Michael Ritscher (MLL Legal), a driving force behind the Zurich IP Retreats, welcomed the participants, whom he affectionately referred to as «IP nerds». He emphasised the continuity and relevance of the Retreats and opened the event by reflecting on whether universal assumptions in IP law actually hold true. Referring to a study by Florent Thouvenin et al. analysing thousands of trademark oppositions, he pointed out that the influence of fictitious persons in decision-making may be overestimated.​¹ Introducing the concept of a «noise audit»,​² he set the stage for the day’s discussions on reducing subjectivity and variability in IP law through empirical and technological tools.

Stefan Bechtold (ETH Zurich) delivered a keynote speech that laid some theoretical groundwork for the retreat. Bechtold’s presentation explored the historical development and theoretical underpinnings of legal constructs in intellectual property law, such as the «person skilled in the art» and the «average consumer». Bechtold highlighted the need for such constructs to be adapted in response to contemporary challenges. He argued that while these fictional characters have served as valuable tools for bringing consistency to IP judgments, they are not without limitations, e.g. due to oversimplification of the nuanced and context-specific factors that influence how real-world stakeholders act. To address these shortcomings, Bechtold suggested incorporating empirical data and advanced technological tools into the decision-making process to enhance objectivity and reduce variability.

Drawing parallels with antitrust law and the evolution of the Sherman Act (1890), Bechtold noted how legal frameworks have increasingly shifted their focus from theoretical constructs to the real-world effects of regulatory decisions. He suggested that IP law could benefit from a similar shift, using modern analytical tools to better understand the practical implications of different legal standards.

One of the innovations discussed was the use of citation networks in patent law. Bechtold explained how these networks allow legal professionals to calculate the «distance» between different artefacts based on their citations in patent documents. By analysing these relationships, courts and patent offices can gain a more accurate understanding of the prior art landscape and assess whether an invention represents a significant advance. Such tools might not only make decisions more objective but might also allow for greater transparency in how judgments are reached.

Bechtold also introduced the concept of «asking the real expert», which he linked to the integration of artificial intelligence (AI) into IP decision-making processes. He discussed how AI systems are already being used to assess the distinctiveness of trademarks or the likelihood of confusion between competing marks. These tools, he argued, can process vast amounts of data with remarkable speed and consistency, making them invaluable in cases where subjective human judgment could introduce bias or error.

The discussion following Bechtold’s keynote was lively, with participants raising questions about the practical implementation of the ideas presented. These exchanges set the stage for the subsequent sessions, which delved deeper into the practical and theoretical dimensions of the basic constructs of intellectual property law.

Panellists: Christopher Sprigman, Mark Schweizer (Moderator: Stefan Bechtold)

The session entitled «Tools for Making Decision-Making in IP More Objective» addressed one of the key challenges in IP law, i.e., how to reduce subjectivity and variability in legal judgments. Mark Schweizer, President of the Swiss Federal Patent Court, set the tone for the session by highlighting the difficulties of adjudicating «mixed questions» of law and fact. Mixed questions arise when elements of both legal principles and factual determinations overlap. Schweizer noted that this may lead to (perceived) uncertainty and (perceived) inconsistent outcomes.

Schweizer began his presentation with approaches to define the term «objective», i.e. that a decision should not be influenced by personal feelings, interpretations or prejudices; rather based on facts and being unbiased. Since objectivity hinges on the facts of the case, decision-making would be (more) objective if empirical evidence were used whenever possible. However, Schweizer noted that there are considerable downsides of this approach. First, the extensive use of empirical evidence is very costly and time-consuming. Secondly, empirical evidence does not help in case of mixed questions of facts and law, such as obviousness in patent law. Schweizer argued that the introduction of «hard-and-fast rules» could further reduce subjectivity in IP decisions. For example, the presumption of acquired distinctiveness for trademarks after ten years of continuous use provides a clear, objective criterion that simplifies legal analysis. Similarly, in patent law, if a combination of three or more pieces of prior art documents is necessary to arrive at a claimed subject-matter, one could acknowledge non-obviousness. Schweizer noted that such rules remove the human factor from the equation, which would be in line with what objectivity is all about. Further, such rules allow for fast and cost-effective decisions and make the outcome predictable for users of the IP system. Schweizer drew a parallel between «hard and fast rules» and large language models (LLMs): Both are algorithms. Schweizer explored the use of LLMs (and AI in general) and their potential to standardise decision-making processes. AI systems, such as Google’s Bard,​³ have demonstrated the ability to assess the likelihood of confusion between two trademarks – not necessarily with the same outcome as a court, but with an elaborated reasoning and about 14 million times faster.​⁴ A further key benefit of such AI systems is consistency: Unlike humans, who can vary in their interpretations, AI systems apply the same algorithms and criteria across cases, ensuring consistency. However, the discussion also touched on the limitations of AI. Concerns were raised about the «black box» aspect inherent in many AI systems, where the underlying decision-making processes are not transparent. In appeals contexts, this lack of transparency poses significant challenges for legal review. In addition, AI systems are only as good as the data with which they are trained. Biases in training datasets can perpetuate or even exacerbate existing inequalities in decision-making, and variations in prompting might have undesired impacts on the outcome.

Christopher Sprigman (New York University School of Law) gave an inspiring presentation with numerous examples. For example, he presented the then ongoing Pennsylvania State University v. Vintage Brand, LLC case, in particular with respect to the ornamental use defence.​⁵ Sprigman critically compared this case with the Arsenal decision of the CJEU.​⁶ With reference to the Jack Daniel’s Properties, Inc. v. VIP Products LLC saga, Sprigman touched upon trademark dilution through blurring, i.e., the association arising from the similarity between a junior mark and a famous mark that impairs the distinctiveness of the famous mark.​⁷ He criticized that often the mere fact that consumers associate the junior mark with the famous mark is sufficient in court to establish blurring. In fact, research has shown that even when consumers associate a junior mark with a famous senior mark, this association does not necessarily result in any impairment of the ability of the senior mark to identify its source and associations.​⁸

Panellists: Klaus Grabinski, Christopher Floyd, Cyrill Rigamonti (Moderator: Mark Schweizer)

This session focused on one of the most intricate issues in IP law: the distinction between questions of law and questions of fact. Panellists explored the implications of this divide for judicial decision-making. The discussion revealed different perspectives on how legal principles and factual determinations intersect in areas such as patent law, trademark disputes, and design rights.

Cyrill Rigamonti (University of Bern) opened the session by emphasising the fundamental importance of the distinction between questions of law and questions of fact. He explained that even though there is nothing factual about the fictive «person skilled in the art», the knowledge of this person still is a question of fact. Rigamonti explained that judges need to be informed of such knowledge; the knowledge of the skilled person is a question of law with factual underpinnings. He reminded the audience that the Swiss Federal Patent Court had dismissed a nullity defence merely because the skilled person and the respective common general knowledge of that person had not been pleaded by the defendant.​⁹

Rigamonti also warned against what he called the «creeping disempowerment» of judges. Over-reliance on expert testimony, he argued, risks delegating critical legal decisions to experts, thereby undermining judicial authority and accountability.

Klaus Grabinski (Court of Appeal of the Unified Patent Court) first outlined the dogmatic difference between questions of fact and questions of law. Facts are open to evidence and, if contested by the adversary party, must be proven by the party bearing the burden of proof. Moreover, the facts must be established by the court, to the extent necessary to decide the case. On the other hand, questions of law concern the interpretation of legal provisions and the subsumption of the established facts under the legal provisions.

Grabinski then offered a deep dive into how issues of law and fact manifest themselves in various aspects of patent law. First, he made clear that claim construction is a question of law that cannot be decided by a court-appointed expert. However, claim construction has factual underpinnings because it must be done from the skilled person’s perspective. Grabinski then explained that in the assessment of infringement under the doctrine of equivalents, all three questions (same effect; accessibility; claim orientation) are mixed questions of law and fact. Likewise, the prior art (Formstein) defence is a question of law, with factual underpinnings.​¹⁰ The same holds true for novelty, obviousness,​¹¹ and sufficiency of disclosure.​¹²

Sir Christopher Floyd (former Lord Justice at the Court of Appeal, England and Wales) explored the central role of legal constructs in IP law, focusing on their purpose, application, and limitations. He began by emphasizing how constructs such as the «person skilled in the art» in patent law, the «average consumer» in trademark law and the «average consumer with an interest in designs» are intended to normalize decision-making in their respective field. More generally, he referred to these legal constructs as «the man on the Clapham omnibus.»​¹³ These constructs serve to strike a balance between competing interests, such as protecting innovation and promoting free trade, while distancing decisions from the subjective opinions of individual judges or tribunals. Floyd detailed the attributes ascribed to these legal constructs. For instance, the «person skilled in the art» is portrayed as technically knowledgeable but devoid of inventive capability. This person is highly informed, methodical, and willing to engage with any material placed before them – but the degree of interest may vary, depending on the facts of the case. Similarly, the «average consumer» in trademark law is deemed reasonably well-informed and observant, yet neither unusually insightful nor careless; i.e., as a matter of law, the «moron in a hurry» is excluded.​¹⁴

Floyd′s presentation emphasized the distinction between questions of law and questions of fact in IP cases. Legal questions, such as defining the characteristics of the skilled person, are distinct from factual inquiries, such as determining the common general knowledge at the relevant time. Floyd explained that establishing common general knowledge, for example, relies on evidence such as textbooks, technical materials, and expert testimony, whereas legal interpretation remains the responsibility of the court. Several illustrative cases were used to highlight these principles. Floyd discussed the case of Dyson v. Hoover, where the reluctance of industry professionals to adopt bagless vacuum cleaner technology influenced the court’s assessment of obviousness.​¹⁵ In another example, involving extendible garden hoses, he contrasted two separate judgments concerning the very same patent. In one, the skilled person was limited to a designer of garden hoses,​¹⁶ while in the other, the definition was broadened to include expertise in technical hoses.​¹⁷ These cases underscored the significant role of evidence in shaping how the skilled person is defined and how this impacts case outcomes. Floyd addressed the use of expert evidence, emphasizing that experts are not called to embody the skilled person but to provide insights into prevailing practices and attitudes. Their testimony assists the court in understanding the technical and contextual backdrop of a case, but the ultimate legal conclusions remain the responsibility of the court.

The presentation also explored the nuanced interpretation of the term «average.» Floyd argued that the construct should be understood as «typical» rather than strictly «average» in a mathematical sense, reflecting practical realities and common behaviours. For instance, the «average consumer» would exhibit varying levels of attentiveness depending on the context, such as when purchasing everyday items versus specialized equipment.

Panellists: Klaus Grabinski, Christopher Floyd, Giovanni Pricolo (Moderator: Tobias Bremi)

The panel on «The Measure of All Things in Patent Law,» moderated by Tobias Bremi (Second Ordinary Judge at the Swiss Federal Patent Court) delved into the pivotal role of the «person skilled in the art» in patent law. Ritscher introduced the panel with the observation that harmonization in patent law has quite likely progressed faster than in other areas of intellectual property. Still, Bremi highlighted quite some procedural differences among jurisdictions, such as the absence of cross-examination of experts in Germany, in contrast to the situation in the UK. Further, the European Patent Office (EPO) does not even define the skilled person on a regular basis, unlike, e.g., the Swiss Federal Patent Court. These differences framed the discussion on how the construct is interpreted and applied across systems. Grabinski set the stage with an overview of definitions of the «skilled persons» for the various aspects in patent law. For the assessment of obviousness, Grabinski defined the skilled person as the one to whom one would have turned to solve the problem underlying the invention. With respect to sufficiency of disclosure, the skilled person is the one to whom one would have turned to carry out the invention. For claim construction, the skilled person is the one to whom one would have turned to understand what is protected by the claim. Notably, Grabinski made clear that the fictional skilled person «does not have good and bad days» and cannot give testimony.

Grabinski also expanded on the topic of how AI might fit into the skilled person construct. In his view, treating AI as a potential tool in the hands of the skilled person could be the right approach, for the time being. Interesting questions that could arise in this respect are whether AI was actually used in the respective field of technology on a regular basis and at the relevant time, and whether it would have been necessary to configure the AI to solve the relevant problem.

Floyd began his presentation with reference to Protagoras: The Greek philosopher said that «man is the measure of all things: of the things that are, that they are, of the things that are not, that they are not.» This is commonly referred to as the philosophy of relativism, i.e., that truth is a relative concept. By way of example, what feels warm for one person does not feel warm for another person. It would be undesirable if issues in a patent case were determined from such a subjective perspective. Floyd described the skilled person as «incapable of a scintilla of invention» while being methodical in solving the technical problem underlying the invention.​¹⁸ Floyd highlighted several oddities of the skilled person, including the assumption that they are «very boring» and «very forgetful.»​¹⁹ Moreover, Floyd cited Nicholas Pumfrey with the observation that the skilled person has knowledge «like a walking library who cannot see its way past the next corner.» But what if those who were involved in the art at the relevant time were all inventive people?​²⁰ Putting it differently: How inventive must a step be for it to be regarded non-obvious? It was held in the discussion that this is a very slippery slope that should be avoided. Floyd finally mentioned that even though the skilled person is not a patent lawyer, he still has knowledge of certain drafting conventions.​²¹ Floyd also touched upon the question whether the skilled person is the same for all purposes. While this will be true in most cases, there are (probably rare) cases where the invention lies in bringing in new expertise, but where the introduction of that expertise into the skilled team is not justified on the basis of the prior art.​²² Finally, Floyd questioned whether the common general knowledge is always that at the priority date. He gave the example of the second question in the assessment of infringement under the doctrine of equivalents: Would the skilled person, on being told what the variant does, consider it obvious that it achieved substantially the same result as the claimed invention in the substantially the same way?​²³ Even though that question is to be considered at the priority date, it also applies to variants which rely on, or are based on, developments which have occurred after the priority date. It is therefore necessary to imbue the skilled person with more information than he would have had at the priority date. On the same token, Grabinski mentioned that, exceptionally, also knowledge that became obvious after the priority date may matter, namely in case of unforeseeable equivalents.​²⁴

Giovanni Pricolo echoed Bremi’s observation that it is rarely critical in EPO proceedings who exactly the skilled person is. In most cases, the skilled person is one from the specific technical field of the invention who might also consider neighbouring technical fields. Or the skilled person is one in another technical field if prompted towards that technical field by the problem underlying the invention.​²⁵

Panellists: Estelle Derclaye, Ivàn Sempere, Alina Waldmann (Moderator: Peter Schramm)

The session «The Measure of All Things in Design Law» focused on the central role of the construct of the «informed user» in assessing the validity and scope of design rights. Panellists explored how the informed user is defined and applied in different jurisdictions, the challenges associated with this legal fiction, and its adaptability in the context of evolving design practices and industries.

Estelle Derclaye (University of Nottingham) began the discussion by outlining the characteristics of the «informed user» from a UK and EU perspective. With reference to the European Court of Justice, Derclaye explained that «that concept must be understood as lying somewhere between that of the average consumer, applicable in trade mark matters, who need not have any specific knowledge and who, as a rule, makes no direct comparison between the trade marks in conflict, and the sectoral expert, who is an expert with detailed technical expertise.»​²⁶ Accordingly, the informed user is not a user of average attention, but a particularly observant one, either because of his personal experience or his extensive knowledge of the sector in question. She noted that the informed user typically makes a direct comparison between the designs at issue, but courts can take into account the actual products corresponding to the respective designs for illustration purposes. With respect to the approach of courts in the UK, Derclaye outlined that the infringement test is meanwhile well established. First, it is the overall impression from the viewpoint of the informed user that matters. It is sufficient to avoid infringement if the accused product produces a different overall impression; i.e., the test is «different», not «clearly different». As Lord Justice Jacob held: «If a design differs, that is enough – an informed user can discriminate.»​²⁷ Derclaye concluded her presentation with an overview of exemplary case law to illustrate the practice.​²⁸

Iván Sempere (University of Alicante) broadened the picture and illustrated the same principles with case-law from Spain.​²⁹ He reminded that neither the product indication nor the classification affects the scope of protection of a Community design as such; Art. 36(6) CDR. He added that expert reports are admitted in Spain, and surveys may be part of such reports.

Alina Waldmann (District Court of Dusseldorf) expanded further on the case-law of the CJEU and courts in Germany. She explained that, according to CJEU case-law, it is not necessary for a design to fall within the scope of protection of an earlier design that the informed user knows the earlier design.​³⁰ Accordingly, there is an absolute presumption that the informed user knows the earlier designs, even across various industry sectors. This underlines that the informed user is a fictional person. Waldmann further expanded on how judges step in the shoes of the informed user (usually without hearing experts or carrying out market surveys), based on the parties’ submissions regarding the existing prior art in the relevant product market and the resulting degree of freedom of the designer.​³¹ Waldmann concluded with the finding that courts focus on what the informed user knows and observes, rather than asking who the informed user is. The courts’ assessment in this respect depends on being provided with the relevant information by the parties.

Peter Schramm (MLL Legal) expressed reservations about tendencies in Swiss case-law to mention «average attention» – which is a fiction borrowed from trademark law. Schweizer noted that scope of a design goes hand in hand with the prior art. In assessing the overall impression, it was recommended to begin with the similarities, and to give those features less (or no) weight where there is little or no design freedom, i.e., features that are dictated by function.

Panellists: Estelle Derclaye, Philippe Gilliéron, Christian Peukert (Moderator: Michael Ritscher)

The session «The Measure of All Things in Copyright Law» explored the role and implications of the «author’s own intellectual creation» (AOIC) standard – a cornerstone of copyright law in Europe and many other jurisdictions.

Derclaye explained that the originality test in EU copyright law does not depend on a fictional person. She explained that, since the Infopaq judgment of the CJEU, the AOIC standard applies to all works: for originality and infringement.​³² She then turned to the Painer judgment of the CJEU.​³³ In that case, the CJEU made it very clear what it meant by the AOIC standard, i.e. a substantive creative requirement for copyright to apply. The standard is met if the author was able to express his creative abilities in the production of the work by making free and creative choices. Painer also held that there is no support for the view that the extent of protection should depend on possible differences in the degree of creative freedom in the production of various categories of works. Specifically with respect to a portrait photograph, the protection conferred cannot be inferior to that enjoyed by other works, including other photographic works. Derclaye then gave a sneak preview of her book, co-authored with Gilles Stupfler, that will provide the first comprehensive comparative and empirical analysis of the state of harmonization in EU copyright law in the 27 Member States, and the UK, at the level of national courts.​³⁴ Derclaye held that the current practice is far from harmonized, and many courts use the prior art to judge originality – which they should not.

In conclusion, Derclaye argued that there is still a need to objectify the AOIC test. Who ultimately carries it out? It is questionable whether judges or experts can be truly objective.

Christian Peukert (University of Lausanne) presented parts of his research on copyright levies and cloud storage.​³⁵ Copyright levies are surcharges on devices to compensate right holders for harm caused by private copying exceptions. Adaptations of levy schemes are regularly discussed, e.g., whether cloud storage should be included in such schemes. His estimates indicate that any (realistic) scenario of levies for the cloud would be welfare decreasing, and it seems unlikely that there are large enough dynamic effects to compensate for the welfare loss. Additionally, the administrative costs are significant. Peukert concludes that, from an economic perspective, adapting the copyright levy system in response to technological change such as cloud storage does not seem advisable. More efficient and potentially welfare-neutral approaches should be developed if the goal is to balance the interests of rightsholders, consumers and cloud storage operators.

In the subsequent discussion, the red bus decision was highlighted.​³⁶ Do obvious similarities suffice, if these similarities are due to the fact that the infringer saw the original work before creating his own work? According to the red bus decision, the answer is yes – when one copied too much of the original work. The tricky question is where to draw the line. A pointed remark from the audience: «Copyright infringement requires copying!»

Panellists: Barton Beebe, Anne Niedermann, Jürg Simon (Moderator: Jürg Simon)

The Trademark Law session focused on the elusive and central concept of the «average consumer» in the assessment of distinctiveness and likelihood of confusion in trademark cases. Panellists discussed how this construct is defined and applied in different jurisdictions, the evidentiary challenges involved, and the evolving role of empirical data and market research in shaping legal outcomes.

Jürg Simon (Lenz & Staehelin) set the stage with some basics of trademark law. He drew the conclusion that it (always) matters how signs are perceived: in the assessment of whether they are distinguishing, and when absolute and relative grounds of exclusion are assessed. The measure of all things in the EU is the average consumer who is reasonably well informed and reasonably observant and circumspect.​³⁷ In Swiss jurisprudence, the leading case referring to the «relevant public» is Kamillosan.​³⁸ Simon walked the audience through some Swiss Supreme Court judgments to get closer to what the relevant public actually is, knows, does, and thinks. For example, the descriptive character of a sign must be «immediately recognizable by the addressed public without any special effort of thought or imagination.»​³⁹ The distinctive character of a sign is to the assessed «on the basis of the meaning which, from the point of view of the relevant public, is paramount in connection with the product claimed.»​⁴⁰ Moreover, it is the «overall impression that the sign leaves in the mind or the relevant addressees» that is decisive.​⁴¹ Simon then turned to the traditional «escape routes» from the difficulties of consumer assessment. One escape route is to treat an issue as a question of law. Questions of law are not open to evidence, and the decision-making power is thus shifted to the court. Examples of questions of law include the delimitation of the relevant group of addressees; how that group perceives the sign on the basis of the expected attention, and whether there is a likelihood of confusion.​⁴² Yet another escape route is to apply rules of experience. Simon questioned whether rules of experience are challengeable, i.e., whether it is possible to prove – based on empirical evidence – that a rule of experience is wrong or not applicable in a specific case. Next, Simon expanded on the use of surveys in trademark matters. Surveys have historically been treated as mere party allegations but are nowadays well accepted as means to provide proof.​⁴³ The Swiss Federal Institute for Intellectual Property (IPI) guidelines emphasize that demoscopy must be carried out professionally and observe the criteria of repeatability, verifiability and |comprehensibility. A discussion arose whether the almost standardized practice of demoscopic surveys in trademark matters might be useful in patent, design or copyright cases as well.

Barton Beebe (New York University School of Law) liked the fact that likelihood of confusion is being treated as a question of law, as Simon had explained. He held that Europe is further ahead here as compared to the U.S. Separately, he wondered whether the notion of «consumer» makes practitioners think like economists – which «is not good for lawyers.» He proposed to just refer to citizens.

Anne Niedermann (Allensbach Institut) expanded on how surveys for legal evidence differ significantly from market research. She referred to the EUIPO trademark guidelines where it is held that opinion polls and market surveys are the most suitable means of evidence gathering for providing information about the degree of knowledge of a mark, the market share it has, or the position it occupies in the market in relation to competitors’ goods. The Swiss Federal Supreme Court held that a survey that has been scientifically conceived and correctly conducted «is the most suitable means of proof.»​⁴⁴

Panellists: Stefan Bechtold, Tobias Bremi, Michael Ritscher, Mark Schweizer

The closing discussion of the Zurich IP Retreat 2023 provided an opportunity for panellists and participants to reflect on the insights gained during the two-day event. Moderated by Michael Ritscher, the session focused on synthesising the key themes from the panels, exploring the practical implications and considering future directions for research, policy and practice.

From a practical perspective it was argued that the actual technical experience of the skilled person in patent matters is rarely important. Floyd reminded the audience of an ad hoc survey that had been done on the first day to figure out whether practitioners feel comfortable with the current state of play in their fields. This was answered in the affirmative, to an overwhelming extent. «None said we are doing it badly.» Floyd noted that this attitude is not helpful for improving systems, and he expressed hope that the Unified Patent Court (UPC) will get things moving.

A patent practitioner held that there was little found in common amongst the various IP rights that the Measure of All Things have in common. Even in patent matters, it is not always the same fictious person – which is not bad per se. If one runs into a new issue, one has to re-think anyway.

Getting back to the questions of law topic, and that these questions of law are not open to evidence, it was remarked that this is a perhaps too strict assumption. It would be helpful to come up with a more tangible definition.

On the general goal of making decisions more objective, it was questioned whether this is even helpful at all. The fictional person of average skill in the art is, by definition, unrealistic. But the system still works – not because the fiction of the skilled person makes decisions more realistic, but more predictable at least.

Ritscher closed the discussion with the remark that the Measure of All Things in all IP fields are fictitious persons. Even though this was not much discussed at the event, this very likely also applies in copyright law. Otherwise, judges should say it is they who decide.