Guest Post: Book Review of ‘Inclusive Innovation in the Age of AI and Big Data’

Next week I will publish the third and final installment of my three-part series on  the Court of Appeals for England and Wales’ recent decision on disgorgement of profits, Lufthansa Technik AG v. Astronics Advanced Electronic Systems, [2026] EWCA Civ 964.  Today, however, I am happy to publish a guest post by Dr. Ashish Bharadwaj, the Founding Pro-Vice Chancellor of  WPU GŌA and the author or editor of several books on law and technology.  As you can see, the guest post is a review of a new edited volume by Professors Daryl Lim and Peter Yu.

*                    *                   *  

Book Review of ‘Inclusive Innovation in the Age of AI and Big Data’

 Edited by Daryl Lim and Peter K. Yu (OUP 2026, 432 pp, ISBN 978-0-19-779941-3)[1]

 Written by Ashish Bharadwaj, India

Innovation policy has long operated on the implicit assumption that economic growth, technological advancement, and human well-being advance in tandem. Yet, as Daryl Lim and Peter K. Yu demonstrate in their ambitious edited collection, Inclusive Innovation in the Age of AI and Big Data, unguided technological expansion frequently entrenches structural exclusion and concentrates capital.

The central idea of this edited volume is to reframe the normative architecture of intellectual property (IP) law around three simple questions. Each is deceptively simple, but demands deep academic thought – innovation by whom, for whom, and to what end? Historically, Western IP paradigms were constructed around formal, individualized, and documentable commercial outputs. This framework systematically marginalized informal, collaborative, or traditional knowledge structures. As artificial intelligence and big data analytics become the primary engines of creative and inventive activity, these historical blind spots do not simply persist—they are automated at scale. Machine learning models trained on historically biased data inevitably reproduce those biases under a guise of technological neutrality. Rather than proceeding through a mechanical chapter-by-chapter summary, this review evaluates how the volume’s fifteen chapters intersect across four thematic axes.

A key strength of the volume is its refusal to rely on anecdotal critiques of IP inequality. Instead, several contributions offer empirical assessments of where and why demographic leaks occur across creative and inventive pipelines. The book begins with Brent Lutes, Michael Palmedo, and Ryan Safner examining United States Copyright Office datasets from 1978 to 2020. Their findings reveal a nuanced picture: while female authorship reached parity in non-dramatic literary works (51.2% by 2020), persistent deficits remain in technical domains such as software (20.4%) and sound recordings (22.0%). Econometric modeling indicates that registration volumes correlate strongly with educational attainment and urban geography, demonstrating how socioeconomic baseline disparities dictate access to formal IP rights.

Focusing on academic patenting in Chapter 4, W. Michael Schuster, Miriam Marcowitz-Bitton, and Deborah R. Gerhardt analyse two decades of data from top U.S. research universities. Female researchers account for only 14% to 19% of named inventors annually. Crucially, the authors uncover structural isolation: over 64% of all-female patent applications involve solo inventors, whereas male inventors disproportionately benefit from collaborative team networks. This network disparity directly deprives female academics of institutional commercialization support and downstream citation impact. Chapters 2 and 3 complicate how scholars measure these disparities. In the former, the authors critique the name-disambiguation and algorithmic gender-attribution tools used in databases like USPTO's PatentsView. These methodologies fail to capture non-binary identities, cultural naming variations, or post-marital surname changes. Evaluating diversity solely through granted patents introduces survival bias: female inventors experience a significantly lower application-to-grant conversion rate (69.44%) than male inventors (73.54%), driven largely by higher rates of application abandonment after initial rejections.

Complementing this, Carlotta Nani, Martin Correa, and Julio Raffo utilize the WIPO Pulse Survey (50 countries) in Chapter 3 to differentiate between objective IP knowledge and self-reported awareness. Their analysis uncovers a confidence gap: women in high-income economies consistently perform well on objective tests regarding design and copyright, yet underreport their own subject-matter confidence compared to male peers. Beyond pipeline metrics, the volume addresses the institutional mechanics that mediate legal protection.

Two chapters are particularly interesting in how they illustrate discretionary human gatekeeping intersecting with algorithmic tools. Jessica C. Lai presents a qualitative critique of patent attorneys as critical actors in defining patentability. Patent attorneys do not merely translate technical disclosures into claims; they construct legal boundaries. Legal constructs such as the Person Having Ordinary Skill in the Art (PHOSITA) have historically been modelled on male, Western archetypes. Consequently, practitioners may implicitly undervalue innovations originating in female-dominated sectors or advise female inventors to accept narrower claim scopes. Lai warns that integrating generative AI tools into claim drafting threatens to institutionalize these biases by training automation tools on historically skewed patent specifications. This institutional failure is illustrated in the next chapter by Jordana Goodman, Yan Li, Regan Murphy, and Khamal Patterson through their study of the NuDred hair sponge—an invention designed for styling Black hair. The authors trace how communication breakdowns between non-Black examiners and Black inventors resulted in inappropriate rejections based on irrelevant prior art, such as bath sponges and antiperspirant applicators. The examination process failed to comprehend the technical behavior of polyurethane foam when applied to Black hair textures. While Goodman et al. suggest that generative AI could assist practitioners in translating culturally specific technical language, they stress that technical tools cannot replace cultural competency within patent offices.

Moving from diagnosis to prescription, Chapters 7 through 13 evaluate administrative interventions and regulatory frameworks designed to foster equity across various organizational interventions and metrics – Targeted Interventions, Capacity Metrics, and Behavioral Nudges. Margo A. Bagley and Colleen V. Chien examine empirical data from the Innovator Diversity Pilots Initiative, demonstrating how blind invention disclosures, mentorship pipelines, and simplified filing workflows reduce entry barriers for underrepresented inventors. On the other hand, Suzanne Harrison and Bowman Heiden critique traditional metrics that rely on patent counts, arguing that counts reflect past output rather than latent inventive capacity. Drawing on corporate diversity pledge programs, they argue that activating latent inventorship among underrepresented employees expands overall organizational productivity.

Paola Cecchi-Dimeglio presents randomized corporate experiments showing that behavioral nudges—such as simplified submission portals, storytelling videos, and inclusive language—increase both the volume and quality of patent disclosures from female employees. In Chapter 10, Deja Workman and Christopher L. Dancy offer a critique of the AI engineering lifecycle. Drawing on Sylvia Wynter’s concept of the "biocentric Man," they illustrate how standard AI development workflows privilege Western paradigms while reproducing techno-colonialism. They argue that true inclusivity requires community-cantered design and, where necessary, the refusal to deploy harmful AI architectures. Statistician David R. Hunter provides a methodological baseline in Chapter 11. Revisiting classic cases such as Gratz v. Bollinger and blind orchestral auditions, Hunter warns against the "streetlight effect"—the tendency to measure what is easy to quantify rather than what is substantively meaningful. He urges caution when using statistical models as definitive proof of structural discrimination. Addressing legal reform, Daryl Lim formulates "equitable progress" as a guiding norm for AI regulation. Synthesizing Rawlsian distributive justice and Amartya Sen’s capability approach, Lim argues that regulatory oversight across the U.S., EU, China, and Singapore must balance commercial incentives against social equity and worker displacement. W. Keith Robinson translates these governance principles into administrative procedures. He proposes a Responsibility, Transparency, and Accountability (RTA) framework for the USPTO, advocating for mandatory algorithmic impact assessments, enhanced disclosure of AI training datasets, and post-grant audit mechanisms for AI-assisted patents.

The final section widens the analysis to international political economy, examining how global technology shifts risk and exacerbates North-South inequalities. Lee Jyh-An and Liu Jingwen analyze the U.S.-China AI rivalry. They contrast China’s state-directed model—characterized by centralized data aggregation, state subsidies, and flexible copyright standards for AI outputs—with the U.S. market-driven approach centered on venture capital, proprietary models, and export controls on advanced semiconductor hardware. This geopolitical competition fragments international standards, reducing complex equity questions to instrumentalities of national security. Closing the volume, Peter K. Yu addresses the widening digital divide separating the Global North from the Global South. Mainstream policy discourse often ignores the infrastructural realities of developing nations, such as deficits in compute capacity, energy infrastructure, local-language datasets, and technical capital.

To prevent digital neo-colonialism, Yu outlines a range of concrete mechanisms worth mentioning. These are (a) Expanded IP flexibilities and statutory fair-use exceptions for model training; (b) Mandatory technology transfer frameworks under international law; (c) An international Global Fund for AI infrastructure; (d) Shared, multi-national computational resource pools; (e) Open-source foundational models tailored for public development; and (f) Institutional support for localized and indigenous innovation ecosystems.

Inclusive Innovation in the Age of AI and Big Data provides a comprehensive, multi-methodological examination of how legal institutions regulate emerging technologies. By pairing rigorous empirical data with theoretical critiques, Lim and Yu have compiled a volume that advances legal scholarship and innovation policy. The volume implicitly highlights an underlying tension between its contributions, namely between the incremental administrative reform and the structural critique.

Authors such as Bagley, Chien, Harrison, Heiden, and Cecchi-Dimeglio propose actionable internal adjustments within existing institutional frameworks (e.g., patent office procedures, corporate diversity metrics, behavioral nudges). For structural critique, contributions such as Workman and Dancy’s deconstruction of the "biocentric Man" suggest that existing legal and market structures are fundamentally configured to perpetuate historical power dynamics. This tension raises an important question for ongoing scholarly inquiry: Can incremental procedural adjustments meaningfully address structural inequities, or do they risk legitimizing systems that are inherently exclusionary? Similarly, while Yu’s policy framework in Chapter 15 offers a clear blueprint for bridging the global AI divide, its implementation faces significant political-economy hurdles. In a geopolitical environment marked by rising techno-nationalism, export controls, and aggressive enforcement of proprietary IP by multinational corporations, securing global consensus for mandatory technology transfers or a Global Fund for AI will be challenging. Future research must examine how developing nations can utilize regional trade agreements, data sovereignty frameworks, and South-South initiatives to build technological capacity independently of Global North concessions.

These conceptual tensions do not diminish the contribution of the collection; rather, they demonstrate its capacity to frame the research agenda for technology policy and IP scholarship. Lim and Yu have produced a foundational text that demonstrates technological neutrality is a myth and unguided market dynamics risk entrenching structural inequalities. Inclusive Innovation in the Age of AI and Big Data provides scholars, practitioners, and policymakers with empirical grounding and analytical tools to design innovation systems centered on equity, transparency, and broad-based human capability. It is an essential reference for legal scholars, economists, and technology policymakers addressing the societal impacts of AI governance.

References

 

1.    Daryl Lim and Peter K Yu (eds), Inclusive Innovation in the Age of AI and Big Data (Oxford University Press 2026).

2.    ibid xx1–xxiii; see also Robyn Klingler-Vidra, Alex Glennie and Courtney Savie Lawrence, Inclusive Innovation (Routledge 2022) 1.

3.    Colleen V Chien, 'The Inequalities of Innovation' (2022) 72 Emory Law Journal 1.

4.    Peter K Yu, 'Intellectual Property, Global Inequality, and Subnational Policy Variations' in Daniel Benoliel, Peter K Yu, Francis Gurry and Keun Lee (eds), Intellectual Property, Innovation and Economic Inequality (Cambridge University Press 2024) 81.

5.    UN General Assembly, 'Transforming Our World: The 2030 Agenda for Sustainable Development' (25 September 2015) UN Doc A/RES/70/1 (SDG 10); Universal Declaration of Human Rights (adopted 10 December 1948) UNGA Res 217 A(III) art 27.

6.    Daryl Lim, 'AI, Equity, and the IP Gap' (2022) 75 SMU Law Review 815, 843–44; Peter K Yu, 'Cultural Relics, IP and Intangible Heritage' (2008) 81 Temple Law Review 433.

7.    Lim (n 6) 831–32.

8.    Brent Lutes, Michael Palmedo and Ryan Safner, 'The State of Inclusivity in Copyright and Creative Ecosystems' in Daryl Lim and Peter K Yu (eds), Inclusive Innovation in the Age of AI and Big Data (Oxford University Press 2026) 3; see also Robert Brauneis and Dotan Oliar, 'An Empirical Study of the Race, Ethnicity, Gender, and Age of Copyright Registrants' (2018) 86 George Washington Law Review 46.

9.    US Copyright Office, Women in the Copyright System: An Analysis of Women Authors in Copyright Registrations from 1978 to 2020 (Office of the Chief Economist 2022); US Copyright Office, The Resilience of Creativity: An Examination of the COVID-19 Impact on Copyright-Reliant Industries and Their Subsequent Recovery (2024).

10.  Michelle Saksena and Gauri Subramani, 'Understanding Demographics in Patent Data' in Daryl Lim and Peter K Yu (eds), Inclusive Innovation in the Age of AI and Big Data (Oxford University Press 2026) 27.

11.  Abhay Aneja, Oren Reshef and Gauri Subramani, 'Attrition and the Gender Patenting Gap' (2025) 107 Review of Economics and Statistics (forthcoming); US Patent and Trademark Office, Progress and Potential: 2020 Update on U.S. Women Inventor-Patentees (Office of the Chief Economist 2020).

12.  Carlotta Nani, Martin Correa and Julio Raffo, 'Gender Differences in Intellectual Property Awareness: Evidence from a Global Survey' in Daryl Lim and Peter K Yu (eds), Inclusive Innovation in the Age of AI and Big Data (Oxford University Press 2026) 45.

13.  Elodie Carpentier and Julio Raffo, The Global Gender Gap in Innovation and Creativity: An International Comparison of the Gender Gap in Global Patenting over Two Decades (World Intellectual Property Organization 2023).

14.  W Michael Schuster, Miriam Marcowitz-Bitton and Deborah R Gerhardt, 'The Gender Gap in Academic Patenting' in Daryl Lim and Peter K Yu (eds), Inclusive Innovation in the Age of AI and Big Data (Oxford University Press 2026) 73; see also W Michael Schuster, Miriam Marcowitz-Bitton and Deborah R Gerhardt, 'The Gender Gap in Academic Patenting' (2022) 56 UC Davis Law Review 759.

15.  Waverly W Ding, Fiona Murray and Toby E Stuart, 'Gender Differences in Patenting in the Academic Life Sciences' (2006) 313 Science 665.

16.  Jessica C Lai, 'Patent Attorneys and the Increasing Use of Artificial Intelligence: A "Thought Experiment" on Our Human and Technological Gatekeepers' in Daryl Lim and Peter K Yu (eds), Inclusive Innovation in the Age of AI and Big Data (Oxford University Press 2026) 97.

17.  Jessica C Lai, Patent Law and Women: Tackling Gender Bias in Knowledge Governance (Routledge 2022); Dan L Burk, 'Do Patents Have Gender?' (2011) 19 American University Journal of Gender, Social Policy & the Law 881.

18.  Jordana Goodman, Yan Li, Regan Murphy and Khamal Patterson, 'Inventing Fairness: Exploring AI's Role in Patent Reform' in Daryl Lim and Peter K Yu (eds), Inclusive Innovation in the Age of AI and Big Data (Oxford University Press 2026) 115.

19.  Jordana R Goodman and Khamal Patterson, 'Access to Justice for Black Inventors' (2024) 77 Vanderbilt Law Review 109; Anjali Vats, The Color of Creatorship: Intellectual Property, Race, and the Making of Americans (Stanford University Press 2020).

20.  Margo A Bagley and Colleen V Chien, 'Inclusive Innovation in an Age of AI: Insights from the Innovator Diversity Pilots Initiative' in Daryl Lim and Peter K Yu (eds), Inclusive Innovation in the Age of AI and Big Data (Oxford University Press 2026) 139.

21.  Suzanne Harrison and Bowman Heiden, 'Improving Diversity and Inclusivity Measurements in Inventorship: A Competitiveness Perspective' in Daryl Lim and Peter K Yu (eds), Inclusive Innovation in the Age of AI and Big Data (Oxford University Press 2026) 171.

22.  Paola Cecchi-Dimeglio, 'Bridging the Gender Gap in Innovation: A Behavioral Approach to Inclusivity' in Daryl Lim and Peter K Yu (eds), Inclusive Innovation in the Age of AI and Big Data (Oxford University Press 2026) 185.

23.  Deja Workman and Christopher L Dancy, 'Identifying Potential Inlets of the Biocentric Man in the Artificial Intelligence Development Process' in Daryl Lim and Peter K Yu (eds), Inclusive Innovation in the Age of AI and Big Data (Oxford University Press 2026) 211.

24.  Sylvia Wynter, 'Unsettling the Coloniality of Being/Power/Truth/Freedom: Towards the Human, After Man, Its Overrepresentation--An Argument' (2003) 3(3) CR: The New Centennial Review 257.

25.  David R Hunter, 'Taking Stock: What Statistics Can and Cannot Do' in Daryl Lim and Peter K Yu (eds), Inclusive Innovation in the Age of AI and Big Data (OUP 2026) 231.

26.  Gratz v Bollinger 539 US 244 (2003).

27.  Daryl Lim, 'Equitable Progress and the Regulation of AI' in Daryl Lim and Peter K Yu (eds), Inclusive Innovation in the Age of AI and Big Data (Oxford University Press 2026) 245.

28.  John Rawls, A Theory of Justice (Harvard University Press 1971); Amartya Sen, Development as Freedom (Oxford University Press 1999).

29.  W Keith Robinson, 'Responsibility, Transparency, and Accountability in AI Patents' in Daryl Lim and Peter K Yu (eds), Inclusive Innovation in the Age of AI and Big Data (Oxford University Press 2026) 279.

30.  Lee Jyh-An and Liu Jingwen, 'Navigating Turbulence: The Challenge of Inclusive Innovation in the U.S.-China AI Race' in Daryl Lim and Peter K Yu (eds), Inclusive Innovation in the Age of AI and Big Data (Oxford University Press 2026) 301.

31.  Peter K Yu, 'Bridging the Global Artificial Intelligence Divide' in Daryl Lim and Peter K Yu (eds), Inclusive Innovation in the Age of AI and Big Data (Oxford University Press 2026) 331.

32.  UNCTAD, Technology and Innovation Report 2025: Inclusive Artificial Intelligence for Development (United Nations 2025) 79; UN Secretary-General's High-Level Advisory Body on AI, Governing AI for Humanity: Final Report (United Nations 2024).

---

[1] Ashish Bharadwaj writes on technology innovation and patents. His writings can be accessed on www.ashishbharadwaj.in and he can be contacted on ab.ashish@gmail.com

 

Wrongful Patent Assertion: Free First Chapter and Discount Code

My book Wrongful Patent Assertion:  A Comparative Law and Economics Analysis (Oxford Univ. Press 2026) is now available for purchase online and in hard copies in all major markets.  Readers can download the first chapter for free using this link, through the end of this month.  In addition, you can use the promotion code listed in the picture below (AUFLY30) for a 30% discount. 

From the book description:

Patents play an important role in inducing the investment needed to transform basic discoveries into the practical innovations that contribute to long-term economic growth. But patents also can generate a variety of social costs which, if unchecked, can unnecessarily impede competition, inhibit innovation, impair the integrity of the marketplace, and reduce overall social welfare.

 

Wrongful Patent Assertion provides the first comprehensive, comparative overview of how the world's leading jurisdictions for patent litigation employ different bodies of law and legal doctrines-including antitrust and unfair competition law, as well as other generally-applicable legal principles-to regulate the enforcement of patent rights, and how these efforts might be improved. Among the topics discussed are the circumstances under which owners are or should be liable for the actual or threatened enforcement of patent rights; what sort of harms should matter in assessing liability; and the circumstances under which liability should depend, in whole or in part, upon the patent having been adjudicated invalid or not infringed ex post, or upon evidence that the assertion of rights enables the owner to extract benefits beyond the patent's probable ex ante scope; or alternatively, upon some reason independent of the patent's actual or probable validity or infringement.

Lufthansa v. Astronics: Disgorgement of Profits, Part 2

This second post on the Court of Appeals for England and Wales’ recent decision in Lufthansa Technik AG v. Astronics Advanced Electronic Systems, [2026] EWCA Civ 964, will focus on Lord Justice Arnold’s analysis of the lower court’s award of a portion of the profits earned by defendants Astronics, Panasonic, and Safran.  For the facts of the case, see my first post in this series, published last Thursday.

The principal issue under consideration, as the court states, is how to “distinguish between ‘the profits derived . . . from the infringement’ and profits which are not ‘derived . . .  from the infringement’” (para. 78).  To answer this question, the opinion sets out to analyze the relationship between a “differential profits” approach and apportionment.  Paragraphs 80 through 82 state the parties’ competing positions:

80. Lufthansa contends that, in most cases, the correct way in which to distinguish between “profits derived … from the infringement” and profits which are not “derived … from the infringement” is by ascertaining the difference between the profits which the defendant made from its infringing activities and the profits which the defendant would have made from the nearest available NIA [noninfringing alternative]. This type of analysis is most commonly referred to in the case law and the academic literature as a “differential profits” analysis, although other terms (such as “incremental profits”) are sometimes used. Once a differential profits analysis has been undertaken, Lufthansa contends that there is no legal or economic justification for applying an apportionment of the profits: these are alternative approaches to the isolation of profits caused by the infringement. Lufthansa accepts that there may be cases in which apportionment is appropriate, but it argues that a differential profits analysis is usually preferable. What is not legitimate is to apply both. Lufthansa also accepts that, having carried out a differential profits analysis, it remains necessary to ask whether any of the resulting profits were too remote (in the broad sense) from (i.e. not legally caused by) the defendant’s acts of infringement. Lufthansa argues that this addresses the problem of long chains of causation, and does not justify apportionment.

 

81. Lufthansa contends that, since the judge found that the Defendants had failed to establish any NIA, the result of the differential profits analysis is that all of the profits in question “derived … from the infringement”. The Defendants’ failure to establish an NIA demonstrates the Patent was a “gateway” patent which controlled access to the relevant market. Accordingly, Lufthansa says, it is just that Lufthansa should recover all of the profits generated as a result.

 

82. Astronics and Panasonic contend that apportionment is a well-established approach in English law to determining what profits are “derived … from the infringement”, and that the use of apportionment is not precluded by a finding that all of the profits in issue were factually caused by the infringing acts. . . . It would be unjust for Lufthansa to recover all of the profits in issue, because there were multiple causes for the generation of those profits. Although infringement of the Patent was necessary for those profits to be generated, it was not sufficient.

The opinion then spends several pages (paras. 84-132) discussing the relevant common-law precedents, including (among many others) Celanese International Corp v BP Chemicals Ltd [1999] RPC 203, a case I discuss and critique at some length in my book Comparative Patent Remedies, and Nova Chemical Corp v Dow Chemical Co [2022] SCC 43 (Can.), a decision that Norman Siebrasse discusses and critiques in two papers, Nova v. Dow: Intuition or Principle in the Accounting of Profits Remedy, Part I, 35 I.P.J, 249 (2023); Norman V. Siebrasse, Nova v. Dow: Intuition or Principle in the Accounting of Profits Remedy, Part II, 36 I.P.J. 81 (2023).  The opinion also discusses and critiques some academic works that Siebrasse and I, among others, have either authored or coauthored (paras. 133-42).  The opinion’s principal critique of some of this work is that it doesn’t answer the question of how a court should proceed when the evidence does not indicate what the NIA would have been.  (Later in the opinion, as noted below, the opinion rejects the argument that courts should resolve this issue simply by allocating the burden of production or burden of proof to one party or the other.)  I commend all of this to readers’ attention,  but in the interest of conciseness I will cut to Lord Justice Arnold’s resolution of the issue, which begins at paragraph 143:

143. Analysis of the law. As is common ground, factual causation is not enough on an account of profits any more than it is on an inquiry as to damages. The reason is simple: factual causation proves too much. Applying the usual “but for” test of factual causation, the infringer may have made profits which it would not have made but for the infringement, but which are not in truth derived from the infringement as opposed to other factors. . . . Accordingly, as is also common ground, legal causation is required as well as factual causation. . . .

 

145. Lufthansa argues that a correct application of the “but for” test of factual causation involves the identification of a counterfactual world in which the infringements did not take place. That requires identification of the nearest NIA available to the defendant. The difference between the profit the infringer in fact made and the profit the defendant would have made had it adopted that NIA (i.e. the differential profit) represents the profit derived from the infringement. On this argument, a differential profit analysis identifies with precision the profits attributable to infringement as opposed to other factors. Thus it accounts for the causal potency of the different factors. Accordingly, the only role for legal causation is to police the length of the causation chain, and to exclude profits which are too remote from the infringing acts (e.g. profits made by reinvestment of the profits from the infringing acts into a cryptoasset which performs very well). . . .

Lord Justice Arnold, however, expresses agreement “with Astronics and Panasonic that the ‘but for’ test of factual causation does not necessarily require the identification of a specific NIA,” and “that the role of legal causation in this context is not limited in the manner contended for by Lufthansa” (para. 148).  In my view, the next several paragraphs are the most important portion of the opinion, so I’m going to quote them with only minimal editing for concision. 

149. As Astronics and Panasonic accept, there may be cases in which a differential profit analysis is useful for this purpose. As they submit, however, differential profit analysis is fraught with difficulty. . . .

 

150. The reason why differential profit analysis is fraught with difficulty is that, in order to identify with precision the profits attributable to the infringement, as opposed to other factors, it is necessary to identify a counterfactual in which all other factors are held constant and the minimum change is made to ensure that the patent is no longer infringed, so that the economic impact of the infringement can be isolated from the economic impact of other factors. In theory, this should present no difficulty. In the real world, the opposite is true.

 

151. The first question is who bears the burden of identifying and proving the NIA. The Canadian courts have held the burden rests on the defendant, but it is not clear to me why this should be so. One could argue that the burden should lie on the claimant, since the claimant is claiming the profits derived from the infringement, and if it relies upon differential profits analysis to quantify those profits, then the claimant must prove the NIA which should be used for that purpose. A potential difficulty with that approach is that it would enable the claimant to skew the differential profits analysis by selecting a very unprofitable NIA. As I understand the jurisprudence of the US courts on this question, they apply a shifting burden of proof under which the claimant must first identify an NIA, and then the onus is upon the defendant if it wishes to rely upon a different NIA as being a better one. That is a principled approach, but in many cases it will lead to an evidential burden on both parties.

 

152. If the defendant bears the burden of proof either in full or in part, the next question is what happens if the defendant proposes an NIA which the claimant contends is not an NIA because it also infringes the patent in suit? This is what happened in the present case. The first problem with this is that it required the judge to undertake a patent infringement trial as part of the account of profits, with all the attendant complexity and expense.

 

153. The next problem is what happens if the claimant turns out to be correct, and the proposed NIA actually infringes. Lufthansa argues that, because the Defendants failed to prove their chosen NIA, Lufthansa can claim all the profits. But all this shows is that the supposed NIA is inapposite for a differential profit analysis because it is not actually an NIA. It should not mean that the court is relieved from the burden of identifying an NIA, because differential profit analysis requires an NIA. One answer to this would be for the defendant to plead and prove a series of alternative potential NIAs, each further away from the claimed invention than the last, but that would simply compound the first problem.

 

154. The next difficulty is that a question may arise as to whether the defendant could have undertaken the NIA. Suppose that the NIA requires access to a particular raw material or part, but the defendant did not have access to that material or part at the relevant time. As I understand Lufthansa’s argument, this means that the defendant cannot rely upon the NIA, but I question why not. The availability of that material or part does not alter the inventive contribution of the patent. The object of a differential profit analysis is to identify what profits are caused by the use of invention, not what profits are caused by adventitious commercial factors. A similar problem arises if the defendant is prevented from using the NIA by regulatory factors unrelated to the invention. . . .  

 

155. Furthermore, a question may arise as to whether, even if the defendant could have undertaken the NIA, it would probably have done so. This seems to me to even less relevant, since ex hypothesi we are considering a counterfactual. A counterfactual is a thought experiment whose purpose is objectively to identify the consequences of what the defendant actually did. It does not depend on the probability of the defendant doing the alternative in the counterfactual world. . . .

 

156. The next difficulty is the one I mentioned when discussing the academic literature. In adversarial litigation courts depend on the parties to adduce evidence. What happens if the evidence does not enable the court to identify a suitable NIA? Counsel for Lufthansa argued that it is always possible to postulate an NIA, even if it is simply not producing the product in question at all. The problem with this argument is that an NIA only serves the purpose of the differential profit analysis if it enables the court to distinguish between the profits derived from the infringement from the profits derived from other factors. As explained above, this requires the identification of an NIA in which the minimum change necessary to avoid infringement is made, but all other factors are held constant. As the present case illustrates, the evidence may not permit this satisfactorily to be done.

 

157. The final problem I will mention is the one touched on by Laddie J in Celanese v BP at [43] (paragraph 102 above). In the real world, it is often the case that a complex product or process is covered (or arguably covered) by multiple patents relating to different aspects of the product or stages of the process. How does differential profits analysis work on a claim for an account of profits for infringement of just one of those patents? It cannot be correct to treat all of the profits generated by the manufacture and sale of the complex product or process as attributable to that infringement and none as attributable to the use of the other inventions. It could be argued that this depends on whether the other patents are (a) valid and (b) infringed, but that raises the spectre of determining the validity and infringement of each of those patents, without the participation of the owners of those patents, for the purposes of an account of profits. . . .

 

158. In short, while differential profit analysis has much to be said for it in terms of legal and economic theory, applying it in real world litigation is at best difficult, costly and uncertain.

 

159. The conclusions which I draw from this discussion are as follows. First, as Astronics and Panasonic accept, there are some cases in which it is possible to say that all of the profits in issue are derived from the infringement. As Laddie J explained in Celanese v BP at [47] (paragraph 104 above) and Lewison LJ noted in Abbott v Design & Display at [28] (paragraph 108 above), these are cases where, without the infringement, the infringer’s product or process would not have existed at all or where the invention was the essential ingredient in the creation of the infringer’s whole product or process. It may be possible, as discussed above, to reconcile such cases with differential profit analysis on the basis that the NIA is not manufacturing products at all, but that does not seem satisfactory. In any event, that is not how they have been analysed in the English or Australian case law. I shall return to this question in the context of ground 2.

 

160. Secondly, as Astronics and Panasonic also accept, there may be some cases in which differential profit analysis is a useful tool to identify the profits derived from the infringement, rather than from other factors. These will be cases where there is a well-defined and uncontested NIA which only changes the defendant’s product or process to the minimum extent necessary to avoid infringement and holds all other factors constant. For the reasons given in paragraphs 149-157 above, I am sceptical as to whether there are likely to be many such cases.

 

161. Thirdly, the English and Australian case law demonstrates that there is a well-established alternative to differential profit analysis, which is for the court to make a fair apportionment of the profits in issue. This approach is to be adopted when the case does not fall into either of the two categories discussed in paragraphs 159-160 above, that is to say, it is not a case where all of the profits are derived from the infringement or where a differential profit analysis can readily be undertaken. I will discuss how apportionment is to be carried out when I come to ground 4.

 I’ll stop there for now, and comment a bit on the preceding paragraphs.

First, I agree with Lord Justice Arnold that, in the real world, and particularly in cases involving complex products, it often may be unduly difficult or impossible to calculate the profit attributable to the infringement by means of the differential profits approach; and that, in recognition of these difficulties, what courts often tend to do is to apportion, as best they can, the profit attributable to the invention in comparison with the other features of the accused product.  This is a point I do make when I teach my IP remedies course, but I probably have not made it sufficiently clear in my scholarship, which has tended to emphasize why I believe that, in general, the differential profits approach is correct in theory and should be employed when it is feasible to do so.  A related point that I have tried to emphasize, however, in papers such as Patent Damages Heuristics, is that there often is a tradeoff between (theoretical) accuracy and administrability; and that sometimes the net benefits of a more easily administered rule outweigh the net benefits of (what might seem to be) a more precise, but also more costly to implement, approach.

Second, I agree with much, though not all, of Lord Justice Arnold’s analysis of the problems that can arise when trying to carry out a differential profits analysis.  It is, of course, often difficult to determine precisely what the NIA was (or to disentangle how its use might have affected other aspects of the accused product).  There is also the difficulty of determining how to proceed if a proposed NIA itself turns out to be patented, which is a matter my coauthors and I briefly noted at pages 20-22, 62 of Patent Remedies and Complex Products, but didn’t attempt to resolve. Where I might respectfully disagree with Lord Justice Arnold is in his discussion at paragraphs 154-55 of whether a differential profits analysis should take into account “adventitious commercial factors” or what the defendant would have done but-for the infringement as opposed to what it could have done.  More generally, I tend to agree with Norman Siebrasse, in his critique of the Canadian Supreme Court decision in Nova v. Dow (a case discussed in the present decision at paras. 127-32), that the correct approach when applying the differential profits analysis is to identify what the defendant would have done but-for the infringement, even if that noninfringing option consists of deploying its resources to make an entirely different product.  The goal of the disgorgement remedy should be to determine how much the defendant benefited from the infringement, and that means taking into consideration what action the defendant would have taken had it not infringed and estimating what benefits, adventitious or not, it would have derived from doing so.  For further discussion of this issue, see Siebrasse's article on Nova v. Dow, Part 1, particularly pp. 299-301 (arguing, inter alia, that "[a]n effort to determine the true value of the invention, apart from such happenstance, is akin to an effort to determine the true harm from negligent driving by awarding damages according to some ideal or average harm that would be caused by a negligent driving accident, rather than the accident that actually happened"). 

Third, and related to the preceding points, the differential profit approach can run into problems if there are two or more patents that are essential to the production of the product in question.  Imagine, for example, that two patentees each own an essential patent that is infringed by the maker of the accused product, and that each patentee files its own independent infringement action.  Each might claim that, absent the use of the patented technology in suit, the defendant would have earned zero profits; but surely it would make no sense to award each patentee the entire profit earned from sales of the infringing product.  Some sort of apportionment would therefore seem to be necessary instead.  (Perhaps the correct theoretical approach solution in such a case would be to apply some version of Shapley Pricing to isolate the inventive contribution of each essential patent, as Siebrasse and I proposed in The Value of the Standard--though we cautioned there that our analysis was not “intended to describe how we think a real world royalty setting process should work,” but rather as presenting “conceptual benchmarks for assessing a FRAND royalty” (p.1199).  I would also call interested readers’ attention to a somewhat analogous issue discussed by Jason Reinecke in his article Lost Profits Damages for Multicomponent Products:  Clarifying the Debate, 71 Stan. L. Rev. 1621 (2019), in the context of lost profits and multiple essential patents.) 

Returning to the Lufthansa decision, the next portion of Lord Justice Arnold’s opinion concludes, in brief, that the trial court was correct to apportion profits, despite some language in the lower court opinion suggesting that it was doing so despite having found that those profits were not legally (proximately) caused by the infringement (see paras. 162-76).  In fact, there was sufficient evidence that an allocable portion of the profits were legally caused by the infringing use of the patent (paras. 177-90).

That leaves for discussion Lord Justice Arnold’s analysis of the method of apportionment, the double recovery issue, and interest, as well as the two short concurring opinions by Lord Justices Nugee and Lewison.  I will return to these in a subsequent post or posts.