Use of Predictive Coding was “Reasonable Inquiry,” Motion to Compel Additional Discovery Denied

Dynamo Holdings Ltd. P’ship v. Comm’r of Internal Revenue, No. 2685-11, 8393-12, 2016 WL 4204067 (T.C. July 13, 2016)

In September 2014, the court approved Petitioners’ use of predictive coding to identify potentially responsive and privileged data contained on two backup tapes, despite Respondent’s objection that the technology was “unproven.” (Read a summary of that opinion here.)  At that time, the court indicated that Respondent could move to compel additional discovery in the event he believed that Petitioners’ response was insufficient.   Accordingly, after Petitioners denied Respondent’s request for production of additional documents containing certain specified search hits, Respondent moved to compel.  Concluding that Petitioners’ reliance on predictive coding satisfied the requirement for a “reasonable inquiry,” the court denied the motion.

Over Respondent’s objection, the court approved Petitioners’ use of predictive coding to identify responsive materials. Thereafter, the parties worked together to craft a predictive coding protocol, with the court stepping in to resolve areas of disagreement.  Upon completion of the process, Petitioners “delivered a production set of approximately 180,000 total documents” for Respondent’s review, 5,796 of which he retained and returned the rest to Petitioners.  In June 2016, Respondent filed his motion to compel, seeking production of documents identified by specified terms which were not previously produced.

Following an informative discussion of “recall versus precision” (two important concepts in predictive coding), the court indicated it would “assume” (as Respondent had argued) that the predictive coding response was “flawed” (because the level of recall was too low) and turned to the question of whether relief should therefore be afforded.

The court began its analysis with the premise that “Respondent’s motion [was] predicated on two myths”: the “myth of human review” and the “myth of a perfect response.” The “myth of human review” (per the court, citing the Sedona Conference) is the idea that manual review by humans “constitutes the gold standard by which all searches should be measured.”  The “myth of a perfect response”—perhaps more obvious on its face—is the idea that a discovery response should be perfect.  Regarding the first “myth,” the court cited a 2012 report from RAND Corp. establishing that “human review is far from perfect” where humans frequently disagree regarding what constitutes a  responsive document.  Regarding the second “myth” the court made clear that the Tax Court Rules (and the Federal Rules of Civil Procedure) “do not require a perfect response,” but rather “require that the responding party make a ‘reasonable inquiry’ before submitting the response.” Moreover, the court reasoned that “[t[he fact that a responding party uses predictive coding to respond to a request for production does not change the standard for measuring the completeness of the response.”

Accordingly, the court concluded:

There is no question that petitioners satisfied our Rules when they responded using predictive coding. Petitioners provided the Commissioner with seed sets of documents from the backup tapes, and the Commissioner determined which documents were relevant. That selection was used to develop the predictive coding algorithm. After the predictive coding algorithm was applied to the backup tapes, petitioners provided the Commissioner with the production set. Thus, it is clear that petitioners satisfied our Rules with their response. Petitioners made a reasonable inquiry in responding to the Commissioner’s discovery demands when they used predictive coding to produce any documents that the algorithm determined was responsive, and petitioners’ response was complete when they produced those documents.

Respondent’s motion was denied.

A full copy of the court’s order is available here.