Predictive coding in electronic data discovery — aka TAR (technology-assisted review) or CAR (computer-assisted review) — had not one, but three separate sessions at the ninth annual 2012 Georgetown Law Advanced E-Discovery Institute, which wrapped up December 7 at The Ritz-Carlton Tysons Corner in McLean, Va.

The third session was moderated by Conor Crowley, of the Crowley Law Office (and an activist with The Sedona Conference), and featured Brian Calla (member, Eckert Seamans), Wendy Butler Curtis (e-discovery counsel at Orrick, Herrington & Sutcliffe), Conrad Jacoby (senior attorney, Winston & Strawn) and Ralph Losey, partner and national e-discovery counsel at Jackson Lewis.

The group identified and discussed five key 2012 e-discovery cases, and then, to further illustrate hot topics in e-discovery, discussed a hypothetical case where each panelist took a specific role in the discussion (not necessarily reflecting their own views). With humor and many “inside baseball” e-discovery jokes, the panel deftly covered numerous issues confronting practitioners.

Probably the case of the year was Da Silva Moore, et al. v. Publicis Groupe SA and MSLGroup under Magistrate Judge Andrew Peck of the U.S. District Court for the Southern District of New York. In the controversial case, Peck authorized the parties to use computer-assisted technology, believed to be the first time a judge has so ruled. The major aspects of DSM included its emphasis on transparency, the panelists agreed, and the efforts for both sides to reach agreements on how they will proceed with document review.

TAR can be used effectively in employment disputes where employees are familiar with a defendents’ internal jargon and information repositories, they noted. It also can be used to focus on key search terms to segregate documents used to create a seed set.

The second discussed case was Kleen Products v. Packaging Corp. of America (No. 10-5711), before the U.S. District Court for the Northern District of Illinois, where the defendants wanted to use sophisticated iterative Boolean keyword search with quality control samplings. The platintiff argued that defendant’s proposed approach would only capture 25 percent of relevant documents, the panelists explained. The case was settled, with no ruling on the predictive coding issues, but Nolan (who has just announced her retirement in January) had expressed support for Sedona Principal #6 — that the producing party should have a primary say in preserving and producing its electronically stored information.

Global Aerospace v. Landow is believed to be the second case where a court permitted predictive coding. The plaintiff had objected to the use of TAR, arguing that it was not as effective as human review. Virginia Judge James Chamblin recognized that the producing party selects the review methodology; while the receiving party can raise issues. The court’s order allowed the plaintiffs to later object to the completeness of contents, or ongoing use of predictive coding.

In Re: Actos, was heard before Judge Rebecca Doherty of the U.S. District Court for the Western District of Lousiana. The judge entered a case management order outlining the ESI protocols that the parties must follow during discovery, said the panel. In that case, it was proposed that each side nominate three “experts” to jointly review a control set — and the plaintiff’s expert reviewers were required to sign a non-disclosure agreement. The exports reviewed random sample training sets of 40 documents each, that the system selected using active learning approach. The process is continued until “stable” training status is reach.

The EORHB Inc. v. HOA Holdings LLC (C.A. No. 7409-VCL), case was a complex multimillion dollar commercial indemnity dispute involving the sale of Hooters. The judge suggested that the case might be appropriate for predictive coding, with a shared vendor, and ordered the parties to “show cause” for not adopting that technology. That generated extensive blogging about the judges requirement. See “Impatient Judges Push Lawyers to Dive Into E-Discovery.”

As to the bottom line, the panel agreed that TAR is not a “one-size-fits-all” situation.

After the quick survey of these 2012 cases, the panel dove into the hypothetical exemplar, with no small dose of humor. Calla cracked up the audience when he poked fun at Recommind, identifying himself as the head of Callamind, claiming that he owned predictive coding and had invented the internet. Ralph Losey followed suit, saying he represented Droll OnTruck (In real life, his firm has brought in Kroll OnTrack to handle litigation support.) But despite the good-natured digs, the fake discussions effectively discussed some of the e-discovery connundrums facing litigators, such as sampling methodologies, recall and precision metrics, and confidence levels.

Monica Bay is editor-in-chief of the Law Technology News, a Legal affiliate based in New York.