Description Amendments - Finally a Referral?

We have been waiting a long time for a Board of Appeal to finally refer questions to the Enlarged Board on the long-running controversy of whether, or to what extent, amendments to the description are required for European patents and patent applications. Given recent developments, we may now be getting near the point where this actually happens.

In brief, the issue of description amendments is, as far as the EPO is concerned, a settled matter. In their view it is necessary for the description to be in line with the claims to avoid any inconsistencies. According to the current version of the Guidelines for Examination, F-IV, 4.3, "The applicant must remove any inconsistencies by amending the description either by deleting the inconsistent embodiments or marking appropriately so that it is clear that they do not fall within the subject-matter for which protection is sought". In many cases, this leads to extra time and trouble being expended when attempting to get an application allowed. Many have also objected that the supposed justification for this requirement simply is not there in the EPC. Article 84, which is used by the EPO in support, simply requires the claims to be "clear and concise and be supported by the description". If the description provides support for the claimed invention, does it even matter if it also includes support for things that are not part of the claimed invention? There are also cases where it may not be clear where the line should be drawn between what is described in the application as a whole as embodiments and what is actually within the scope of the claimed invention. Forcing applicants, and their attorneys, to make a decision during prosecution that can affect how the invention may be interpreted post-grant (since courts invariably insist that the description is always used to interpret the claims) may be considered unfair, if not actually unjust. 

Those of us who have been keeping a close eye on this thought that we were going to get a referral last from the appeal case of T 56/21. In the end, however, the Board decided that a referral was not needed because in their view Article 84 did not provide any justification for amendments to the description at all. I wrote last year about this lengthy decision here. The EPO, however, has been taking no notice at all of the decision (which is, of course, only binding on the case in question) and maintains that their practice is justified. The Guidelines therefore remain unchanged. 

However, in another appeal case, T 697/22, there has been some movement on the matter. Following oral proceedings held in December 2024, the Board indicated that they were minded to refer questions to the Enlarged Board. A communication has recently issued with some proposed questions, which the Board has asked the parties involved to comment on before they make a decision on a referral. Unlike in T 56/21, the Board found that there was divergent case law on the subject of description amendments and a referral may be needed to resolve the question of whether these were in fact required. The proposed questions are the following:

1. If the claims of a European patent are amended during opposition proceedings or opposition-appeal proceedings, and the amendments introduce an inconsistency between the amended claims and the description of the patent, is it necessary to comply with the requirements of the EPC, to adapt the description to the amended claims such that the inconsistency is removed? 

2. If the first question is unconditionally answered in the affirmative, which requirement of the EPC necessitates such an adaptation? 

3. If the first question cannot be unconditionally answered in the affirmative, what are the conditions under which adaptation of the description is necessary to comply with the requirements of the EPC? 

4. Would the answer to these questions change if an inconsistency existed between the claims and the description of a European patent application?

The parties have been given until 26 May 2025 to provide comments, following which the Board will decide whether a referral is to be made, and if so what the final form of the questions should be. 

For now, all we can do is wait and see what happens. It is, however, interesting to note that the draft questions avoid mentioning Article 84 EPC at all. This is clearly deliberate, since question 2 asks whether there is any provision in the EPC that requires the description to be adapted to the claims. If T 56/21 is to be followed, there is in fact no such provision. The Enlarged Board, however, may find differently. 

Faketoshi Patent Oppositions - The Story So Far

For the past few years, in my spare time I have been working on oppositions against three European patents granted to nChain Licensing AG, each of which originated from GB priority applications filed in April 2016. Each patent named Craig Wright and fellow Australian Stephane Savanah as co-inventors. For those few people who may still be unaware, Wright has since 2015 been falsely and fraudulently claiming to be the person behind the pseudonym Satoshi Nakamoto, the creator of Bitcoin. Wright's implausible, and easily disprovable, claims were disputed in late 2015 almost as soon as they became public through engineered leaks to the press. It was, however, only after several lengthy, and extremely costly, legal battles involving many others (including myself), that his claims were finally and comprehensively demolished by Mr Justice Mellor in a mammoth judgment handed down on 20 May 2024 (Crypto Open Patent Alliance v Craig Steven Wright [2024] EWHC 1198). The judgment was appealed by Wright, but permission was denied by Lord Justice Arnold in, for him, an unusually brief 3 page judgment issued on 29 November 2024. There are further strands to the Craig Wright story, including a contempt of court finding, alleged tax fraud in both the UK and his native Australia and a likely upcoming criminal prosecution for perjury, but here is not the place to go into them.  

Going back to the beginning, the company nChain was set up in 2016 by Wright and his business partner Stefan Matthews, with help from Canadian businessman Robert MacGregor, and enabled by substantial financial backing from Antiguan-based Canadian online gambling tycoon Calvin Ayre. The stated business aim of the company was to patent and commercialise inventions arising from Wright's alleged extensive knowledge of Bitcoin and related technology by virtue of his being Satoshi. As explained in the 2016 article by Andrew O'Hagan, "The Satoshi Affair":

They would complete the work on his inventions and patent applications – he appeared to have hundreds of them – and the whole lot would be sold as the work of Satoshi Nakamoto, who would be unmasked as part of the project. Once packaged, Matthews and MacGregor planned to sell the intellectual property for upwards of a billion dollars. MacGregor later told me he was speaking to Google and Uber, as well as to a number of Swiss banks. ‘The plan was to package it all up and sell it,’ Matthews told me. ‘The plan was never to operate it.’

The infamous "1Feex..." paper wallet.

Even at that time, it was evident to anyone who took the time to look into the matter that Wright was clearly not Satoshi but was a rather inept forger with a particular talent for bare-faced lies. Some, however, apparently remained persuaded (or perhaps were willing to go along with the lies for their own reasons), including Mr Ayre who continued to fund Wright's Satoshi project on the promise of a big payout in the future, not only in respect of the supposedly valuable patentable Satoshi inventions but also on some very large stashes of Bitcoin that Wright claimed to own and which were apparently used as collateral.  

nChain GB filings 2016-2024

So, after having conveniently departed Australia in October 2015 following a raid by the Australian Tax Office, Wright began working as Chief Scientist for nChain in his new base in London to get his  supposed inventions patented. Wright was also incentivised by getting a substantial share in the company. The patent filings started flowing in to the UK IPO, the first one being filed on 23 February 2016 in the name of EITC Holdings Limited (later assigned to nChain). 76 applications in total were filed in 2016 and applications continued to be filed at a similar rate each year right up to the end of 2023, at which point they appear to have come to an abrupt and almost complete halt. 

Wright and friend (not a patent attorney)

Wright, of course, did not do this all by himself but was assisted by people who supposedly knew what they were doing with preparing and prosecuting patent applications. We would normally term these people patent attorneys but I am reluctant to do so in this case because, as it turns out, the main person involved in preparing and filing Wright's applications, at least in the early days, was not actually a patent attorney. She was instead a fairly recently graduated computer scientist PhD with a few years' experience working with patents but had not yet passed the UK examinations. Wright, however, got on very well with her and continued to employ her services right up until at least last year, even though she has still not passed the UK exams. Although the inventions themselves may not have been of the highest quality given their provenance, it certainly did not help that they were turned into patent applications that, at least to some extent, were somewhat handicapped by fairly poor drafting. As it turns out, one of the cases was also not helped during prosecution by some poor handling by others leading up to it being granted. 

I started looking into Wright's patenting activities in around 2020 and first wrote about them here in February 2021, noting that nChain had already built up a substantial patent portfolio with over 300 priority filings, which then resulted in other filings that were prosecuted more widely in other jurisdictions, including at the European Patent Office. At the time I wondered about their technical relevance, given that Wright's relevant knowledge was clearly not going to be important or relevant for any developments in Bitcoin or related technologies. Nevertheless, nChain was already demonstrating that whatever strategy was being used was starting to bear fruit, with patents being granted by the EPO. Many of these patents appeared to me, based on a limited review, to be examples of what I termed 'cargo cult patenting', as they appeared to contain overly technical-sounding, and sometimes lengthy, claims that were able to get through the examination process but which were highly unlikely to be infringed by anyone. It appeared, however, to be mainly a numbers game at that point, where the number of patents being granted was the main aim and not whether the patents covered anything actually useful. The reasoning may well have been that, if a sufficiently large patent portfolio could be built up, it would almost inevitably be considered worth a lot of money to someone and would then be bought up by a big player in the field who wanted to get a position in this up and coming blockchain technology area. It almost didn't matter what the patents covered, as long as they sounded sufficiently technical and their importance could be puffed up by Wright with his brash salesman bravado. After all, Theranos managed to do something similar, at least for a while. 

As a result of the technical obfuscation technique used to get many of their patents through the system, the prospects of finding any of the patents that might be a good enough target for opposing seemed to me unlikely. A particular group of applications, however, came to my attention that were an example of one of Wright's nonsense ideas, which was that Bitcoin could be made "Turing complete". How this was possible when the language used to run Bitcoin transaction scripts did not allow for loops was at the time unclear to anyone actually knowledgeable in the field, but Wright continued (and still continues) to insist that Bitcoin could be made Turing complete by using transactions as logic gates and doing loops off the blockchain. What the point of all this was is still not clear, but nChain's patent attorneys (real ones this time) managed to get 3 patent applications through the EPO's normally rigorous examination system, resulting in EP3449450 granted in June 2022, EP3449451 granted in July 2022 and EP3449452 granted in June 2022. I wrote about the first one of these patents, and the issues around Turing completeness, in November 2022, shortly after an opposition was filed by Arthur van Pelt with my assistance and with financial help from various generous Bitcoin supporters. 

Craig Wright's opinion of the first opposition. 

The first opposition, which was labelled "Faketoshi01" for ease of reference, was all about added matter. For the full details, see the post mentioned above. In short, the problem with the patent was that claim 1 had been amended during prosecution in a way that added matter (not allowed under Article 123(2) EPC). This was not noticed by either the patent attorney who did the amending or by another attorney who signed it off, nor was it noticed by the 3 member examining division at the EPO, who allowed the application. Once granted, the patent was fatally flawed because the only way to correct the added matter issue would have been to remove it. Unfortunately for the patentee, this was also not allowed (under Article 123(3) EPC) because this would inevitably broaden the scope of the patent. The patent was therefore stuck in an added matter trap that was impossible to get out of. Once the opposition was filed, it was inevitable that the only outcome could be revocation. Nonetheless, nChain's patent attorneys tried anyway and ended up making things worse. In the end, shortly before oral proceedings were due to be held to decide the matter, nChain switched patent attorneys and their new representatives clearly saw that the problem was insurmountable. Instead of fighting a battle they knew they could not win, they threw in the towel and asked for the patent to be revoked. The patent was then revoked in May 2024 without any fight. This decision was final and no appeal was filed. 

The second opposition (Faketoshi02) was filed in March 2023 against EP3449452. This time, the issues were not about added matter but about novelty and inventive step. Claim 1 of the patent defined a method that would cover a process carried out by a conventional Bitcoin node that was well known before the 2016 priority date. In the opposition, the primary source of prior art was the 2014 book "Mastering Bitcoin" by Andreas Antonopoulos. This is such an important and useful piece of prior art because it goes into technical detail of how Bitcoin was known to work at the time, including details of the various opcodes used in Bitcoin transactions. Although it had already been cited during prosecution by the EPO examiner, nChain's representative had somehow managed to convince the examiner that there was some kind of difference in what was claimed compared to what was disclosed in the book. As argued in the opposition, this turned out not to be correct.

This time the proprietor fought back. An initial, rather inept, response was first filed by the representatives who got the patent granted, which again ended up causing more trouble than it was worth because it left nothing on file that was even admissible, let alone allowable. As with the first case, the attorneys then got switched and the new attorneys got to work by filing 15 sets of amendments to the claims in response to a preliminary opinion that was very much against them. At oral proceedings held on 26 September 2024, there was quite a fight over the first 7 of these sets of amendments but at this point the representatives decided to give up and throw in the towel, presumably realising they had nothing left. The EPO opposition division then decided to revoke the patent. A written decision issued on 4 November 2024.

The third opposition (Faketoshi03) was filed in April 2023 against EP3449451, again based on novelty and inventive step grounds. The proprietor's representatives, which were again switched after a failed first attempt, filed even more sets of amendments. At the oral proceedings on 24 September 2024, the Opposition Division surprisingly came up with added matter as a new ground of opposition. The proprietor's attorneys, being justifiably surprised at this (as was I), got the proceedings rescheduled to give them time to think a bit more. At the rescheduled proceedings held on 10 December 2024, the new added matter ground was upheld and the patent was revoked after only about 30 minutes of argument. Unfortunately for the patentee in this case, the added matter problem was the same for all their numerous requests so, once the first request fell all the others followed for the same reason. The written decision revoking the patent then issued on 23 December 2024. 

A hit rate of 3 out of 3 is not bad going, but for two of these we are not yet at the end of the story. For the second and third cases, the patentee decided to file appeals against the revocation decisions. I was slightly surprised at this because they had taken the precaution of filing divisional applications on each of them, all of which are currently still pending at the EPO, meaning that they could have another go at getting a patent without the limitations of post-grant amendments. However, given that appeals have been filed there is still the possibility, however remote, that they could rescue something and perhaps save face. So far, grounds of appeal have been filed on Faketoshi02, which have already been responded to, and grounds on Faketoshi03 are due in the next few days (Update 29 April 2025: grounds have now been filed and responded to). So far, the arguments from the proprietor/appellant do not look very impressive but we will probably have to wait at least another couple of years before we find out what the Board of Appeal thinks of them, given the large backlog of cases they have to deal with. More excitement is therefore yet to come. I shall, of course, keep you posted. 

A problem with dates

Knowing when a divisional application (or, more correctly, a new application under section 15(9)) can be filed is a tricky business in the UK. I have written previously here about how applicants and their representatives can sometimes get caught out by assuming that the same rules apply to those at the EPO. Unlike at the EPO, where the general rule is that a divisional can be filed provided there is a current application pending, in the UK the latest date for filing a divisional is 3 months before the end of the compliance period, provided the application is still pending. This additional requirement can cause problems when applications are still being prosecuted close to the end of the compliance period, especially if it's not yet clear what will get allowed. 

According to Rule 30, the compliance period is 4 years and 6 months from the filing or priority date, or 12 months from the date of the first substantive examination report if this is later. The end of the period can be extended by 2 months under Rule 108 without providing any reasons, and may be extended further if evidence is provided. This means that, if the end of the compliance period is getting closer than 3 months, a divisional can still be filed if an extension request is filed in time. 

Let's say we are exactly one month away from the end of the compliance period, which for the sake of this example is 30 December 2024. Can we still file a divisional application? The current date is 30 November 2024, which is a Saturday. We have unfortunately left this rather late, as we knew a while back that a divisional application would be needed. As a busy patent attorney, however, we are used to dealing with things at the last minute and we are sure that all will be fine because we can just file a request for a 2 month extension at the same time as filing the divisional application and all will be in order. 

Unfortunately, things will not be in order. A further quirk of the UK patent system is that dates on which things are deemed to be filed differ depending on what they are and whether the Office is open. For the purposes of most things the Office is closed at the weekend, meaning that anything filed gets a deemed date of the next working day (see here). This applies to everything apart from new applications that do not claim priority, which get the filing date of the day on which they are actually filed, regardless of whether the Office is open. 

The upshot of all this is that the divisional application we filed on 30 November (which did not claim priority) gets that date of filing but the extension request we filed at the same time gets a deemed filing date of 2 December. The result is that the divisional was filed out of time because on the date of filing it was within only 1 month of the end of the compliance period. The only way to rescue this would be to get a further discretionary extension to the compliance period to allow another purported divisional application to be filed within time, but we would have to adequately explain what happened and why this would be justified. 

This all sounds very implausible and not something that should happen in the real world. Any competent patent attorney with a reliable docketing system will be warned in good time beforehand about the upcoming deadline for filing any divisional and will not resort to the risky business of filing things on a Saturday evening. This is not, however, a theoretical example but actually happened. The full details can be found in a recent decision from the UK IPO in BL O/0308/25. To prevent further embarrassment to the patent attorney involved, I will not go into any more detail. It is, however, highly recommended reading for all UK patent attorneys, and especially for those involved in marking papers for the patent attorney examinations.