Over to Peter:
"In a GuestPost three months ago I shared some thoughts on the recent Court of Appeal decision in Comptroller - General of Patents, Designs and Trade Marks v Emotional Perception AI Limited [2024] EWCA Civ 825. At the end of that article, we shared a poll asking whether it is time for an overhaul of the exclusions to patent protection. The results are now in, and they reveal that 59% of participating readers (105) are in favour of a rehauling exclusions to patent protection.
The list of non-inventions in Art. 52(2) EPC is now more than 50 years old. The list comprises a disparate group of topics with programs for computers lumped in with discoveries, mathematical methods, mental acts and aesthetic creations. As Peter Prescott QC observed in CFPH LLC v Comptroller-General of Patents, Designs and Trade Marks [2005] EWHC 1589 (Pat), it is miscellaneous rag-bag that cannot be provided in a logical group.
The legislative intent behind the list of non-inventions is difficult to follow. In Aerotel Ltd v Telco Holding Ltd and others, and Neal William Macrossan's application [2006] EWCA Civ 1371, the Court of Appeal noted that the travaux preparatoires to the EPC do not help much in understanding any overarching reasons behind the exclusions, nor how they should be interpreted. The best explanation for the computer program exclusion is probably that they were generally regarded as analogous to literary works which would benefit from protection by copyright. There was also a sense (acknowledged in CFPH) that searching the prior art would be problematic, given that binary code was not intelligible in a manual search. Whatever the intent, the 1970s legislators could not have anticipated how computer programs would have subsequently evolved and become an enduring feature in all aspects of technology. Today, the Courts and stakeholders are left with an exclusion whose original rationale cannot be entirely explained. We are left to try and make sense of it, and to find an interpretation that deals with developing technologies like machine learning and quantum computing.
The majority view is that it is now time for an overhaul and for a patent system that more fairly rewards investment in innovation, regardless of whether an invention involves an arbitrary technical effect. For those in favour of the status quo, I would ask – for how long should these exclusions be maintained without scrutiny? Another 50 years? All things must change eventually, and now is the time to grasp this nettle."
What sayeth dear legislators?
ReplyDeleteWhat say the legislators? They’ll look at past attempts to tweak this aspect of patentability and run a mile.
The ambiguity in the law is the whole point - it is helpful to allow the case law to adapt what is 'technical', rather than by restricting it to declarative statements as to what is and is not patentable subject matter.
ReplyDeleteThe only overhaul suggested in the post actually boils down to accept programs for computers as patentable inventions.
ReplyDeleteIt has first to be reminded that the Diplomatic Revision Conference of 2000 decided that the patentability of computer programs would be left for a second basket to be dealt with by a future Diplomatic Revision Conference.
I would say that a new Diplomatic Revision Conference is a certain as a Conference of Ministers of the Contracting States according to Art 4a EPC2000. A conference according to Art 4a should by now have been convened twice.
It is only by a revision of the EPC that computer programs as such could become patentable.
Originally, the reason for not allowing computer programs as such as patentable inventions, was indeed to be seen in the impossibility to carry out a significant search. There are lots of computer program languages and it is not possible for an examiner to be aware of all of them and of their different syntax. In the 80ies and early 90ies, the EPO even allowed short excerpts of programs, but the nub of the invention had to be disclosed, not in the program, but in plain language in the description.
The case law developed by the boards of appeal of the EPO, going from Vicom to Comvik, just to name a few, has long accepted Computer Implemented Inventions.
As a patent is prima facie a technical solution to a technical problem, inventions in which a device is controlled by a program or when a method is implemented with the help of a program are more and more usual. It is just necessary to provide a connection with the real world or a specific technical effect inside a data processing machine. In order to be patentable, an invention need to involves a technical effect achieved with technical means. The requirement of technicality is thus anything but arbitrary.
Machine learning and quantum computing are perfectly patentable, provided there imply a technical effect on the real world. As far as machine learning is concerned, the problem lies more in sufficiency. In order to obtain a patent, it is not enough just to disclose the correlation rules and the weighing factors of a neural network, but as the same time the learning data has to be disclosed. As data are in our modern world a precious asset, I cannot see the holder of data happily disclosing learning data.
In view of the above considerations, the need for a revision of Art 52(2) in respect of computer programs as such appear simply unnecessary.