On July 13, 2016, in Dynamo Holdings Limited P’ship v. Comm’r, the U.S. Tax Court strongly defended the taxpayer’s use of computer-assisted review in a dispute with the IRS. In a 2014 decision in the same case, the Tax Court had already endorsed computer-assisted review, namely predictive coding, as a general matter. “Predictive coding is an expedited and efficient form of computer-assisted review that allows parties in litigation to avoid the time and costs associated with the traditional, manual review of large volumes of documents.” Dynamo Holdings, 143 T.C. 183 (2014).

After that 2014 decision, the taxpayer and the IRS jointly developed a process for applying predictive coding to a large set of data in discovery. The IRS was unsatisfied with the results of the process and in 2016 moved the court for an order to compel the taxpayer to use a traditional Boolean search. The Tax Court did not take the motion well. The court found that the IRS’s motion identified no deficiency with the methodology that the taxpayer used and instead attacked only the adequacy of the results of the search. The court held that the IRS “is seeking a perfect response to [its] discovery request, but our Rules do not require a perfect response.” Instead, the Tax Court rules require that the responding party make a “reasonable inquiry” before submitting the response. The court found that the IRS’s arguments were based on the “the myth of [the superiority of] human review” and “the myth of a perfect response.”

This decision is gratifying for its acknowledgment of the indisputable truths that (1) no review will catch every responsive document (complete recall) and exclude every nonresponsive document (complete precision), (2) computer-assisted review may be the best course when dealing with large volumes of data, and (3) systems must focus on the quality of processes, not results, if they are to be at all workable.