Ever since Magistrate Judge Peck’s decision last year in Da Silva Moore v. Publicis Groupe SA, 2012 WL 607412 (S.D.N.Y. Feb. 24, 2012), there has been an increasing stream of orders and opinions weighing in on the use (or proposed use) of predictive coding. With each opinion, a new wrinkle appears, further shaping how parties are conducting technology assisted review during discovery.

Last month, Judge Robert Miller Jr., of the United States District Court for the Northern District of Indiana, joined this group with his order in In Re: Biomet M2a Magnum Hip Implant Products Liability Litigation (MDL 2391). There he found reasonable the process that had been undertaken by defendant Biomet in a multi-district litigation, involving the use of keyword searches followed by predictive coding, despite complaints from the plaintiffs’ Steering Committee.

The Steering Committee complained that Biomet had started the document culling process with keyword searches before the application of predictive coding, and then had reviewed and produced millions of pages prior to the transfer of the cases to the MDL. Their specific grievance with keyword searching relied on (1) studies indicating that “Boolean searches identify less than a quarter of the relevant documents,” and (2) a New York Journal article referencing the Global Aerospace case in which the court accepted defendants proposal that “predictive coding at a 75 percent responsive rate would be sufficient.” In short, the Steering Committee argued that predictive coding is better than keyword searching and the Court should order Biomet to restart the process, apply predictive coding to the original population of 19.5 million documents, and collaborate throughout the review process. Biomet submitted evidence that this “do over” would cost in the “low seven figures” on top of document review expenses already predicted to be between $2 and $3.25 million. The Steering Committee justified this additional cost as the price Biomet deserved to pay for beginning its document review before the cases were consolidated in the MDL process, and thus before the Steering Committee was formed and available for “collaboration,” and failing to observe what the Steering Committee argued is the proper procedure for predictive coding reviews.

Judge Miller rejected the Steering Committee’s request for a “do over”, stating that Biomet fully complied with the Federal Rules of Civil Procedure 26(b) and 34(b)(2). He accepted Biomet’s defense of its discovery process and various metrics it provided to support its position, noting that “even in light of the need of the hundreds of plaintiffs in this case, the very large amount in controversy, the parties’ resources, the importance of the issues at stake, and the importance of this discovery, [he could not] find that the likely benefits of the discovery proposed by the Steering Committee equals or outweighs its additional burden on, and additional expense to, Biomet.” The Court held that the Steering Committee’s request “sits uneasily with the proportionality standard in Rule 26(b)(2)(C).” He also relied on available scholarly sources in reaching his decision, including The Sedona Conference® Commentary on Proportionality in Electronic Discovery and the Seventh Circuit Principles Relating to the Discovery of Electronically Stored Information.

Judge Miller also specifically addressed the keyword search question, adopting Biomet’s position that the use of such searches remains an effective component of the document review process. As described in detail in its submission to the court, Biomet used fairly broad search terms and phrases, i.e., “metal on metal”, “all metal hip”, cobaltism, “metal /2 toxic”, in order to reduce the initial collection from 19.5 million documents to 2.5 million after deduplication (constituting nearly 1.5 terabytes of data). Biomet’s brief explained how statistical sampling established that between 0.55 and 1.33 percent of the “unselected” documents (the “null set”) of the remaining 19.5 million documents would be responsive. From there, Biomet deployed a well-known predictive coding tool, in order to refine the “selected” document set and identify documents for further contract attorney review and production.

Judge Miller also rejected the Steering Committee’s position that Biomet should have waited until the centralization of the lawsuits before starting its document review process, noting that that there was no evidence presented that Biomet could wait without facing judicial repercussions in the individual actions. After noting the benefits of the transfer process by the MDL Panel, Judge Miller noted that “to hold that a party that behaves as the transferee court directs, or that follows the transferee court’s standing procedures, does so only by forfeiture of the proportionality provision of Rule 26(b)(2)(C), seems an uncongenial exercise of whatever discretion [he had].”

The Biomet decision is an important example of the application of the “proportionality” principles of Rule 26(b)(2)(C) to evaluate a party’s discovery obligations. As Judge Miller held:

The issue before me today isn’t whether predictive coding is a better way of doing things than keyword searching prior to predictive coding. I must decide whether Biomet’s procedure satisfies its discovery obligations and, if so, whether it must also do what the Steering Committee seeks.

The Court went on to hold that:

It might well be that predictive coding, instead of a keyword search … would unearth additional relevant documents. But it would cost Biomet a million, or millions, of dollars to test the Steering Committee’s theory that predictive coding would product a significantly greater number of relevant documents. … I can’t find that the likely benefits of the discovery proposed by the Steering Committee equals or outweighs its additional burden on, and additional expense to, Biomet. Fed.R.civ.P. 26(b)(2)(C).

An additional point worth mentioning has to do with the Steering Committee’s proposal that in the predictive coding “do-over” process, the plaintiffs and defendants should jointly be involved in coding the training rounds of documents (i.e., entering the “find more like this” commands). Judge Miller recognized that Principle 1.02 of the Seventh Circuit Principles Relating to the Discovery of Electronically Stored Information requires “cooperation,” but he does not “read it as requiring counsel from both sides to sit in adjoining seats while rummaging through millions of files that haven’t been reviewed for confidentiality or privilege.”

There is much discussion currently about whether technology assisted review is better than the alternatives and how much “transparency” and “cooperation” should be required of parties that choose to utilize such tools. See, e.g, Predictive Coding: How Much Transparency and Cooperation Is Required When Using Technology Assisted Review In Litigation? The Biomet decision shows that courts are willing to take a reasonable approach to concepts of proportionality and cooperation in evaluating discovery obligations.