The Sedona Principles for Electronic Document Production
Sedona Principle 6:
Responding parties are best situated to evaluate the procedures, methodologies, and technologies appropriate for preserving and producing their own electronically stored information.
In the predictive coding world, ‘pro-choice’ advocates are hailing two recent wins.
In two recent orders — one from a federal court in Indiana, the other from a state court in California — each court considered a motion filed by one party seeking to force the other party to use a particular predictive coding methodology to identify documents for production. In both instances, the court rejected the request and permitted the producing party to employ its chosen method for identifying and producing documents. In each case, the producing party invoked Principle 6 of The Sedona Principles: Best Practices Recommendations & Principles for Addressing Electronic Document Production (2d ed. 2007), which states that “responding parties are best situated to evaluate the procedures, methodologies, and technologies appropriate for . . . producing their own electronically stored information.” That principle — along with the tenets of reasonableness and proportionality — carried the day in both cases.
Biomet M2a Magnum Hip Implant Products Liability Litigation
In the federal case, In re: Biomet M2a Magnum Hip Implant Products Liability Litigation (MDL), No. 3:12-MD-2391 (N.D. Ind. Apr. 18, 2013), defendant Biomet first began producing documents to comply with discovery demands in various cases that eventually were consolidated in the MDL; Biomet then proposed to produce those same documents in the MDL in response to almost identical discovery requests. To identify documents for production, Biomet used a combination of techniques. First, it used keyword searches to cull the collected document set from 19.5 million documents and attachments down to 3.9 million. Then it used a predictive coding tool on the culled (and de-duplicated) data set, identifying almost 2 million documents for production. According to Biomet, it had already spent $1 million on discovery, and expected to spend up to $2.25 million more.
Plaintiffs challenged Biomet’s methodology. They argued that keyword searches should not be used to cull the collection, and requested instead that Biomet be ordered to run predictive coding on the entire 19.5 million-document collection. According to the plaintiffs, keyword searches are less accurate than predictive coding, and Biomet’s efforts were tainted by first using keyword searching before deploying predictive coding. Plaintiffs also asked the court to impose a protocol for predictive coding similar to the one used in Actos, in which the parties worked collaboratively to make relevance decisions about documents and train the predictive coding tool.
The court rejected the plaintiffs’ requests. According to the court, Biomet’s methodology satisfied the standard set out in Federal Rules 26 and 34, namely, that its efforts must be “reasonable.” And although Sedona Conference principles and local discovery rules encourage parties to cooperate in discovery, such cooperation does not require “counsel from both sides to sit in adjoining seats while rummaging through millions of files that haven’t been reviewed for confidentiality or privilege.” Order at p. 4. Finally, the plaintiffs’ demand that Biomet go back to “square one” and re-do its entire search and production using predictive coding at an earlier stage would violate the proportionality standard in Rule 26(b)(2)(C). Id. at 5. Through statistical testing of its search terms, Biomet demonstrated that a relatively small number of documents were left behind in the culling. Given the small number of potentially missed documents, the cost and effort required to find those additional documents would not be warranted. As the court explained:
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 [plaintiffs’] theory that predictive coding would produce a significantly greater number of relevant documents. Even in light of the needs 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 in resolving the issues, I can’t find that the likely benefits of the discovery proposed by [plaintiffs] equals or outweighs its additional burden on, and additional expense to, Biomet.
Id. at 6.
The Biomet order advances the judicial discussion of predictive coding in a few key respects. For starters, this is the first case in which a court found reasonable the use of keyword searches to cull a collection, followed by the application of predictive coding on the culled data set. We often counsel clients that in many cases predictive coding can replace – not augment – key word searching, but Biomet illustrates the flexibility of predictive coding workflows. So long as the initial culling by key words is tested and validated to counsel’s satisfaction, shrinking down the volume of documents put through predictive coding can be a significant cost saver. And speaking of cost, Biomet also recognized that predictive coding is not always a cheap solution, and a request that a producing party use predictive coding must be considered in light of the cost and burden of doing so. Finally, the court in Biomet recognized that an Actos-like protocol, in which parties sit at the same table to code documents by agreement, is not suitable for every matter. Parties should not be dragged to the collaboration table unwillingly.
Fosamax-Alendronate Sodium Drug Cases
In the California matter, Fosamax/Alendronate Sodium Drug Cases, No. JCCP 4644 (Orange Co. Cal. Sup. Ct.) (Apr. 18, 2013), the court issued a short minute order without much analysis, but based on the parties’ briefing, the judge considered arguments similar to those in Biomet. In Fosamax, the producing party Merck used a keyword search methodology to identify and produce more than 11 million documents in a series of related product liability cases. Plaintiffs sought to require Merck to re-collect and re-search documents in the current matter using predictive coding technology instead.
Like Biomet, Merck argued that it should be permitted to use any reasonable methodology of its choosing to identify documents for production, and that its search term process constituted the “reasonable search” and “diligent inquiry” required by California rules of procedure. The “do-over” requested by the plaintiffs would impose significant expense and effort, and would not provide sufficient benefit in return.
Again, the court agreed with the producing party. Plaintiffs had not shown any material deficiency in the search term process used by Merck to produce documents in other cases. Nor could the plaintiffs show that the additional benefit from a new production would outweigh the significant burden to Merck of re-collecting and re-searching using predictive coding, especially considering the already available responsive documents.
Although Fosamax does not provide much judicial analysis, the order nevertheless is significant for its recognition that keyword searches – if carried out reasonably and defensibly, with adequate supporting documentation and testimony – are still legitimate means of finding responsive information. It also reinforces the notion that proportionality will play an increasing role in the analysis of discovery disputes in this era of massive volumes of electronically stored information.
At DiscoverReady, we embrace the use of predictive coding and other automated strategies for culling, searching, and identifying documents in discovery. But like the producing parties in Biomet and Fosamax, we also maintain that there is no one tool, technique, or protocol that fits all cases. The available technologies must be considered in light of the facts of each unique matter, and considerations of cost, time, and effort must be taken into account. What may be reasonable and proportional in one case may not be in another. We agree with the Sedona Principles, especially Sedona Principle 6, and strive every day to collaborate effectively with our clients to “evaluate the procedures, methodologies, and technologies appropriate” for their discovery matters.
A recognized thought leader in e-discovery, Maureen collaborates with the company’s clients and operations teams to develop innovative information strategies for legal discovery, compliance, and sensitive data protection. She speaks and writes frequently on significant issues in e-discovery and information governance, and participates actively in the Sedona Conference Working Groups on Electronic Document Retention and Production and Data Privacy and Security. Prior to DiscoverReady, Maureen was a partner at Paul Hastings LLP, where she represented Fortune 100 companies in complex employment litigation matters.