While e-discovery software is commonly used by law firms to cut down on the time and costs associated with document review, an evolving technology called “predictive coding” is making some firms uneasy about the defensibility and reliability of producing documents.

Unlike previous e-discovery methods that relied on humans to develop and conduct keyword searches, predictive coding (also known as computer-assisted review) is an automated process that culls through electronic documents and analyzes them for attributes other than keywords such as context and word frequency. It’s not, however, without human input, because the software requires attorneys familiar with a case to set the search parameters ["The Electronic Eye," Litigation Supplement, Fall 2011].

But even though there are tremendous cost-saving advantages to computer-assisted review, some firms have been skittish about it. Concerns have largely been twofold: The software could miss relevant material, thus leaving firms open to malpractice claims; and defendants could tamper with the search specifications in order to hide potentially relevant documents.

Until recently, the courts have been mum about the issues surrounding computer-aided review. But in February the silence was broken when Manhattan federal magistrate judge Andrew Peck weighed in with the first judicial opinion about the reliability of the automated coding.

The ruling came in Da Silva Moore v. Publicis Groupe et al., a gender discrimination suit brought by five female employees against the advertising giant and its subsidiary, MSL Group, a U.S. public relations firm. (Morgan, Lewis & Bockius and Jackson Lewis are representing the companies, respectively.) The plaintiffs, represented by Sanford Wittels & Heisler, allege that both companies have a glass ceiling that limits women to entry-level positions, and that their male counterparts are paid more and promoted more often.

The discovery phase in this case generated some 3 million e-mails. After several holding discovery conferences and rulings, federal district court judge Richard Sullivan, who had been assigned to the case, referred it to magistrate Peck for pretrial supervision. (Peck has written about computer-assisted review for sibling publication Law Technology News ["Search, Forward," LTN, September/October 2011].)

During Peck’s first discovery conference, both parties mentioned that they had been discussing an electronic discovery protocol, but that the the plaintiffs were reluctant to utilize predictive coding to cull down the 3 million electronic documents produced during discovery. However, since both parties expressed a willingness to use the software — Axcelerate by Recommind, Inc. — Peck ordered the two sides to draft a protocol that they could both agree to.

On February 24, Peck issued what appears to be the first judicial opinion on predictive coding. “The decision to allow computer-assisted review in this case was relatively easy — the parties agreed to its use (although disagreed about how best to implement such review),” he wrote. Peck continued: “The court recognizes that computer-assisted review is not a magic, Staples-Easy-Button solution appropriate for all cases. The technology exists and should be used where appropriate, but it is not a case of machine replacing humans: It is the process used and the interaction of man and machine that the courts need to examine.” Peck further noted that the bar should take note of his opinion, and that attorneys no longer had to worry about being the “guinea pig” for judicial acceptance of computer-assisted review.

While defenders, and vendors, of automated coding technology cheered Peck’s decision, not everyone was happy about it, including the plaintiffs’ attorneys who initially objected to the ruling. Three weeks after Peck issued his opinion, they filed a brief in support of their objections to Peck’s ruling with U.S. district court judge Andrew Carter (now the presiding judge). At press time Carter had not yet ruled on the objections. •