What you need to know
- The Supreme Court of Victoria recently handed down a decision which considered and approved the use of predictive coding for document review in a discovery exercise. This is the first time an Australian Court has sanctioned the use of this technology.
- In line with the global trend in common law jurisdictions, the decision supports the technology as an acceptable method of conducting a reasonable search across large volumes of electronic documents for the purpose of discovery.
- Predictive coding is likely to become a key feature of the case management of large commercial disputes in the near future
What is predictive coding?
Predictive coding or technology assisted review allows a computer system to be used to identify relevant documents in large electronic data sets. The system learns from decisions made by human reviewers as to relevance of an initial set of documents and then applies what it has learned to predict the relevance of other documents. It is now widely accepted that predictive coding can be at least as accurate as a traditional manual review.
What happened in the case?
McConnell Dowell Constructors (Aust) Pty Ltd v Santam Ltd & Ors (No 1) [2016] VSC 734 concerned a contract for the design and construction of a natural gas pipeline in Queensland. The number of electronic documents potentially relevant to the proceeding was reduced from approximately 4 million to approximately 1.4 million using de-duplication technology. The Court was concerned with the time and costs that would be incurred reviewing this volume of electronic documents.
Vickery J appointed a special referee to report on the appropriate management of discovery as the traditional manual method would be too costly and inefficient. The referee reported that the parties had agreed to predictive coding and had established a protocol. Vickery J's reasons adopted the report and endorsed the use of predictive coding technology to assist in the review of documents for discovery. His Honour also made a number of references to decisions in other jurisdictions.
The global trend
The use of predictive coding is gathering real momentum in common law jurisdictions where the proportionality of discovery costs is considered to be an issue.
Predictive coding has now been used for over five years in the US, where the first reported decision was by Magistrate Judge Andrew Peck in the case of Da Silva Moore v Publicis Groupe in 2011. Judge Peck is well known in the US as a strong advocate of predictive coding and his decision in Da Silva Moore provides a comprehensive account of the nature of the technology, its benefits and its limitations as well as the reasons for ordering predictive coding in that case.
Following Da Silva Moore there were cases in the Republic of Ireland and England in which the courts in both those jurisdictions approved the use of predictive coding. In Irish Bank Resolution v Quinn, the Irish High Court ordered predictive coding over the objections of the party that had requested discovery. The English courts considered the issue for the first time in Pyrrho Investments v MWB Property, where both parties submitted that the technology should be used. This was taken one step further in that jurisdiction in David Brown v BCA Trading where, like in Quinn, the court ordered predictive coding where the parties disagreed as to its use.
One of the key messages running through the judgments in these cases is the need for discovery orders to be adapted and fashioned in a way that achieves proportionality in respect of discovery costs and discharges the overriding obligations on litigants and courts to strive for expediency and economy.
The position in Australia
How an Australian court would treat the use of predictive coding had been, until this decision, unclear. The recent endorsement in McConnell Dowell is not a surprise. This is because there have been a number of recent extra-judicial comments—including from the Chief Justice of the Federal Court of Australia—made about the potential benefits of the technology.
Consistent with this, the Federal Court of Australia recently published a new set of practice notes including Practice Note GPN-TECH: Technology and the Court, which makes explicit reference to the use of predictive coding. Similarly, as is noted by his Honour in McConnell Dowell, The Supreme Court of Victoria has now published, as part of a set of revised practice notes, Practice Note SC Gen 5, Technology in Civil Litigation, to be issued on 30 January 2017 which contains guidelines for the use of technology assisted review.
The future
It is worth emphasising that the parties in this case were in agreement about using predictive coding. It is possible that, if presented with a contested application, a court might take a different view.
However, in cases where there are vast amounts of electronic data requiring review, we consider that the Australian courts and courts in other relevant jurisdictions will become increasingly reluctant to entertain objections to the use of predictive coding, particularly given the prominence of proportionality as a consideration.
We would therefore expect that in time predictive coding is likely to become a common feature of the litigation landscape.
Not a silver bullet…
Not all cases or review populations are appropriate for predictive coding. Analysis prior to commencing the exercise should inform, for example, whether the types of document are appropriate for predictive coding or whether separate approaches need to be taken to reflect characteristics within the document population to be reviewed.
Predictive coding also does not provide a one size fits all solution. It is just one of several ways in which technology can assist with the review burden. When used effectively however, predictive coding can deliver real cost savings and efficiencies, both in the context of discovery exercises and in regulatory investigations.
Ashurst has extensive experience throughout its global network assisting clients make use of this technology in contentious matters. If you would like to discuss predictive coding further, please contact any of the contacts below.
Authors: Mark Elvy, Partner; Wen Ts'ai Lim, Partner; Tim West, Senior Associate; Nathan Bellgrove, Senior Associate
