Last week, the Tax Court moved to the cutting edge in addressing the cost and efficiency associated with discovery of electronically stored information (ESI). The Tax Court issued a division opinion in Dynamo Holdings L.P. v. Commissioner, 143 T.C. No. 9, which allowed the taxpayer to use "predictive coding," an algorithm-driven, computer-based method for identifying responsive and potentially privileged documents in ESI.
What Is Predictive Coding?
In a typical ESI discovery case the Commissioner and the taxpayer agree on relevant terms or "keywords". The taxpayer uploads data to an electronic database, filters the ESI using the keywords, and then taxpayer's counsel reviews (typically using junior associates or contract attorneys) the filtered documents for responsiveness and privilege. Using keywords can often reduce the number of potentially responsive documents but taxpayer's counsel must review the remaining filtered documents, usually at significant expense to the taxpayer.
While there are several methods of predictive coding, generally speaking predictive coding relies on algorithms and the review of a limited number of documents to determine the responsiveness and potential privilege of the entire pool of ESI. Taxpayer's counsel reviews and codes a representative sample of the documents. As the reviewer codes the documents, the computer "learns" from the attorney and, using complex algorithms, applies the learned coding to the remaining pool of documents. The reviewing attorney then reviews a subset of the computer-coded documents to ensure that the algorithm is correctly filtering responsive and privileged documents. This process continues until the reviewer's coding and the computer's coding of the reviewed documents are sufficiently similar. Studies on predictive coding have found that a reviewer needs to review only about 2% of the documents to achieve the desired level of computer-reviewer agreement. The computer's learned coding is then applied to the remaining documents. The taxpayer's counsel must then review the pool of responsive documents for privilege before providing them to the Commissioner. The predictive coding process requires transparency and cooperation to ensure that both parties are satisfied with the scope and performance of the computer-aided review.
The Dynamo Decision
In Dynamo, the Commissioner filed a motion to compel production of documents. The motion to compel requested that the taxpayer produce relevant ESI stored on two back-up tapes. Alternatively, the Commissioner requested that the taxpayer provide complete copies of the back-up tapes, subject to an agreement that the taxpayer could claw back privileged documents. The taxpayer objected because of the cost and time necessary to comply with the Commissioner's request. Alternatively, the taxpayer requested that the Tax Court allow the taxpayer to use "predictive coding" to "efficiently and economically identify the nonprivileged information responsive to [the Commissioner's] discovery request." The Commissioner objected to the use of predictive coding, arguing that it was "unproven technology."
The Tax Court rejected the Commissioner's argument and agreed with the taxpayer's expert, who testified that predictive coding was the most reasonable method for responding to the Commissioner's discovery requests. Relying on a number of federal district court opinions, articles, and the aforementioned expert testimony, the Tax Court held that predictive coding was appropriate where a taxpayer: (1) reasonably requests to use predictive coding to conserve time and expense; (2) represents to the Court that the taxpayer will retain electronic discovery experts to meet with the Commissioner's counsel or his experts; and (3) agrees to work with the Commissioner to develop mutually acceptable procedures for the ESI review. The Tax Court held that the use of predictive coding complied with the Tax Court's discovery Rules, which must be "construed to secure the just, speedy, and inexpensive determination of every case."
Predictive coding should be a boon to taxpayers litigating in Tax Court. Taxpayers will need to work with the Commissioner to cooperate on its implementation, but predictive coding can save taxpayers time and significantly reduce ESI discovery costs. The Tax Court's decision is a welcome recognition of the advancement and utility of computer-based discovery techniques.
For additional information, please contact any of the following lawyers:
Kevin Kenworthy, email@example.com, 202-626-5848
Steve Dixon, firstname.lastname@example.org, 202-626-5844
Jarrett Jacinto, email@example.com, 202-626-5820
Nicholas Metcalf,* firstname.lastname@example.org, 202-626-5961
*Admitted in Illinois only. Not admitted in the District of Columbia. Practicing under the supervision of a member of the D.C. Bar.