Nature abhors a vacuum. So, too does the challenging field of e-Discovery, where data sources are growing in number and complexity in inverse proportion to storage costs. The phenomenon has some litigants grappling with the astronomical costs associated with mining mountains of electronically stored information (ESI) to find what they need. Filling that vacuum is a nascent methodology called technology-assisted review that has the potential to modernize document review processes and significantly reduce associated costs.
TAR (also known as computer-assisted review or predictive coding) is a process for prioritizing or coding a collection of documents using computer programs to quickly and efficiently search large amounts of data for data that meet specific requirements. Some TAR methods use training sets to establish algorithms capable of distinguishing relevant from non-relevant documents. Other TAR methods systematically devise rules that emulate human reviewers’ decision-making process. Both use sampling techniques or statistical models to guide the process and measure effectiveness.
In a 2012 RAND study ("Where the Money Goes: Understanding Litigant Expenditures for Producing Electronic Discovery") document review accounted for 73 percent of e-Discovery costs in the 49 cases studied, which included traditional lawsuits and regulatory investigations. The study indicated that although there is scant research on the issue, evidence suggests that the costs of predictive coding are likely to be substantially lower than the costs of human review due to the reduction of in-person hours needed for large-scale production.
The study also concluded that the “exponential growth in digital information, which shows no signs of slowing, makes computer-categorized review strategy, such as predictive coding, not only a cost effective choice, but perhaps the only reasonable way to handle many large-scale productions.”
Last year, the courts began recognizing computer-assisted review as an acceptable way to search for relevant ESI in appropriate cases. In one notable case last October (EORHB v. HOA Holdings LLC), the court directed the parties to use predictive coding, and in a later ruling directed both parties to use the same vendor and document repository.
“These decisions on TAR indicate a court’s willingness to entertain new technology such as predictive coding to expedite and make the discovery collection process more cost-efficient,” said Paul Burns, senior counsel at Procopio, Cory, Hargreaves & Savitch LLP, and a nationally recognized e-Discovery expert. “It reflects a trend that the court will approve a stipulation that the parties will use TAR.”
Burns cautioned that TAR should not be applied across the board, but in appropriate cases it may be a reasonable way to conduct discovery. Lawyers and support teams will also need to achieve a certain comfort level as they learn the pros, cons, limits and benefits.
“As the artificial intelligence aspects of the software improve, the number of cases where it’s appropriate will increase over time,” he said.
Burns also serves as pro-bono general counsel for The Sedona Conference, a legal think tank. A member of its executive committee, he said he is working with an e-Discovery board comprised of various constituents, including defense and plaintiff attorneys, academics and advisory board judges, to draft practical guidelines and establish a protocol that will satisfy the requesting parties' concerns while allowing the responding party similar latitude in deciding on how to conduct its own discovery process.
To address defensibility, one of the biggest concerns of using TAR, The Sedona Conference commentary, which Burns estimates will be available for public review and comment in the fall, will also establish guidelines that define when the challenge to results is appropriate and what the appropriate scope of judicial review should be.
“Part of the process for counsel in deciding whether to use TAR is to consider the probability that the use is likely to be challenged if you don’t have consent or stipulation from the opposing side, the extent to which there is anticipated costly litigation over having to defend the methodology, and the risk of having to re-do discovery collection search and review a second time if the judge doesn’t like the way it was done with the TAR tool,” Burns said. “It’s one of the risk factors that has to be considered.”
Although TAR is a nascent technology, according to Burns there is an ever-increasing demand from in-house counsel to use it.
“They recognize the potential for savings and are demanding their outside counsels become knowledgeable, familiar and comfortable with TAR and to realize that we’re not going to get perfection in review and production of discovery.”
Burns dismissed the notion that one barrier to widespread use is resistance from law firms that would stand to lose a historical revenue stream.
“I believe my colleagues at the bar will always want to do what is best for the client and wouldn’t want to be in a position where they’re charging clients more than what needs to be done in a case prudently,” Burns said. “Faced with the challenge of reviewing millions of documents, lawyers would welcome a reliable — capable of validation and verification — technology tool to make the process more efficient. I don’t think the driver is loss of revenue to a firm. I think the driver is fear of whether or not the software reviewing process is going to work and be accurate.”