Predictive coding AKA Technology assisted review is the latest and greatest marquee topic among different e-discovery and legal bloggers, conference break out sessions, lunchroom banter around the vendors' water cooler and most recently the opinion written by a certain New York Federal judge who I have on very good information is a HUGE New York Yankees fan.
But what is an appropriate moniker for what we have thrown around as the next loaf of sliced bread?
Every time I hear someone call it "Predictive Review", all I can think is "Isn't that what a review is?" Any one who has setup a document review has had to train document reviewers whether attorneys or paralegals. How do we train them? We show them examples of documents and describe content that we are looking to code and how we want it coded. Based on inputting that data, we can "predict" how they will code the document.
Yes, I know. That is not the definition that the technologists want you to think of when they roll out their "better than the last guy's" black box software. They want you to believe that they can save you millions and do it perfectly. Well I hope they deliver both.
But isn't this more of a situation of technology assisted review? Based on the information we input into the system, we expect certain coding to happen. Once again, just like inputting data into human reviewers, the results of your technology assisted review is only as good as how well you "train" your software.
Let us not get away from the objective here and that is to reduce cost of discovery which is something whose day has come. I am very much in favor of any technology that can help my clients save money and be able to respond to litigation in a very effective manner.
Let's just figure out what to call it.
Comments, please?
No comments:
Post a Comment