Epiq Systems London judicial panel points to future of technology-assisted review

The original invitation to Epiq Systems’ panel debate on judicial attitudes to technology assisted review said that the speakers would be Senior Master Whitaker and US Magistrate Judge Andrew Peck.  That was enough to have me book my place, but Epiq subsequently rounded out the panel with the addition of Bob Lewis from Barclays Bank and barrister Shantanu Majumdar which made it even more interesting. Others obviously thought so too, because the room was packed despite (or perhaps because of) an 8:15 start. Epiq’s Laura Kibbe moderated with her usual style.

I will stick in this article to the term predictive coding, if you don’t mind, despite our host’s preference for one of the many other terms which describe broadly the same thing; I quite like my articles to be found in Google, and “predictive coding” is the term which people search for.

Bob Lewis leads Barclays’ initiative to bring much of the eDisclosure process in-house, acquiring software, building teams and developing processes to make the task as efficient and cost-effective as possible.  I know Shantanu Majumdar because I have taken part in video seminars with him. What do you get if you add together the best in modern technology, a largely self-sufficient client, a barrister who understands eDisclosure and two of the most eloquent judicial advocates of technology as an aid to proportionate disclosure?  Well, at the immediate level you get a well-informed  panel; in the distance,  a solicitor would see a cloud no bigger than a man’s hand which will one day, and not too far off, pose a significant threat to his business or, at least, to those members of the profession who cannot see which way the wind is blowing.

The technology reduces the need for manpower, and if you need bodies, then Epiq is amongst those who offer managed document review services. Sure, you will need a lawyer to give input on the law, the issues, the tactics and strategy. But does that have to be a firm of solicitors with all the lumbering apparatus which seems to come with the breed? Why not just find a switched-on barrister who will give input when you need it? If as a client you employ someone like Bob Lewis and his team, the attachment to solicitors to manage your eDisclosure begins to seem a luxury, a handling charge paid to people who are not adding much value. All you need then is a judge who understands a little – not a lot – about electronic disclosure.

All this, you understand, is my own musing derived from seeing this assembly of people on Epiq’s panel. It did not form part of the discussion, which was far more about the present and the immediate future. The best panels, however, make you lift your eyes beyond the next step and encourage you to look round the corners ahead. This one certainly did that.

The session was recorded on video which is available here. That makes it unnecessary for me to summarise the whole event. Instead, I will pick a handful of key points, particularly those which support the idea that the present reluctance to make use of the power of predictive coding is to ignore the best hope there is of making litigation affordable.

Epiq - Laura KibbeLaura Kibbe (pictured right) opened with the unarguable proposition that volumes are going up at a great rate – “the neck of the funnel keeps getting bigger” she said, and we need a more drastic reduction in the funnel. Why, she wondered, are lawyers reluctant to consider the use of technology assisted review – and she and the panel accepted that there is some reluctance. One of the challenges, she said, was to explain what was happening in terms which were easily understood

Epiq Systems - Judge PeckJudge Peck (pictured left) observed that lawyers are creatures of precedent and that US lawyers have been waiting for judicial approval of predictive coding. They now have it, in the form of his own opinion in Da Silva Moore, the pending opinion in Kleen Products and the Global Aerospace case from Virginia.  He could not, of course, say anything about Da Silva Moore and in any event his point was made by the mere report that three cases are going through US courts at the moment. Judge Peck reminded us that no particular technology, even keywords, had ever been expressly approved by a US court; what mattered was the process, and in particular the validation process, which was wrapped around the technology. This, of course, was one of the key messages of Judge Peck’s Da Silva Moore Opinion.

Epiq - Master WhitakerMaster Whitaker (pictured right) drew attention to the misapprehension on the part of many lawyers and judges that they must understand how predictive coding works. It is necessary to explain the training process and to emphasise both the ability to cross-check results and the defects of manual review.  We do not, he said, need to look under the hood. Predictive coding is not a “universal panacea” but it gets the job done better and faster than humans can. The biggest problem, as he saw it, was that people do not know what the costs are of using such technology. This is not a message directed exclusively at the suppliers – the lawyers should make more serious attempts to understand the costs of different ways of managing the eDisclosure process, as they will have to when costs budgeting passes into the rules in April 2013.

Epiq - Shantanu MajumdarShantanu Majumdar (pictured left) observed that lawyers are always nervous of trying things. He is involved, he said, in a number of cases in which the use of predictive coding is being discussed internally and with the other side. There is a fear, he said, which is “exacerbated by cases you hear about”, a reference, I suspect, to the West African Gas Pipeline case whose (to my eye) unhelpful outcome threatens to blight attempts towards narrower disclosure. That case also has the pernicious potential to encourage challenges, and not necessarily meritorious challenges, to the other side’s disclosure. Shantanu said that we must avoid such satellite litigation and the spending of money on picking holes in this way.

Bob Lewis continued the “under the hood” line by reminding us that none of us knows how Microsoft Office works yet we are prepared to entrust our businesses to its several components. Of predictive coding, he said that Barclays will move sooner or later to using predictive coding alongside the other tools which it had. It requires, he said, a “good understanding of what you want” together with good input and the acceptance that it will not do everything.

Master Whitaker drew attention to the ability provided by predictive coding tools to decide on an acceptable margin of error and to measure output accordingly. The error rate will almost certainly be better than with human review – to the extent that you can ever measure the shortcomings of human review.

Epiq Panel 2

Bob Lewis, second left, makes a point

An interesting observation was made from the floor at this point in the context of shifting people from their view of human review as the putative “gold standard”.  There is a human instinct not to blame human error, and technology provided a useful scapegoat.

It was some discussion about the extent of cooperation expected between the parties. Judge Peck said that he had met some resistance to this, quite apart from the gamesmanship which characterised much US litigation. You have to design a process, he said, and be able to explain it and how the technology works within the process. Master Whitaker thought that UK parties would have less difficulty in agreeing, for example, where to draw the line beyond which further review was disproportionate.

Master Whitaker also pointed to the merits of having senior lawyers give their considered input at the front end of the process which, coupled with cooperation with opponents, meant that high-quality decisions were being made from the outset. That, he said, was very much better than having juniors make the primary decisions without discussion – a recipe for specific disclosure applications.

Other good points were made which I report in short form:

Predictive coding can be used to test an incoming production, for early case assessment and first-pass review, for identifying potentially privileged documents, for quality control and, perhaps most of all, for relevance ranking. The ability to do immediate cross-checks – “this document is similar to one which you discarded yesterday” – cannot be matched by any purely human process.

The human element comes in also with the discussions with the client which are no less important with technology. You get a feel for the most obvious custodians, the date ranges and for the location of documents which might otherwise be missed. Only by such questioning will you discover, for example, that the project now known as X was, in its early days, known alternatively as Y,  or that a custodian had a nickname which would not be clear from his or her HR record or e-mail address.

Intelligence is needed in framing searches whatever technology is being used; Master Whitaker was scornful of the parties who wished to include the word “road” as a keyword, and Bob Lewis said that lawyers often ask stupid questions – get me everything  containing the words “Barclays” and “confidential”, for example.

Both judges emphasised that (in Judge Peck’s words) “reasonableness is all we are looking for”, not perfection.

It is important to document the process, partly in order to be able to justify steps which were taken, but also to supplement corporate memory – associates and partners move on and you cannot be sure that their oral evidence will be available to support the decisions which were made during the review. The mere act of making such a record gives confidence that you can defend what has been done.

Shantanu Majumdar mentioned the importance of getting in experts to help, referring to a case in which such an expert had quickly demonstrated the flaws in the searching which his side had already done.

If one wanted a single word which made a difference, it could be the word “early”. Shantanu said, for example, that counsel used to be involved very late in the disclosure process but are increasingly being brought in early to help decide the scope of disclosure.

These last two points brings me back to my opening – if experts are flushing out by a combination of experience and technology documents which the lawyers missed, and if barristers are being involved in early discussions about the scope of disclosure, then what are the solicitors bringing to the party?  Later in the same week I heard Richard Susskind speak at the LawTech Camp London event. He posed the same question – what are solicitors uniquely good at in the context of litigation? Tactics and strategy was his answer, in a world in which a combination of technology and third-party providers of outsourced services, to say nothing of barristers, can between them do all those other tasks which have hitherto been the preserve of solicitors. And that takes no account of initiatives by clients such as Barclays – these scale down, and you do not have to be the size of Barclays to bring home part of the process which has traditionally been sent out to solicitors.

Epiq - Laura Kibbe 2This was a wide-ranging and important session which went far beyond its advertised remit, at least in its implications. Laura Kibbe is a great moderator, skilfully giving her panellists their head whilst keeping to the subject and the clock.

I make two recommendations: one is that you watch the video which Epiq have provided; the other is that you make prompt arrangements to meet one of the providers of predictive coding software and services. You cannot reject what you have not seen.

Larger versions of the photographs seen here, with others, can be found here.


About Chris Dale

I have been an English solicitor since 1980. I run the e-Disclosure Information Project which collects and comments on information about electronic disclosure / eDiscovery and related subjects in the UK, the US, AsiaPac and elsewhere
This entry was posted in Discovery, eDisclosure, eDiscovery, Electronic disclosure, Epiq Systems, Predictive Coding. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s