Bloomberg Law
Jan. 21, 2016, 1:30 PM UTC

Leapfrogging Technology and the Slow Acceptance of Predictive Coding

Caragh Landry

Editor’s Note: The author of this post works at a legal process outsourcing company.

By Caragh McGovern Landry, Global Head of Onshore Managed Review, Integreon

Legal Tech 2016 is right around the corner and a quick look at the schedule of sessions and seminars begs the question - what happened to Predictive Coding? In 2012, with more than 20 sessions focused on predictive coding, and even more centered on general TAR/CAR, Legal Tech was allaboutpredictive coding. It was what everyone was talking about and sessions on this topic were full. But here we are four years later and a quick search on the 2016 schedule will bring back only four hits for the phrase.

Some say that predictive coding is dead. That it peaked too soon, before anyone was ready for it, and now it’s just faded into the background. Predictive Coding has become a bad word in the lexicon of lawyers and many corporate counsel and we now talk instead about analytics, prioritized review and categorization, and seldom mention predictive coding, unless specifically asked.

I was on a Women in eDiscovery panel in mid-2015 and I sat with 3 other peers, all from technology companies with predictive coding tools. We tried to impress upon the audience that predictive coding was a cost and time saving approach, defensible through heavy reporting and good workflow design and simpler than expected, if you had the right guide and process.

We, the technologists, sat on one side of the table and on the other side of the table sat lawyers, paralegals and legal admins of varying levels. There must have been at least 50 long-term professionals in the room, experienced men and women, but not one person onthatside of the table had any level of comfort with predictive coding.

There are only 2 sessions focused on Technology Assisted Review/Predictive Coding and one of them is a use case session, geared to show us again that technology that can be helpful in a friendly, approachable narrative approach.

The other panelists and I quickly shifted the conversation to the use of analytics and stayed away from the topic of Predictive Coding for the remainder of the session. Instead we discussed ideas for ECA using concepts and concept searching, using exemplars to group similar documents and using Find Similar or More Like This tools on the Priv Log. After removing the fear and animosity commonly evoked by the term Predictive Coding, the audience seemed to breathe a collective sigh of relief and became really engaged in new workflow ideas.

At LegalTech 2016 the eDiscovery conversation also seems to be changing and the conversation has shifted in the same general direction. Gone is almost any mention of Predictive Coding and in its place we now have several session on ECA, uses of Analytics in processing and review, and blending of people, process and technology. There are only 2 sessions focused on Technology Assisted Review/Predictive Coding and one of them is a use case session, geared to show us again that technology that can be helpful in a friendly, approachable narrative approach. Stories from the field.

ECA and strategies for using Analytics are big topics in the educational tracks and this harkens back to 2007-2010 at Legal Tech where everyone was talking about clustering, concept searching, threading and near duplicate technology. Many of the analytics tools in play today came from that same era and haven’t really changed much since then. Legal Tech 2016 seems to be the resurgence of technology assisted review tools and features, blended into talk about the importance of people, process and technology integration. It is a similar conversation from 2007 and on, but it is now a more mature and richer conversation because we have 9 years of experience under our collective belts.

I personally feel that same leapfrogging is a main reason for the failing of general acceptance of Predictive Coding in the market.

So, again, what happened to Predictive Coding? I think it was that we, the technologists, got so excited about this wholistic way to employ the technology that we neglected to think of our audience, the lawyers tasked with managing and presenting the cases. Back in 2011, I was struck by how quickly the eDiscovery industry had dropped the usage of thread grouping and clustering to drive a review and focused instead on the emergence of Predictive Coding as the end-all, be-all in the future of eDiscovery dialogue. None of my clients were ready for it and I think the leapfrogging of the tools and features embedded in, and at the base of, Predictive Coding, to full on ‘produce without looking at everything’ steered the conversation in a bad direction. I personally feel that same leapfrogging is a main reason for the failing of general acceptance of Predictive Coding in the market. The fact that we are lawyers and consultants trying to win cases doesn’t help either, nor does the fact that we need to use statistics to defend/define the workflow.

Predictive Coding seems to be a case of too much too soon. Had we slowed the pace for implementation of the complete workflow and focused instead on adoption of the smaller pieces, perhaps four years on wewouldbe ready to use Predictive Coding whenever possible. Instead, we are now going back to where we should have focused four years ago and starting again. Will Predictive Coding be the next big thing at LegalTech 2020? If we’re smarter then, the answer will be yes.

Learn more about Bloomberg Law or Log In to keep reading:

Learn About Bloomberg Law

AI-powered legal analytics, workflow tools and premium legal & business news.

Already a subscriber?

Log in to keep reading or access research tools.