Conference Report: Zurich IP Retreat 2017 - Patents and Hindsight (Part I)

The first Zurich IP Retreat was held on Friday/Saturday 8/9 September 2017 on the shores of lake Zurich, organized by INGRES and ETH Zurich, in honour of Dr. Dieter Brändle, the first President of the Swiss Federal Patent Court who will retire at the end of 2017. It brought together opinion leaders in the law of innovation and technology from all over the world, including judges, litigators, patent attorneys, in-house lawyers and academics, to discuss one of the fundamental problems in patent law: the need to assess the prior art, and most notably inventive step (non-obviousness), without hindsight and knowledge of the invention. Is this even possible from a psychological perspective? Likely how successful are the approaches to avoid hindsight bias developed in different jurisdictions? Presentations and panel discussions over the course of two short days shed some light on these questions.

Psychological Dimensions

The opening presentation and the first panel looked at hindsight bias from a psychological perspective.

Aileen Oeberst (Professor, University of Mainz, Germany) gave an overview of the psychological research on hindsight bias. “Hindsight bias” has two components: on the one hand, after the event is known, it appears more inevitable than before the event (“It had to happen”). On the other hand, it appears more foreseeable than it actually was (“I knew it all along”).[1] A third aspect is that the event distorts the memory – after the event, one even remembers (falsely) that the one predicted the event. Since this latter aspect seems not to play a role in the context of patent law, it is not further addressed. Which aspect of hindsight bias is more important in the context of obviousness assessments in patent law is an unresearched question. It is potentially important because the underlying process informs which debiasing strategies, if any, could be successful. Prof. Oeberst also raised the question whether knowledge of an invention is more alike an “event” or almanac knowledge, which are the categories commonly studied in hindsight experiments, and hinted that it might not fit into either of these categories.

Aileen Oeberst

Hindsight bias is a very robust phenomenon with small to mediate effect sizes.[2] Small effect sizes may still have an important effect on binary choice if the decider is on the threshold; i.e., in close cases. Whether domain expertise can safeguard against hindsight bias is not entirely clear, experts – specifically judges – are certainly not immune to hindsight bias.[3] Deliberation in groups does not seem to reliably reduce hindsight bias, but the research is limited and restricted to small groups (three people).

Debiasing hindsight bias is generally considered very difficult.[4] Instructions to “ignore the ex post knowledge” are useless.[5] The best strategy is to withhold the knowledge of the outcome from the decision maker, but this may not be feasible for obviousness assessments. Short of shielding the decision maker from information, “consider the opposite” strategies show promise.[6] It is, however, important that the decision maker comes up with his or her own reasons why the opposite may have happened. It is therefore unclear whether advocacy demonstrating reasons for other outcomes is a successful strategy for debiasing.

Greg Mandel (Professor, Temple University Law School, USA) summarized his original research on hindsight bias in patent law.[7] Mandel found, in an experimental study with mock jurors, a strong effect of hindsight bias. While only between 23% and 49% of participants (depending on scenario) found that a solution to the technical problem presented – with a selection of pertinent prior art – was obvious when they did not know the patented solution, 59% to 85% found the solution obvious when they were aware of it. In other words, about a third more decision makers found the invention obvious when they were aware of it versus those who were only told of the problem, despite having been informed of the same prior art references. Standard jury instructions for patent cases and explanation of the “teaching suggestion motivation” test did not reduce hindsight bias.

Marco Kleine (Senior Research Fellow, MPI for Innovation and Competition, Munich, Germany) added some thoughts from an experimental economist’s perspective, but had to admit that the economic research in this area largely relied on the psychological studies introduced by Prof. Oeberst. Several studies indicate that hindsight bias increases over time,[8] which might be important when obviousness is judged years after the invention was made. The type of ex post information matters – almanac questions exhibit greater hindsight bias than real world events or case scenarios.

Mark Schweizer (President elect, Federal Patent Court, Switzerland) added that people tend to incorrectly believe that others are more susceptible to cognitive biases than they are themselves (“bias blind spot”). Cognitive sophistication, as measured by SAT (Scholastic Aptitude Test) and the Cognitive Reflection Test (CRT) does not reduce the bias blind spot.[9] In other words, smart people also believe they are less susceptible to biases than others, but are in fact equally biased.

Avoiding Hindsight Bias In Patentability Assessment Around The World

The second panel sought to show which approaches to reduce hindsight bias different jurisdictions had developed.

Szepler, Yu, Klett, Arnold, Kalden, Bremi, O'Malley, Meier-Beck, Ashley (2nd Panel)

Graham Ashley (Chair, EPO Board of Appeal) briefly summarized the EPO’s problem-solution approach (PSA).[10] Selection of the closest prior art is done with knowledge of the invention, potentially introducing hindsight bias. The attempt to avoid this is the requirement that the prior art must address the same or a similar problem than the invention. A critique of having a “closest” prior art, i.e., conducting inventive step assessment only starting from one prior art document, is that the invention must be inventive over the entire state of the art (Article 56 EPC), so it should be irrelevant what the starting point is.

When defining the objective problem to be solved, care must be taken that the problem is defined as achieving the effect of the distinguishing features, not the distinguishing features. If the distinguishing feature is X, the problem is not to find X, but how to achieve the effect that X has.

Another safeguard against hindsight bias is a strong emphasis on whether the skilled person “would” have found the solution. The “would” question has three aspects: (i) would the skilled person having started from D1 (= closest prior art) have actually considered D2; (ii) would he or she actually found the solution in D2; and (iii) would the skilled person have applied the suggested solution to the teaching of D1? Graham explained that “his” Board of Appeal often sketched the argument for inventive step as well as against inventive step, assuming first that the invention is based on inventive step, then assuming that it is not based on inventive step. Whichever argument reads more convincingly prevails.

Peter Meier-Beck (Presiding Judge, 10^th [Patent] Senate, Federal Court of Justice, Germany) emphasized that a retrospective view cannot be avoided, but one should strive to avoid being biased. There is a risk to read into the text what one has only learned subsequently. It is essential that any piece of prior art is read in its own context. Reducing the reading to elements or paragraphs which are or seem to be similar or close to the invention is highly dangerous and seduces to interpreting the text in accordance with an understanding which is predetermined by the patent in suit, i.e. by hindsight.

Starting the analysis by determining the closest piece of prior art is also highly dangerous. We need to know the invention to determine what is the closest piece of prior art. But ante inventionem the skilled person does not necessarily know the most promising springboard. Taking this springboard for granted is pure hindsight. Defining the problem by determining the difference between the invention and the closest piece of prior art is also hindsight. Sometimes it is plausible that the skilled person would have tried to solve that problem, sometimes the problem construed that way may be artificial.

Of course, we need some reduction of the complexity of prior art to find a reasonable starting point. But is makes sense to ask the question whether the starting point was realistic ante inventionem and whether it was plausible that a skilled person would have tried to solve a problem which is the result of our hindsight-led “problem construction” and not mentioned in a prior art document.  It is not enough that the skilled person could have started from the closest prior art, it is required that he or she would have – a similar question like the one posed in the third step of the PSA, but now posed in a different context, namely the choice of the starting point for the development that led to the invention.

Kathleen M. O’Malley (Judge, Court of Appeals for the Federal Circuit, USA) explained that currently, there were three avenues to challenge patents in the United States – through the District Courts up to the CAFC, through the International Trade Commission, and through the USPTO Patent and Trademark Appeal Boards (PTAB) to the CAFC. The approaches in the different venues were not necessarily congruent, which leads to problems. The PTAB has been instructed to “get rid of bad patents” – it has been told to be (hindsight) biased. PTAB combine a lot of references in finding obviousness, unlike District Courts. They also do not consider secondary, objective criteria, that traditionally plays an important role in obviousness determinations in US law.

Hindsight bias is particularly difficult to deal with in patent cases. In other cases, the court can limit the information the jury receives, e.g., changes to product design after an injury occurred. This is not possible in patent cases; the jury must know the invention. O’Malley believes that (constitutionally mandated) twelve-member juries lead to better decisions, as the discussion becomes more nuanced and violations of jury instructions less likely.

Additional “biases” may be introduced by attorneys. Counsel for patentee tries to paint the inventor as heroic and sympathetic. Copying, which is considered a secondary indication of inventiveness, often helps the patentee.

Relevant prior art may also be from related fields, not necessarily from the same field as the invention. There is not necessarily a “closest” prior art except in chemical/pharmaceutical cases, where the assessment must start from a lead compound. Expert evidence is used to understand what the prior art actually taught back at the priority date.

It is important to show that there was motivation to combine references. Such motivation can come from outside the field of the invention. Another important consideration is whether there was a reasonable expectation of success. Teaching away, often difficult to prove, may safeguard against hindsight bias.

Secondary evidence such as a surprising effect of the new feature, a longfelt need, scepticism in the field and failed attempts by others to solve the same problem (ideally by the defendant), industry praise, commercial success and copying areimportant elements of hindsight analysis and help avoiding hindsight bias.

Inventions in technological fields that are familiar to the general public may be found obvious more readily, while complex technology makes a combination of references seem less obvious to a jury of lay people.

Rian Kalden (Judge, The Court of Appeal of The Hague, Netherlands) notes that hindsight affects inventive step by definition, because the judge is supposed to go back in time before the contribution by the invention was known. This assessment requires knowledge of the contribution.

Simple solutions to difficult problems are often underappreciated. Dutch courts generally employ the the PSA, but it is not required. Cases are often framed in PSA terms by the lawyers. One criticism is that the PSA is not suitable for “problem inventions”, where the creative act lies in defining the problem. Well – if it is not suitable, do not use the PSA, it is not required by law. While the PSA may not be immune to hindsight bias, neither are other approaches.

One safeguard against hindsight bias is that under the PSA, the closest prior art must be a document that skilled person would actually have looked at – in the same field, trying to solve the same problem. The closest prior art may only be from a different field of technology if there is a good reason choose it. One should not simply count the number of identical features – this may be decisive only when there are several realistic starting points.

Formulating the objective problem determines the outcome of the case. The great risk of the PSA is formulating the wrong problem – such as including a feature of the invention in the problem to be solved rather than an effect of that feature. The objective problem must not contain any pointers to the solution. There is almost always disagreement on the formulation of the problem to be solved between the parties. Judges need to be very careful. If the closest prior art is the same as in the application, the problem should be the same as the one formulated in the application, too. However, if the closest prior art is different, the objective problem will be different, too.

Would the skilled person have arrived at the solution at the priority date? Only a combination of two documents, possibly with common general knowledge (CGK), is allowed. A combination of more than two documents is an indication of inventiveness. No combination with just any other document that happens to share the missing feature is sufficient. There must be a pointer to the document to be combined and an incentive to find it, read it and use it. Not everything that can be found will be found. If the title and abstract of a reference point to another direction, the skilled person will probably not read it. Is there a reasonable expectation of success that the disclosed solution will work? Generally yes, unless there is teaching away and there are other options without teaching away.

Hindsight may also favour patentees, when further positive effects of the invention are found only after the filing date. It is difficult to ignore such effects, but they should be ignored. The plausibility test requires that the effect must be made plausible in the application for subsequent evidence to be considered and seeks to safeguard against this form of hindsight bias.

Richard Arnold (Judge, Patents Court, UK) explained that English law is clear – hindsight bias must be avoided when assessing obviousness. Moulton LJ, in British Westinghouse Co v Braulik (1910) 27 RPC 209, 230, states “I confess that I view with suspicion arguments to the effect that a new combination, bringing with it new and important consequences in the shape of practical machines, is not an invention, because, when it has once been established, it is easy to show how it might be arrived at by starting from something known, and taking a series of apparently easy steps. This ex post facto analysis of invention is unfair to the inventors and, in my opinion, it is not countenanced by English Patent Law.”

In Mölnlycke AB v Procter & Gamble Ltd [1994] RPC 49, 113 Sir Donald Nicholls VC emphasized the importance of considering secondary (contemporary) evidence to avoid hindsight bias: “What with hindsight seems plain and obvious often was not so seen at the time. It is for this reason that contemporary events can be of evidential assistance when testing the experts’ primary evidence. … Secondary evidence of this type has its place and the importance, or weight, to be attached to it will vary from case to case.”

Technip France SA’s Patent [2004] RPC 46 at [15] reminds everybody that the reasons an expert gives for obviousness, not his or her conclusions, are most important.  Similarly, SmithKline Beecham plc v Apotex Europe Ltd [2005] FSR 23 at [53] teaches that not sympathy towards the expert, but the fundamental reasons for their opinions should determine the credibility of experts.

MedImune Ltd v Novartis Pharmaceuticals UK Ltd [2011] EWHC 1669 (Pat) at [118] teaches that “sequential unmasking” should be used in the instruction of experts. First, the expert should only consider the prior art, then the priority documents and only finally the patent(s). HTC Corp v Gemalto SA [2014] RPC 9 at [274] cautions that this is not always possible and not required by law.

In Schlumberger Holdings Ltd v Electromagnetic Geoservices AS [2010] RPC 33 at [77], [79] Jacob LJ exhorts the benefits of secondary evidence: “It generally only comes into play when one is considering the question ‘if it was obvious, why was it not done before?’ That question itself can have many answers showing it was nothing to do with the invention, … . But once all other reasons have been discounted and the problem is shown to have been long-standing and solved by the invention, secondary evidence can and often does, play an important role. If a useful development was, in hindsight, seemingly obvious for years and the apparently straightforward technical step from the prior art simply was not taken, then there is likely to have been an invention. … “

The attacker may rely on any prior art, even completely obscure references. This is hindsight, but not bias. The law allows this kind of hindsight for sound policy reasons.

Ritscher, Brändle, Stauder

Dariusz Szleper (Attorney, Paris, France) explains that hindsight bias is only explicitly addressed in French judgments since about 15 years ago under the influence of the PSA. He points out that the wording of Article 56 EPC in French is not precisely the same as in English (“ne découle pas d'une manière évidente” versus “is not obvious”).

It is not clear whether judges discuss the problem of hindsight bias in chambers, because French judgments are very concise, making it difficult to assess the underlying reasoning. Assessment of inventive step is considered a question of fact, outside the scope of appellate review, so appellate courts hardly ever, if at all, address the issue. Dariusz cannot find traces of the problem in the written opinions. During legal conferences, it is emphasized that the PSA is applied in France and whether the skilled person “would” find the solution, not whether he or she “could”, is relevant. While the reasoning may be opaque, the final outcome of French cases is often similar to those in other jurisdictions.

Dariusz concluded by adding that motivational biases such as the French “egalitarian mood”, which considers patentees to be asking for too much, may disfavour patentees.

Kathrin Klett (Presiding Judge, 2^nd Civil Chamber, Federal Supreme Court, Switzerland) reminded everybody that hindsight bias is a general problem, not limited to patent cases. It plays an important role in liability cases in the assessment of negligence (foreseeability of damage).

The Swiss Federal Supreme Court has not developed specific methods or approaches to avoid hindsight in patent law. It has always been aware, however, of the saying that knowledge after the event is always easy and problems solved present no difficulties. An invention involves inventive step when the solution is beyond the the zone situated in between the state of the art and what a person skilled in the art on the basis of his knowledge and skills would have found with only minimal mental effort. The formula differs from the legal practice of the EPO, but is designed to mean essentially the same.[11]

The choice of the starting point for the assessment of inventive step should be irrelevant, as the invention must be inventive starting from any reference. Therefore, an appeal based on the argument that the lower court chose the wrong starting point will be unsuccessful. It is the definition of the person or the team skilled in the art, their knowledge and specific technical skills at the priority date which are crucial for the decision whether the claimed technical solution was obvious or not. This seems to Kathrin Klett not so much a question of law, but of the expert knowledge guided by the appropriate questions.

Xiang Yu (Professor, Huazhong University of Sience & Technology, Wuhan, China) explained that under Chinese law, inventive step requires that compared to the prior art, the invention has prominent substantive features and provides notable progress. The assessment of inventive step is a three-step test. First, the prior art has to be determined. It must be from the same or similar technical field, which is defined increasingly narrow. Secondly, the distinguishing features must be identified. There was a risk that the examiner fails to look at the solution as a whole and misses the gist of the invention. Thirdly, it must be assessed whether the solution as such, not the distinguishing features alone, are obvious to the skilled person.

Chinese knows a figure of speech for hindsight bias – “Zhuge Liang after the fact” (“事后诸葛亮”), Zhuge Liang being a famous prime minister of ancient China known for his great wisdom and resourcefulness. To avoid hindsight bias, market success may be considered as secondary evidence for lack of obviousness.

---

[1] Blank, Nestler, von Collani & Fischer, How many hindsight biases are there?, Cognition 2008, 1408-1440.

[2] Guilbault , Bryant , Brockway & Posavac, A Meta-Analysis of Research on Hindsight Bias, Basic and Applied Social Psychology 2004, 103-117.

[3] Oeberst & Goeckenjan, When being wise after the event results in injustice: Evidence for hindsight bias in judges’ negligence assessments, Psychology, Public Policy, and Law 2016, 271-279.

[4] Giroux, Coburn, Harley, Connolly & Bernstein, Hindsight Bias and Law, Zeitschrift für Psychologie 2016, 190-203.

[5] Kelman, Fallas & Folger, Decomposing Hindsight Bias, Journal of Risk and Uncertainty 1998, 251-269.

[6] Roese & Vohs, Hindsight Bias, Perspectives on Psychological Science 2012, 411-426.

[7] Mandel, Patently Non-Obvious: Empirical Demonstration that the Hindsight Bias Renders Patent Decisions Irrational, Ohio State Law Journal 2006, 1391-1453.

[8] Bryant & Guilbault, “I Knew It All Along” Eventually: The Development of Hindsight Bias in Reaction to the Clinton Impeachment Verdict, Basic and Applied Social Psychology 2002, 27-41.

[9] West, Meserve & Stanovich, Cognitive sophistication does not attenuate the bias blind spot, Journal of Personality and Social Psychology 2012, 506-519.

[10] EPO, Guidelines for Examination in the European Patent Office, November 2016, Part G, Chapter VII, Section 5.

[11] Sutter, Der bundesgerichtliche Begriff des Erfinderischen, sic! 2004, S. 469 sqq., 472.