106 Geo. L.J. 619, 622 (2018); Jonathan Masur, Patent Inflation, 121 Yale L.J. 470, 510 (2011).Show More In recent years, Congress and the courts have increased the difficulty of obtaining and enforcing patents, meaning that third parties can more safely ignore patents—a deliberate policy intended to alleviate some of the roadblocks that patents can pose to innovation and the economy.38 38.Gugliuzza, supra note 37, at 624.Show More This Article argues that private actors, in making patents more salient, are moving patent law away from values espoused by public actors.39 39.See infra Section III.C.Show More

Despite these challenges, technological developments in patent law are not inherently negative. Software’s ability to draw information from millions of patents is exciting and may improve patents’ ability to fulfill their disclosure function.40 40.One way in which patents incentivize innovation is by providing information about cutting-edge inventions to the public. E.g., Sean B. Seymore, The Teaching Function of Patents, 85 Notre Dame L. Rev. 621, 622 (2010).Show More It is important for e-commerce platforms to have some form of patent enforcement mechanism.41 41.Terry & Margonis, supra note 14.Show More But these technologies can be improved. This Article suggests avenues for using artificial intelligence (“AI”) to expand in-depth analysis of patents and also highlights where AI is unlikely to work.42 42.See infra Section IV.A.Show More The Article additionally recommends strategies to alter patents to better interface with AI.43 43.See infra Section IV.B.Show More And, doctrinally, the Article suggests reviewing the implications of patent law doctrine and theories such as rational ignorance, research exceptions, the application of damages and other remedies, and reliance on under-enforcement—all areas that may be impacted by the new salience of patents.44 44.See infra Section IV.C.Show More

The Article proceeds as follows. Part I explores why patents have historically been ignored and, for those few patents that are not, why in-depth analysis is essential to understand the enforceability of any patent. Part II provides three case studies of technologies that render patents newly salient. Part III turns to the implications of this shift towards technologically-driven patent impact (Section III.A). It further discusses the consequences of platforms that avoid in-depth patent analysis (Section III.B), the divergence between the public trend towards easier invalidation and the private trend towards easier enforcement (Section III.C), and the displacement of substantive law by private choices (Section III.D). Part IV suggests policy reform.

  1.  Colleen Chien, Predicting Patent Litigation, 90 Tex. L. Rev. 283, 289 (2011); Gaia Bernstein, The Rise of the End User in Patent Litigation, 55 B.C. L. Rev. 1443, 1452–53 (2014). ↑
  2.  Mark A. Lemley & Carl Shapiro, Patent Holdup and Royalty Stacking, 85 Tex. L. Rev. 1991, 2027 (2007). ↑
  3.  U.S. Patent No. 6,368,227 (filed Nov. 17, 2000). ↑
  4.  U.S. Patent No. 6,360,693 (filed Dec. 2, 1999). ↑
  5.  In re Seagate Tech., LLC, 497 F.3d 1360, 1368 (Fed. Cir. 2007). ↑
  6.  Mark A. Lemley, Rational Ignorance at the Patent Office, 95 Nw. U. L. Rev. 1495, 1497 (2001). ↑
  7.  Mark A. Lemley, Ignoring Patents, 2008 Mich. St. L. Rev. 19, 25. This is analogous to many other areas of law—torts, criminal law—where the system is characterized by pervasive under-enforcement. See Richard Abel, The Real Tort Crisis—Too Few Claims, 48 Ohio St. L.J. 443, 447 (1987); Richard Frase, The Decision to File Federal Criminal Charges: A Quantitative Study of Prosecutorial Discretion, 47 U. Chi. L. Rev. 246, 246 (1980). ↑
  8.  See, e.g., Jonathan M. Barnett, Property as Process: How Innovation Markets Select Innovation Regimes, 119 Yale L.J. 384, 392 (2009) (noting that criticisms of the subject-matter expansion of patents as excessive propertization are overblown because most patents are ignored); Tun-Jen Chiang, Fixing Patent Boundaries, 108 Mich. L. Rev. 523, 542 (2010) (suggesting that the notice functions of patent claims work poorly in part because competitors ignore patents); Lemley, supra note 6, at 1510–11 (arguing that low-cost, error-prone patent examination is rational because most patents are ignored). ↑
  9.  See infra Part II. ↑
  10.  See infra Part III. ↑
  11.  See infra Section II.A. ↑
  12.  Sara Szymkuc et al., Computer-Generated “Synthetic Contingency” Plans at Times of Logistics and Supply Problems: Scenarios for Hydroxychloroquine and Remdesivir, 11 Chem. Sci. 6736, 6736 (2020). ↑
  13.  Ganda Suthivarakom, Welcome to the Era of Fake Products, N.Y. Times: Wirecutter (Feb. 11, 2020), https://www.nytimes.com/wirecutter/blog/amazon-counterfeit-fake-produc‌ts/ [https://perma.cc/B3LJ-UACW]. ↑
  14.  Tammy Terry & Lisa Margonis, Unpacking Amazon’s Patent Infringement Evaluation Process, Law360 (Mar. 19, 2021), https://www.law360.com/articles/1366714/unpacking-amazon-s-patent-infringement-evaluation-process [https://perma.cc/TQ48-KTBY]. ↑
  15.  Leonidas Aristodemou, Frank Tietze, Nikoletta Athanassopoulou & Tim Minshall, Exploring the Future of Patent Analytics: A Technology Roadmapping Approach, at Abstract (Univ. of Cambridge Ctr. for Tech. Mgmt., Working Paper No. 5, 2017). ↑
  16.  Chem. Abstracts Serv., Am. Chem. Soc’y, Hydrogen Fuel: Insights into a Growing Market 12 (2019). ↑
  17.  See infra Subsection II.A.1. ↑
  18.  See infra Subsection II.B.1. ↑
  19.  See infra Subsection II.B.1. ↑
  20.  See infra Subsection II.C.1. ↑
  21.  35 U.S.C. § 282. ↑
  22.  Lemley, supra note 7, at 27. ↑
  23.  See infra Part II. ↑
  24.  Terry & Margonis, supra note 14; Lemley, supra note 6, at 1502 (“Virtually every patent infringement lawsuit includes a claim that the patent is either invalid or unenforceable due to inequitable conduct (or commonly both).”). ↑
  25.  See infra Sections II.A, II.C. ↑
  26.  See infra Sections II.A, II.C. ↑
  27.  Lemley, supra note 6, at 1497. ↑
  28.  Id. at 1500. ↑
  29.  See infra Section III.B. ↑
  30.  See infra Section IV.C. ↑
  31.  See infra Section IV.C. ↑
  32.  See infra Part II. ↑
  33.  See infra Section III.D. ↑
  34.  See infra Section III.D. ↑
  35.  E.g., Matthew Sag, Internet Safe Harbors and the Transformation of Copyright Law, 93 Notre Dame L. Rev. 499, 499 (2017). ↑
  36.  See infra Section III.D. More specifically, technologies that rely on AI do not always disclose the data used to train the AI, making it difficult to predict bias in output. See, e.g., Shlomit Yanisky-Ravid & Sean K. Hallisey, “Equality and Privacy by Design”: A New Model of Artificial Intelligence Data Transparency via Auditing, Certification, and Safe Harbor Regimes, 46 Fordham Urb. L.J. 428, 474 (2019) (recommending increased disclosure of data inputs in order to prevent discrimination). ↑
  37.  Paul Gugliuzza, Quick Decisions in Patent Cases, 106 Geo. L.J. 619, 622 (2018); Jonathan Masur, Patent Inflation, 121 Yale L.J. 470, 510 (2011). ↑
  38.  Gugliuzza, supra note 37, at 624. ↑
  39.  See infra Section III.C. ↑
  40.  One way in which patents incentivize innovation is by providing information about cutting-edge inventions to the public. E.g., Sean B. Seymore, The Teaching Function of Patents, 85 Notre Dame L. Rev. 621, 622 (2010). ↑
  41.  Terry & Margonis, supra note 14. ↑
  42.  See infra Section IV.A. ↑
  43.  See infra Section IV.B. ↑

  44.  See infra Section IV.C. ↑

©Virginia Law Review Association. All rights reserved.