What You Need to Know About Predictive Coding

Written By: Elizabeth M. Midgley, Esq., Partner at Anspach Meeks Ellenberger

Like many attorneys, when it comes to disclosing client documents in litigation, I can be a bit of a… control freak.  I like to look at every single line of every single page of every single document that we’re about to produce.  Only then, when I’m sure that we’re not divulging protected information, can I sleep soundly after a document production.

So, what happens when us control-happy attorneys are faced with having to produce thousands upon thousands upon thousands of pages of electronically stored information (“ESI”)?  Luckily the computer geniuses of the world have come to our rescue (again), and have brought us computer-assisted review.

© Inok / Fotolia
© Inok / Fotolia

Forget the Sci-Fi robot movies – computer-assisted review is not a computer taking the place of humans, it’s humans and computers working together to manage electronic discovery.  What it means for you is an expedited document review process; for your clients, it means manageable eDiscovery costs.

Computer-assisted review, a/k/a predictive coding, uses human input on samples of documents to predict which of the documents are relevant.  The (human) reviewer works with the computer to review a small fraction of the overall document population.  The reviewer’s determination of whether or not these documents are responsive, in essence, trains the computer to predict how each document should be classified.  The computer is then able to use this training and applies it across all of the thousands of documents in question to determine what is responsive.  Then, the reviewer is able to analyze only those documents which the computer predicted are relevant – saving hours upon hours of review time, and (in some cases) hundreds of thousands in review costs.

In the past two years alone, the amount of cases using predictive coding for discovery has increased exponentially as firms across the country are learning the great benefits to computer-assisted review.  Reviewers have learned that computers are more than just a tool for spreadsheets and word documents – they are now a partner in the document review process.

Although it may sound like a process that is too good to be true, predictive coding is quickly becoming a Court-accepted way to search for relevant ESI in certain cases.  During a local ESI discovery conference in Gordon v. Kaleida Health, the Court “pointed [the parties] to the availability of predictive coding, a computer assisted ESI reviewing and production method”, where the parties had been less-than successful at resolving ESI production on their own.  Gordon v. Kaleida Health, 2013 U.S. Dist. LEXIS 73334 (W.D.N.Y. 2013).

In working towards resolving the discovery issues in Gordon, Magistrate Judge Foschio pointed the parties to the ground-breaking decision in Moore v. Publicis Groups & MSL Group, 287 F.R.D. 182 (S.D.N.Y. 2012).  Magistrate Judge Peck’s decision in Moore became the first judicial opinion to “[recognize] that computer-assisted review is an acceptable way to search for relevant ESI in appropriate cases.”  Moore, 287 F.R.D. at 183.  In concluding that predictive coding is a tool that should be considered, along with existing and more established tools, in large volume ESI cases, the Courts have opened the door to using this new wave of technology.

So the next time you are facing discovery of a mind-numbing amount of ESI, you aren’t looking at a sleepless (and spouseless) month of 18-review-hour days.  It’s totally (court) acceptable to ask for help – in computer form.

If you’re still not sold on all of this, just try it once.  As they say – seeing is believing.  And I bet that if you ask nicely, Avalon will even give you a great deal on it to show you just how great it can be.

If you liked this blog you might also be interested in reading: “Take Your Geeks to Court” & Other Advice from the Federal Judges Panel on eDiscovery (Part I)

Learn how Avalon’s Managed eDiscovery Services helped one litigation support manager increase his realization rate to 90%

Managed eDiscovery Case Study

author avatar
Ian Gattie