InterDigital, Inc & Ors v Optis Cellular Technology LLC & Ors

This appeal relates to the proper scope of redactions of a non-party’s confidential information in a FRAND judgment. The case is about the judgment of Marcus Smith J ([2023] EWHC 1095 (Ch)) dated 10 May 2023. That decision resolved a dispute between Optis and Apple concerning the terms of a Fair Reasonable and Non-Discriminatory (FRAND) licence from Optis to Apple of a portfolio of patents held by Optis which are or are said to be essential to certain mobile telecommunications standards (so called Standards Essential Patents or SEPs). Those findings were appealed to this court (Newey, Arnold and Birss LJJ) and are addressed in the public judgment on the main appeal at [2025] EWCA Civ 552 dated 1^st May 2025.

In the proceedings disclosure was given of various patent licences between Optis or Apple on the one hand and other companies on the other. These other companies were not parties to these proceedings. They have been referred to in this case as third parties and I will refer to them as such, although they were not additional parties within the terms of Part 20 of the Civil Procedure Rules. They are parties to these appeals but that is all.

The third party licences were relevant and some of the terms are referred to in the full version of the judgment, which is confidential. That full version was sent in draft to the parties in March 2023 under embargo and the final form was handed down in private to the parties on 10 May 2023. There was a further full version on 17 May 2023 but nothing turns on that. A non-confidential version of the full judgment was handed down in public on 7 June 2023. All potentially confidential material had been redacted, erring on the side of caution. It took time to address the proper scope of the final redactions, which included submissions and evidence from some of the third parties. The issues were resolved in a judgment dated 14 February 2024 ([2024] EWHC 197 (Ch)) which held that a “less-redacted” version of the public judgment should be published. As its name suggests, the less-redacted version still redacted some information in the full judgment but the redactions were less extensive than in the public version handed down in June 2023. The form of that less-redacted judgment was settled and it was annexed to the order dated 17 June 2024. The publication of the less-redacted judgment was stayed pending any appeal, if permission was granted. Neither party to the proceedings – i.e. neither Optis nor Apple – sought permission to appeal at this stage. They both maintained a neutral stance. A number of third parties did seek permission to appeal. Other third parties, including at least one who had made submissions, did not. Permission to appeal was given by Falk LJ on 27 September 2024.

The main appeal having been heard and decided earlier this year, this confidentiality appeal came on to be heard on 9 July 2025. During the hearing it became clear that to deal with this matter properly and fairly, the third party appellants needed to amend their grounds of appeal. In addition Apple served an appellant’s notice, essentially to address the position of information relating directly to other third parties who were counterparties to licences with Apple but had not appealed. The reason there is no corresponding appellant’s notice from Optis is simply because the only counterparty to an Optis licence with relevant confidential terms is Google, who had already appealed. We gave permission for these amendments and for the fresh appellant’s notice to be filed out of time.

The jurisdiction to settle the terms of a FRAND licence in a SEP dispute is well established (see Unwired Planet v Huawei [2020] UKSC 37 in the Supreme Court followed in the Court of Appeal in InterDigital v Lenovo [2024] EWCA Civ 743 and in the May 2025 judgment in this case (above)). The main question to be settled is the royalty rate to be paid by the licensee to the licensor. This may be a single rate for the world for all standards or it may involve a mix of rates. One important source of evidence relevant to the rate is evidence of the terms of other licences which have been entered into by one or other of the parties. Disclosure of some of them will be given. Each side will select the licences it contends are the best comparable licences. Evidence will be given about these licences including, if need be, accountancy evidence which “unpacks” the licence terms to make the comparison more reliable. This might involve converting a licence expressed as a single lump sum payment into a sum expressed as a percentage of sale price per unit (a so called ad valorem royalty) or as a fixed sum per unit (DPU or dollar per unit). Evidence about the comparability of the licences, which may involve the situation of the parties, the other licence terms and the degree of hold out or hold up involved, may also be given.

The court’s conclusion about the rate is likely to be based on a so called “bottom up” approach starting from what the judge decides is the best comparable licence (or licences). The judgment may also involve a cross-check based on a “top down” approach starting from a figure to be paid as an overall licence for a notional stack of all relevant patents essential to a given standard and taking a view about what the licensor’s share of that stack might be. For the purpose of this appeal these concepts do not require further elaboration. More details in the context of this case are in the judgment in the main May 2025 appeal (above) - see the introductory paragraphs at [1] to [21] and also [34]-[42] and [90]-[92].

These proceedings started as a series of technical trials addressing the validity and scope of the Optis patents in issue. At least some Optis UK patents were found to be valid and essential (including on appeal) and so by the relevant time both sides wished to enter into a licence but could not agree on its terms. Disclosure included a large number of licences. By the time of the trial Optis had identified 9 licences in which it was a licensor which it contended were relevant. In 7 of these licences the royalty was or included a sum to be paid per unit expressed as a percentage of the sale price. In the other licences the licensee’s royalty was or included a lump sum. Google was the licensee in the only Optis lump sum licence. Also by the trial Apple had identified 19 licences in which it was a licensee, 14 of which it contended were relevant. In all 19 of these licences the royalty involved a lump sum to be paid by the licensee (Apple). Many of the Apple licences were in fact cross-licences and that needed to be taken into account, but for present purposes one can treat them as licences in which Apple is the relevant licensee. As part of the management of the proceedings and as in the previous FRAND cases, an extensive confidentiality regime was established. The result was that only representatives of the two parties – and the court – were able to see all the information.

The full judgment necessarily grappled with the various licences and, as I say, it included information derived from them. The judge’s approach involved rejecting both sides’ accountancy evidence and applying his own methodology to grapple with the material. That was the subject of the main appeal but subject to one point about unpacking, those issues do not matter in this appeal.

The full judgment contains a number of tables in which the terms of the various licences are addressed and summarised. Starting at [230] the judgment refers to a summary of the terms of each of the 28 licences in issue which forms a table at Annex 3 to the full judgment. The four columns include information such as the date of the licence and a description of its main provisions, including the financial terms. Where the licences are or involve lump sums, these figures are set out. Some percentage rates are included but many are not. The Annex 3 table also includes an entry for my judgment in Unwired Planet [2017] EWHC 2988 (Pat) because that was being relied on as part of Optis’s case.

Pausing here, in the public version of the judgment published in June 2023 everything in Annex 3 was redacted save the column headings and the reference to Unwired Planet. It is now common ground that almost all of Annex 3 does not need to be redacted. In the less-redacted form of the judgment, for the reasons given in the February 2024 judgment, the only aspects of Annex 3 which would remain redacted are the various lump sum figures. There is no appeal against that conclusion. The appeal relates to some text in Annex 3 which describes an aspect of the terms of the licences, and two figures relating to the Qualcomm licence, a per unit percentage and a DPU.

Returning to the full judgment, the next important aspect is at [465] which introduces Table 9. In this table the judgment lists all 28 of the licences and provides three columns of numbers. One gives the percentage share of the patent stack held by the relevant licensor. The other two columns set out percentage royalty figures. For certain Optis licences which were set at Optis’s announced licensing rates, the information is not confidential (the Doro 2018, Blu 2018, Gigaset 2021, Cyrus 2021, Doro 2021, and Blu 2021 licences). For all the other licences, which were or included lump sum arrangements, the figures in the table are those produced by the expert Mr Bezant by the process of unpacking. In two further cases, which do not now matter, no percentages were given in Table 9.

In the public June 2024 version of the full judgment, as with Annex 3, the headings of Table 9 are included but every row is redacted. It is now common ground that for each row in Table 9 the stack share does not need to be redacted (nor the party names and date). In the less-redacted form of the judgment, the two percentage rates would be revealed for each licence. Maintaining the redactions (save for the ones with the Optis announced licensing rates) is one of the purposes of the appeals.

Next, at [466] of the full judgment, in Table 10 the judge reworked Table 9 with a view to showing what each of the percentage rates implied if those rates were pro-rated to the same stack share. In terms of redactions, subject to what follows, Table 10 is the same as Table 9. So in the public June 2023 version, every row was redacted, and the appeal relates to the two percentage rates for each licence. The less redacted form of the judgment would make them public and the appellants maintain they ought not be.

However Table 10 differs from Table 9 in that it has additional rows in which the data in the table are aggregated in various ways. These include total rates overall and totals broken down for Apple licences and Optis licences; as well as average rates, again averages for all and averages broken down for Apple and for Optis licences. One of the problems which only emerged with clarity on this appeal was as follows. Each appellant (or at least most of them) had only sought to appeal the decision not to redact specific information about the licence to which it was a party. However even if a given appellant’s case was accepted in relation to the figures in the body of the table, what emerged was that this case would in all probability be undermined by a combination of the absence of appeals by some non-parties and what would happen on publication of the averages and totals. This is why the applications were made by the appellants to amend the grounds of appeal to cover this issue and by Apple to serve an appellant’s notice out of time. Although I must say it seemed to me at the hearing that this was a problem which must have been obvious for a long time, the submissions satisfied me that the complexity of the issues and the manner in which confidentiality was dealt with between June 2023 and the order of 17 June 2024, and the course of this appeal, led to a state of affairs in which the issue only emerged at the hearing. As a result it was in the interests of justice to give permission to take the steps identified. The totals and averages are within the scope of the amended grounds of appeal and the further appellant’s notice.

Returning to the full judgment, having examined the percentage rates the judge decided not to use them to reach a conclusion but rather to use an approach based on lump sums. This led to Table 11 at [482]. This table addresses the Google licence with Optis and all 19 Apple licences. All 20 licences are based on lump sums. As I explained in a bit more detail in the main appeal judgment at [62] to [66], essentially for each licence Table 11 sets out the actual lump sum and the term of the licence. From this an annualised annual payment is calculated and then, using the stack shares (and catering for cross-licensing if relevant), an implied annual value for the whole stack is calculated expressed as a lump sum. In the less-redacted judgment the actual lump sum and the various calculated lump sum figures would remain redacted while the stack shares and the term of each licence would be revealed. Nothing in Table 11 is in issue on this appeal.

Table 12 reworks the results in Table 11 and adds nothing on this appeal. The next important aspect is Table 13, introduced at [483](ii). The first relevant column of Table 13 shows the implied lump sums of Table 11 and then the further columns show further lump sum figures produced as a result of various adjustments. As for Table 11, the less-redacted judgment would maintain the confidentiality of all the lump sum figures in the body of Table 13. However Table 13 also contains two final rows of averages and totals which would be published in the less-redacted judgment and which are now subject to the amended scope of these appeals and the further appellant’s notice.

This summary explains the main aspects of the full judgment from the point of view of this appeal. The full judgment also includes various individual figures which are subject to the appeal. They either come directly from the tables or are derived from those figures and so they stand or fall with the corresponding data in the tables. For example the figures in [467] to [470] and [485] are based on average figures in Table 10. Also [470](iii)(b) includes a figure based on Table 10.

After the public June 2023 judgment was published the third parties received partially redacted copies of the full judgment in order to allow them to make representations about confidentiality. They attended and were represented at a hearing in July 2023. At that stage the judge rightly made the point that any further version of the judgment would not be handed down without them being heard. However as things developed the third parties were not given that opportunity at the hearings in December 2023 and February 2024 which led up to the February 2024 judgment. They ought to have been given that opportunity and that was part of what precipitated the late amendments to the grounds and the further appellant’s notice.

The relevant part of the February 2024 judgment is Section B from [10] to [57]. In this section at [22] the redactions are classified into 6 classes. The classes now relevant are classes 1, 3, 4 and 5, as follows:

Class 1 (slips) is addressed at [22](i) and [24] to [25]. It is convenient to address this aspect of the appeal after other issues.

Class 3 (lump sums) is addressed from [30] to [44]. Here the judgment draws a distinction between the approach described as the old test ([30]-[35]) and the new test ([36]-[44]). The old test is based on my judgment on the confidentiality redactions in a FRAND case in Unwired Planet v Huawei [2017] EWHC 3083 (Pat) and was followed and applied by Mellor J in June 2023 in InterDigital v Lenovo [2023] EWHC 1577 (Pat). The new testis derived from the judgment of Arnold LJ (with whom Sir Andrew McFarlane P and Elisabeth Laing LJ agreed) in the Court of Appeal in July 2023 in JC Bamford Excavators v Manitou [2023] EWCA Civ 840, based on the Trade Secrets (Enforcement, etc) Regulations 2018 (SI 2018/597).

The judgment characterises the old test as involving a balance of factors to decide what if any information should be redacted, and the conclusion is that if the old test applied then the lump sum figures in class 3 would not be redacted, nor would any of the other information. However, the judgment goes on to hold that this old test has been replaced by the new test, which works in a different way. By the new test if the information falls within the definition of a “trade secret” as defined by the 2018 Regulations then the court is required to redact that information from a judgment without any consideration of a balance. Applying that new test, the class 3 lump sum figures qualify as trade secrets within the 2018 Regulations, and must therefore be redacted without any further consideration.

Class 4 (other terms) is addressed at [45] to [51]. The conclusions are that this information does not qualify as a trade secret and so does not fall within the scope of the new test, requiring redaction. Nor would its publication undermine the protection of the lump sum figures in class 3. Applying the old test, the information was not confidential ([47]). On that basis most of the text descriptions in Annex 3, which describe other terms of the agreements, would be revealed on publication of the less-redacted judgment.

Class 5 (derived data) is addressed at [52] to [53]. The approach taken here is to examine whether publication of the derived data would undermine the class 3 redactions. If it would then it should be redacted, otherwise not. Applying this approach the material conclusions are (a) that the totals and averages in Table 13 can be disclosed ([53](i)) and (b) that the unpacked percentage values in Table 9 do not need to be redacted. The latter conclusion is because the risk that the (redacted) lump sums could be derived accurately by a process of reversing the unpacking exercise (so called re-packing) starting from the percentage values was fanciful ([53](ii)), and inaccurate lump sums produced by re-packing would not be confidential ([53](ii)(a)).

Standing back from the detail I have just described, an important aspect of the February 2024 decision was the distinction between the treatment of two different kinds of financial information: lump sums and unpacked per unit rates. Confidentiality in the third parties’ lump sum data in the full judgment was maintained while publication of the unpacked percentage and DPU data (both of which are kinds of per unit rate) was to take place. The distinction applied whether those lump sums were the actual figures read off the documents or were derived lump sum data produced in the proceedings. The soundness of this approach is the main practical issue on this appeal and it is convenient to deal with it now.

In my judgment that distinction is flawed for a number of reasons. Part of this debate turns on the risk that the unpacked per unit rates could be reverse engineered to derive the lump sums. The judge did not accept that repacking to achieve an accurate view of the lump sums was a true practical likelihood. Key to this conclusion was the judge’s view that unpacking was “a subjective and unreliable process” (see the February 2024 decision at [53] as a whole; the quotation is from [53](ii)(a)). That passage in the February 2024 decision refers back to [301] of the main judgment which expresses this conclusion. In the main judgment this was a fundamental point. It had been common ground between the two accountancy experts that the way to evaluate the various licences was to adopt some kind of unpacking and evaluate the licences in per unit terms (main appeal [115]). The judge’s view in [301] of the main judgment was the basis for the rejection of all this accountancy evidence, which was in turn the evidence which had produced the unpacked per unit rates now under consideration. It was also the reason why in the main judgment the judge’s own method to arrive at a FRAND rate did not use unpacked per unit data. However Optis’s appeal against the rejection of the expert’s unpacking was upheld in the main appeal ([40]-[41] and [86]-[96]). The unpacking exercise produced useful data and indeed the FRAND rate reached in the Court of Appeal’s judgment was based on the unpacked per unit data ([129]-[146]).

There is a point on the accuracy of repacking or, as the judge put it, the risk of being wrong. There are three different reasons why it does not matter whether another company looking at the unpacked per unit rates would be able to back calculate the lump sum for a given licence with pinpoint accuracy or certainty that this was the “right” answer. First a rough estimate of the lump sum is all that is needed and would do just as much harm to the negotiating position of the third party whose licence the figure derives from. The fact that the person doing the unpacking cannot check the result does not matter. The second is that in fact, as the judge acknowledged at [53](d), there is information available (albeit for a price) which would be useful in that exercise. It was used in the worked example given to the judge. The third is that, as Nokia point out on this appeal, even inaccurate data puts them in difficulty because the only way to rebut a contention by someone who has carried out the repacking is to reveal the relevant confidential information.

A different point, emphasised by the third parties on this appeal, is that the unpacked rates themselves will be relied on by other would be licensees negotiating with the third parties for a licence even if they could not reliably be reverse engineered to the lump sums. These figures are derived from the true rates and do themselves represent the view of the professional experts called at trial as to the value of these licences. That data has just the same kind of value as the lump sum figures themselves would have, if they were published. In other words the status of the unpacked rates ought not to have depended solely on reverse engineering.

Finally there is a distinct point about two further per unit rates which appear in Annex 3 relating to Qualcomm and were not the result of unpacking. They are not mentioned specifically in the February 2024 judgment and appear to have been treated in the same way as the unpacked per unit rates, whereas these are rates from the licence itself. There is no rational distinction between the status of those rates and the lump sum in that licence.

Therefore in my judgment there is no sound basis for drawing a distinction between the lump sum figures and the per unit figures here. Either they ought both to be redacted or neither should be, whereas the judge’s approach rejected redaction of both sets of data on the old test and only allowed redaction of the lump sums but not the per unit rates, on the new test.

Looking ahead, in summary I have reached the conclusion that there is no new test and that the right approach is in effect the old test, but that when that test is applied to these figures, then they all ought to be redacted, essentially for the same reasons that the same kind of information was redacted in Unwired Planet and InterDigital v Lenovo. However to explain my reasons for this will require some detail to be gone through, and that is the next task.

As amended, InterDigital’s appellant’s notice raised 7 grounds of appeal. InterDigital Ground 1 submitted that the right approach in law was the old test. It remained the right approach despite JC Bamford v Manitou and if that approach had been applied then the various items InterDigital contended should be redacted, would have been. InterDigital Ground 2 submitted that the judge wrongly applied the test in JC Bamford v Manitou and if he had applied that approach correctly then again, it would have led to the redactions sought by InterDigital. InterDigital Grounds 3 and 4 related to the redactions in class 1, contending that the legal test identified in the judgment was wrong in law, that on the right test the information would be redacted, and alternatively even if the legal test identified was the right approach, it was misapplied. InterDigital Ground 5 challenged the conclusion on reverse engineering as a reason for not redacting the unpacked per unit rates, which I have already dealt with. InterDigital Ground 6 (by amendment) contends that the judge ought to have redacted all of the figures for unpacked rates in Tables 9 and 10 (save for certain Optis licences) as well as the figures for totals and averages in Table 10, because failure to do so would undermine the confidentiality of InterDigital’s unpacked rates in the tables. InterDigital Ground 7 (by amendment) makes a further (confidential) point relating to the same redaction the subject of Grounds 3 and 4 and is best dealt with in that context.

Qualcomm advanced two grounds of appeal, in which Qualcomm Ground 1 broadly corresponds to InterDigital Ground 1 (that the right approach, still applicable, was the old test) and in which Qualcomm Ground 2 broadly corresponds to InterDigital Ground 2 (the judge erred in applying JC Bamford v Manitou since there is no valid distinction between lump sums and per unit rates). By a letter from their solicitors dated 11 July 2025 Qualcomm also explicitly supported the grounds advanced in Apple’s appellant’s notice of 10 July 2025.

Nokia’s Grounds 1 and 2 are the same as InterDigital Grounds 1 and 2. Nokia Ground 3 is the same as InterDigital Ground 5 (that the issue ought not to turn on the ease of reverse engineering). Nokia Ground 4 adds a further point, that the appearance of the unpacked rates in a judgment will mean that other people will treat the rates as reliable, and Nokia could not easily rebut that without revealing confidential information.

Google, Huawei and LG Electronics (together “GHL”) were represented by the same legal team for the purpose of this appeal and filed a single appellant’s notice. As amended GHL advanced 10 grounds. Grounds 1 to 6 relate to the same issue as InterDigital Ground 5 and Nokia Grounds 3 and 4. Grounds 7 and 8 (as amended) relate to the impact of other data in the tables along with the averages and totals. A point is also made about the order of the listing in Table 13 and the possible derivation of information from that. Grounds 9 and 10 (renumbered by amendment) relate to the similar issues which are the subject of InterDigital Grounds 3 and 4 and are best dealt with in that context.

In terms of grounds, Apple’s appellant’s notice seeks permission to advance the submissions which were in fact set out in a joint note by counsel for Apple and Optis filed with the court a few days before the hearing and which first raised the issue about the impact on the appellants of other data in the tables along with the averages and totals.

In oral submissions the third party appellants placed the emphasis slightly differently. Mr Campbell KC, who made the main submissions, started with the submission that a simple way to resolve this appeal was to apply JC Bamford v Manitou, i.e. the new test, and that while the judge did so and rightly redacted the lump sums, the correct application of JC Bamford ought also to have led to redaction of all the various items of information in issue on the appeal (derived lump sums, per unit figures, totals and averages, and certain other licence terms). Taking that approach the distinction between the new and the old test did not matter. However, in addition and in the alternative, the correct application of the old test (i.e. Unwired Planet v Huawei and InterDigital v Lenovo on confidentiality) should have led to the redaction not only of the lump sums (in class 3) but also all the various items of information in issue on the appeal. Finally, on the distinction between the old and new tests, the approach taken in Unwired Planet v Huawei and InterDigital v Lenovo on confidentiality was correct and either had not been affected by JC Bamford v Manitou or, to the extent it had been, JC Bamford increases rather than reduces the protection given to confidential information. In any event, applying the correct test both the lump sum and per unit figures would be redacted, as well as the other terms.

There was a question at one stage whether the third party appellants had standing to appeal at all. They plainly do, see e.g. George Wimpey UK v Tewkesbury BC [2008] EWCA Civ 12 in which Dyson LJ (with whom Lloyd LJ agreed) held that under the CPR an appellant did not have to have been a party to the proceedings below (at [19]) which was followed in In Re W (A Child) (Care Proceedings: Non Party Appeal) [2016] EWCA Civ 1140 (see [41]).

The correct approach

Transparency and open justice are crucial in a democratic society and so justice is almost always done in public. Nevertheless there are exceptions to this which arise when the yet more fundamental principle, that courts work in the interests of justice, applies to displace the usual requirement for publicity. This explanation of how exceptions to open justice arise from the interests of justice itself was famously provided by Lord Haldane in Scott v Scott [1913] AC 417 at pp437-438. One example of its application, given by Lord Haldane in that passage, was litigation about a secret process. The effect of publicity would be to destroy the subject matter of the dispute and so it can be kept confidential. As Arnold LJ noted in JC Bamford at [77], three of the other judges in Scott v Scott made the very same point and in doing so described the information in question as a “trade secret”. For reasons which will emerge below I will use the term technical trade secret to refer to this kind of confidential information concerning technological secrets such as a secret formula or secret manufacturing process.

More recently in R (Mohamed) v Secretary of State for Foreign and Commonwealth Affairs (No. 2) [2010] EWCA Civ 65, Lord Neuberger MR at [134], identified a “very strong presumption indeed” that a judgment should be fully available for all to see but also held, citing Scott v Scott, that this fundamental principle must occasionally yield to other factors. Then at [176] he held that all parts of a judgment should be public unless there is a “very powerful reason to the contrary” or, as Lord Judge CJ put it at [41] such redactions should be “rare indeed”.

The nature of patent cases, which often involve technical trade secrets, means that the judges who deal with them are familiar with examining with particular care what are often unopposed claims to confidentiality in support of requests to sit in private, seal up documents or redact passages from judgments. The court rarely has the benefit of a party putting forward arguments against the submissions made by those seeking redactions. Nevertheless this is a context in which the judges themselves have a role to play in ensuring justice is done in public, subject only to properly substantiated exceptions. In the present case the judge had sought an amicus on this topic but none was provided.

In November 2017 the Unwired Planet confidentiality judgment (cited above) was the first time these issues arose relating to the terms of a full judgment in a FRAND case. The position on the facts was that, as is usual in this industry, each of the licence contracts contained a clause providing that its terms were confidential. The two parties (Unwired Planet and Huawei) as well as two former parties (Ericsson and Samsung) sought redactions from the full judgment. The information sought to be redacted related mostly to the licence terms: largely financial information such as royalty rates and lump sums, as well as a small amount about other terms in licences. All of the information related to licences to which at least one of Unwired Planet, Huawei, Ericsson or Samsung was a party, but third parties were involved too. No distinction was made in this judgment between financial information produced by unpacking or which was read off a licence. In fact (although one would need to read the full public FRAND judgment to see it) the financial information in issue there included both kinds.

Before going further it is worth making the obvious point that none of the information in issue in Unwired Planet was a technical trade secret. In such a case it is fairly easy to see why the right thing to do is to protect the secrecy of that information so that open justice yields (as referred to in Scott v Scott). There was no suggestion in Unwired Planet that the information in issue was a trade secret at all, I expect because at that time practitioners assumed (see e.g. Scott v Scott itself) that trade secrets were technical in nature. As Arnold LJ explains in JC Bamford at [53] under English law (prior to the Trade Secrets Directive) trade secrets, as a particular category of confidential information, received greater protection than confidential information in general. The question in Unwired Planet was how to grapple with information which was confidential and said to be commercially valuable but was not (then) regarded as a trade secret.

I addressed the principles as I saw them starting at [7], which highlights R (Mohamed) v Secretary of State and refers to some of the passages cited above. At [8] I referred to the common occurrence in patent disputes of having cases involving technical trade secrets and the public form of a judgment having such material redacted from it. I also noted that the fact this happens relatively often is not because the principles of open justice are different for patents, it is simply that these circumstances come up relatively frequently. I then said:

“Even then however the redactions from judgments (or confidential annexes which amount to the same thing) will be kept to the absolute minimum and claims to confidentiality have to be justified with cogent evidence focussed on the specifics rather than on generalities.”

The next important paragraph is [10] in which I noted that the information the court was being asked to redact had three relevant characteristics. First, it was not a technical trade secret, second it was (almost all) the terms of licences and third, these were licences in which at least one of the parties were involved but third parties were involved too. All three factors are present here too.

I then addressed other similar cases at [12]-[13] noting that in patent damages enquiries that sort of information is not as far as I know routinely redacted whereas in competition cases it is. Then at [14] I noted that more transparency about rates would be a good thing in the FRAND context.

Then at [15]-[16] I accepted the submission that one reason for keeping certain information out of the public domain would be if publication could be anti-competitive, because it could weaken the competitive position of a party relative to others in the market. Nevertheless because it is easy to make generalised assertions that publication would harm a company’s interests there is a risk of a lack of specificity. As I said at the end of [16] “Just because a company regards the information as confidential and would prefer that it not be disclosed is not enough.”

Finally, and after dealing with other aspects which do not now require elaboration, I summarised the applicable principles as I saw them at [23] and [24], as follows:

“[23] Unless the public can see and understand a judge’s reasons they cannot hold the courts to account. There is therefore a strong principle that all parts of a judgment should normally be publicly available. Nevertheless there are occasions on which judgments may be redacted. Redactions will require powerful reasons, supported by cogent evidence which addresses the details. Generalities will not do. Although redactions will be rare indeed when looking across the legal system in general, certain kinds of proceedings may regularly involve redactions due to the nature of the proceedings and the material involved. In any event however redactions must be kept to the bare minimum.

[24] Factors which will be relevant include:

(i)

The nature of the information itself: for example cases in which some redaction may more readily be accepted could include technical trade secrets and private information about family life.

(ii)

The effect of the publication of the information. This will be a critical factor. If publication would be truly against the public interest then no doubt the information should be redacted. If publication would destroy the subject matter of the proceedings – such as a technical trade secret – then redaction may be justified. The effect on competition and competitiveness could be a factor but will need to [be] examined critically.

(iii)

The nature of the proceedings: for example privacy injunctions and competition law claims may require some redaction while an intellectual property damages claim may not. The point is not that different kinds of case demand a different approach, it is that the balance of factors will change in different cases (e.g. the need to encourage leniency applications in competition law).

(iv)

The relationship between the information in issue and the judgment (as well as the proceedings as a whole). Obviously judges do not deliberately insert irrelevant information into judgments but not every word of a judgment is as important as every other word. It may be that some sensitive information can be redacted without seriously undermining the public’s understanding of the reasons.

(v)

The relationship between the person seeking to restrain publication of the information and the proceedings themselves (including the judgment). For example, a patentee seeking damages for patent infringement on a lost profit basis knows that they will have to disclose their profit margin in the proceedings and that those proceedings are public. A third party whose only relationship with the case is that they are a party to a contract disclosed by one of the parties to the litigation is in a different position.”

These principles expressly contemplate a weighing of factors in a balancing exercise. That in my judgment is also what was being described in R (Mohamed) v Secretary of State, see e.g. Lord Neuberger’s [176] which refers to factors “to be placed in the balance”.

Applying these principles to the facts, at [35] I decided that publication of the commercial licensing information (subject to exceptions) would materially weaken the competitive position of the relevant party in each case, particularly the relevant licensor. That harm was a sufficiently powerful reason to justify redaction. The public could understand how the conclusions were reached without seeing those details. Although, for example, the public could not decide for themselves whether they would agree that the specific figures arrived at are justified by the data, to provide that information would substantially weaken the position of various companies mentioned in the judgment, particularly as licensors but also as licensees.

One exception (at [41]) was about other terms in a historic licence which expired 5 years before the judgment. I accepted that the specific figures should be redacted but was not persuaded why it was that revealing the other terms referred to would really be damaging.

Next, in June 2023, Mellor J handed down his judgment on the same issue in InterDigital v Lenovo (above). On the applicable principles he followed the Unwired Planet confidentiality judgment and at [8] he agreed with them and accepted a submission from Apple (in effect a third party in those proceedings) that this approach involved undertaking a balancing exercise relating to the specific circumstances of the case. At [9] Mellor J noted that whatever was made public in a FRAND judgment would be analysed in great detail, with links being made between different pieces of information to draw inferences as to royalty rates, terms and licensing practices more generally.

As he explained in summary at [52], Mellor J decided to adopt a similar line to that taken in Unwired Planet on the basis that the way to strike the right balance in that case was to redact the rates (which included unpacked rates) and conditions in current and recently expired licences, while ensuring that the public version of the judgment sets out the important parts of his reasoning, recognising that that might require some details of licences key to the reasoning to be made public. At the end of that paragraph the judge noted the cost of dealing with these confidentiality issues and expressed the hope that by adopting a consistent line in these cases, the costs would be reduced.

Pausing here, in my judgment the articulation of the principles and their application in Unwired Planet and InterDigital v Lenovo, involving as it does a balance, follows directly from and is in line with the existing authority at that time, including Scott v Scott and R (Mohamed) v Secretary of State. The balance involved is a fact sensitive evaluation to find out whether, in the particular circumstances, the interests of justice outweigh the principle of open justice.

This brings me to JC Bamford v Manitou, which was in July 2023 in the Court of Appeal. This concerned an annex to a judgment in a patent case which set out a detailed description of the defendant’s machine, including a key criterion used by the machine in its operation called criterion X. The parties did not agree whether the information was confidential or not. The defendant contended that criterion X was confidential and so that part of the annex should be redacted from the judgment before publication. The claimant denied this. The judge held that the information was indeed confidential but that open justice outweighed the defendant’s claim to preserve confidentiality. There was an appeal about the conclusion that the information was confidential but that failed. The defendant appealed the decision not to redact that information from the public judgment. That appeal succeeded. The Court of Appeal held that the proper way to describe the defendant’s information in issue was that it was not merely confidential information in a general sense but was a technical trade secret (see Arnold LJ at [35] and [108], in the latter describing this as the crucial point). On that basis, referring to Scott v Scott at [110], Arnold LJ made the point that “where it is necessary to protect trade secrets, open justice must give way to the still greater principle, which is justice itself”, and then he said “This may make it impossible for the public to understand the details of the court’s reasoning but that is the price that must be paid for the proper protection of trade secrets.” I agree with this. I also agree, as Arnold LJ explained earlier at [106] and [107] that it was irrelevant that the machine did not infringe or that the trade secret belonged to the defendant.

In this part of [110] Arnold LJ also described the application of Scott v Scott as being a situation in which the “court is not engaged in an exercise of trying to balance incommensurables.” This is the origin of the judge’s view in the case under appeal that JC Bamford represented a new test which involves no weighing of interests at all and is therefore different in kind from the old test. I do not agree with that interpretation of this sentence, particularly read in the context of the judgment as a whole. All that was being described there was the point that if it was indeed necessary to protect the trade secret by restricting publication then, and only to the extent it was indeed necessary, open justice has to give way. The idea of one interest giving way to another makes the point.

Also, at [111] Arnold LJ noted that the JC Bamford case was quite different from an earlier case in the Court of Appeal, Lilly ICOS v Pfizer (No 2) [2002] 1 WLR 2253, which concerned confidential financial information. In that case (summarised in JC Bamford at [93] - [97]) the Court of Appeal made an order under CPR r31.22(2) preventing disclosure and subsequent use of a document containing confidential financial information which had been read and referred to at trial, despite the principle of open justice, owing to the limited role the document had played at trial. At the end of [111] Arnold LJ observed that it was doubtful the information in the document could properly be described as a trade secret. He observed that Lilly ICOS showed that even lower grade confidential information was entitled to protection when its publication was not necessary for open justice. There is no suggestion here that Arnold LJ was taking the view that Lilly ICOS was the application of a fundamentally different legal approach to one he had just described in the previous paragraph.

In JC Bamford Arnold LJ made it clear that he regarded his reasoning as in line with and based on existing English law, however he also explained and addressed a new aspect, the impact of the European Directive 2016/943/EU of 8 June 2016 on “the protection of undisclosed know-how and business information (trade secrets) against their unlawful acquisition, use and disclosure” and the Trade Secrets (Enforcement, etc) Regulations 2018 (SI 2018/597) which implement the directive. As Arnold LJ noted at [110] recitals (24) and (25) and Article 9 of the Directive, the relevant parts of which have been transposed into national law by Regulation 10 of SI 2018/597, lend support to the approach under Scott v Scott in that they explicitly provide that the court has power to publish non-confidential versions of judgments in which the trade secrets have been redacted. Again, so far nothing here appears to indicate a divergence in approach between the existing English law and the approach under the Directive. There was no need for this to be noted in JC Bamford but I observe that the terms of Article 9 and Regulation 10 are permissive in nature (both Art 9(2) and Reg 10(4) provide that courts “may” take the measures concerned).

In his analysis of the Trade Secrets Directive and Regulation Arnold LJ drew attention to the definition of “trade secret” there. That definition (Reg 2 of SI 2018/597, applying Art 2 of the Directive) is not limited to technical trade secrets but encompasses any information which in summary is (a) secret, (b) has commercial value because it is secret, and (c) has been subject to reasonable steps to keep it secret. The scope of this definition is wide and this is not the case to examine it. This wider scope did not matter in JC Bamford, which is no doubt why in [111] although Arnold LJ doubted that the information in Lilly ICOS was a trade secret within that wider definition, he did not have to resolve that doubt. In the present case the judge interpreted what he regarded as the new test as being applicable to any information which qualifies as a “trade secret” under the Regulation. However I do not believe JC Bamford reaches that conclusion nor, if the matter was free from authority, would I interpret the Trade Secrets Directive and Regulation as requiring that result. Therefore in my judgment the conclusion that there is a new test, different in kind from the old test is not right. There is only one test. The question is whether in the circumstances the principle of open justice gives way to the interests of justice itself, as described in Scott v Scott and set out above. In cases about technical trade secrets it will almost always be necessary to protect such a secret because of the nature of that kind of information, and so open justice will yield. However, as was identified at [111] of JC Bamford, even lower grade financial information is entitled to protection in some circumstances. The fact that information falls within the wide definition of a trade secret in the Trade Secrets Directive and Regulation does not lead to a step change in the approach.

It is never enough simply to say, as the judge did at [44], that because the information is a trade secret as defined in the Trade Secrets Directive and Regulation then necessarily and directly open justice will yield. On the contrary the circumstances will always be relevant.

Apply the right approach to the financial data

The financial information in question here is really nothing more complicated than the price the parties to a given licence were willing to accept to enter into the contract. The reason the corresponding financial information was redacted in Unwired Planet and InterDigital v Lenovo is because it was accepted that this pricing information is kept confidential because to publish it would weaken the competitive position of the parties, particularly the licensor (i.e. the third parties in all the Apple licences). That harm was a sufficiently powerful reason to justify redaction because the public could understand how the conclusions were reached without seeing those details.

The judge came to a different conclusion. Applying the old test the judge had accepted that at least for the lump sums (Class 3) the information was confidential and that that confidentiality was justified because the price at which a party is prepared to do a deal is likely to be helpful to others in the market, and its disclosure is likely to be correspondingly disadvantageous to the parties to the given licence ([31](iii)]). However he then gave four reasons why although this was a factor in favour of redaction, it was not a strong factor.

The first ([31](iv)(a)] was that the information was not “especially commercially significant” because it was very dated in a market which was very dynamic. This was not correct on the facts. Clearly the more recent the licensing information, the more sensitive it is and in the case of truly historic information, then the situation might be different. Licences are typically on 5 year terms so even 2014 information (the earliest start date of a licence here) can represent a good data source. In addition, as InterDigital pointed out, if these licences were relevant to determine a FRAND rate in litigation (as they were) then for the same reason they must be at least potentially relevant in FRAND negotiations. The fact that many of the licences had expired by the time of the consequentials judgment (February 2024) does not mean that they do not have value. Also as Qualcomm pointed out in submissions, its licence period has been extended and the licence is therefore current even though the judge appears to have thought it was not.

The second reason ([31](iv)(b)] was that there was no specific evidence of harm from the relevant third parties. However there was specific evidence from Apple, the counterparty to all the Apple licences, on exactly this point. The fact that evidence did not come directly from the numerous third party counterparties to those Apple licences was not a logical distinction. The evidence of harm was before the court and was as specific as the subject matter allows.

The third reason is given in [31](iv)(c). This passage states that the commercial disadvantage to the third parties of publication would be minor and that they could look after themselves. However, on the contrary, there was clear evidence (e.g. from Ms Mattis, Vice President and Chief Licensing Counsel of InterDigital) which addressed the very large revenues at stake in this licensing business and which explained, in detail, the problems which would be caused to a licensor like InterDigital by disclosure of this material. Licensees naturally want to seek the lowest rates possible and if these rates are made public the licensees will push for lower rates. The court will always examine evidence of this kind critically but this is cogent evidence about the problems caused by disclosure of the rates. It goes too far to dismiss the commercial disadvantage described there as minor.

It is worth noting that if the judge’s conclusion here about the minor disadvantage was correct, it would demonstrate the importance and oddity of the distinction between the new and old tests as the judge saw it. As applied the new test demanded redaction, to which open justice had to give way, even though on the findings as they were the commercial disadvantage was minor.

The fourth and final reason is given in [31](iv)(d). This is the FRAND background. Sub-section (i) questions the commercial importance of these rates at all given the existence of the FRAND regime, however that is answered by the fact that the evidence before the court did establish that the rates were highly important commercially, and this was all given in the context of FRAND. Sub-section (ii) makes the point that given the FRAND regime there is a self-evident interest in publishing the rates. I made the same point in Unwired Planet (at [14]). However as it seemed to me then, that approach was a matter for the ETSI IPR Policy and it was not for the courts to unilaterally take a different line. We were informed that since then ETSI has considered the issue and decided not to amend its IPR policy to require transparency in rate setting.

Standing back, none of the reasons justify refusing to redact the financial details of these licences. This information is the third parties’ confidential information and to publish it would cause them real commercial harm. Redaction is necessarily the only way to avoid that harm. That will inevitably mean that the full detail of the reasoning in the main judgment (and for that matter the main appeal judgment) will not be published. However the public can still understand how the conclusions were reached without seeing those details. Therefore in these circumstances the right thing to do in the interests of justice itself is to redact that material albeit this involves a derogation from open justice.

It follows from this conclusion that the averages and totals in the tables ought to be redacted too. The reason is because the nature of the information in the tables means that if the averages and totals were revealed, particularly with some individual entries being revealed too, it would be relatively easy to infer good approximations of values for redacted entries. In other words the publication of the averages and totals would undermine the redaction of the financial details.

The remaining issues

There are two further related issues to resolve. They relate to a provision of the InterDigital licence about whether or not the licence allows for sub-licensing and a provision of the LG Electronics licence about whether or not the licence involves a cross-licence. The statements in the full judgment (Annex 3) about these two details are wrong. The less-redacted judgment would publish both wrong statements. The judge was asked to correct these mistakes but refused. Both parties contend that the mistakes ought to be corrected and at least the wrong information should not be published. LG Electronics does not on this appeal seek to keep the correct position (that there is a cross-licence) secret. InterDigital contends the correct information should not be published either, for the same reasons as apply to the financial details and so for the InterDigital licence I will not spell out the error or the correct position.

I will start with the correction of errors. To recap, in the judge’s judgment under appeal, class 1 was defined as “immaterial objective errors in the Judgment to be corrected under the ‘slip’ rule” and addressed at [22](i) and [24]-[25]. The issue arose because a number of third parties, as well as InterDigital and LG Electronics, had identified what they contended were a few factual errors in the judgment. This happened only after the full judgment had been handed down because the third parties had not seen the embargoed draft (as the judge recognised at [22](i)). The judge also rightly observed at [22](i) that this is not a question of redaction at all. He then referred to the “slip” rule.

The judge set out the principles he was going to apply in [24]-[25]. In summary these were that third parties have no right to seek to engage in the drafting of a judgment that is handed down and final, that he was only prepared to entertain these because they are incidental to the redaction issues, and that the general answer to such corrections ought to be a firm “No”. However at [25] he explained that he was prepared to make corrections under the “slip” rule provided (i) the correction is of an objective error which all parties concerned agree is an error, (ii) the correction is entirely immaterial to the judgment, (iii) the correction can be achieved without any form of rewriting. In footnote 26 the particular errors the parties had asked to be corrected were described as sitting just above “trivial” and properly called minor.

Before this court there are only two errors in issue. They were put to the judge, but he did not accept the proposed correction.

The approach set out at [24]–[25] is not right. The slip rule is at CPR r40.12 as follows:

Correction of errors in judgments and orders

40.12

(1) The court may at any time correct an accidental slip or omission in a judgment or order.

(2)

A party may apply for a correction without notice.

There is nothing in the rule which prevents a third party from raising what appears to be an accidental slip or omission. The fact rule 40.12(2) refers to a party being able to apply without notice is permissive, it is not there to prevent a third party raising the matter. Turning to [25](i), no doubt normally all parties will agree and if so and the matter is within the rule, then the court will make a correction, but agreement is not a prerequisite. The court simply needs to be satisfied the correction should be made under the rule. As for the need for immateriality (at [25](ii)) and no rewriting (at [25](iii)), on the contrary the accidental slip or omission may be highly material and might well require rewriting. The important limits to the slip rule, which can be seen in the authorities but do not arise in this case, are that this rule is not there to allow second thoughts, reconsideration or as a substitute for an appeal.

The errors in this case are obviously simple slips and ought to be corrected. The error relating to LG Electronics can be corrected by deleting the relevant words. The error relating to InterDigital can be corrected by replacing the relevant words with the words used elsewhere in Annex 3 for other licences which would be a correct reflection of the InterDigital licence.

Should the corrected information, about whether or not the InterDigital licence allows for sub-licensing, be redacted? Given the Apple appellant’s notice, this issue now relates to all of the licences. I am not convinced. As I have held above, the evidence before the judge, particularly from InterDigital, was sufficiently specific to justify redaction of the rates, but it did not focus on this feature at all. No doubt in some very general sense a licence which permitted sub-licensing might be more valuable than one which did not, but that is really going to depend on specifics and detail. Without specific evidence there is no justification for redaction.

This conclusion, to include in the less-redacted judgment the correct statement about sub-licensing, therefore applies to all the entries in Annex 3.

I therefore allow the appeals, save for the point on redacting the corrected information in Annex 3.

I would invite the parties to agree arrangements for sorting out the consequential matters.

There will also be a need to address the redactions in the judgment on the main appeal, which involved a separate constitution (with me in common). I would invite the parties to write to the court on that matter with proposals on the way forward.

I agree.

I agree.