Reasonableness and Proportionality Win Another Fight for Predictive Coding

On July 22, 2014, Judge Joe B. Brown issued a discovery order in Bridgestone Americas, Inc. v. Int’l Bus. Machines Corp. (Case No. 3:13-1196 M.D. Tenn.), in the plaintiff’s favor, allowing the use of predictive coding to further narrow a document set that was previously filtered from a broader universe of documents using keyword search terms.  The defendant objected to the plaintiff’s use of predictive coding after search terms were agreed to and applied, arguing that the application of predictive coding ex-post would be an unwarranted change to the case management order and unfairly advantageous to the plaintiff.

Magistrate Judge Andrew Peck, who broke the ice for predictive coding in his seminal order in Da Silva Moore v. Publicis Groupe et al. (No. 11-civ-1279 S.D.N.Y.), could use the Bridgestone Court’s Order as promotional material since it largely relies on references to Peck’s “various best practice suggestions,” “excellent article” and numerous rulings on predictive coding.  But the Bridgestone Court ultimately did get it right, stating that “[i]n the final analysis, the use of predictive coding is a judgment call, hopefully keeping in mind the exhortation of Rule 26 that discovery be tailored by the court to be as efficient and cost-effective as possible.”  In other words, reasonableness and proportionality are what matters in a court’s analysis of discovery disputes, including as to predictive coding, whether or not contemplated in a case management order.  Aptly stated, “[t]here is no single, simple, correct solution possible under these circumstances,” which in Bridgestone included the need to review millions of documents even after filtering by search terms.

So predictive coding got the green light from the Court.  The plaintiff was directed to be transparent and open with defendant throughout the process.  In a sense, the Court hedged its Order on another touchstone of Rule 26—cooperation.

The Bridgestone Court reached the right conclusion, but did so on general principles and by riding M.J. Peck’s coattails.  Sure, all is well that ends well, but the Court demonstrated an incomplete understanding of predictive coding.  Indeed, predictive coding often incorporates initial keyword filtering before applying the coding software itself.  Eric Seggebruch of Recommind, a pioneer in the field, noted that “using search terms in the first instance to cull a document set initially is viable, even common,” and that “predictive analytics is a viable method to employ in any situation where keyword culling is otherwise appropriate.”  Seggebruch added, “whether a keyword cull is appropriate prior to implementing predictive coding can depend on a number of factors, such as the type of data, the value of the case juxtaposed to the cost of using advanced analytics, and other factors that are matter specific.”

Keyword culling without predictive analytics is commonplace, and utilizing search terms before applying predictive coding software can increase efficiency and effectiveness in culling large document sets for purposes of preempting identification of the “junk” folder that sits in the corner of every document review dungeon that junior associates hope to never see.

In short, the Bridgestone Court (along with counsel) reached the appropriate conclusion, but without complete knowledge of the realities of how predictive coding is actually used.  Practitioners should make a concerted effort to understand predictive coding for what it is and is not—the failure to do so undermines the very objective that ever-evolving predictive coding tools seek to achieve.

About The Author

Jason Bonk is an experienced litigator in the firm's New York office. He represents Fortune 500 companies along with middle-market businesses in a variety of high-stakes matters, including complex commercial cases involving contract claims as well as fiduciary and other equitable claims, class actions, white collar investigations, labor and employment disputes, and bankruptcy litigation. Prior to joining Cozen O'Connor, Jason spent most of his career at Weil, Gotshal & Manges, and practiced, most recently, at Kleinberg, Kaplan, Wolff and Cohen.

Posted in Predictive Coding

Leave a Reply

Your email address will not be published. Required fields are marked *


Subscribe To Our Posts


Cozen O’Connor Blogs