Is Predictive Coding Passe Already?
On an otherwise ordinary Friday in February of 2012, Judge Peck changed the landscape of eDiscovery forever when he approved the use of Computer Assisted Review (Often thought of as Predictive Coding). Ever since that day, Predictive Coding has been the buzzword at the tip of everyone’s tongue. Articles upon articles written on the topic, It’s so hot in fact, it’s already suffering from backlash.
Let’s dial it back for a minute: we all understand the basics of how predictive coding works, right?
Can I be of Assistance?
Something to remember about computer assisted review, is that it is simply that – assisted. It is still hugely dependant on human reviewers. The common misconception is that it’s automated and all one must do is press “on.” This couldn’t be further from the truth. Predictive coding, just like any TAR tool, relies entirely on human reviewers. The reviewer/s must first create training sets and the success of the rendered search results hinges entirely on the intelligence and precision with which these initial sets are created. Reviewers must have a clear and thorough understanding on what it is they need assistance finding. They must think outside the box and have a strategy. So as with anything performed by humans there is room for error. Especially when accuracy depends on a review that takes place without feedback or any kind of automated error checking. The fact is that recall will have its limitations without ESI content analysis and relationship identification.
Let’s get it straight.
Experts on the forefront of technology will say that predictive coding alone, is not that exciting anymore, that it’s so yesterday! People often confuse Computer Assisted Review or Technology Assisted Review for predictive coding. This is not entirely the case. Predictive coding is just one tool in an ever increasing TAR or CAR toolkit. In fact, according to Law Technology News, there is a lot of energy and resources within the industry being poured into improving predictive coding, something that is inherently limited. Energies now are being focused on refining predictive coding and the tools used in conjunction with it, such as computational linguistics, data mining, language translation, corpus-based content analysis and case specific information supplied in the form natural language inquiries.
So where does that leave predictive coding, and what does its future look like?
With the work being done by industry specialists, predictive coding will not become passé, it will instead become stronger when paired with other technologies. Content analysis will allow powerful categorization of ESI based on data mining and language translation techniques. If technology continues the way it has been, we will become less and less dependent on human reviewers to create the training sets, leaving our lawyers more time to check for errors and get to reviewing. The latest buzzword in our industry, Information Governance, is something that absolutely relies on machine learning technologies and predictive analytics. This will perpetuate the ever-evolving quest to save time and money in the world of litigation and beyond.