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

The second part of this conference report on the 2017 Zurich IP Retreat - Patents and Hindsight summarizes the panels on the closest prior art and the definition of the problem (to be solved). Part I of the conference report is here.

Closest Prior Art

This panel, the third in the conference and the last on Friday, looked at the closest prior art and how hindsight might influence the choice of the starting point for the assessment of inventive step.

Dana Beldimann (Professor, Bucerius Law School, Hamburg, Germany, and UC Hastings College, San Francisco, USA) briefly introduced the framework for obviousness analysis under US law. For an invention to be patentable under 35 USC § 103, it must contribute more than obvious advances to the state of the art. There is no concept of “the” closest prior art in US law. The “analogous prior art” must be considered as a whole. Prior art is analogous if it belongs to the same field of endeavor, or if not, is “reasonably pertinent“ to the particular problem, so that it logically would have commended itself to the inventor's attention (Circuit Check Inc. v. QXQ Inc. (Fed. Cir. 2015)). The analogous prior art can comprise multiple references, except in chemical and pharmaceutical fields, where there must be a lead-compound.

Beldimann, Brändle, Wilming, Hoyng, Bausch, Blumer (3rd panel)

References may only be combined when there is a justification to do so. The justification lies in teaching, suggestion or motivation (TSM) to combine or modify references and must be found in the prior art. The TSM standard is the main safeguard against hindsight bias. KSR International Co. v. Teleflex (US Sup. Ct. 2007) held that the TSM test is a helpful insight into reasons to combine, but should not be treated as a “rigid mandatory formula”. In addition to TSM, common sense can also constitute a path to proving obviousness, since a skilled person may “be able to fit the teachings of multiple patents together like pieces of a puzzle”. KSR v. Teflex may lead to a larger influence of hindsight than the traditional TSM test. There is some pushback from the CAFC. In Circuit Check v. QXQ, the CAFC held: “An alleged infringer should not be able to transform all systems and methods within the common knowledge into analogous prior art simply by stating that anyone would have known of such a system or method. The question is not whether simple concepts such as rock carvings, engraved signage, or Prussian Blue dye are within the knowledge of lay people or even within the knowledge of a person of ordinary skill in the art. Rather, the question is whether an inventor would look to this particular art to solve the particular problem at hand.”

Dieter Brändle (President, Federal Patent Court, Switzerland) noted that patentees always complain about hindsight when the patent is found lacking inventive step. Since the court must rely on references introduced by the parties, the choice of prior art is not determined by the court. Only what the parties present to the court can be considered.

A patent must have inventive step irrespective of the starting point of the analysis. The law requires that the invention is inventive over the entire prior art. If a patent is found inventive starting from one prior art reference, it must be assessed whether it is also inventive starting from another starting point, if such starting point is alleged. Unless a reference can be excluded “right away” as a valid starting point, it must be accepted as starting point. The selection of the starting point is not subject to hindsight bias; the hindsight bias only comes into play in the third step of the PSA, when it is assessed whether the skilled person would have modified the starting point such as to arrive at the claimed invention.

Willem A. Hoyng (Attorney, Hoyng, Rokh, Monegier, Amsterdam, the Netherlands) reminded everybody that the PSA is not found in the European Patent Convention. It is a tool developed by the European Patent Office to deal with its massive case load. It may be useful, but it is artificial and may also lead to wrong results. The choice of “the” closest prior art is often arbitrary. Courts should not be obliged to follow the PSA, which in fact there are not according to UK, German and Dutch case law.

Art. 56 EPC requires consideration of the whole prior art. That whole prior art may give indications (pointers) towards the invention or away from the invention. Closest Prior Art and PSA may have a tendency towards obviousness if not applied correctly. So one should consider the art as a whole and how much incentive gives the art as a whole the skilled person to come to the invention and what is his/her expectation of success. Incentive and expectation of success should be communicative vessels when deciding the obviousness question.

Willem ended his presentation with the statement that he considered the way the EPO deals with Art. 54(3) EPC a big problem which leads to multiple patents for the same invention (and – via divisionals – unacceptable uncertainty of third parties). Why can EPO and Dutch courts not read “the content of European patent application” broader? The skilled person should always read (when reading for Art. 54(3) EPC purposes) with the common general knowledge and consider each combination with the common general knowledge as disclosed. This would avoid the grant of multiple divisionals for essentially the same invention.

What you get for writing the conference report - an elephant

Thorsten Bausch (Patent Attorney, Hoffman Eitle, Munich, Germany) reiterated that Germany does not have a concept of closest prior art, but the Bundesgerichtshof demands that the starting point for inventive step analysis must be justified (BGH, 16 December 2008 - X ZR 89/07 – “Olanzapine”). If, for example, a document discloses 200 compounds, it is impermissible to take one of the compounds and argue that its formulation as a sustained release formulation was obvious when there were no specific reasons ex ante for the skilled person to choose this compound among the many disclosed.

Torsten’s conclusion from the unavoidability of a certain hindsight bias is that judges should be twice as careful and cautious before revoking a patent for lack of inventive step. On the other hand hindsight bias may favour the patentee when it comes to questions of claim construction and the doctrine of equivalents, as the interpretation of the claim with knowledge of the infringing embodiment may lead to a broader interpretation, and equivalents may be found obvious once they were employed.

Fritz Blumer (Member of the Legal Board of Appeal, EPO) stated that the case law of the Board of Appeals of the EPO emphasized continuously that hindsight should be avoided in the choice of the closest prior art. “[…] in order to avoid ex-post facto considerations, the closest state of the art is not generally that merely showing superficially the most similarities, but rather that conceived for solving the same primary problem or aiming at the same objective as the claimed invention and which requires the minimum of structural and functional modifications.” (T 026/04). Fritz gave an example of an invention of a multilayer panel for aircraft interiors. D1 discloses a panel for aircraft interiors, but with a different layer sequence. D2 discloses a panel for furniture with a layer sequence like the claimed one. The “same purpose” criteria should lead to the choice of D1 as closest prior art. Choosing D2 is likely based on hindsight and may lead to difficulties in formulating the objective problem (“alternative use for furniture panels” is hardly satisfactory).

The prior art should not be read with the invention in mind (T 970/004, cons. 4.1.2). Similar to the point made by Peter Meier-Beck and Thorsten Bausch, Fritz cautioned that one should not “pick and choose” from a prior art reference unless there were good reasons why the skilled person would choose the specific disclosure over another.

During the ensuing discussion, Robin Jacob interjected that it should be up to the attacker of the validity of a patent to choose the starting point from which he or she wants to attack the patent. Graham Ashley prefers using the term “relevant starting point” rather than “closest prior art”. One can start from several references, but the attacker needs to justify why he wants to start from each one of them. Similarly, Rian Kalden added that there could be several starting points, but not 30. There must be a justification for a specific starting point. Katherine J. Strandburg pointed out that the PSA starting from a single closest prior art reference assumed a cumulative view of innovation, which was not necessarily correct. Innovation may also consist of thinking about a problem. Dieter Stauder made the sensible point that many granted patents were never practiced and only a small fraction ever litigated. The case load at the EPO required a standardized approach for examination that was simple and predictable. Courts, on the other hand, are not and should not be bound by it.

Objective Technical Problem

Fritz Blumer reminded everybody that the PSA was not cast in stone, although in practice it was ruling. Rule 42(1)(c) EPC also requires a description of “the advantageous effect of the invention with reference to the prior art” in the patent application. The PSA goes back to the very beginning of the operations of the EPO (see T 1/80 of 1981). The definition of the objective problem to be solved is a moving target. When the closest prior art changes – for example after the introduction of claim limitations – the problem to be solved also changes.

Dirk Szynka explains the prior art

Dirk Szynka (Patent Attorney, König, Szynka, Tilmann, von Renesse, Munich, Germany) opined that the problem the problem to be solved is particularly prone to hindsight bias because it does exist only with knowledge of the claimed invention, as it is derived from the effects of the differences between the subject matter of the claim and the closest prior art. The choice of the spring board document (on which the problem is based) is very important. The more sophisticated approaches (e.g., BGH, 5 October 2016, X ZR 78/14 – “Opto-Bauelement”) require looking at the technical effect of the springboard reference. However, everyday practice of the first instance and the examination proceedings tend to identify a spring board document based on a more arithmetical maximum structural identity which can lead to an objective patent problem quite independent from the patent description and the original technical approach therein and may often lead to hindsight.

If one starts from a document that is not related to the problem or effect of the invention but the argument clearly leads to obviousness, it’s hard to justify disregarding it. After all, the law requires consideration of all prior art.

From Dirk’s point of view, the main practical hindsight problem is, however, the tendency to be observed in the EPO but also in German proceedings, to formulate a “concrete” technical problem by comparing a spring board document and a claim. Such argumentations explicitly seek to avoid a too “abstract”, “artificial” or “empty” problem. In his opinion the recognition of disadvantages of the prior art is a first step of the invention (if not clearly induced by the prior art or the technical knowledge of the skilled person). This approach is also supported by the Board of Appeal case law, e.g. T 835/00, and, more recently, BGH,  13 January 2015, X ZR 41/13 – “Quetiapin” and BGH, 11. November 2014, X ZR 128/09 – “Repaglinid”. Accordingly, a “general and neutral” technical problem must be chosen which must not contain elements of the solution and, further, no elements found in the elaboration thereof.

Katherine Strandburg (Professor, NYU Law, New York, USA) added that US law knows no requirement to identify a problem to be solved. Identifying or framing a problem can itself be inventive. The problem solved by the invention may play a role in the definition of the analogous prior art, because it must be reasonably pertinent to the problem. KSR v. Teleflex repudiates a narrow approach to problem definition. The problem is not limited to the problem this particular patentee tried to solve. The problem can motivate the skilled person to look at other prior art that was not designed to solve that problem.

Whether hindsight leads to “wrong” decisions is ultimately not clear. Policy should decide which inventions deserve protection, and this should be informed by the outcomes – does the protection of these inventions lead to welfare gains?

Katherine wondered whether the “fundamental attribution error”[1], i.e., our tendency to explain someone's behaviour based on internal factors, such as personality or disposition, and to underestimate the influence that situational factors have, might help the patentee. In the case of obviousness assessments, the fundamental attribution error may lead to an attribution of the invention to the effort and creativity of the inventor, rather than situational factors (state of the field). It may therefore lead to a finding of non-obviousness and counteract hindsight bias.

During the discussion, Christoph Ann (Professor, Technical University of Munich, Germany) said we should hear more about biases other than hindsight if we want to reach unbiased judgments. Stefan Bechtold (Professor, ETH Zurich, Switzerland) pointed out that if hindsight bias had the same magnitude across the board, it was not a huge issue from a policy perspective. If, however, it had different effects in different cases, it was problematic. We simply do not know which one it is.

There was much more at this conference, but the format of a blog post prohibit an exhaustive summary. I therefore will not report the findings of the panels on expert testimony and technically qualified judges, claim construction and the broader perspective (yes, hindsight plays a role in copyright, too!). All in all a very enjoyable conference with contributions from a broad range of highly qualified speakers. And the dinner was excellent, too.

---

[1] Ross, The intuitive psychologist and his shortcomings: Distortions in the attribution process, in: Berkowitz (ed.), Advances in experimental social psychology (Vol. 10), New York 1977, 173-220.