In the world of searching for relevant documents in the recesses of email inboxes and hard drives, a new high-tech tool has appeared that, despite causing trepidation among some attorneys, will likely become commonly used during the discovery process to tame the growing volumes of data.
Predictive coding is the coming wave in electronic discovery. The software filters through electronically stored information and separates the responsive documents from the non-responsive ones. Although the process is not completely without human input, it is touted as reducing the amount of time and money involved in an e-discovery process because fewer lawyers are needed to sift through the reams of data.
This technology is being used or seriously considered in cases in other states, and Indiana attorneys anticipate it will be enlisted here very soon. The present concerns over predictive coding’s reliability may give way to pragmatism since electronic discovery is increasingly becoming part of litigation, and the massive amount of documents now stored on computers has begun to outweigh many law firms’ ability to do reviews by hand.
Indianapolis attorneys Becky Biller and John Papageorge believe predictive coding will appear in litigation in Indiana either this year or next. In fact, Papageorge, partner at Taft Stettinius & Hollister LLP, said his firm has a client “on the cusp” of using the technology.
As they await the arrival of predictive coding, Biller and Papageorge reflect the differing attitudes toward the software.
Biller, associate at Krieg DeVault LLP and co-chair of the firm’s electronic discovery and data management group, said she is excited to use the technology. In litigation with a high dollar value along with a large number of documents to review, the use of predictive coding will make sense.
Papageorge, while not opposed to using the technology, admitted it makes him uneasy because clients, attorneys and judges will be relying on computers to do the work and hoping the algorithms are right.
Even so, it will likely have to be used, he said, because of the “staggering sums of money” that clients will have to pay if traditional methods of discovery are used to review the huge number of documents that are often part of e-discovery. As an example, Papageorge estimated that five to seven lawyers doing discovery all day, every day, for one or two weeks would be able to examine about 10,000 to 12,000 documents which compares to electronically stored information that can easily surpass a million, five million or more documents.
Not hocus pocus
Predictive coding is not about computers making all the decisions and taking jobs from lawyers, said Juliet Hanna and Caragh Landry, executives at Integreon, an international company that provides legal, research and business support to law firms, in-house legal departments and corporations. Rather, it is based on the pre-existing technology of clustering and concept searches to prioritize the documents so the attorney can better review them.
“You have to have people involved in this,” Hanna, vice president of legal discovery services, said. “I think sometimes people get really scared because computers are taking over the world. … It’s not succeeding control to a computer. It’s using technology to work better.”
They compared the work of predictive coding to that of a paralegal. Just like a paralegal would take a box of documents into a conference room, separate the pieces of paper into different piles, and place a Post-it note labeling each stack, predictive coding categorizes electronic documents and, thereby, enhances an attorney’s ability to review the responsive information.
Both Papageorge and Phillip Fowler, partner at Bingham Greenebaum Doll LLP, outlined how predictive coding sorts the documents. Basically, an attorney will examine a seed set of documents and separate the relevant from the non-relevant. Then using this template, the algorithm is adjusted until the software can similarly identify relevant and non-relevant data.
Afterward, the attorney can review all documents that the computer selected as responsive or review a sample of those documents to test for quality – and to guard against privileged information being released – before turning the information over to opposing counsel.
“I see it as an opportunity for us to provide services we need to provide to our clients but at a reduced cost,” Fowler said.
He also speculated that predictive coding could have a significant impact on litigation by sending more cases into the courtroom. Currently, clients often decide to settle because both sides get tired of paying the fees that come with protracted discoveries. With this technology’s ability to pare down a large volume of documents into a more manageable size, thus cutting the time and expense of review, more disputes may go to court where they will be judged on their merits.
A much publicized court opinion regarding predictive coding came in February from the U.S. District Court for the Southern District of New York. U.S. Magistrate Judge Andrew Peck approved the use of predictive coding in Monique Da Silva Moore, et al. v Publicis Groupe & MSL Group, 11 Civ. 1279 (ALC) (AJP), concluding that computer-assisted review is an available tool and should be considered in large data cases where its use might save money.
U.S. Magistrate Judge Mark J. Dinsmore of the U.S. District Court for the Southern District of Indiana has already had a case come before him where predictive coding was considered but ultimately not used because many of the electronic documents in the case were PDFs, and the software was not as successful in culling the responsive documents.
Dinsmore shied away from describing himself as “comfortable” with the technology but said he is “open-minded about it.” At this point, he said he could not imagine objecting to predictive coding if both sides agreed to its use.
Landry and Hanna argued that attorneys do not have to disclose when they are employing predictive coding in the e-discovery process. The technology underlying the software has been around for years, and other methods of reviewing documents, like keyword searches, are not often disclosed to either the opposing counsel or to the court.
However, Indiana attorneys advised against keeping predictive coding a secret. Mostly because of the cost of the technology, which runs into six figures, disclosure is prudent to avoid having the other side raise challenges after the expense has been incurred. In addition, getting agreement early may limit objections later.
Biller views the federal courts as the best place for using predictive coding since the meet and confer requirement gives attorneys a forum to discuss and agree on the e-discovery process. Getting prior agreement would enhance the discovery process, she said, and eliminate disputes.
“It would be a real tragedy to invest in predictive coding only to have the other party in court challenge it and say, ‘You can’t use it’ because it is expensive,” Biller said.•