E-Discovery and Information Management: Courts Continue to Endorse Technology Assisted Review To Control Rising Costs of Electronic Discovery

May 3, 2015

The rapid pace of technological development, particularly in the area of data storage, has created significant burdens on all businesses, large and small. In litigation, the cost of locating relevant Electronically-Stored Information (ESI) has become increasingly prohibitive as the volume of ESI stored by businesses has ballooned. Against this landscape, Predictive Coding, or Technology Assisted Review (TAR), has emerged as an approved methodology for reducing the costs of e-discovery.

In TAR, a computer algorithm utilizes human input on a small sample of documents to predict the relevancy of a much larger pool of documents. The parties (typically through their counsel or experts) begin by selecting a so-called "seed set," a sample of documents from the universe of documents to be searched by using search criteria. As the human reviewers code the seed set, the computer algorithm analyzes the coding patterns. Rather than reviewing every document in the pool, the human reviewers typically only need to review a few thousand documents before the computer algorithm becomes knowledgeable enough to code the remaining documents. Thus, TAR combines the power of a trained human expert who is familiar with the subject matter of the litigation with a powerful computer algorithm to maximize both recall and precision, and to significantly reduce the time that is required to locate documents that are responsive to the requesting party’s discovery demands.

Courts have increasingly endorsed TAR in lieu of manual document review. Indeed, one of the first judges to endorse TAR, Magistrate Judge Andrew J. Peck of the United States District Court for the Southern District of New York,(1) recently went so far as to proclaim that a producing party’s right to utilize TAR for high-volume ESI cases is "now black letter law."(2) However, Judge Peck’s discussion in Rio Tinto also highlights a number of unresolved issues that are likely to be litigated as the use of TAR becomes more widespread. For example, to what extent can a requesting party demand transparency with respect to the seed set used by the producing party? In both Da Silva and Rio Tinto, the parties stipulated in their e-discovery protocols that they would disclose all non-privileged documents used in their seed sets. Although Judge Peck endorsed this transparency, he declined to opine whether such transparency was required in all cases.

The growing judicial acceptance of TAR evidenced by decisions like Rio Tinto is likely to impact all producing parties, even those who would prefer to forgo TAR in favor of traditional keyword searching. For example, in a recent multidistrict litigation, the Plaintiffs’ Steering Committee sought to compel the defendant to utilize predictive coding to ensure higher recall of responsive documents.(3) Although the court denied the Committee’s motion because the defendant had already conducted considerable review through traditional keyword searching, the decision might have been different if the defendant’s review had not already been well underway. As more courts tout the precision of TAR, requesting parties are ever more likely to demand that producing parties use this sophisticated technology to ensure that the maximum number of responsive documents are being unearthed.

The decision in Rio Tinto and the other cases cited by Judge Peck should eliminate any doubt whether TAR can be defensibly employed in any large-volume ESI case. For cases in which TAR is appropriate, litigants should utilize this technology from the outset of the litigation, and develop a protocol to ensure that the selected predictive technology is both defensible and transparent. Doing so will both maximize cost-savings and minimize the likelihood of expensive discovery disputes.

To learn more, contact the author Kate I. Reid, as well as Brian J. Butler or Clifford G. Tsan, Co-Chairs of Bond’s E-Discovery Practice Group.

Kate I. Reid

Brian J. Butler

Clifford G. Tsan


(1) Da Silva Moore v. Publicis Groupe & MSL Group, 27 F.R.D. 182 (S.D.N.Y. 2012).

(2) Rio Tinto PLC v. Vale S.A., No. 14 Civ. 3042, 2015 U.S. Dist. LEXIS 24996 (S.D.N.Y. March 2, 2015).

(3) In re Biomet M2a Magnum Hip Implant Prods. Liab. Lit, No. 3:12-MD-2391, 2013 U.S. Dist. LEXIS 84440 (N.D. Ind. April 18, 2013).