Can Predictive Coding Identify Relevant and Non-Privileged Data on Back-Up Tapes?

3 Oct 2014

In our last post, we discussed the consolidated cases of Dynamo v. Commissioner of Internal Review and Beekman v. Commissioner of Internal Review, Docket Nos. 2685-11, 8393-12 (US Tax Ct. Sept. 17, 2014). In a quick review of the facts, Respondent sought back-up tapes of ESI with metadata. Petitioner admitted the tapes contained relevant electronic data, but claimed that a full privilege and relevancy review was prohibitively expensive. Petitioners offered to use predictive coding to review for privilege and cull the production. Respondents objected and wanted the tapes in full, offering a clawback order to address the privilege concerns.

Did the Tax Court Allow Predictive Coding in this Case?

The court agreed that Petitioners need not turn over the back-up tapes without first reviewing them. It noted that while a clawback order is a good clause to have in an eDiscovery agreement, it certainly would not prevent a party from seeing privileged information in the first place. The court also agreed that human review would be unduly expensive.

Therefore, the court agreed a fair compromise would be to use predictive coding. Predictive coding is a form of computer-assisted review which combines human guidance with computer-controlled searching. A human-reviewed “seed set” trains the eDiscovery software to recognize relevant information within a larger group of documents. Through the coding of this small sample of documents, computers can predict the relevance of documents to a discovery request and then identify which documents are and are not responsive in the full production. There is also a log which details the records that were withheld and why.

Is Predictive Coding “Unproven Technology” As Respondent Asserted?

The court concluded predictive coding was not unproven technology, and it cited cases and articles about how this form of computer review was now accepted in the technology industry as effective. Finding Petitioner’s offer to use predictive coding to be a reasonable compromise for the dispute, it ordered the Petitioners to respond to the Respondent’s eDiscovery requests. If Respondent later believes the ESI production to be incomplete, it can file an appropriate motion to compel at that time.

ILS – Plaintiff eDiscovery Experts