By Ann Marie Gibbs
Daegis
In rapid succession, the month of April saw two courts approve the use of technology assisted review (TAR), also known in the industry as predictive coding. In both cases, the judge noted that the receiving parties can continue to avail themselves of traditional e-discovery methods if they choose to seek relief by challenging the "completeness" of the production.
In Da Silva Moore v. Publicis Groupe in the U.S. District Court for the Southern District of New York, Judge Andrew Carter affirmed U.S. Magistrate Judge Andrew Peck's order instructing the parties to proceed with a disputed protocol for implementing TAR. Carter observed that " … even if all parties here were willing to entertain the notion of manually reviewing the documents, such review is prone to human error and marred with inconsistencies … " This decision from the highly regarded Southern District of New York will undoubtedly accelerate the adoption of technology to relieve corporate clients of the pain of costly document reviews.
The second decision in Global Aerospace v. Landow Aviation, (Loudon County Circuit Court in Virginia) echoed Da Silva in reminding the parties that the option of challenging the completeness of the production was always available. It is also the first case to order the use of TAR over a party's objection as the parties in Da Silva initially agreed to use TAR.
Almost immediately after Peck issued his ruling, the plaintiffs filed a motion asking that it be overturned. They also issued a warning that "Judge Peck sets a dangerous precedent that is likely to deter future litigants from even considering predictive coding, lest they be bound by a protocol that contains no measure of reliability." This ominous warning cuts to the heart of a debate that has been raging in the e-discovery industry for some time now. While there have been many e-discovery advancements, no new technology or methodology has generated as much debate as predictive coding.
As Peck recognized in his ruling, predictive coding remains a viable solution to a growing problem — the skyrocketing cost of document review. Research consistently shows that current e-discovery methodologies cannot keep up with the ever-increasing volumes of data. Vendors, litigants and the courts must agree upon ways to embrace the benefits of evolutionary technology, such as predictive coding, rather than decrying its risks. Just as OCR (optical character recognition), concept clustering and relevance ranking have been accepted as viable tools, predictive coding will soon be seen as a preferred solution to a growing problem. TAR provides a defensible alternative to costly and time-consuming manual review, while providing the reasonable results demanded by law.
Practitioners must start by understanding the way predictive coding works. Predictive coding engines have three things in common: a representation of the data, reasoning about responsiveness and interaction between humans and technology. Any system that incorporates predictive coding must have robust quality control and accuracy measurements. Any system that does not include these components cannot be categorized as TAR.
Second, we must recognize that effectively implementing TAR is all about process. As with any other technology tool used in e-discovery, a proper workflow must be in place to create an efficient, defensible and repeatable process. For TAR, this means creating an implementation plan as well as a rigorous quality control workflow. Many of the elements of a traditional document review will be utilized in a successful technology-assisted review.
Third, it is important to remember that this is not technology review, but rather technology assistedreview. No matter which process is used, all predictive coding technologies assume that a human is going to be reviewing documents at some point in the review process. The documents reviewed by humans will be used to "train" the predictive coding software to categorize the unreviewed documents into different groups. Using the intelligence gained from the human reviewers, the predictive coding software will create a map of the unreviewed population, segregating the documents into categories. Different applications will accomplish this task in different ways.
No matter how the predictive coding software categorizes them, additional human review of the documents will be necessary. This step is critical to improving the quality of the final documents being produced because it supplies the computer with additional data points to use when making calculations about what is responsive and what is not. Furthermore, it will ensure that privilege and responsiveness calls are not overlooked.
Attorneys are responsible for ensuring the document review is defensible. To demonstrate the defensibility of a review, you should have a thorough understanding of the technology, a repeatable process and demonstrable methods for measuring accuracy. It also creates a blueprint for repeatable success.
Ann Marie Gibbs is the national director of consulting at Daegis. She is an attorney, author and speaker on numerous topics related to electronic discovery.