knowledge 12 April 2018 |

UK Computer Assisted Review Decision likely to Influence Irish Courts

Triumph Controls UK Ltd v Primus International Holding Co¹ was a multi-million dollar breach of warranty claim which arose following the sale of the defendants’ aerospace business to the claimants. The defendants were dissatisfied with the claimants’ disclosure and sought production of a list of 860,000 folders and file paths which the claimants had identified on a shared drive so that the defendants could check if there were any files which should have been searched but which were not. The defendants also sought an order that the claimants undertake a manual review of 220,000 particular documents.  Using agreed keywords, 450,000 documents had originally been identified as potentially disclosable. However, the claimants subsequently, without agreement, decided not to manually review 220,000 of the 450,000 documents.

The Decision

The court was satisfied that the claimants’ approach to the shared drive was reasonable and proportionate in circumstances where only two further folders or file paths were subsequently disclosed and the defendants were unable to identify any additional missing folders or file paths.

However, the court expressed “greater concerns” on the claimants’ failure to manually review the 220,000 keyword responsive documents. In their Electronic Documents Questionnaire, the claimants had agreed to manually review all 450,000 of the documents. However, they unilaterally decided that it would be disproportionate to review 220,000 of the documents. This was based on a 1% sample of those particular documents using Computer Assisted Review (“CAR”)² which indicated that only 0.38% of them would be relevant. However, the claimants gave no information on how they carried out this sampling exercise.

The court emphasised that a senior lawyer had not overseen the CAR process, referring to to Pyrrho Investments Ltd v MWB Property Ltd ³ which set guidelines for training the CAR technology to correctly identify disclosable documents.  In Pyrrho the court stated that “the best practice would be for a single, senior lawyer who has mastered the issues in the case to consider the whole sample”.  Here, instead of a senior lawyer training the system, ten paralegals and four associates had trained the system. The court considered that this meant that the system might not be as well “educated” as it might have been given the number of those involved, particularly in respect of the criteria of relevance.

The court concluded that both the claimants’ CAR and the sampling exercise were not transparent and independently verifiable. It was reasonable and proportionate to require them to take further steps and having regard to the costs involved and an imminent trial date, the parties should agree a methodology for a sample of 25% of the 220,000 documents to be manually reviewed within three weeks. The results should be put into an agreed letter for the court.

Irish Position

This judgment will be of interest for litigants before the Irish courts where the need for transparency in the CAR (or TAR) process and consultation between the parties has been recognised⁴.   In any case involving TAR, the methodology for making discovery must be “transparent and reliable” and the court is likely to require expert evidence regarding any statistical sampling in the event of a dispute. The involvement of experienced lawyers in the TAR process and consultation between the parties in relation to the methodology adopted are key requirements, bearing in mind that as the Court of Appeal stated on appeal in IBRC –v- Quinn⁵, the plaintiffs' solicitors McCann FitzGerald had correctly confirmed on affidavit that the use of TAR did not in any way derogate from the obligation to make discovery in accordance with Order 31 Rule 12 and that it was clearly being done in accordance with the obligations of clients in conjunction with their solicitors to swear affidavits in accordance with the Rules of the Superior Courts.

Triumph Controls shows the importance of early engagement between the parties prior to making discovery.  While the UK protocols do not apply in Ireland such engagement is recommended by the CLAI Good Practice Discovery Guide V2.0 and can lessen the risk of challenges to discovery later on.