Incorporating electronic discovery technology into litigation workflows has remained a hot topic of discussion, and much its recent focus has been on a clear trend towards a rise in judicial acceptance of predictive coding. Recent cases include a mix of examples in which courts have evaluated litigants' actual use of predictive coding to meet document discovery obligations; examples in which courts have suggested that litigants evaluate the viability of predictive coding to meet particularly burdensome document production obligations; and examples in which government agencies have evaluated how to integrate predictive coding into regulatory review workflows. Each of these themes, along with the cases elaborating on them, is addressed in turn below:
In re Actos (Pioglitazone) Products Liability Litigationi, a case order went into almost extraordinary detail documenting how a predictive coding workflow would be devised. The Western District Court of Louisiana instructed adverse litigants to collaboratively review the non-privileged population of documents from which a sample set was selected to "train" the software to make automated responsiveness determinations. It required that the parties agree on "Relevance Scores" to be assigned to certain categories within the sample. Further, it required the parties to determine a threshold relevance score for documents that would require manual review as opposed to automated review; and required that the parties collaboratively review a random sample of documents determined by the software to be non-responsive to relevant document requests.
In an update on the Kleen Products caseii, plaintiffs agreed to withdraw a demand for defendants to use predictive coding (which plaintiffs had initially contended to be preferable to defendants' Boolean search term-driven methodology for generating potentially responsive documents). Plaintiffs' withdrawal was ultimately attributed to a better understanding of defendants' proposed methodology after a series of meet and confers, and this portion of the outcome was ultimately described by the Northern District Court of Illinois as a testament to the "collaborative" manner in which the adverse parties resolved this issue.
The very recent in re Biomet has generated attention for its emphasis on proportionality.iii In this case the defendant had initially attempted to meet its document production obligations using predictive coding software to help identify Relevant and Non Relevant documents. When the plaintiff raised subsequent concerns about the universe of documents to which defendant had applied the software, the Northern District Court of Indiana ruled that the defendant would not be required to essentially re-conduct a substantial portion of its review in order to appease plaintiff's concerns. The court noted: "It might well be that (Plaintiff's proposed approach) would unearth additional relevant documents. But it would cost (Defendant) millions of dollars to test (plaintiff's) theory that (its approach) would produce a significantly greater number of relevant documents. Even in light of... the importance of the issues at stake, and the importance of this discovery in resolving the issues, I can't find that the likely benefits of the discovery proposed by (Plaintiff) equals or outweighs its additional burden on, and additional expense to, (Defendant)."
Global Aerospace, Inc. v. Landow Aviationiv has been viewed as a potential impact case because a Virginia Circuit Court allowed defendants to proceed with the use of a predictive coding-driven document review process over plaintiff objections.
In a notable update on the now famous Da Silva Moore ruling,v the Southern District Court of New York rejected plaintiff objections to a court order endorsing predictive coding, stating that the initial order was "well-reasoned" and that "there is insufficient evidence to conclude that the use of predictive coding software will deny plaintiffs to a liberal discovery"; and further concluding the use of predictive coding software would be more appropriate than keyword searching given the circumstances of that case.
EORHB, Inc., et al. v. HOA Holdings, LLCvi has been viewed as a potentially landmark decision based on the Delaware Chancery Court essentially requiring litigants to use predictive coding to meet their document production obligations in a pending matter, prior to either party actually proposing its use. The court's sua sponte order further required both parties to choose a single technology vendor (that portion of the ruling was subsequently revised to allow the adverse parties to use different e-discovery providers), and commanded the attention of practitioners with comments such as:
"I would like you all to talk about a single discovery provider that could be used to warehouse both sides' documents to be your single vendor. Pick one of these wonderful discovery super powers that is able to maintain the integrity of both side's documents and insure that no one can access the other side's information. If you cannot agree on a suitable discovery vendor, you can submit names to me and I will pick one for you."
Chevron Corp. v. Donzigervii, while not focusing on a litigant's active use of predictive coding, was noted by practitioners for the Southern District Court of New York's comment that a third party whose documents were subject to subpoena had erred in not adequately considering the viability of predictive coding as an alternative where that third party had argued that the subpoena request would impose an overly burdensome production requirement.
National Day Laborer Organizing Network v. U.S. Immigration and Customs Enforcement Agencyviii similarly did not address a litigant's actual use of predictive coding, but was noted by practitioners for the Southern District Court of New York's suggestions that parties to a FOIA request for documents from a group of government agencies should evaluate the viability of predictive coding as an alternative where the potential scope of the request could result in a "massive" document production obligation for the agencies.
Harris v. Subcontracting Conceptsix, similar to Chevron and National Day Laborer, did not address a litigant's active use of predictive coding, but was noted by practitioners for the Northern District Court of New York's suggestion that the increasing availability of predictive coding and other document review technology for large datasets would potentially allow litigants to meet document discovery obligations that may have been considered to be overly burdensome to comply with via less technology-enhanced processes.
In the proposed merger of Anheuser-Busch InBev and Grupo Modelo, the DOJ approved a request to use predictive coding to review documents related to the antitrust review associated with the merger.x While commentary surrounding the Anheuser review indicates that government agencies evaluate the potential use of applications like predictive coding on a case by case basis, the SEC's recent licensing of a preferred predictive coding software provider indicates that regulatory agencies are actively vetting emerging e-discovery technologies as well.xi
Taken collectively, the cases above reflect a clear progression towards continued judicial acceptance of predictive coding. In an environment in which practitioners are required to anticipate potential scenarios, these examples are potentially instructive. For example, suggestions by the Harris and Chevron courts that litigants evaluate the viability or predictive coding to assist in meeting particularly burdensome discovery obligations, taken jointly with the possibility of a court ordering predictive coding as in EOHRB, suggest that court intervention may be conceivable as a potential response to a litigant's objections to certain types of document requests (a number of other variables would obviously affect the likelihood – or lack thereof – of this type of situation occurring as well). These issues reflect the number of different paths that this landscape could unfold in, and considering the various permutations will undoubtedly remain a focus for practitioners going forward.
Ashoke Prasad serves as an Assistant Vice President with Pangea3 and has an extensive background in financial services, tax, and financial accounting related matters. His role at Pangea3 has primarily involved assisting financial institutions with complex data reviews on derivatives and OTC securities-related matters. Ashoke is admitted to practice law in Illinois and has also passed all four parts of the Uniform CPA (Certified Public Accountant) examination (licensing pending). He received his J.D. from the University of Michigan Law School, an LL.M. in Taxation from Georgetown University Law Center, and his B.S. in Economics from the University of Michigan. Prior to joining Pangea3, Ashoke worked with Ernst and Young National Tax in Washington, D.C. and Huron Consulting Group, focusing on financial investigations and restatements.
Reagan Pereira serves as a Senior Manager with Pangea3 and has extensive document review management experience with various leading investment banks, a leader in the common carrier industry, a pharmaceutical manufacturer and a leading international tobacco company. Prior to joining Pangea3, he worked as a Senior Financial Advisor in JPMC providing financial solutions to delinquent credit card-holders. Reagan holds an LL.B. from V.M. Salgaonkar College of Law, Goa, India.
i 6:11-md-2299 (W.D. La. July 27, 2012)
ii F.Supp 2d, 2012 WL 449846
iii 2013 WL 1729682 (N.D. Ind. Apr. 18, 2013)
iv L.P., No. CL 61040 (Vir. Cir. Ct. April 23, 2012)
v 868 F.Supp.2d 137(S.D.N.Y. June 15, 2012)
vi C.A. No. 7409-VCL (Del. Ch. Oct. 15, 2012)
vii No. 11 Civ. 0691, 2013 U.S. Dist. LEXIS 36353 (S.D.N.Y. Mar. 15, 2013)
viii 2012 WL 2878130 (S.D.N.Y. July 13, 2012)
ix LLC, Civ. No. 1:12 – MC -82(DNH/RFT)(N.D.N.Y. Mar. 11, 2013)
x Joe Pallazzolo, "Software: The Attorney Who Is Always on the Job", The Wall Street Journal, May 6, 2013
xi Michael Roach, "SEC Implements Recommind Axcelerate for Its Enforcement Division", Law Technology News, January 31,2013