6. Ex post facto analysis

Many decisions of the boards of appeal warn against an ex post facto approach when assessing inventive step (see also the EPC Guidelines G‑VII, 8 – April 2025 version). This applies especially to inventions which at first sight seem obvious, to combination inventions and where the proposed solution is supposedly "simple". Correct application of the problem-solution approach avoids this inadmissible ex post facto analysis which draws on knowledge of the invention (T 24/81, OJ 1983, 133; T 564/89; T 645/92; T 795/93).

When assessing inventive step, an interpretation of the prior art documents as influenced by the problem solved by the invention, where the problem was neither mentioned or even suggested in those documents, must be avoided, such an approach being merely the result of an a posteriori analysis (T 5/81, OJ 1982, 249; T 63/97, T 170/97, T 414/98, T 2622/19, T 2004/21).

In T 970/00 the board stated that any ex post facto analysis, and in particular any conclusion going beyond what the skilled person would have objectively inferred from the prior art, without the benefit of hindsight knowledge of the invention, is of necessity at variance with a proper application of the problem-solution approach. Any attempt to interpret the disclosure of the closest prior art so as to distort or misrepresent, based on hindsight knowledge of the invention, the proper technical teaching of the disclosure in such a way that it artificially meets specific features recited in the claim under consideration must fail, especially as this would risk unfairly and tendentiously concealing the technical contribution of the invention and prejudice the subsequent objective determination of the technical problem solved by the claimed invention (see also T 266/07, T 1486/10).

In T 2201/10, the board found that the examining division had taken an ex-post-facto approach to analysing the facts. Even assuming that the proposed solution could be derived from common general knowledge, it went against the very essence of the teaching in the closest prior art and, therefore, would not realistically have been chosen. The problem-solution approach required that documents not relating to the technical field of the invention be disregarded, and it also precluded any analysis according to which the skilled person would have modified the closest prior art in a manner contrary to its very purpose. In other words, a finding that a claimed invention deviated from what was, in light of the aim pursued, the essence of a prior-art disclosure was enough by itself to conclude that it was inventive over that prior art (see also T 2057/12).

In T 855/15 the board considered that for the assessment of obviousness a consideration as to what the skilled person would do, in particular whether the skilled person "would select" a particular document in order "to arrive at the invention as claimed" would amount to hindsight reasoning, because the skilled person would have to be assumed to know the invention before an argument could be made as to what they would do in order "to arrive at" it. The board in T 2057/12 agreed with the board in T 855/15 insofar as the closest prior art belonged to the same or a neighbouring technical field of the skilled person or to their common general knowledge. Argumentation or evidence should be provided as to why the skilled person in a specific technical field would have indeed envisaged selecting a document in a remote field of technology as closest prior art or whether they would indeed have considered adapting a prior-art disclosure originating from their technical field to implement it in a remote technical field.

In T 1087/15 the board pointed out that knowledge of the claimed invention was absolutely necessary in order to formulate the objective technical problem, irrespective of how 'close' the prior art document was to the claimed invention; the features differentiating the claimed subject-matter from the starting document must be identified, their technical effect determined and thus the problem to be solved formulated. In as far as such knowledge of the claimed invention was labelled as 'hindsight', this particular type of hindsight was indeed a necessity in order for the technical problem to be formulated on an objective basis. As long as there was no reason (primarily a technical reason) why the content of the document selected as the closest prior art would itself not be suitable or compatible for use with respect to the subject-matter claimed, any analysis of 'how close' the prior art starting document was judged to be to the claimed invention was not something which should exclude it from being the closest prior art.

In T 1349/19 the board found that the appellant's (opponent's) arguments involved a convoluted set of sequential steps conceived starting from the compositions defined in claim 1 and working backwards, in an attempt to bridge the considerable gap with the composition described in D19 (closes prior art). Since these steps were not suggested by the prior art, they could only be taken by exercising hindsight. See also T 1750/19 in which the board found the combination of incompatible prior art documents was only possible with hindsight, and T 2643/16 of 16 February 2023 date: 2023-02-16 in which a necessary two-step combination of prior art documents was found to be based on hindsight, with numerous other instances of hindsight identified with regard to alternative combinations of prior art documents.

In T 1126/19 the closest prior D1 disclosed a list of about 60 pharmaceutically acceptable salts of rucaparib that could be used in therapy, including camsylate, although this was not preferred. The invention was based on the finding that crystalline rucaparib camsylate was particularly suitable for the preparation of solid dosage forms. The board held that the starting point should be the whole list in D1 rather than the specific option of the camsylate salt. As explained in T 970/00, the disclosure of the closest prior art had to be considered on the basis of its technical information, without distorting or misrepresenting it by the knowledge of the invention. The isolation of one of the non-preferred options (camsylate) would distort the teaching of D1, putting an inappropriate weight on that option.

In T 2196/21 the board defined the objective technical problem to be "how to enable a reliable verification of the warning messages of D9 at the receiving device in a bandwidth-efficient way". It found that D9 taught away from the invention and thus in order to arrive at the claimed invention, the skilled person would have had to dismiss the explicit teaching of this document and make impermissible use of hindsight. Thus the subject-matter of claim 1 involved an inventive step starting from D9.

In T 1888/21 the board did not agree that in order to avoid hindsight an inventive-step analysis should necessarily start from the general teaching of a document as e.g. specified in the claims. This was case dependent. A preferred embodiment such as a specific example of a patent document which unambiguously described a combination of concrete and promising measures toward the achievement of the goals set out in the patent under examination constituted a realistic approach of the skilled person.