Lawyerbots take the strain

By | 24th July 2012

Software that sifts through millions of documents for relevant information gets the green light to replace human lawyers

When you see such “Shock! Horror!” headlines many lawyers will throw up their hands in despair. They have just about come to terms with the idea that technology exists which, in appropriate circumstances, may shorten the time taken to search through a document population to identify the relevant material to be perused, but the idea that they will have to give way to an army of “Lawyerbots” will be a step too far.

I am not sure, however, that this is the case or that the headline above is as scary as it may appear at first sight. Let me explain.

The headline appears in New Scientist of 28 June11 – Lawyerbot takes the drudgery out of law  – and a further article on the same subject may be found in Gizomodo: Lawyerbots Given the Green Light in the US.

The case referred to is the Landow Aviation case in Pittsburgh, which I wrote about in a recent post.  [Keeping up with the Joneses, 29th June, 2012]

Lawyer Thomas Gricks had to examine over 2 million emails and attachments, a task which he estimated would take 20,000 person hours and cost in the region of $2 million. He wanted to use predictive coding to cut the task to a couple of weeks and a fraction of the cost (said to be 1%) and in the face of objections from the plaintiffs in the case, the judge agreed.

The main concern usually expressed by lawyers is whether a computer can do as well as a team of lawyers. In fact there are studies which suggest that computers are much better at reviewing documents than humans (see the study by Maura Grossman, a lawyer at New York firm Wachtell, Lipton, Rosen & Katz, and Gordon Cormack of the University of Waterloo in Ontario, Canada.)

“Both lawyers and software examined a collection of over 800,000 emails and attachments gathered as a result of the collapse of Enron. Predictive coding software was able to match and sometimes beat humans at finding relevant documents, while also being better at avoiding returning irrelevant material. In one case, when asked to flag emails related to document shredding, the human reviewers managed to miss one that said: “I’ll be shredding ‘ till 11am” – but the software caught it (Richmond Journal of Law and Technology, vol 17, issue 3).”

But, crucially, this does not mean that computers are doing away with the lawyers. It is a vital part of any predictive coding exercise that a senior lawyer on the team has the job of deciding which documents should be discarded and which retained in a relevance pass. Once those decisions have been made it is that same lawyer and the team on the case who review the remaining documents as they have always done in any disclosure exercise. This is not doing away with the lawyers but making best use of their time and skills so that the disclosure exercise is more efficient and more cost effective to the benefit of the client and the speedier resolution of the case. It just cannot be done without the lawyers! Of course, this does not mean that there will not be changes in working practices. For example, the concept of teams of paralegals shuffling paper is now as outdated as writing with a quill pen (and anyone who thinks differently will get a nasty shock when they try to run a case in the old way and when they seek to recover the costs of so doing). Lawyers will have to adapt and they will, and some have changed already.

We have direct evidence of this change in practice as Millnet is currently in receipt of a steady stream of cases where the technology is in use and for a wide variety of law firms, both UK based and from the US.

And why do lawyers want to reduce costs? Gricks answers the question as follows:

“To attract new clients. We can try cases that weren’t being tried because it was too expensive to get through discovery.”

And that is our experience too. We have advised the use of predictive coding on cases which simply would not have proceeded if the “normal” costs of disclosure had applied. As a result, the case has proceeded and the lawyers have earned fees which would have been denied them had the case folded.

There will be further ups and downs before the use of predictive coding is accepted as the norm. There will be arguments about suitability and defensibility but the technology is achieving a major transformation in the field of e-disclosure as can be seen by the article in The Lawyer of 16th July and the views expressed by the lawyers and litigation support specialists quoted.

Not everyone will be convinced at the outset and some cases will not lend themselves easily to the use of the technology; for example, where the documents are largely non text files. However, I have little doubt that the technology is here to stay and that the lawyerbots have definitely not yet taken over the field.

Frankly, I doubt they ever will.

Illustration: Lawyerbot (Neopets)