Predictive coding is machine-learning technology that enables computers to automatically predict how documents should be classified based on limited human input. Some commentators argue that predictive coding will cure all of E-Discovery’s woes. In Da Silva Moore v. Publicis Groupe, Magistrate Judge Andrew Peck of the Southern District of New York — a leading proponent of the technology— is overseeing one of the leading cases involving predictive coding. The case is, essentially, a Title VII gender-discrimination action. The parties initially agreed to use predictive coding, but negotiations over a mutually agreeable protocol for programming the coding technology broke down. Plaintiffs argued that the court erred in accepting defendant’s recommended protocol, which they contend is flawed. As a result, discovery has been partially stayed and plaintiffs have now moved for Judge Peck’s recusal, arguing his support of the technology and relationships with its vendors have made him biased.
In Kleen Products v. Packaging Corp. of America, plaintiffs in an antitrust matter asked that defendants redo their document production and all future productions using predictive-coding technology. Plaintiffs argued that use of the technology would realize more thorough results than those provided by defendants’ keyword-search technology. This request came after defendants had already spent thousands of hours reviewing documents and had produced more than one million documents. After hearing expert-witness testimony on the sufficiency of their document productions, Magistrate Judge Nan Nolan asked the parties to reach a compromise on keyword searches. In comments on the record, Judge Nolan confirmed that parties may not dictate what technology their opponent may use without showing how the production results are insufficient or inaccurate.
- Predictive coding is E-Discovery’s latest buzzword. The momentum behind predictive coding comes from possible cost-saving advantages the technology offers by searching documents electronically. While courts, litigators, and in-house counsel are still assessing the effectiveness and accuracy, it is most often used in collecting client documents.
- Predictive coding isn’t the only answer for collecting client documents. There are other ways to do what predictive coding tries to do. For example, consider doing targeted collections instead of using predictive coding. Focus on the key custodians who have unique information and actually have the individuals show you where they keep their information. If a custodian is unorganized, develop search terms with that individual unique to that information. You’ll find that this will create an accurate search and greatly reduce the amount of information that needs to be collected and ultimately reviewed—resulting in cost savings.
- Predictive coding hasn’t defeated the detail devil. Operating protocols for predictive coding remain open to dispute in litigation. Even when parties and the court agree on use of predictive coding technology in discovery, negotiations might still break down over protocols for programming the technology. These kinds of disputes could delay discovery, burden courts with additional motion practice, and drive up costs.
Parties in litigation may not dictate what E-Discovery technology their opponents may use without first identifying how document production results are insufficient or inaccurate.
The articles on our Website include some of the publications and papers authored by our attorneys, both before and after they joined our firm. The content of these articles should not be taken as legal advice.