Editor’s Note: The author of this post is a discovery and information governance consultant.
In recent weeks, courts have shattered the silence on technology-assisted review (TAR) that has pervaded eDiscovery practice in the U.S. for most of 2016.
Last month, a U.S. Tax Court issued its second opinion in Dynamo Holdings v. Commissioner of Internal Revenue, with Judge Ronald L. Buch approving a production of documents made using TAR. And just this week in Hyles v. New York City, U.S. Magistrate Judge Andrew Peck rejected a requesting party’s demand that a producing party use TAR instead of search terms to complete a production of documents.
Both Dynamo Holdings II and Hyles are instructive on various issues surrounding TAR including: (1) the efficacy of human review and search terms; (2) the limits of cooperation; and (3) the standard a producing party must meet to satisfy discovery production requirements.
Efficacy of Human Review and Search Terms
A recurring question throughout eDiscovery jurisprudence is whether TAR is as effective as human review or search terms in identifying responsive information. While various studies and court decisions have shown that a properly functioning TAR process is superior to human review and search terms, requesting parties still rely on the “myth” that human review and search terms can generate more inclusive productions of relevant documents.
In Dynamo Holdings II, the court rejected this myth. In approving a production made using the parties’ jointly developed TAR process, Judge Buch explained that “human review is far from perfect” because document reviewers frequently misapply the criteria used for determining responsiveness. Indeed, human error – compounded by fatigue – results in reviewers “disagree[ing] with each on more than half of the responsiveness claims.”
Dynamo Holdings II also referred to the fact that search terms are often “flawed,” a theme that Judge Peck spotlights in Hyles. In Hyles, the requesting party (Hyles) argued that TAR is a “more cost-effective and efficient method of obtaining ESI” than keywords.
In response, Judge Peck cited various cases supporting this notion, i.e., that a properly designed TAR methodology would generate better production results than search terms. Though he ultimately declined Hyles’ request, Judge Peck clarified the primacy of TAR over search terms: “To be clear, the Court believes that for most cases today, TAR is the best and most efficient search tool.”
Cooperation and Its Limits
Cooperation is an essential aspect of discovery, particularly in the context of TAR. This is because a cooperative and transparent approach to developing a TAR process may help insulate the producing party from attacks by litigation adversaries on the adequacy of that process. The result from Dynamo Holdings II confirms as much as Judge Buch rejected the requesting party’s demand for a production do-over using search terms. Since it participated in both the development and training of the TAR algorithm, the requesting party would not be permitted to seek a broader subset of materials based on its speculation that other responsive documents had yet to be produced.
Like Dynamo Holdings II, Hyles emphasizes the importance of cooperation. Nevertheless, it also places reasonable limits on its application. A court cannot use a requesting party’s willingness to cooperatively and jointly develop a TAR process, explained Judge Peck, as leverage “to force the responding party to use TAR.” Such a position runs counter to the well-established rule reflected in Sedona Principle Six that the responding party is best situated to determine how it should search for, review, and produce its responsive documents. While it may be “impossible to argue against cooperation” in discovery, Hyles confirms that a court cannot coerce “a responding party to use TAR when it prefers to use keyword searching.”
The Federal Rules of Civil Procedure (FRCP) generally require that productions satisfy standards of relevance, proportionality, and reasonableness. While these traditional discovery touchstones remain unchanged in the age of eDiscovery, “the myth of a perfect response” continues to spawn expensive and futile satellite litigation as evidenced by Dynamo Holdings II and Hyles.
Dynamo Holdings II debunked this myth and pointed to the “reasonable inquiry” standard under FRCP 26(g) as the correct benchmark for evaluating a production. The parties’ jointly developed TAR process adequately satisfied that standard given that it resulted in a proportionate production of responsive documents. While additional responsive information might have been located with search terms, that position was of no moment given the adequacy of the producing party’s underlying response.
In like manner, Hyles emphasized that productions need not be perfect. Instead, they must be “reasonable and proportional.” Given this backdrop, Judge Peck refused to order the producing party to use TAR in lieu of search terms. Even though a production made using search terms “may not be as complete as it would be if TAR were used,” it can still satisfy notions of relevance, proportionality, and reasonableness.
Developing a proper understanding regarding the above-referenced issues is essential for ensuring more cost effective and efficient uses of TAR, together with other discovery search methodologies. Indeed, lawyers must grasp and follow judicial guidance on these issues if we realistically expect TAR to be “widely used” one day as Judge Peck envisions.