As everyone knows by now, the Supreme Court handed down its judgment in the Emotional Perception case last week. There is plenty of commentary around on the judgment itself, so I won't go into the detail of it here. For background, I wrote about the case at the UK IPO stage in the September 2022 CIPA Journal, at the High Court here and at the Court of Appeal here. In brief, the invention is about the use of an artificial neural network (ANN) to compare two media files to determine how similar they are semantically, the main described application being for recommending a music track based on similarities with other music tracks. The application was refused by the UK IPO Hearing Officer in June 2022. The applicant then appealed to the High Court, which overturned the decision (and in my opinion got it completely wrong - see here for why). The Court of Appeal then reversed this finding (correctly in my view) and the Supreme Court has upheld in part the Court of Appeal's decision but decided that now is the time to abandon the Aerotel approach for determining patentability. What this is to be replaced with, however, is not yet clear.
 |
| A new hurdle and some steps. |
For the time being, we have to try to figure out what all this means in terms of the test we have to apply. To a first approximation, given that the main gist of the Supreme Court's decision was to make the UK approach more in line with that at the EPO, it would be reasonably safe to assume that following the EPO's problem-solution approach will more or less work, although some parts will need to be adjusted to align with UK law. Taking the relevant parts from the EPO approach, as set out in T 258/03 (Hitachi), T 641/00 (Comvik) and T 154/04 (Duns Licensing) and combining this with the Pozzoli four-step test for inventive step, a possible test could be something like the following.
1. Construe the claimed invention and determine its technical character. This is a necessary first step because it establishes whether the claimed invention gets over the Article 52(2) / Section 1(2) 'hurdle', as for example set out in Hitachi. This is a low bar, as the presence of any hardware (even pen and paper) will enable a claimed invention that would otherwise be excluded to get over the hurdle. The step is also necessary to establish what the nature of the technical character of the claimed invention is, as this will need to be looked at later.
2. Determine the closest prior art. This is step 1 of the Pozzoli test, and includes an assessment of the identity of the skilled person and their common general knowledge. It also corresponds to the first step of the problem-solution approach, although we are not going to use this to assess what technical problem is to be solved.
3. Determine the features of the claimed invention that contribute to the technical character. This is a requirement that was set out in Comvik, which applies to so-called 'mixed' inventions, i.e. those comprising a mixture of technical and non-technical features.
4. Identify the inventive concept. This is step 2 of the Pozzoli test, but with the difference that only those features that contribute to the technical character (step 3) can contribute to the inventive concept. This requirement is set out in part (F) of the summary in G 1/19 of Duns (point 30 of the reasons), in which features that do not contribute to the technical character of the invention are to be ignored for assessing both novelty and inventive step.
5. Identify what, if any, differences exist. This is step 3 of the Pozzoli test, but again adapted such that only those features identified as being part of the inventive concept can be taken into account. If there are no such features, the claimed invention is not patentable.
6. Would these differences be obvious to the skilled person? This is the final step of the Pozzoli test, in which any differences, if these have been identified at step 5, are assessed in light of what the skilled person knows and would find obvious. Even if an invention has been found to have features that do contribute to the technical character, they could still be found to be obvious in light of what the closest prior art is and what the skilled person is presumed to know.
The above is only a first pass at what I guess the new test will need to involve. The details and the particular order in which some of the steps are taken may vary (steps 1 and 2, for example, could probably be done the other way around) but I think all of these steps will be needed in some form to take into account everything required for an assessment of patentability that also includes novelty and inventive step.
Although all the Court of Appeal level decisions on patentability leading up to Aerotel are to be abandoned, some other features of existing UK case law may usefully be taken into account, for example the 'signposts' set out in AT&T/CVON (not mentioned in the Supreme Court decision), which could apply when considering step 3. Although these signposts were provided to help decisions on the final Aerotel step, i.e. whether the contribution is technical, they should also be useful for assessing whether any feature contributes to the technical character of the invention since similar criteria apply.
We will have to wait and see what the UK IPO come up with, but I think it will have to be something along these lines. If they do come up with something similar, in the particular case in question my educated guess is that the claimed invention will probably not get past step 3 or may fail at step 5. I think it very unlikely that it will get all the way to step 6. I also think that, whatever happens, the boundary of what is patentable in the UK will not get significantly extended as a result.
It seems to me that we're essentially back where we started with the first instance decision in Merrill Lynch [1988] RPC 1:
ReplyDelete"...on its proper construction the qualification in section 1(2) does require of the Patent Office an initial enquiry and assessment as to whether the inventive step resides in the contribution of that excluded matter alone - if only that contribution of the excluded matter is the inventive step, the invention is not patentable by virtue of section 1(2). I endorse, therefore, the view of the principal examiner that in the determination of whether or not an application relates to an excluded thing it is necessary to take into account whether the non-excluded features are already known or obvious."
Now, the EPO's thinking has obviously developed a little in the decades since, and we're now all comfortable with the fact that features that are on the face of it excluded can contribute, and features that are on the face of it non-excluded might not contribute. And the correct ground of rejection is not exclusion by virtue of s.1(2) but lack of inventive step under s.3.
As an aside, I think that deals with any conflict with the Genentech decision which said that Falconer J's reasoning was wrong, as you can therefore have a "discovery" (like an amino acid sequence) that is recognised for its interaction with the technical means in the claim (a method of synthesising the protein). To Jacob LJ's credit, this was noted in Aerotel but they felt bound by their own precedent.
As to the test, the Supreme Court gave a very strong hint as to how they think it should work, and I think the summary here is pretty much all you can now do. I think those wanting wholesale importation of problem-solution will be disappointed, especially given the direction of travel at the UPC.
I think it's important though to separate out the issues of inherent patentability under s.1, from the other requirements of ss.2-4. This ensures we're not getting bogged down in things like "contribution" "technical character" etc. when assessing whether any exclusions are engaged. Indeed the Supreme Court said that this is not appropriate (para. 59).
The UKIPO may feel more constrained to employ the exact 4 steps of Pozolli, given it has been approved in that form right up to Supreme Court level in Actavis. It may be enough to simply roll all of the steps on excluded and non-excluded features into guidance on application of Pozzoli step 2. Perhaps it could be called the "inventive concept with the rider" approach. I'm not sure there is a named test for novelty but it needs to also feature there, too, if we are to follow Duns (F) as the SC says we should.
I'm going to differ on the end result - I think it passes the any hardware test, and I think that the "fundamental insight" (pairwise comparisons of the property and semantic vectors of files to provide convergence of semantically similar files in property space during the ANN training stage) interacts with the technical features and is not disclosed in the prior art. I'd argue it's the central inventive concept (forget all the other fluff after step (a) of claim 4, it reminds me of the claim in PKTWO - the applicant may do well to file a divisional with much more focussed claims directed to the method of training and rely on product by process protection). We'll have to wait for a top up search but, based on the prior art of record, I'd be inclined to think that this was not obvious at the priority date.
Excellent points, thank you. I agree that the invention in this case passes step 1 but I wonder about the alleged novel features contributing to the technical character of the invention, which is what made me think of the AT&T/CVON signposts. For example, do the differences make the ANN a better and more efficient ANN? The end result of providing a recommendation is not part of the technical character because this is merely cognitive content (which, according to EPO case law, is not technical), so the difference have to make the computer in some way a better computer. I have not looked into the details of the invention enough to properly figure that out.
DeleteThank you for this very useful analysis - I had been trying myself to think through what the Monster might look like. This is helpful.
ReplyDeleteI'm not sure that I yet agree about the “AT&T signposts” though.
The signposts don’t really make sense anymore, do they? Even if sense can be made of them, is there authority for using them as you suggest?
A few other thoughts/questions - what is your view on the impact this will have prevalence of refusals to search at the UKIPO? I am cautiously optimistic that we should see fewer of these. How is the office to justify a refusal to search if they need to assess the patentability of “mixed” features?
Finally – in my experience the ability to argue on the basis of the interaction between technical and non-technical features is the thing which makes the biggest difference in outcomes at the EPO as compared to the UK office. If those arguments are available to us now at the UKIPO, won't that make things significantly more "EPO-like"?
You may be correct about the signposts. I thought it might be a bit of a stretch to include them, but I think they are nevertheless still useful.
DeleteOn the question of whether this will result in the IPO doing more searches, I think this is one of the current bugbears with UK attorneys and it should get reduced because determining the closest prior art (or whatever it is called under the Pozzoli test) should always be needed, unless UK IPO examiners can refer to 'notorious' prior art like general purpose computers.
On the final point, I agree that it might make a difference being able to argue for 'non-technical' features making a contribution to the technical character, but I think that should already be the case.