Nevada District Court Considers Late Request for Predictive Coding
Is predictive coding, also known as automated issue coding, a feasible alternative to reviewing numerous ESI documents solely by human review if requested to modify an Agreed Order? Such were the issues considered by Nevada’s District Court in the case of Progressive Casualty Insurance Company v. Jackie K. Delaney, et al., 2014 WL 2112927 (D.Nev. May 20, 2014).
Progressive was saddled with manually reviewing 565,000 potentially responsive documents. After quickly determining that the manual review would be too time intensive and costly, Progressive sought to utilize predictive coding techniques to review the ESI documents.
Progressive argued that the court should allow for a modification to the stipulated ESI protocols that would permit predictive coding, as opposed to strictly human review. Progressive maintained that predictive coding would achieve a more accurate measure of relevant responsive documents, while saving time and money, with recall rates of 70.80%. Id.
Defendant, FDIC-R, maintained that the predictive coding protocol was overly complicated and carried the potential for numerous satellite disputes. Secondly, FDIC-R argued that other courts that have allowed predictive coding have stressed the need for transparent and cooperative engagement by all parties, of which Progressive had not abided.
The court found that:
- While studies may have determined its increased accuracy, predictive coding requires a heightened degree of transparent cooperation among parties in the review and production of ESI documents as they relate to discovery requests.
- In the few cases that have approved technology assisted review of ESI, courts have consistently required the producing parties to provide the requesting parties with full disclosures about the technology being used, processes, methodology and documents used to “train” computers.
- A point of contention is that litigators are very hesitant to reveal their methodological decisions because of the risk of revealing work product, redundant discovery about discovery and the risks of revealing sensitive non-responsive documents.
Further, the court found that Progressive has been unwilling to engage in the cooperative transparency that is necessary for predictive coding protocol to be accepted by the court or opposing counsel. Approving Progressive’s predictive coding proposal, under such circumstances, would only result in more disputes and delays.
Automated issue coding is an incredibly effective method of document review if done by experts in the field. In this case, Progressive should have done a better job of preparing a strategy from the beginning – the main problem here was that Progressive sought to modify an agreed order too late in the game. Further, they must be open and transparent about the methodology to be used in the issue coding; had it done so, this case would case would have likely come out much differently.
Did you know: Global eDiscovery markets will have a year-over-year growth of around 15% through 2017, increasing from $1.7bn to $2.9bn.