You appear to be using an older version of Internet Explorer. We suggest you upgrade your browser for the best viewing experience. Upgrade to a Modern Browser
Two recent decisions of an EPO Technical Board of Appeal emphasise the need for a proper technical context to be set out in claims to render an abstract idea patentable. The two appeals, by Ab Initio Technology LLC, related to inventions in the field of parallel processing, a field that seems on the face of it highly technical. However, the inventions were claimed as processes for converting computation graphs into a form better adapted for parallel processes but without specifying any steps of performing any processing, parallel or otherwise. At this level of generality the Board considered the inventions to be non-technical as merely mathematical methods. Claims to methods involving actual execution of the converted graphs were not sought, possibly because of lack of good basis in the applications as originally filed, but ought to have saved the day. This emphasises the need for detailed consideration of issues of technicality at the drafting stage.
T 1125/17 (Parallelizing computation graphs/AB INITIO) of 2.4.2019 and T 1337/17 (Sort value indicator/AB INITIO) of 2.4.2019 were heard by Technical Board of Appeal 3.5.06 on the same day, a common practice for related cases, in this instance a parent and divisional respectively. Although various requests in the two cases were refused for lack of clarity or the presence of added matter, the interesting parts of the decisions relate to inventive step of the 5th Auxiliary Request in T 1125/17 and obiter comments in T 1337/17.
In the 5th Auxiliary Request of T 1125/17 the mapping of an input graph to the parallel computation graph was based on metadata in the input graph, so the Board took the view that the mapping added no information and the potential advantages of parallelisation of the generated graph were intrinsic in the input graph. The Board commented that:
Thus, the mapping of one graph to another, which is equivalent to expressing the same computation, is, essentially, a mathematical operation on computation graphs, which does not contribute to the effect of parallelization. The transformation itself therefore does not contribute towards inventive step[.]
In its preliminary opinion in T 1337/17, the Board suggested “that no […] technical effect would be established without the claims specifying the execution on a parallel hardware” but did not formally decide on this point since all requests were refused for lack of clarity.
In both cases it seems conceivable that a claim could have been drafted to a method of parallel processing or to hardware implementing the results of the methods but for a lack of good basis in the parent application as originally filed. In this document there was no detailed description of an actual parallel processing system, simply a reference to an earlier US patent. Whilst such a reference is usually enough to support a sufficient disclosure, EPO case law rarely allows a referenced document to be mined for basis for amendments.
Thus the applicant was forced to rely on the potential for a technical effect if the outputs of the claimed methods were executed. The Board however took the view that the claimed advantage required execution on a specific platform, which was not explicit in the claim, hence the technical effect is not inherent in the program. The Board did however imply that in some cases it might be possible that the “required execution platform was implicit in the program claim” but otherwise the hardware platform necessary to achieve any advantage must be explicit.
The possibility of hardware being implicit in a program claim seems likely to be rare but is consistent with T 0163/85 (Colour television signal) of 14.3.1989 which held that the structure of a television receiver was implicit in a claim to a colour television signal.
The concerns of Board 3.5.06 in these cases to ensure that all features necessary to achieve a technical effect are explicit in a claim, are also consistent with Board 3.5.07 in T 0489/14 (Pedestrian simulation/CONNOR) of 22.2.2019 which has referred questions on simulation to the Enlarged Board (see our news item for more details).