Matthew Field is a Director at FTI Consulting in London. I interviewed him recently about the increase in the use of the technology which is known variously as technology-assisted review and predictive coding (and which I will call “TAR” for the purposes of this post).
At the time of the interview, FTI Consulting owned the Ringtail eDiscovery platform, one of the best-known eDiscovery tools including TAR, which FTI has since sold to Nuix. FTI has a three-year agreement with Nuix for consultancy and knowledge transfer in relation to Ringtail. What Matthew Field says about Ringtail here emphasises that FTI’s primary role is a consultative one.
Matthew Field said that there were two main reasons why UK lawyers were now becoming comfortable with the use of TAR. One is that the end clients keenly understand the cost centres and see an opportunity to save money by the use of this kind of technology on what seems to them to be a low-value part of the exercise. The other was the UK judgments which are favourable towards the use of TAR. Pyrrho, and then the heavily-contested BCA Trading case, have opened the doors to wider use of TAR.
Ringtail was in fact the technology which was the subject of the BCA Trading judgment. It was used in-house by Berwin Leighton Paisner (now Bryan Cave Leighton Paisner) who were highly skilled and who needed little consultative input from FTI to achieve success both in the arguments about the use of TAR and at trial. Apart from firms like this, there is much-increased awareness of the technology, but less developed expertise in applying it. This, Matthew Field says, offers a huge opportunity for FTI to add its skills and experience to the lawyers’ expertise. FTI’s focus, he said, was on the defensibility of TAR. Whilst its use is not being criticised in judgments, there is an increasing level of judicial criticism aimed at the management of discovery. FTI can help head this off.
A further development is that the technology itself is improving. The original iterations of TAR were thought suitable only for big cases, not least because of the high front-end costs involved in training. Newer iterations of TAR, generally known as Active Learning or Continuous Active Learning, do not need the training phase and could potentially be used on smaller cases, particularly in those where it was necessary to find the needle in the haystack among large volumes of junk data.
These uses did not necessarily have to involve disputes – CAL opens the door to a more pre-emptive approach where the client simply wants to know what data it has, perhaps for an internal investigation or for a due diligence exercise.
The right question is not “Is this a case for TAR” but “Which tools should we use for dealing with this stage of this matter?”
One overlooked ability is the possibility of reusing algorithms for multiple cases. FTI was involved in a case for a bank client who had developed an algorithm for a particular case and then, some time later, faced a similar one. The idea of “banking” predicted coding algorithms was one which is likely to develop.
I asked Matthew Field what he anticipated would be coming in the next year. TAR will be used more widely and more flexibly, he said, married with the long-established concept clustering technology, a combination which promised an increased ability to get quickly to the key documents.