2 September 2016
Disclosure is often the most expensive part of the litigation process and document review is often the most expensive part of disclosure. Two cases this year have given strong support to the use of technology assisted review.
The first was the Judgment given by Master Matthews in Pyrrho Investments v MWB Property  EWHC 256 (Ch) where he gave permission to the parties to use predictive coding in support of the review exercise. Orders permitting the use of predictive coding had previously been made in the US and, more recently, in Ireland, but there had been no published decisions in the English courts.
In summary, predictive coding is a process whereby computer software is trained to identify relevant documents for disclosure and then conducts the review process instead of humans. Initially, a lawyer, who is likely to be a senior lawyer, will review a small subset of the overall universe of documents potentially reviewable for disclosure and the results of that review will be fed into the relevant proprietary computer software. The process will be iterative and subject to an agreed protocol between the parties. When, after a number of such iterations, the program demonstrates the agreed level of accuracy, the rest of the documents will be reviewed using the program without, save for certain quality control tests, human intervention.
Master Matthews identified a list of factors which he said supported the use of predictive coding, including favourable experience of the use of such tools in other jurisdictions in terms of the accuracy and consistency of disclosure as well as the beneficial impact on costs. As to the latter, in the Pyrrho case, the costs of human review were estimated to be least several million pounds. On the other hand, the cost of using the predictive coding software was estimated as somewhere between £180,000-£470,000 plus monthly hosting costs of £15,000-£20,000.
The significance of the second case, Brown v BCA Trading  EWHC 1464 (Ch), was that, unlike in the Pyrrho case, the application to use predictive coding (by BCA, the respondent) was opposed by the petitioner. However, the Registrar allowed the application, citing favourably the factors identified in Pyrrho.
While these cases support the use of technology assisted review, in particular predictive coding, they indicate that such decisions must be made on a case by case basis, assessing whether the features of the particular case make such an approach appropriate. In Brown v BCA the Registrar stressed the importance, where disclosure was potentially extensive, of the parties' lawyers identifying the "true issues" in the case and the anticipated categories of documents relevant to those issues, and entering into discussions with a view to minimising the burden of disclosure.