Jersey Unreported Judgments

Discovery - application by the plaintiffs to vary their approach to providing electronic discovery

[2021]JRC320

Royal Court

(Samedi)

20 December 2021

Before     :

Advocate Matthew John Thompson, Master of the Royal Court.

 

Advocate J. P. Speck for the first Plaintiffs.

Advocate D. M. Cadin for the Defendant.

CONTENTS

		Paras
1.	Introduction	1
2.	Background	2-24
3.	Submissions	25-56
4.	Decision	57-73

judgment

the MASTER:

Introduction

1.        This judgment contains my detailed reasons in relation to an application by the plaintiffs to vary their approach to providing electronic discovery.

Background

2.        The present proceedings were commenced by an order of justice dated 27^th February 2015.  The proceedings make allegations of breach of trust by the defendant in particular focusing on the failure to give or take tax advice or to advise of the need to take tax advice.

3.        The present proceedings arose as a result of HMRC making certain enquires of the second plaintiff, in particular in relation to the tax treatment of three payments made to facilitate the acquisition of a property in London and secondly, the tax treatment of certain payments made into certain trusts by various third parties in 2008 for the benefit of the second plaintiff.

4.        Initially the proceedings were adjourned to enable the second plaintiff to continue to engage with HMRC.  Further adjournments took place due to the ongoing engagement with HMRC and also because of various health reasons relating to the second plaintiff.  However, matters were not resolved with HMRC who therefore issued various assessments against the second plaintiff.  The issue of those assessments by HMRC is being challenged by the second plaintiff.  The amount of tax assessed is, in round figures, £10.75 million.

5.        On 10^th December 2019, directions were given requiring the defendant to file an answer.  The answer ultimately filed denies that the defendant was holding itself out as a tax adviser, denies it owed a duty to give or take tax advice, avers that the second plaintiff was taking tax advice from others and avers that such tax advice was paid for out of trust assets.  Any breach of any duty alleged was therefore also denied.  A counterclaim is also made against the second plaintiff seeking damages based an alleged breach of a contractual undertaking by him because it is said he failed to explain the circumstances of the gifts made to the trusts which are now the subject of tax assessments by HMRC.  It is not otherwise necessary to set out the detail of the pleaded allegations or the detail contained in the answer for the purposes of this judgment.

6.        It is necessary, however, to refer to certain parts of the procedural history of this dispute before focusing on the present application.

7.        At a hearing on 20^th August 2019, when I permitted the plaintiffs to amend their order of justice, I indicated that I was minded to limit discovery to issues of liability as it was apparent at that stage that discovery was going to be extensive and complicated.

8.        At a hearing on 8^th July 2020, while granting a further stay, I encouraged the plaintiffs in the strongest terms to provide documents referred to in communications from HMRC to the defendant because such documents would be disclosable.  I also encouraged the defendant to provide documents requested by the plaintiffs for the purposes of the HMRC appeal.

9.        On 3^rd February 2021, I made orders requiring the filing of pleadings as well as orders requiring the parties to exchange with each other the information required by Practice Direction RC17/08 in relation to electronic discovery.  The context of this was that I was being asked to order discovery before the filing of pleadings which I refused.  However I indicated that at the next directions hearing I wanted to be addressed on how discovery should be dealt with.

10.      On 7^th May 2021 I approved a consent order varying the timetable set out in the Act of Court of 3^rd February 2021, including varying the time limit by which the parties were to exchange with each other the information required by Practice Direction RC 17/08 in relation to electronic discovery.

11.      The question of what discovery to order was dealt with on 23^rd August 2021 where I made the following orders: -

"1.       by 5:00 p.m. Friday, 29th October 2021 the Parties shall each make discovery to each other by a provision of a list of documents verified by affidavit.  The discovery to be provided shall be as set out in this order;

2.        the Defendant's electronic discovery shall be as set out in its discovery protocol dated 26th June 2021 save that

a.        the period of searches for each of the claims identified in paragraph 1 shall commence 3 months before the date identified by the Defendant and shall end 3 months after the end of the relevant period, again identified by the Defendant; and

b.        for the third tranche the relevant dates shall 3 months either side of 1st January 2008 and 23rd January 2009 by reference to the dates pleaded at paragraph 34 of the re-amended order of justice;

3.        the Defendant shall further provide discovery of all relevant documentation concerning the Albion Trust accounts referred to in paragraph 26 of the re-amended order of justice and what is described in the re-amended order of justice as the second tranche payment;

4.        Other than pursuant to paragraph 3 of this order the Defendant shall not be required to provide discovery of any hard copy files;

5.        the Plaintiffs shall provide discovery in accordance with its discovery protocol dated 6th August 2021 save that the commencement date for any searches shall be 1st October 2006 being the date 3 months prior to the date of the transaction described as tranche one;

6.        following the provision of discovery by the Parties pursuant to paragraph 1 of this order, the Plaintiffs shall, either every 6 weeks or when any major step is taken by either of the Plaintiffs, whichever occurs first  in relation to either the Plaintiffs' review of penalties imposed by HMRC or in relation to any appeal to the first tier tax tribunal,  provide within 14 days of the expiry of the said 6 week period or the occurrence of a major step , all relevant documents created during the said 6 week period or in relation to the said step..."

12.      I made a discovery order as I could not see the parties agreeing each other's discovery protocols which they had exchanged.  Instead I focused on the parties producing documents core to the dispute while at the same time taking a reasonably wide ranging approach into what searches would be carried out by each party.

13.      In relation to the defendant, I broadly adopted its protocol subject to broadening the time periods that the defendant had to search.

14.      In relation to the plaintiffs, I narrowed their discovery so that it commenced three months prior to the date of the first tranche payment referred to in the order of justice.

15.      I also set a date for a further directions hearing to deal with various matters including any applications concerning the other party's discovery provided pursuant to the discovery order or for any specific discovery application.  I took this approach so that, if the discovery approach adopted by one party turned out to have significant flaws or relevant material was clearly missing, I could be addressed on that with full analysis of any defects and any necessary evidence.

16.      The other relevant order was paragraph 6 which was designed to ensure that the plaintiffs kept the defendant informed following discovery on a regular basis of what was happening between the plaintiffs and HMRC.

17.      On 13^th October 2021 Ms Nicola Le Boutillier from Mourant Ozannes (Jersey) LLP ("Mourant") wrote to Advocate Cadin and stated the following: -

"We write by way of courtesy to inform you that it has become clear to us that we are not going to be in a position to exchange discovery by that deadline. This is because the data collection exercise from our clients and third party data sources has taken significantly longer than anticipated and the number of documents responsive to the search terms set out in the Plaintiffs' discovery protocol is excessive. As a result, it will be necessary to revise the search terms to be applied to the data in order to reduce the number of responsive documents requiring review. You will recall that, at the time the Court order was made, the Plaintiffs were not yet in possession of the data from the relevant data sources and had not conducted keyword search terms and so included a caveat at paragraph 6.6 of their discovery protocol notifying the Defendant that subsequent amendments to the search terms may be required. Once we have determined alternative search terms that return a proportionate number of documents to be reviewed, we will of course provide you with our revised protocol to seek your consent to the same before updating the Court as to the required revisions."

18.      On 20^th October 2021 Advocate Speck wrote to Advocate Cadin.  Paragraph 4 of his letter summarised the position as follows: -

"4. These issues have been compounded by the fact that, upon running the requisite searches against the data that was available to them, TransPerfect informed us that there were over 1.5 million responsive documents. Clearly, this is an excessive number of documents and it would be disproportionate for us to review all of these documents. As a result, and in light of the overriding objective, we have sought to limit the number of responsive documents by narrowing the search terms set out in the Plaintiffs' discovery protocol. Having conducted the narrower search terms against all of the data currently available, TransPerfect has confirmed that these terms return just over 92,000 documents. We should be grateful for your confirmation that you agree to the Plaintiffs' use of the revised search terms, as set out in the enclosed example search term report that was conducted against the largest data set held by TransPerfect. Once confirmed, we will prepare a consent order for the parties to sign, annexing the revised search terms for the Master's endorsement."

19.      Advocate Cadin replied on 22^nd October 2021 complaining as follows: -

"You indicate at paragraph 4 of your letter that TransPerfect has run the requisite searches in accordance with your client's proposed discovery protocol and this has resulted in over 1.5 million responsive documents. You are now proposing an (unspecified) amendment to the protocol to reduce the number of responsive documents to 92,000. This is a substantial reduction and one that will need to be scrutinised properly. Yet you have only given us 2 days in which to respond. Moreover, the information provided to us in order that we might consider the request is singularly lacking (and in the case of page 2 of the search terms report, entirely illegible). For example, you have not provided (amongst other things):

(i)        any search terms report in respect of the original search exercise which leads to 1.5 million responsive documents;

(ii)       any explanation or methodology used to determine the appropriateness of the revised search terms; or

(iii)      any comparative document showing the difference between what was originally proposed and what is now proposed (in terms of search terms, hits, and discovery protocol)."

20.      Advocate Speck replied providing the information requested in his letter of 26^th October 2021.  Attached to that letter was an amended discovery protocol and a table showing the changes to the search terms approved by me on 23^rd August.  Bedell Cristin replied the next day suggesting a general extension of time to exchange discovery terminable by either party on 7 days' notice or by further order of the court.

21.      However, when this proposal was put to me, I was not willing to agree an unspecified date by which discovery was to be provided because such an approach (of leaving the timeframe for discovery to the parties) was inconsistent with the principles of active case management.  This was set out in a letter from me to the parties dated 29^th October 2021.  In the same letter I noted that in their communication to me the plaintiffs considered that they would be ready to provide discovery by Friday, 17^th December 2021 if their revised terms were agreed.  I therefore suggested this would be an appropriate date for the provision of discovery.  However, I also suggested that the defendant was given two weeks to decide whether it objected to the revised search terms and if there was such an objection then the application should be heard before me as soon as possible.

22.      On 5^th November 2021, Bedell Cristin wrote to me, copied to Mourant, objecting to the plaintiffs' revised protocol.

23.      This led to the present application with the plaintiffs' summons seeking a variation to the order of 23^rd August as follows: -

"1. The deadline for discovery at paragraph 1 of the Act of Court dated 23 August 2021 (Act of Court) be extended to 5:00pm on Friday, 17 December 2021. Such order to be a final order, without liberty to apply.

2. Paragraph 5 of the Act of Court be removed and replaced with the following wording:

"the Plaintiffs shall provide discovery in accordance with their revised proposed discovery protocol annexed hereto, hereinafter referred to as the "Plaintiffs' Further Revised Protocol", subject to paragraph 6 below".

3. A new paragraph 6 be added into the Act of Court with the following wording:

"to the extent that any of the documents responsive to the Plaintiffs' Further Revised Protocol are allocated a predictive relevance ranking of between 0 and 0.02 out of 1 using Brainspace analytics, the review of such documents shall be limited to a review of a 3% random sample, such sample to be determined by Brainspace's random selection software"."

24.      The application was brought by the plaintiffs because I had already made an order for discovery and therefore what the plaintiffs were seeking was a variation of that order.  Accordingly, it was appropriate for the plaintiffs to bring an application.

Submissions

25.      Advocate Speck for the plaintiffs in his extensive written submissions supplemented at the hearing made the following points.

26.      Firstly, he referred me to paragraph 6.6 of the original discovery protocol where Mourant had set out that they were not in possession of all relevant data from each of the data sources.  They had therefore noted that it might be necessary for the search terms to be revised if excessive documents were returned for review.

27.      What had led to the present application was that the data sources turned out to be larger than anticipated.  The plaintiffs' skeleton at paragraph 14 suggested the amount of data recovered was 709 gigabits which in correspondence was described as being four times the amount anticipated.  The skeleton further explained that, during the first month allowed for discovery, problems were not anticipated about the amount of data to be collected.

28.      Once data had been extracted from the first plaintiff, applying the search terms as ordered (subject to some minor modifications made to take into account observations by the defendant) led to the identification of  some 1.5 million relevant documents.  This figure was arrived at by a search term report from the plaintiffs' e-discovery provider.  The plaintiffs regarded this as excessive and therefore sought to revise the search terms used.  The proposed revised search terms produced from the first plaintiff and two other sources if permitted would produce 92,165 potentially relevant documents.

29.      How the court should approach the application should be determined by the application of the overriding objective and the detail guidance in Practice Directions RC17/07 and RC17/08 as to what was a reasonable search for documents.

30.      In relation to why the revised protocol was justified, the plaintiffs considered it highly unlikely that there would be over 1.5 million documents potentially relevant to the matters in issue given the issues in dispute.  As this figure related only to the first plaintiff's data, apart from its Chief Executive Mr Crosby who was dealt with separately, the total pool was likely to be larger.  Whatever the final size of the pool of documents, using existing search terms meant it would obviously be impossible to review 1.5 million documents or indeed any larger pool of documents within 10 weeks.

31.      The plaintiffs contrasted the number of documents it was being provided to review with those that the defendant was required to review, which amounted to around 10,000 relevant documents.

32.      Part of the problem the plaintiffs faced was due to the use of generic terms or search terms that threw up too many irrelevant documents.  The plaintiffs' approach was therefore to take into account the relevance or importance of each of the terms in the context of the pleadings.  This meant that search terms in relation to tax were not modified.  However, to address the issue of documents being returned that concerned unrelated or irrelevant trusts, the terms ("Busson or BFT or AB or Arki or Arpad or Andre or Beaulieu or Albion or Penn or Tabalke) were added to the vast majority of the search terms.  This produced much more focussed responses.

33.      In relation to data received from the second plaintiff and his business known as LumX, the protocol was further amended again to avoid an excessive number of documents.  In relation to the second plaintiff, his data had only been provided on 19^th November 2021 and so it was still not possible to identify how many documents the review would produce.  However, the plaintiffs considered that if the modifications made to the first plaintiff's documents were applied to the second plaintiff an excessive number of documents would not be returned.

34.      Advocate Speck further explained that Miss Katrina Phillips, since the original discovery protocol was provided, had refused to provide access to emails on her personal email account.

35.      He also explained that in September 2021 Saffron Tax had informed Mourant  that all their files and records for the second plaintiff had been archived years earlier and had since been destroyed as a consequence.

36.      Advocate Speck accepted that the plaintiffs' affidavit of discovery should provide an explanation of these matters. 

37.      By the time the skeleton arguments were exchanged the total number of documents that had been produced as potentially relevant applying the revised search terms totalled some 190,000 documents.  This figure reduced to 155,000 documents by addressing duplicates which the plaintiffs argued was a more propionate number.

38.      The plaintiffs criticised the defendant for not engaging with the plaintiffs in relation to this revised proposed proposal although it had been provided on 26^th October 2021.

39.      Without modifying the plaintiffs' discovery protocol, it was likely that the plaintiffs would have to review over 2 million documents.

40.      To the extent that the defendant had made suggestions to limit the scope of searches required to be carried out, these did not work because they still led to 1.5 million documents having to be reviewed.  The defendant's approach, which suggested only starting searches in 2011, did not significantly alter the number of responsive documents that were produced.  By contrast, maintaining the earlier search dates but applying the revised search terms did produce a much more proportionate response.  This confirmed that the cause of a large number of returns was the search terms as used, not the period being searched.

41.      If the plaintiffs were required to review documents either based on the original proposal or the defendant's revised protocol this would cost somewhere between and £2 and £3 million for a discovery exercise just to determine relevance.  There would also be further costs with the e-discovery provider and subsequent reviews.

42.      In relation to paragraph 3 of the plaintiffs' summons, the plaintiffs argued that when reviewing documents, the plaintiffs' e-discovery provider was using a technology known as "Brainspace".  The combination of analysis of documents produced as being relevant and the technology within "Brainspace" (applying artificial intelligence to learn from reviews what is relevant) allowed Brainspace" to analyse the 150,000 documents identified to determine based on a predictive relevance rating which of those documents were not likely to be relevant.  The plaintiffs wished to use a relevance rating of 0.03 so that any documents with a predictive relevance of 0.03 and 0.2 would remain to be reviewed.  This would further reduce the number of documents to be analysed by another 100,000 documents.

43.      In relation to the 100,000 documents that Brainspace would determine as being irrelevant, the plaintiffs then proposed to take a random sample of 3% of these documents (i.e. approximately 2000) as a quality control measure to ensure that the plaintiffs could safely on rely on Brainspace's predictive rating.

44.      The plaintiffs did not object to certain minor modifications to the search terms that I proposed where there appeared to be inconsistency on the face of the document that led to certain required changes.

45.      Advocate Cadin for the defendant contended that the application should be refused.

46.      He was particularly critical of the plaintiffs' approach for a number of reasons.  Firstly, he complained that as at 16^th November 2021, the second plaintiff's data had not yet been retrieved from his personal account.  Yet this was in respect of litigation that had commenced in 2015 and when the second plaintiff was under an obligation to file documentation with HMRC in relation to his challenge to their tax assessments by the end of October.  Advocate Cadin suggested it suited the plaintiffs to delay discovery in Jersey so that embarrassing or difficult documents would not have been produced and therefore did not have to be dealt with in the second plaintiff's submissions to HMRC.

47.      Advocate Cadin also criticised the plaintiffs for the lack of any explanation as to why the current situation had arisen, without any evidence or any apology.

48.      This was part of a wider pattern of the plaintiffs have been reluctant to provide the defendant with documents for information about the second plaintiff.

49.      The defendant also criticised the plaintiffs for pursuing an expensive discovery protocol and yet only raising the current problems late in the day.  This led the defendant to accuse the plaintiffs of playing games.

50.      The defendant's central complaint was that the plaintiffs' approach would leave undisclosed relevant documents in the sole possession of the plaintiffs.

51.      The defendant was also critical of the plaintiffs for varying the search terms for other data sources when the problem related to the first plaintiff only.  The defendant in particular complained that these revised search terms were being applied to data that had not yet been extracted.

52.      According to the defendant, its e-discovery provider discouraged a sole focus on the volume of data collected by applying key words alone and suggested instead that a technology assisted review be applied to the larger set.  I was also informed that the e-discovery provider considered that it was highly unusual for search terms to be eliminated in the manner proposed.

53.      In relation to the use of "Brainspace" this had not been raised previously and was contradicted by the discovery protocol advanced by the plaintiffs on 23^rd August 2021 when they stated they did not intend to use a technology assisted review.  The defendant suggested that the court could not have any confidence that the plaintiffs' approach would lead to relevant documents being produced.

54.      The wider context of the defendant's criticisms was that a discovery protocol was not a discussion document.  Once ordered that document existed for a reason.  The defendant in particular emphasised that endorsing the plaintiffs' approach ran the risks of the court's time and parties' money being wasted, that directions hearings relating to the scope of discovery might otherwise become pointless, and that orders made by the court would be nothing more than expressions of hopeful expectation.

55.      All the issues the defendant was raising now should have been considered in August.  What was required instead was an unless order.

56.      Finally, the defendant was critical of the plaintiffs for only providing some 155 pages of detailed analysis of the proposed changes two days before skeletons were due to be exchanged.  The defendant therefore had no opportunity to respond to this information.

Decision

57.      In relation to the decision I reached, there was significant force to the defendant's criticisms of the plaintiffs.

58.      Firstly, it was not explained to me why the issues now identified could not have been identified as part of the production of the discovery protocol leading to the hearing in August this year.  As noted in the chronology above, issues about discovery were being explored from 2019 onwards.  By the time of the directions hearing on 23^rd August 2021 the plaintiffs should therefore have identified how much data was going to be involved in order that a realistic timetable could have been set.  As the plaintiffs had significantly underestimated how much data had to be extracted for review, this meant that in effect the analysis for the hearing that took place on 23rd August 2021 has had to be repeated.  Although the plaintiffs reserved their position that the amount of data collected might change, this reservation was not sufficient to account for why more than 4 times the amount of data envisaged had now been extracted using the plaintiffs' original search terms.  This was also data held by the first plaintiff only which appeared to have been analysed in some manner albeit unexplained.

59.      At the very least, even if the plaintiffs had not identified how much data was to be extracted by the date I made a discovery order, I would have expected the data to have been extracted immediately after the order was made on 23^rd August.  Yet there was no evidence before me explaining why the present difficulty had arisen so late in the day.  I did not receive any affidavit in support explaining when searches had been carried out.  Nor was there any affidavit from any e-discovery provider.  In particular it was not explained to me when data was extracted and why the problem was only appreciated shortly before the 12^th October 2021 even though I made a discovery order on 23^rd August 2021.

60.      Secondly, in relation to what approach I should take, I was only provided with skeleton arguments relaying the e-discovery providers' suggestions.  This was wholly inadequate.  In future on any issues about what approach to take and whether or not a technology assisted review should be carried out or search terms applied and/or any combination of these approaches, where that is in issue as it is in this case, then evidence should be provided by such an e-discovery provider. 

61.      While I understand there may have been a concern about a party waiving privilege by its e-discovery provider providing evidence, such concerns are misplaced.  Such e-discovery providers are effectively experts assisting the court.  While they are assisting the court in respect of a procedural dispute, as distinct from a substantive issue at trial, I consider that the same obligations should apply in terms of their duties to the court.  In relation to privilege, reports can therefore be drafted in such a way that privilege is not waived but the questions the expert has been asked to address are set out.  As this did not occur, I was not able to rely on or take into account any submissions based on observations from the e-discovery providers when those observations were not before me as evidence.

62.      I wish to add that this criticism applies to both parties.  While the plaintiffs should have produced a report from their e-discovery provider to set out their e-discovery provider's view as to why the plaintiffs' proposed approach was proportionate, the defendant in arguing for use of technology only did not produce a report either.

63.      Returning to the question of data extraction, I was also concerned about the lateness of the extraction of data from the second plaintiff which had only been extracted a day or two before the skeleton arguments were to be filed seeking the variation.  There was no evidence again before me explaining why data had not been extracted earlier.  Yet there is an obligation to preserve data once a party is on notice of a dispute.  If data is only being extracted years after proceedings have started, the late extraction of data at its lowest raises concerns whether all relevant data has been secured.  This may become relevant if documents used to exist and can no longer be traced.  It is also of concern when set against the other concerns expressed in this judgment.

64.      The statements that Saffron Tax have had destroyed their records concerning the second plaintiff and that Miss Phillips was no longer willing to provide access to relevant emails on her personal email account were also troubling, both in themselves and again when set in in the wider context of the other criticisms set out in this judgment.

65.      The difficult position I was therefore left with was that requiring the plaintiffs to carry out a review of 1.5 million documents was clearly disproportionate.  From my general experience I am aware that it takes one person a day to review around 500 documents.  Reviewing 1.5 million documents would therefore take one individual 3000 working days, which based on 250 working days per year is around 12 years.  While clearly teams of individuals are required to carry out such reviews the above analysis illustrates why the current search terms needed to be modified.  It would take any team of reviewers many months and at a substantial cost to review 1.5 million documents.  In this case there are also only two main factual issues namely the acquisition of one property in three tranches and some payments into the trusts by third parties.  The factual background is not therefore overly complex and so does not require an expensive review of a large pool of documents.

66.      The approach I therefore took was to accept the plaintiffs' modifications to limit the more generic words that could on their own produce irrelevant results by the addition of search terms linking the more generic words to references to the second plaintiff or various names he or family members used or the names of relevant trusts.  I was able to take this approach because the defendant in its e-discovery protocol had applied very similar core key words to be used across the entire set of electronic documents.  Those core key words are very similar to those words the plaintiffs now propose to link to its remaining search terms.  The size of the smaller pool given what is in dispute is still large enough to lead to most if not all relevant documents being identified.

67.      I reached this view, however, only in relation to documents recovered from the first plaintiff and the tax advisers involved.  I did not reach the same view for documents held by the second plaintiff.

68.      In relation to the second plaintiff, I agreed with the defendant that it was appropriate for the second plaintiff to use whatever technology assisted review he is advised is appropriate with relevant detail being set out in the second plaintiff's affidavit of discovery.

69.      I reached this view for the following reasons: -

(i)        the first plaintiff's documents were those held either by the defendant or were created by successor trustees.  They are therefore documents held by regulated businesses who are all under obligations to preserve relevant records and in my experience have systems in place to enable them to do so.  This gives confidence that the revised search terms will identify any documents relevant to this dispute. By contrast I do not know how documents were held by the second plaintiff or LumX and whether they were held logically to enable any review of documents produced using search terms to ensure that nothing material was missing or had been overlooked.

(ii)       It was concerning that only a day or two before the applications made by the plaintiffs to vary the search terms used, the second plaintiff's data had not yet been extracted.  For a dispute that began in 2015 and when the defendant had been complaining about being kept in the dark, I concluded that the overriding objective including dealing with cases justly and ensuring that parties are on equal footing, meant that I could not accede to the plaintiffs' request to modify search terms for documents held by or for the second plaintiff.  I concluded it was important for the defendant to have confidence that relevant documents had been looked for in the most effective manner possible to ensure that they were identified and had been looked for.  While it is not proportionate for the second plaintiff to have to review 300,000 documents manually, the answer to the difficulties with a manual review of this number of documents is for the second plaintiff to use technology to determine relevance and to explain the process followed.

70.      In reaching this conclusion I do not accept that the conduct of the second plaintiff was deliberate.  I was able to reach this conclusion because Advocate Speck assured me that he and his firm are heavily involved in the discovery process.  What has occurred therefore in my judgment is not a deliberate conspiracy to delay the discovery process, but rather very serious failings which should have been addressed earlier and which now, as set out above, are only explained in part.

71.      In relation to the specific request to use "Brainspace" to determine relevance, to exclude around 100,000 documents and to test the same using a sample review only, I refused this part of the application.  I simply did not have any evidence to determine whether this was a proportionate approach.  While I am aware of "Brainspace" and I have seen it demonstrated, there is a world of difference between understanding generally how a product might work and drawing a conclusion to say that significant categories of documents which have been determined as relevant using search terms need not be reviewed.  A much more detailed analysis is required to arrive at such a position.  The difficulties in trying to limit discovery in the manner proposed have already caused problems for the court (see CMC & Anor v Forster & Ors [2019] JRC 202); it was not appropriate to try and revisit that type of approach without evidence in support.

72.      Finally, in terms of costs, I ordered the plaintiffs to pay the costs of and occasioned by the application on an indemnity basis.  I reached this view firstly because this application could have been avoided, had the plaintiffs identified prior to the August hearing how much data they were dealing with and what sought of response search terms might have produced.  Their failure to approach discovery until now in the manner expected has necessitated this application.  The failure to grapple with what was required in the manner now sought has also delayed the progress of this case to trial.  What applications of this kind require as noted at paragraph 21 of Practice Direction RC 17/08 is co-operation.  The problems I had to grapple with were also only identified 6 weeks after the order was made and revised search terms were only provided on 26^th October 2021 i.e. 3 days before discovery was due to be provided.  The detailed analysis underpinning why revised search terms were required was only provided 2 days before skeletons were due to be exchanged.  None of this was at all satisfactory and justified an indemnity costs order to express my displeasure at the delays and failings that have occurred.

73.      For these reasons the application to extend time was granted as was permission to revise the search terms as far as the first plaintiff was concerned.  The application to revise search terms in relation to the second plaintiff was refused but permission was given to the second plaintiff to determine relevance by use of a technology assisted review provided that such matters were explained in the second plaintiff's affidavit of discovery.  The application to limit reviews of documents identified as relevant by the first plaintiff using technology was also refused.

Authorities

Practice Direction RC17/08.

CMC & Anor v Forster & Ors [2019] JRC 202.