The disclosure of documents is an integral part of litigation in England and Wales and is governed in the practical sense by Practice Direction 31B with which parties must comply when conducting litigation. Whilst Practice Direction 31B makes provision for the extent of the search for electronic documents, it fails to specify the method for that search. As a result, litigating parties tend to be reluctant to fully commit to the use of predictive coding technology to assist with the review of electronic documents, instead preferring the traditional approach of linear review.1
Change may, however, be in the air following the decision of the High Court in Pyrrho Investments Limited and another v MWB Property Limited and others  EWHC 256 (Ch) in February 2016. Master Matthews broke new ground in Pyrrho Investments by not only considering for the first time the method of search for electronic documents, but also endorsing the use of predictive coding software for the disclosure of electronic documents with only a limited manual review of the results.
The purpose of this 57th issue of Insight is to:
- provide a brief overview of electronic disclosure and predictive coding;
- review the decision in Pyrrho Investments Limited and another v MWB Property Limited and others  EWHC 256 (Ch) and the practice points arising; and
- consider the future of electronic disclosure in the construction law context following the court’s decision in Pyrrho Investments.
The obligation to search for documents
Parties who are ordered to give standard disclosure are under an obligation to make a reasonable search for documents (including electronic documents) under CPR Part 31. The Practice Direction to CPR Part 31, PD 31B, states that the search which must be made will depend upon the circumstances of the case. The factors that may be relevant in deciding the reasonableness of a search for electronic documents include (but are not limited to): the number of documents involved; the nature and complexity of the proceedings; the ease and expense of the retrieval of any document; and the significance of any document that is likely to be located during the search. The parties are also obliged to bear in mind that the overriding objective at CPR Part 1 includes dealing with the case in ways that are “proportionate”.
Notably, other than a reference in PD 31B to the use of keyword searches or “other automated methods of searching if a full review of each and every document would be unreasonable”, there is no guidance whatsoever as to how the search and review of electronic documents should be conducted. The TeCSA/SCL/TECBAR Protocol makes a reference to the use of automated or computer-assisted review tools at paragraph 5.3 but its status as a protocol means that it has no legal force (albeit the parties may be ordered to comply with it in default of proposing a suitable alternative).
Electronic disclosure is a term that has been adopted to refer to the disclosure of electronic documents and has become increasingly important over recent years in light of the fact that around 90% of daily business communication is now electronic.
Over the past few years, the court has put specific measures in place to deal with the disclosure of electronic documents in the form of a Menu Option at CPR 31.5. Under the Menu Option, the court has the power (amongst other things) to make any order in relation to the disclosure of electronic documents that it considers appropriate and the court therefore has complete discretion in relation to eDisclosure.
Predictive coding is a document review undertaken by proprietary computer software as opposed to human beings, and was created by eDisclosure providers as a potential solution to the time and cost associated with linear review.
Typically, the parties agree a protocol, after which a representative sample data set of potentially relevant documents is obtained and reviewed for relevance by lawyers who are familiar with the issues in the case and who program the software to review the whole document set. The document review software then applies complex algorithms to the lawyers’ review to suggest similar documents for review, and employs search technology to “score” documents for relevance having regard to the issues in the case, and to predict the degree of relevance of the remaining documents in the data set. Documents that achieve a sufficient score can then either be immediately disclosed, or put through a traditional manual review.
The decision in Pyrrho Investments
Pyrrho Investments was a multimillion- pound case that arose out of allegations of payments that were made in breach of fiduciary duty. Disclosure posed a particular problem as even after deduplication was performed and search terms were run, there were still in excess of 3.1 million relevant documents that remained to be reviewed. Following extensive correspondence, the parties agreed to the use of predictive coding in principle, subject to the approval of the court. The court’s endorsement was considered necessary by the parties for two reasons: first, because the method of predictive coding that was contemplated meant that not all the documents disclosed by one of the parties would be reviewed by its legal team prior to disclosure, and secondly, because there was no English law authority specifically endorsing the use of predictive coding.
Master Matthews in the Chancery Division of the High Court referred to the decision of the US Federal Court in Moore v Publicis Groupe 11 Civ 1279 (ALC)(AJP) and the decision of the Irish High Court in Irish Bank Resolution Corporation Ltd v Quinn  IEHC 174, both of which provide helpful commentary on and approval of the use of predictive coding in those jurisdictions.
The Master approved the use of predictive coding having regard to the facts of the case and because experience in other jurisdictions suggested that predictive coding was useful in “appropriate” cases. Master Matthews found there was nothing to suggest that predictive coding was less reliable than manual review alone, or keyword searches and manual review combined. He also noted that predictive coding can be superior to a linear review as using a computer to apply the approach of a senior lawyer towards the initial sample to the entire document allows for far greater consistency than using a number of lower-grade fee–earners each taking an independent view as to which documents are relevant to which issue. Predictive coding therefore allows for electronic documents to be reviewed at proportionate cost in circumstances where manual review would be much more time-consuming, and probably less reliable.
The Master further pointed out that there was nothing in the Civil Procedure Rules or their associated Practice Directions prohibiting the use of predictive coding; the parties had agreed its use, and the trial was some way off so there was scope for other review methods to be used if necessary. Finally, Master Matthews was of the view that there were no factors pointing against the use of predictive coding in the case.
- The costs of predictive coding (and electronic disclosure in general) are usually considerably less than an equivalent manual review. In Pyrrho Investments, the estimates given for the use of predictive coding in the case ranged from nearly £182,000 plus monthly hosting costs of £15,700, to £468,000 plus monthly hosting costs of nearly £21,000. Master Matthews remarked that this was obviously far less expensive than the full manual alternative which would likely be in the region of several millions of pounds at least.
- Unlike human review, the cost of predictive coding does not increase proportionally to the number of documents, which makes it a costeffective and proportionate method of reviewing a large number of electronic documents.
- The approval of the Master is not currently a prerequisite for a party’s use of predictive coding software, but whether predictive coding will be expressly approved by the court depends upon the particular circumstances of the case.
- Following the reasoning of Master Matthews in Pyrrho Investments, as a general rule, predictive coding may be regarded as being suitable by the court if the quantum is in excess of £3 million, in which case predictive coding would result in a substantial cost-saving and the costs of using the technology would be proportionate to the value of the claim.
The future of electronic disclosure
Despite the availability and merits of predictive coding technology, litigating parties have historically been reticent to use predictive coding because it was not defensible, and the question of whether the English courts would accept predictive coding as a valid electronic disclosure tool had not been answered prior to the decision in Pyrrho Investments. Whether the court’s decision will herald the birth of a new standard order for cases that are suitable for predictive coding remains to be seen, but the implication at first instance is that predictive coding is here to stay. At the very least, Master Matthews’ decision confirms the benefits and use of predictive coding in appropriate cases in England and Wales, and the decision (at its highest) may lead to a body of English judicial authority supporting its use.
Whether predictive coding is appropriate for complex construction disputes which invariably involve numerous technical issues that may develop over time and can be interrelated, is much less certain. It would be almost impossible, for example, to teach the software how drawings, spreadsheets of data, and photographs relate to the issues in dispute as these documents are held in files that the algorithms cannot read.
Whether predictive coding software will find acceptable solutions to these problems in the future remains to be seen.