The use of predictive coding to assist with disclosure in English proceedings

On 17 May 2016, the English High Court took a significant step in the move to accept the use of predictive coding in English civil litigation proceedings. In the case of David Brown v BCA Trading, the Court has approved the use of predictive coding by the Defendant despite this being strongly objected to by the Claimant.

English litigation involves a process known as “disclosure” by which all parties are obliged to locate all documents relevant to the proceedings in their possession and provide them (subject to a number of exceptions) to their opponents. As the use of modern technology in businesses has exponentially increased over recent decades, so too has the number of documents (mostly electronic data) which must be collated, reviewed and processed. As a result it is not unusual for millions of documents to have to be manually reviewed as part of the disclosure process. The most visible practical consequence of this increase in electronic data has been a significant increase in the costs involved in carrying out this process.

English law practitioners have grappled with how best to overcome the tension between ensuring that the process is properly carried out (and therefore that the disclosure obligations are properly fulfilled) and ensuring that the costs of doing so do not spiral out of control.

US practitioners (who faced a similar but not identical problem) developed what is known as “predictive coding” as a response to this tension. Predictive coding involves employing complex algorithms to apply the results of a sample review across a much larger data set. Using this process means that senior lawyers can consider a sample of electronic documents and based on their decision process, millions of documents can then be automatically reviewed and the relevant documents located by the software. Although use of predictive coding software is not cheap, proponents of predictive coding have heralded both the significantly reduced costs of the process and the fact that the risk of human error and inconsistency is vastly reduced.

Predictive coding did not however receive a warm welcome in England. Whilst practitioners have sometimes been willing to use this method as a complement to the manual process there has been significant scepticism to relying solely on an automatic process. This reluctance stemmed in the most part from a concern that the strict obligations imposed by English procedural law to disclose all relevant documents in one’s possession would be weakened and that as a result one of the key features of civil litigation before the English Courts would be diluted.

In a landmark ruling earlier this year (Pyrrho Investments Ltd v MWB Property Ltd, [2016] EWHC 256 (Ch)), the High Court approved the use of predictive coding during the disclosure review process for the first time. This was a case involving the review of more than 3 million documents and both parties expressly agreed that predictive coding should be used. The High Court was asked by the parties to approve the use of predictive coding. The Court agreed to do so in this instance but in doing so, Master Matthews set out a number of factors as to why predictive coding could be adopted and made it expressly clear that whether or not predictive coding would be suitable in other cases would depend on the precise circumstances of each case.

In the most recent case of David Brown v BCA Trading the High Court had to decide whether predictive coding would be appropriate in circumstances where one of the parties expressly objected to its use. The Claimant’s lawyers argued that predictive coding should not be used since the accuracy of the automatic process could not be trusted and there was a serious concern that if the process was used, relevant documents would be missed. The Defendant’s lawyers argued that the automated process was more reliable than a manual review and relied heavily on the fact that the costs of a manual review would be more than double those if predictive coding was used. In what may be seen as an historic ruling, the Court agreed with the Defendant and ruled that it was permitted to use predictive coding in order to carry out the disclosure review.

It is likely that as a result of this most recent decision, the use of predictive coding will become more widespread, although scepticism as to its benefits is likely to remain for some time. Experts are continuing to refine the process with a view to continually improving accuracy which is likely to allay some of these fears. However, the true success (or failure) of using predictive coding in English litigation is only likely to become apparent as these initial cases progress. In time predictive coding may well prove to be the answer to streamlining the process of disclosure and making it more affordable and accurate, which can only be a good thing for both claimants and defendants involved in English litigation. Decisions such as these are also likely to support the increased use of predictive coding in internal investigations.