Triumph Controls UK Ltd & Anor v Primus International Holding Co & Ors

INTRODUCTION

In these proceedings, the claimants claim some US$65 million against the defendants for breaches of warranty following the sale of the defendants’ aerospace business to the claimants via a share capital purchase. It is said that the defendants failed to disclose various aspects of the business, including alleged operational failings at, and lack of accreditation of, their facility at Farnborough. The original trial date was adjourned as a result of the claimants’ late application to re-amend the Particulars of Claim, and the trial is now due to take place for five weeks starting on 18 June 2018.

By an application dated 19 December 2017, the defendants sought wide-ranging orders arising out of what they said were the fundamental deficiencies in the claimants’ disclosure. On the face of it, the application required the claimants to re-do, from scratch, the search exercise that they had undertaken. That was an unrealistic stance and Mr Pepperall recognised that in his helpful skeleton argument.

In consequence, the defendants sought two more focused orders from the court. The first was an order that the list of 860,000 folders and file paths which had been identified by the claimants on the Farnborough shared drive should be provided to the defendants so that they could see whether there were any folders or file paths which had not yet been – but should be – searched. The second was for an order that the claimants undertake a manual review of the balance of 220,000 documents (out of a total of 450,000) which had been identified as potentially disclosable following the keyword search but which, other than a very limited sampling exercise, had not been further searched.

At the end of the hearing, in view of the proximity of the trial date and the overriding need for speed, I set out orally the conclusions noted at paragraphs 24 and 42 below, and gave brief reasons for my decision. I promised to provide a Judgment setting out those reasons in detail.

Accordingly, I deal with the disputes between the parties in this way. I set out the background to the disclosure dispute in Section 2. I set out the relevant law in Section 3. Thereafter, at Sections 4 and 5, I deal respectively with the application in respect of the file paths, and the application in relation to the 220,000 documents. I am very grateful to counsel for their extremely helpful written and oral submissions.

THE BACKGROUND

Disclosure has been something of a running sore in this case. The claimants’ Electronic Documents Questionnaire (“EDQ”) indicated that, following a keyword search, all the documents responsive to those searches would be “manually reviewed”. There was no reference to a Computer Assisted Review (“CAR”).

The parties agreed the keywords. Because of the large amount of potential documents that would have been turned up by the original keywords, those were reduced in number, again by agreement. This refined exercise produced a total of 450,000 responsive documents.

On 8 July 2016, the claimants served their first list of documents. This was based on a review of over 200,000 documents. 12,476 documents were disclosed in that first list.

On 29 July 2016, the claimants served a first supplemental list, disclosing a further 4,163 documents. The covering letter indicated that those had been “manually reviewed”. This made a total of 16,500 disclosed documents.

It has subsequently become apparent that, having reviewed a total of around 230,000 of the 450,000 documents by the end of July 2016, using manual searches aided by CAR, the claimants decided not to search the balance of 220,000 documents which had been de-prioritised for manual review by CAR. The claimants say that they sampled 1% of those documents using a CAR technique which revealed that only 0.38% of the remaining documents would be relevant. They therefore concluded that that was disproportionate and that no further searches were required. The extent to which that was an appropriate course to take is dealt with in Section 5. It was not at any time discussed with the defendants, let alone agreed with them.

Since then, the claimants have produced 10 supplemental lists of documents, together with a separate disclosure exercise relating to valuation documents. Almost 3,000 additional documents have been disclosed as a result, making a total of 19,500 disclosed documents. The evidence suggests that about two thirds of these later documents, so around 2,000 documents, were responsive to the keywords which had been agreed in 2016. On the face of it, therefore, those 2,000 documents should have been disclosed originally.

THE RELEVANT PRINCIPLES OF LAW

The starting point for any consideration of the court’s approach to disclosure disputes is Nichia Corporation v Argos Limited [2007] EWCA Civ. 741. Jacob LJ stressed that standard disclosure would not necessarily give rise to ‘perfect justice’. He said:

“50.

There is more to be said about the change to standard disclosure and indeed to the express introduction of proportionality into the rules of procedure. "Perfect justice" in one sense involves a tribunal examining every conceivable aspect of a dispute. All relevant witness and all relevant documents need to be considered. And each party must be given a full opportunity of considering everything and challenging anything it wishes. No stone, however small, should remain unturned. Even the adversarial system at its most expensive in this country has not gone that far. For instance we do not include the evidence of a potentially material witness if neither side calls him or her. Nor do we allow pre-trial oral disclosure from all potential witnesses as is (or at least was) commonly the practice in the US.

51.

But a system which sought such "perfect justice" in every case would actually defeat justice. The cost and time involved would make it impossible to decide all but the most vastly funded cases. The cost of nearly every case would be greater than what it is about. Life is too short to investigate everything in that way. So a compromise is made: one makes do with a lesser procedure even though it may result in the justice being rougher. Putting it another way, better justice is achieved by risking a little bit of injustice.

52.

The "standard disclosure" and associated "reasonable search" rules provide examples of this. It is possible for a highly material document to exist which would be outside "standard disclosure" but within the Peruvian Guanotest. Or such a document might be one which would not be found by a reasonable search. No doubt such cases are rare. But the rules now sacrifice the "perfect justice" solution for the more pragmatic "standard disclosure" and "reasonable search" rules, even though in the rare instance the "right" result may not be achieved. In the vast majority of instances it will be, and more cheaply so.”

Another important case dealing with general disclosure principles, particularly relating to electronic documents, is Digicel (St Lucia) Limited and Others v Cable & Wireless Plc and Others [2008] EWHC 2522 (Ch). In that case, Morgan J ordered the defendants to redo the expensive search they had already undertaken, in part because they had acted without the agreement of the claimants in the first place. He said at paragraph 81:

“It is unfortunate that this dispute about the extent of the key word search comes to Court after the Defendants acted unilaterally in choosing key words and conducting a search. In acting unilaterally, and in disregarding the clear advice in Part 31 Practice Direction, the Defendants have exposed themselves to the risk that the Court will conclude that their search was inadequate and that the Court should order the Defendants to carry out a further search. The Defendants submit that because they have already carried out a search, taking a considerable time and involving a very large cost, the Court should be most reluctant to order them to carry out a further search. It seems to me that I should approach the issues in relation to key words in two stages. At the first stage, I will attempt to identify where to draw the line between inclusion and exclusion of the suggested additional key words. If as a result of that process I conclude that the Defendants should have, first time round, used additional key words I will then have to consider whether to make an order for a further search under Rule 31.12. It is also possible for me to distinguish between the e-mail accounts of the 16 individuals who have already been the subject of key word searches and any further e-mail accounts found following the restoration of back-up tapes, where there have not previously been any key word searches.”

In that case, Morgan J also dealt in clear terms with the issue of proportionality. He said:

“94.

The Defendants say it would be wholly disproportionate for the Court now to require them to carry out a further search. They refer to the fact that the Defendants, unlike the Claimants, did not bring all the electronic documents together on one database. The result will be that if a further search is required the search will have to be done in situ in the various territories. It is also said that further searches are likely to throw up many further documents which (the Defendants submit) will be largely irrelevant and it will be a burdensome task to review manually so many documents. I have attempted, so far as I think appropriate, to balance the prospect of benefit from the exercise against the burden of the exercise so that where the burden on my assessment outweighs the benefit I have not included the key words asked for by the Claimants.

95.

My overall conclusion is that is appropriate to order the Defendants to carry out electronic searches of the e-mail accounts of the 16 identified individuals, using the additional key words which I have identified. As I understand it, it ought to be possible for the Defendants to run a negative key word search using as negative key words the 10 words they previously used so as to eliminate documents which have already been processed in the earlier round of searches.

96.

When judgment is handed down, I will hear Counsel as to the time which should be allowed for the Defendants to give the further disclosure which may be appropriate following further searches. I will also hear Counsel as to any consequential adjustments which may be appropriate to the directions previously given. At the hearing, the Defendants submitted that I should not order further disclosure because such an order would jeopardise the trial date. I will hear Counsel as to whether the trial will need to be re-fixed. It is not obvious to me at present that that will be necessary. If it is necessary to re-fix the trial date as a result of my further order for disclosure, then the Claimants do not complain because, of course, they seek an order for disclosure. As to the Defendants, unfortunately it was their failure to carry out a reasonable search in the first instance which has lead to the making of this further order.”

Morgan J in his judgment referred extensively to Practice Direction 31B, which deals with the disclosure of electronic documents. In my view, this is a thorough and clear PD. One of the things which it emphasises, just as Morgan J did in Digicel, is the need for discussion and agreement between the parties at the outset: see for example paragraphs 8 and 9, and paragraph 19, which stresses that, if disclosure is given without prior discussions with the other side, “the court may require that party to carry out further searches for documents or to repeat other steps which that party has already carried out.”

Of even greater value, in my view, is the TeCSA/TECBAR edisclosure protocol, used as a guide in every TCC case involving electronic disclosure. This provides the clearest possible guidance as to how this work should be undertaken. Amongst other relevant matters, paragraph 4.2 states that “filtering should be agreed with the opposing party at the earliest possible stage to avoid the risk of having to repeat the exercise later.” Paragraph 5.2 emphasises the importance of co-operation so as to avoid misunderstandings, and paragraph 5.4 deals with the use of CAR.

Finally in relation to edisclosure, I was referred to the decision of Master Matthews (as he then was) in Pyrrho Investments Limited and Another v MWB Property Limited and Others [2016] EWHC 256 (Ch). This was a slightly curious case because the parties were, in accordance with PD 31B, agreed about the use of CAR. But, despite that, Master Matthews gave a judgment anyway. He referred to the TCC edisclosure protocol, but only in passing, describing it as “only a protocol and [it] has no normative force”. With respect, that rather old-fashioned approach misunderstands the importance of protocols in the TCC, where the court’s overarching requirement is that the parties cooperate and comply with them, unless there is a very good reason why not.

Notwithstanding these oddities, at paragraphs 17-24, Master Matthews sets out some useful guidelines. In particular, he said that, in teaching the CAR technology about the document collection exercise in any given case, “it is essential that the criteria for relevance be consistently applied at this stage. So the best practice would be for a single, senior lawyer who has mastered the issues in the case to consider the whole sample.”

ISSUE 1: THE FOLDERS AND FILE PATHS

The Farnborough shared drive is extremely large, containing three Terabytes (“TB”) of data approximating to some 20 million documents in 860,000 folders or file paths. In order to reduce this to manageable proportions, the claimants’ solicitors asked their custodians, listed in Appendix A of their Disclosure List, to identify the folders or file paths that were likely to contain relevant documents. The solicitors also considered, by way of a cross-check, the full list of file paths. In this way, the data collected was reduced to about one TB.

It is certainly right that the claimants took this step without consulting the defendants. It may have been better if they had done so. But I am satisfied that this process was made plain in the claimants’ original List of Documents under the heading “Steps Taken to Collect and Process Electronic Documents”. Thus, even if the defendants were entitled in principle to complain about the process which reduced the numbers of folders and file paths which were searched, their complaint has been raised 17 months after they first became aware of what had been done.

The next question is whether the evidence demonstrates that the claimants’ method of reducing the data collected has been sensible and proportionate. In my view, the evidence demonstrates that it has been. Only two further folders/file paths, beyond those originally identified by the claimants’ custodians, have been subsequently disclosed. That is a statistically negligible amount of further disclosure, thereby demonstrating the reasonableness of the claimants’ original approach.

In addition, I note that the defendants have been unable to identify any particular folders or file paths which are obviously missing or which would contain relevant information. Of course, I accept that, following the share purchase agreement in 2013, the defendants no longer have any direct access to all the documents that were originally their own. But given that they have the assistance of, amongst others, Mr Paul Jerram, the managing director at the relevant time, it seems to me more likely than not that, if there were obviously missing folders/file paths, they would have been identified.

For those reasons, I consider that the claimants have always been clear about what they have done about the Farnborough shared drive; any complaint about it now raised by the defendants is very late; the evidence shows that the claimants’ approach has been reasonable and proportionate; and the defendants are unable to demonstrate that any particular folders/file paths have not been disclosed.

For all those reasons, I conclude that it would be neither reasonable nor proportionate to order any further disclosure of the folders/file paths. That aspect of the defendants’ application must fail.

THE BALANCE OF 220,000 DOCUMENTS

As noted above, 450,000 documents were identified in the original keyword search. The claimants have searched around 230,000 of those documents manually aided by CAR. It appears that those searches were a combination of a manual search and a CAR process. In the first instance, that produced around 16,000 disclosed documents.

As noted above, those searches ceased in late July 2016 because it was felt that there were diminishing returns. The claimants have subsequently said that they undertook a sample of 1% of the remaining 220,000 documents, using the CAR. That produced the prediction that only 0.38% of those documents would be relevant. In those circumstances, the claimants concluded that it was not proportionate or reasonable for those further 220,000 documents to be further searched.

On this aspect of the defendants’ application, I have much greater concerns about the claimants’ approach. First, what they did is not what they said they would do in the EDQ, which promised a manual review of all documents responsive to the keyword searches. Neither is what they did at all clear from their Disclosure List.

Although the broad outline of the searches undertaken by the claimants was subsequently explained in September 2016, it was not a detailed exposition. At no time have the claimants provided relevant details as to how the CAR was set up or how it was operated. In circumstances where the decision to use the CAR was unilateral, and where the defendants had no input into it at all, that is unsatisfactory: see Digicel, Pyrrho, PD 31B and the TCC edisclosure protocol.

This problem has been compounded by the lack of information as to the sampling exercise. All that the defendants, and the court, have been told is that there was a sampling exercise which produced a predictive figure of 0.38%. But there is no information as to precisely how that sampling exercise was conducted. There are, for example, no stated tolerances and no explanation of how many rounds of sampling were undertaken. That again is unsatisfactory. It is not in accordance with the TCC edisclosure protocol. I am bound to say that, given that both the CAR and the sampling were unilateral, I am slightly surprised that there is not now better evidence as to what actually happened. There is always a risk, as Digicel makes plain, that a unilateral decision will be carefully scrutinised by the court at a later date, and a different course may be ordered. That will be more likely if the evidence as to what was done remains vague.

There is a further point about the number of people involved in the CAR process. The evidence suggested that there were perhaps ten paralegals and four associates involved in the searches. It is not apparent that there was any overseeing senior lawyer, and certainly not one undertaking the role advocated in Pyrrho (paragraph 18 above). So whilst the recommendation in that case may be regarded as a counsel of perfection, I think that Mr Pepperall is right to say that the sheer volume of those involved with the CAR system in this case may mean that it has not been ‘educated’ as well as it might have been, particularly in respect of the criteria for relevance.

In all those circumstances, therefore, I agree that both the CAR exercise, and the sampling exercise that it produced, cannot be described as transparent, and cannot be said to be independently verifiable.

But does that matter? Notwithstanding the flaws in the system adopted, is it reasonable or proportionate to require the claimants to do anything more in relation to disclosure of the balance of 220,000 documents? I have considered this question carefully, and I have concluded that it is. There are a number of reasons for that.

First, the total number of documents disclosed (16,500 originally, and 3,000 since, or about 19,500 in total) appears very modest in circumstances where 450,000 documents have been identified as being responsive to the agreed keywords. In my experience, even in an ordinary case, the percentage of disclosable documents, compared to the total of responsive documents, would tend to be larger than the 4.3% or so which the 19,500 documents represent. That potential discrepancy is even starker in the present case, which is based on a wide-ranging series of allegations as to non-disclosure in respect of many different elements of an international business. Such allegations will almost always lead to a document-heavy trial.

Secondly, as I have noted above, around 2,000 of the documents disclosed by the claimants subsequently (i.e. after the first list and the first supplemental list) were responsive to the original keywords. That strongly suggests that they should have been disclosed originally. That raises a further question-mark about the adequacy of the claimants’ disclosure process.

Thirdly, the evidence makes plain that at least some of the total of 3,000 further documents have been disclosed because they were documents that the claimants’ own witnesses and experts have referred to and/or wish to rely on. Something has potentially gone wrong with disclosure when, for example, a witness of fact can remember a particular document when preparing his or her witness statement which, on the one hand, is so important that he/she wants to refer to it in their evidence but which, on the other hand, has not even been disclosed.

Finally, the only sampling exercise which has been done on the 220,000 documents is that which produced the 0.38% prediction. For the reasons that I have given, that figure may well not be reliable. Indeed, I consider that all the other evidence demonstrates that the 0.38% is likely to be an underestimate.

For all these reasons, I conclude that the steps taken by the claimants in relation to the balance of the 220,000 documents have not been adequate. What therefore should be done to correct that, considering all the circumstances of the case?

First, I consider that some form of manual search is required. Whilst I agree with Mr Pillai that, as he put it, there is no magic in a manual review, the absence of any real explanation of the CAR process in this case means that it is difficult to contemplate any alternative which would not be potentially controversial. That is one of the dangers of operating a system without agreement or explanation. Since time is of the essence, only a manual review will do.

Following the late change in the relief sought by the defendants, the claimants’ solicitor very helpfully set out, in their letter of 30 January 2018, their concerns about a manual review of the 220,000 documents. They said it would cost £180,000 and that it would take two months.

As to the costs, whilst of course the court is anxious to keep them to a minimum, that figure must be set against the claim for US$65 million. As to the estimated time, I consider that two months is very much a ‘worst case scenario’ because, whatever the arguments might be as to their status as proper comparators, there was other evidence in the documents which indicated that it would not take anything like as long as two months.

However, I recognise the importance of proportionality. I also recognise that the trial is less than six months away.

Thus, at the hearing on 31 January 2018, having given my reasons for arriving at the conclusion summarised in paragraph 37 above, I ordered that the parties were forthwith to agree a methodology by which a sample of 25% of the 220,000 documents was to be manually searched. That search was to take no longer than three weeks. The results are to be put into an agreed letter. I can then be shown the letter and told about the results of that search at the hearing on 22 February 2018, which is the date on which the parties are back before this court to argue about another aspect of this case.