The Future Has Arrived: For the First Time, Judge Orders Predictive Coding in a Federal Case

Just a few weeks ago, Magistrate Judge Andrew Peck (S.D.N.Y.) spoke to several hundred people at LegalTech New York about the importance of predictive coding for the future of electronic discovery. He expressed his hope that a federal court would, sooner rather than later, officially encourage using the technology in a case.

Shortly after participating in the panel, Judge Peck fulfilled his own wish. Last week, he became what appears to be the first federal judge to order litigants to use the cutting-edge technology in a case.

Let’s look at the details, as well as take a little refresher on predictive coding…

The Law Technology News has the scoop:

Predictive coding, aka computer-assisted review, is an evolving technology that provides litigants an alternative to the time and cost of manual review of large document sets. While the use of predictive coding is promising to reduce document sets and the cost of reivew, the technology’s reliability and defensibility have yet to determined by the courts. However, Magistrate Judge Andrew J. Peck of the U.S. District Court for the Southern District of New York is on the cusp of probing the reliability of predictive coding.

In what appears to be the first federal case to adopt the use of automated coding, Peck, in Da Silva Moore v. Publicis Groupe et al., ordered the parties to adopt a protocol for e-discovery that includes the use of predictive coding as implemented by Recommind’s Axcelerate product.

Paul Neale, CEO of DOAR Litigation Consulting and Gene Klimov, vice president of Discovery Consulting, advised the plaintiffs and the court on developing a protocol for e-discovery that used iterative sample sets of 2,399 documents from a corpus of 3 million documents (95 percent confidence level; plus or minus 2 percent variance). In effect, the parties will review from 15,000 to 20,000 documents to instruct Axcelerate on what documents are relevant in the litigation, which is no easy matter in class actions like Da Silva Moore, or in other cases that plead multiple issues of law and fact.

As we have mentioned before, supporters of predictive coding say it has the potential to cut the discovery costs by massive percentages and drastically increase the accuracy of document review.

Sponsored

Assuming nothing horrendous goes wrong as De Silva Moore continues to progress, simply having a judge-recommended predictive coding case on the books is a big step. it has the potential to give computer-assisted review and air of official recognition that other more established technologies, like keyword searching, don’t even have. (For what it’s worth, Peck strongly dislikes keyword searching, even though, as he explained at the conference, everybody does it anyway.)

It’s worth mentioning, even though it’s not super relevant to the big picture, the case in question is a Title VII gender discrimination class-action against Publicis Groupe, an advertising conglomerate. The plaintiff alleged that she and other women “suffered discriminatory terminations, demotions, and job reassignments”:

Discovery ensued in the case to flesh out common questions of fact, inter alia, whether Publicis Groupe compensated female employees less than similarly situated males via salary, bonuses, or perks; precluded or delayed the selection and promotion of females into higher level jobs held by male employees; and carried out terminations or reassignments when the company was reorganized in 2008 that disproportionately impacted female employees. The common questions typify the problem of using keywords to query e-discovery corpora and retrieve sample sets.

I guess this means the future is now. What will happen if predictive coding takes off? Worst-case scenario: fewer contract attorney document review jobs for recent law school graduates. Best case scenario: fewer contract attorney document review jobs for recent law school graduates.

Judge Peck Orders Predictive Coding in Federal Case [Law Technology News]

Sponsored

CRM Banner