Disclosure is an essential part of litigation and arbitration under English law as it usually provides both parties with access to the contemporaneous documents which support or adversely affect a party’s case. The exponential growth in recent years of the number of electronic documents created during the course of a project has increased the size and, as a result, the cost of the disclosure exercise. Technology has developed to assist with the disclosure exercise, and a decision in the High Court in 2016 has approved the use of such technology thereby providing parties with ways to limit the cost of electronic disclosure. 

For anyone who has been involved in a large disclosure exercise, including the clients who foot the bill, it should come as welcome news that the Lord Chief Justice, Rt. Hon. Lord Thomas of Cwmgiedd, indicated in his report to the UK Parliament in November 20161 that the judiciary is working with lawyers to consider the best approach to electronic disclosure, particularly in light of the first decision allowing the use of predictive-coding-assisted disclosure, Pyrrho Investments Ltd and anor v. MWB Property Ltd & Ors2 (“Pyrrho”), and that further amendments to Civil Procedure Rules (“CPR”) Part 31 are being considered. At this stage, the extent of the possible changes is unclear as Lord Justice Jackson, in his speech at the Law Society of England and Wales’ Commercial Litigation Conference a month earlier3 , indicated that he does not recommend any further reforms of the CPR but suggests greater use of the existing rules.

Use of technology

Technology and indeed a whole industry surrounding it have developed to facilitate the process of disclosure in litigation and arbitration. The process includes identifying, preserving and collecting documents, processing and reducing the pool of documents, review and analysis and finally exchange and inspection.

The use of technology to carry out most stages of the disclosure process has been common over the past 10 years and indeed recognised and encouraged by the Judiciary. Practice Direction, PD31B on the Disclosure of Electronic Documents to the CPR was introduced on 1 October 2010 to encourage and assist the parties to reach agreement in relation to the disclosure of electronic documents in a proportionate and cost-effective manner. PD31B indicates that it may be reasonable to use keyword searches or other automated methods of searching if a full review of each and every document would be unreasonable, but adds that this will often be insufficient and the parties should consider supplementing the search with other techniques.

In line with PD31B, the approach to the review and analysis stage in documentheavy cases has often been to employ a team of junior lawyers to manually review documents returned from searches across the pool of documents between specific date ranges for certain keywords, custodians and domains. However, such a review is often time-consuming and expensive. If a junior laywer reviews 800 documents a day on average, it will take approximately 375 working days to review 300,000 documents. When you consider that the court’s definition of a document includes emails, texts, instant messages, voicemails, photos and videos as well as traditional electronic and hard copy documents, an average shipbuilding or other offshore construction project can easily produce far in excess of this number of documents.

Methods of technology-assisted review have therefore been developed to speed up the process and reduce the costs. They include predictive coding, email threading, near de-duplication, conceptual searching and clustering. An eDisclosure Protocol for the Technology and Construction Court introduced in November 2013 (the latest version was published on 9 January 20154 ), contemplates the use of predictive coding in appropriate cases, but its status as a protocol has no normative force. In the circumstances many lawyers and indeed judges remain sceptical of technology-assisted review but the English High Court gave the use of predictive coding a boost by approving its use in the Pyrrho case.

Pyrrho and predictive coding

Pyrrho, credited as being the first Greek sceptic philosopher, seems like an ironic name for a case which will hopefully allay some of the concerns of clients and lawyers sceptical about the use of predictive coding to review documents.

The Pyrrho judgment arose out of an application seeking the court’s approval of the use of predictive coding in the disclosure process in a matter concerning various alleged breaches of fiduciary duty by the former directors of a company in relation to loans and payments made by the company over a number of years. In this case, the original pool of documents was more than 17.6 million, and it was reduced by de-duplication to some 3.1 million documents.

In brief, predictive coding involves a lawyer reviewing samples of documents and identifying those which are relevant to the dispute (or whichever would fall within the scope of disclosure ordered by the court) thereby training the computer software to do the same. The software then applies the logic to the complete pool of documents and suggests which documents are relevant. The process is repeated until the legal team is happy with the final pool of documents.

The judgment approved the use of predictive coding and gave ten reasons for it including evidence that such a review would be more accurate, more consistent, cheaper and proportionate to the value of the claim than a manual review, and that the timetable would allow for alternative methods to be employed if predictive coding produced unsatisfactory results.

Shortly after this judgment, the English High Court made an order in a separate case5 permitting a party to use predictive coding against the opposing party’s wishes, demonstrating that predictive coding is more than just an alternative to manual review. In that case it was estimated that a review based on predictive coding would cost £120,000, whereas it would be nearer £250,000 if junior lawyers carried out a manual review based on keyword searches.

Our experience and comment

At Haynes and Boone CDG, we regularly employ technology to assist our clients with the collection, processing, review, analysis and production of documents for the purposes of disclosure. Indeed, given the document-heavy nature of shipbuilding and offshore construction disputes whether in court or before an arbitral tribunal, we doubt that we would be able to manage the process efficiently in many cases without it.

Whilst the English Court decisions supporting the use of predictive coding are clearly a very positive development for those involved in large disclosure exercises and a key to keeping costs down, we would also welcome confirmation that the use of other forms of technology-assisted review such as email threading, near de-duplication, conceptual searching and clustering were also the subject of judicial approval.

We therefore look forward to the developments in this area including any amendments to CPR 31 in order that we can undertake the disclosure exercise in the most cost efficient but effective way for our clients.