Predictive Coding Gaining Popularity in Complex Litigation

July 5, 2012Articles
As seen in Business Lexington.

Think about how many emails, text messages and instant messages you send and receive each day. According to The Radicati Group, the average corporate employee sends and receives about 110 emails per day and the number doubles if they are one of the many who now use Instant Messaging at work or send text messages from company phones. Researchers from the University of California, San Diego recently found that the average employee creates or handles 1.6 gigabytes of information – equivalent to 100,000 pages of printed text - on a daily basis. These numbers are going to keep growing and if your company is involved in litigation, you know that courts now impose a high duty on litigants to preserve electronic documents – including emails – whenever they reasonably contemplate litigation. Failure to do so can result in sanctions even if it means preserving hundreds of thousands or even millions of electronic documents.

The cost of first preserving this data and then paying your attorney to review it can be daunting. As a result, some law firms have developed document review strategies that incorporate advanced software to sort and filter potentially relevant or privileged documents. One such technology, called “predictive coding,” uses techniques to filter documents based on a combination of keyword searches and human-guided sampling. That is, an attorney reviews and codes a sample set of documents. Then a computer program identifies the properties of those documents to refine search parameters based on the attorney’s coding. Proponents say that this is a more efficient and cheaper way to review documents because it reduces the hours attorneys must spend manually reviewing records.

Of course, lawyers, and the law in general, can be slow to accept advancements in technology. We tend to be creatures of habit and skeptical of things that alter the way we have (successfully) done things in the past. Most lawyers have spent their careers ensuring that every single document from their clients has been manually reviewed for relevance and expect that their opposing counsel does the same before the two sides exchange records. Even five or six years ago this was probably the norm, but as more courts order the production of vast amounts of electronically stored information, is it reasonable for lawyers to continue this practice?

As always, there are two sides to every coin. Some argue that using key word searches and predictive coding for document review is inherently unreliable because, by definition, some documents that will not get tagged for review. We have all seen examples of the extraordinary things that people put in emails, so what if the “smoking gun” is in one of the documents that did not have the right key words or parameters? Others argue that, when done properly, the algorithms used in computer assisted searches are more reliable than having very junior level associates review hundreds of thousands of documents and then letting them decide which records are relevant.

So, who has the better argument? The answer depends largely on who you talk to and the circumstances of a particular case, but earlier this year, a federal judge in New York, in the case of Moore v. Publicis Groupe & MSL Group, 2012 U.S. Dist. LEXIS 23350 (S.D. N.Y.), became the first judge to recognize computer assisted review as an acceptable way to search electronic information in response to a discovery request. Moore involved a review of three million emails. The attorneys representing the females who had allegedly been discriminated against objected to the employer’s desire to use predictive coding to sort the relevant emails for fear that “key” documents would be missed. The court fashioned a strict protocol for the predictive coding search requiring the employer defendant to take several steps to ensure that the review was reliable such as jointly working with opposing counsel to create boolean search terms and requiring more senior attorneys to create the initial sample set of documents to be coded by the computer program.

Moore may be a turning point for widespread acceptance and use of predictive coding technology. Many of us believe it is the future and have already begun using computer software to assist with document review because of the cost savings benefits. Informed discussions between attorneys and clients about the pros and cons of using these searches are imperative. An open line of communication with opposing counsel is also important. If your opponent has concerns about the process, resolve them early by agreement or with the court’s assistance.