Exterro's E-Discovery & Privacy Breakdown

The world of E-Discovery & Privacy is constantly changing – let us break it down for you with a weekly dose of News, Resources, Case Law, and Humor, all written in a concise and easy to understand format.


Live Blogging at LTNY: Session on Predictive Coding

Created on January 30, 2012

Date/Time: 3:45 – 5:00 (EST), Monday, January 30, 2012
Participants: Daniel Garrie, Partner, FSRDG; and General Counsel, Pulse; Kathryn Goetz, Vice President of Litigation Discovery, Qualcomm; Maura Grossman, Counsel, Wachtell Lipton Rosen & Katz; Joel Mitnick, Partner and Co-Chair of eDiscovery Task Force, Sidley Austin LLP; Moderator: Warwick Sharp, Vice President, Equivio
Session Description: More Predictive Than Ever: Predictive Coding, Predictive Costs, and Predictive Outcomes - New Techniques to Manage and Control Electronic Discovery
In this session, a panel of four e-discovery experts discussed the emergence of predictive coding.  They addressed the benefits and risks associated with the new technology and considered whether the courts will eventually embrace it. 
  • A prevalent myth in e-discovery is that human review is the gold standard.  Panelist Maura Grossman* explained that studies show that computer assisted review (predictive coding) can be just as effective, if not more so.  Moreover, she said that human error in review tends to be random - the result of basic oversights - while computer review errors tend to be systematic, or pattern based.  It's much easier to fix systematic errors than random ones.
  • Predictive coding DOES NOT mark the end of human involvement in review.  Panelist Kathryn Goetz said that humans play a critical role in predictive coding because they "train" the technology to recognize relevance.  The accuracy and thoroughness of the technology tool will only be effective if lawyers provide it with a large enough subset of carefully reviewed documents as a base for the algorithm that the technology relies on.
  • While opinions on this differ, the panel was in agreement that predictive coding technology should not be relied on as a privilege filter.  It is very hard to train technology on privilege issues, which are rarely black and white.  Predictive coding is not intended to replace the human judgemet required of a thorough privilege review.
  • Ultimately, the defensibility of a review process is based on results and not necessarily the technology  that is involved.  People are clamoring for a judge to embrace predictive coding as a defensible review technology.  While such a ruling would be significant, the courts have shown a willingness to accept various technological tools and strategies as long as the results can be adequately defended.
E-Discovery Beat’s Key Takeaway
Certainly, predictive coding is an exciting new technology.  It's benefits are already being felt by corporate legal teams and law firms alike.  However, predictive coding is not an e-discovery panacea.  The human role in predictive coding is critical both in terms of training the technology and assessing the results, which includes frequent sampling, gathering and analyzing metrics and communicating with opposing counsel on how the technology is being used.  In the end, technology is only part of the larger review process and it's the process, not the technology, which determines defensibility.

*Maura Grossman recently spoke on an Exterro webcast, "2012 E-Discovery Case Law Forecast: Hindsight is 20/20," which you can watch here