Embracing predictive â€¨coding for e-disclosure
Gavin Foggo and Shella Khan discuss two recent cases in which the High Court has stood firm on the use of this technology, and advise on how it can be implemented by firms
also referred to as technology assisted review (TAR), is here to stay. This is the process of using a senior lawyer to review a small set of disclosure documents, which is then analysed by the predictive coding software, together with several rounds
of refinement, to produce an algorithm.
This is then applied across
the whole database of disclosure documents to produce a sensible number of documents for manual review. Advocates
for its use argue that it is more accurate than a human review undertaken by armies of paralegals, and that it saves
time and costs.It is clear from two recent decisions - Pyrrho Investments Ltd and another v MWB Property Ltd and others  EWHC 256 (Ch) and the contested case of David Brown v BCA Trading (17 May 2016, unreported) - that courts are attracted to the cost-saving element of the technology and are increasingly sceptical of protestations with regard to its perceived inaccuracies.
In February 2016, Master Matthews approved the use of predictive coding in Pyrrho. That case involved a large-scale disclosure exercise of approximately three million documents.
Matthews applied the
Jackson principles of balancing proportionality and cost management, deciding that
the use of the software was appropriate in this case. He set out ten factors in favour of its use, and in this case the majority were satisfied. However, he warned that the use of the technology should not be taken as a broad-brush approach to disclosure; it depends on the individual nature of each case. An important aspect of Pyrrho was that all parties agreed to use predictive coding.
More recently, in Brown,
the claimant argued against
the use of predictive coding for disclosure involving 500,000 documents, citing serious concerns as to the accuracy of the technology. The defendant disagreed and estimated the costs of a manual review to be more than twice those with predictive coding. Mr Registrar Jones, sitting in the case management conference
(CMC) at the Companies Court, approved its use, affirming the Pyrrho decision. He agreed
with the respondents that the technology is more accurate
and less expensive than a manual review.
Predictive coding is already widely used in the US, following the decision in Da Silva Moore v Publicis Group in 2012, approved in Rio Tinto plc v Vale SA in 2014.
It has also been endorsed in Ireland in Irish Bank Resolution Corporation Ltd v Quinn in 2015. Practical tips
Many of the larger firms
are developing their own in-house predictive coding capabilities. This will not currently be cost effective
for smaller firms, which
will need to establish relationships with trusted suppliers. One of the reasons why the costs quoted by the defendant's lawyers in Brown were so low was because they were able to undertake the predictive coding work in house, rather than paying an external provider;
If predictive coding is to be considered, it is advisable to undertake a cost-benefit analysis as early as possible and to provide this to the other side and to the court
at the CMC;
The upfront costs of predictive coding can be substantial but the cost benefit is gained when spread across a large number of documents, ultimately producing a lower review cost per document. The costs should be weighed against the estimated cost of a manual review;
It will be much easier to
resist the use of predictive coding for smaller disclosure exercises, although predictive coding can be effective when searching for documents on an issue-by-issue basis;
Persuasive technical evidence will be required
to support any argument that predictive coding is less accurate than human review;
Predictive coding is likely
to make it more difficult
to restrict the scope of disclosure reviews and to avoid an order for standard disclosure, particularly in
the larger cases; and
Given the increasing concerns of clients, practitioners, and the judiciary about the costs of e-disclosure, it is likely that courts will become more robust in requiring parties
to communicate and co-operate well in advance of the CMC on this issue. Failure to do so is likely to be challenged by the court.