Opinion Highlights Questions Surrounding Proper Predictive Coding Protocols

Gordon v. Kaleida Health, No. 08-CV-378S(F), 2013 WL 2250579 (W.D.N.Y. May 21, 2013)

In this case, the court considered Plaintiffs’ motion “to compel Defendants to meet and confer with respect to establishing an agreed protocol for implementing the use of predictive coding software” and for an order that if the parties could not agree on an ESI protocol, that each side would submit their proposal for the court’s determination as to which should be adopted.  Although Defendants initially objected to meeting and conferring with Plaintiffs and their experts based on Plaintiffs’ expert’s prior work for Defendants in the same case, the issue was ultimately resolved when Defendants indicated they were prepared to meet and confer (perhaps motivated by the court’s denial of their motion to disqualify Plaintiffs’ experts).  Despite the relatively simple resolution of Plaintiffs’ motion to compel, the parties’ arguments raised interesting questions related to the level of cooperation required when one side has chosen to utilize predictive coding.

After attempting to agree with Plaintiffs on a keyword search protocol, Defendants decided to utilize predictive coding after the court directed the parties’ attention to the recent decision in Da Silva Moore v. Publicis Groupe SA, 287 F.R.D. 182 (S.D.N.Y. 2012), in which predictive coding was approved by the court.  Thereafter, Defendants provided Plaintiffs with their ESI protocol and indicated they would also provide a list of their email custodians.  Plaintiffs objected to the protocol and noted several issues which they thought should be discussed “with the assistance of Plaintiffs’ ESI consultants and cooperatively resolved by the parties” before Defendants’ predictive coding efforts were initiated.  Several days later, Plaintiffs filed their motion to compel Defendants to meet and confer.

Plaintiffs contended (citing Da Silva Moore) that “where a party intends to utilize predictive coding to assist in the review and production of ESI, it is necessary that the parties negotiate a protocol to guide the use of predictive coding software for the case” and argued that Defendants refused to meet and confer with the parties’ respective experts to discuss several issues “important to a cooperatively negotiated ESI protocol needed for this case.”  More specifically, Plaintiffs maintained that Defendants’ position—“that the only issue they need to discuss with plaintiffs . . . is custodians”—“exlude[d] Plaintiffs’ access to important information regarding Defendants’ selection of so-called ‘seed set documents,’” used to “train” the computer for predictive coding.

Defendants, in turn, indicated they had no objection to meeting and conferring with Plaintiffs and their consultants generally, but specifically objected to meeting with those consultants that were the subject of Defendants’ motion to disqualify.  Further, Defendants asserted that courts “do not order parties in ESI discovery disputes to agree to specific protocols to facilitate a computer-based review of ESI based on the general rule that ESI production is within the ‘sound discretion’ of the producing party” and that the court in Da Silva Moore did not require the defendants in that case to provide the plaintiffs with their seed set documents, but rather the defendants volunteered to do so.  Plaintiffs did not “directly contest Defendants’ reading of the specific holding in [Da Silva] Moore regarding the absence of any requirement by that court that the parties meet and confer regarding the producing party’s selection of a ‘seed set of documents.'”  Ultimately, questions surrounding the level of disclosure required by Defendants remained largely unaddressed by the court, in light of Defendants’ willingness to meet and confer with Plaintiffs’ experts.  Nonetheless, the disagreement highlighted important questions surrounding the predictive coding process.