It can safely be said that 2012 will go down as the year of Predictive in e-discovery circles. Not only did several new players enter the predictive technologies arena, but the courts finally began addressing the validity of the technology as a defensible means for identifying relevant documents. Of course, we've seen e-discovery trends come and go in the past, but most experts agree that the predictive phenomenon may actually only be in its infancy because we're only scratching the surface of the technology's full capability. This very trend and topic was addressed during a recent Exterro webcast, where Charles Skamser, CEO and lead analyst of the eDiscovery Solutions Group, and Linda Luperchio, director of information lifecycle governance for The Hanover Insurance Group, discussed how predictive technologies can be leveraged throughout all stages of e-discovery, beyond the review phase, which is where most predictive products are currently applied.
Taking Predictive Outside the Review Bubble
As we've addressed on the E-Discovery Beat in the past, a lot of confusion exists surrounding what we are supposed to call the growing number of predictive applications. Skamser said he likes the term 'predictive technologies' because it reflects the potential flexibility of the technology going forward, unlike more established names like “predictive coding" or “technology-assisted review," which tend to associated purely with document review.
“There are lots of areas within the EDRM where it's extremely important and has been historically cumbersome to go find the documents that are actually going to have to be produced at some time," said Skamser. “As we all know, what's currently being done without predictive technologies is we're collecting a fairly large sample of what we think will eventually have to be produced, and then we're going through a human review process that is costing right now an estimated 70 percent of all e-discovery costs."
Dispelling Myths About Accuracy
Linda Luperchio, who oversees e-discovery operations at the Hanover Insurance Group, said she was recently able to take a pre-collection set of 7,600 documents and reduce it down to just 326 in a mere half hour using predictive methods. While she was pleased with the speed at which the technology was able to analyze the documents set, she was equally impressed with the accuracy. “The documents, when we did review them, were absolutely on point. It's one thing to know that you're cutting back the documents, but the bottom line is you want to produce what really makes sense and what's responsive," said Luperchio.
Despite recent court decisions that have endorsed the efficacy of predictive technologies, concerns about accuracy persist. Skamser underscored the prevailing myth that human review is the accuracy gold standard. He mentioned that several studies have shown that human reviewers are not nearly as accurate as many people assume. In fact, in a recent survey What Does Counsel Really Think About Predictive Coding?, 57% of inside and outside counsel respondents saw predictive technologies as an effective means for quality assurance on human reviewers. One respondent noted that people make mistakes and simply lose concentration while reviewing large document sets. “If you take that into account, predictive technologies are just as accurate and in most cases more accurate than what you're going to get with human review," Skamser said.
Predictive Technologies and Information Governance
The discourse surrounding predictive technologies today is almost exclusively focused around litigation. Skamser and Luperchio, however, agreed that predictive technologies can be applied to numerous other areas, including information governance, internal investigations, regulatory inquiries, even day-to-day protection of IP and other confidential information. Luperchio, who is also charged with managing The Hanover's electronic records, cited Massachusetts' toughest-in-the-nation data protection laws, which require that businesses encrypt all the transmitted personally identifiable information (PII) of their customers, adopt secure user authentication protocols and employ reasonable monitoring of systems. “That's a great use for predictive technologies," she said. “You can choose a subject matter and go through your documents and find where your risk is, where your PII is, what you have to look at. It's part of e-discovery, it's part of records, it's part of privacy. It belongs everywhere."
During the webcast, a question came in from the audience pondering whether we might see a day when predictive technologies are used proactively, similar to anti-virus software, to ensure sensitive information is adequately protected. “Absolutely," Skamser said. “Using the technology proactively to identify documents that are at risk across the board, for protection of confidential information and IP and for privacy concerns and for making sure you're following regulatory compliance and what not makes a lot of sense."
A full on-demand recording of Exterro's webcast “Leveraging Predictive Technologies Across the EDRM" is available here.