Why Is TAR Like Ice Cream?

From my Advanced Discovery blog of July 7:

Short answer: because both have so many flavors.

You wouldn’t ask me to go to the store to “get some ice cream” without telling me what flavor you want.  But everyone these days is talking about TAR (Technology Assisted Review) like it’s a flavor all its own.  By everyone, I mean columnists, bloggers, consultants and sometimes even judges.

Remember the cases Aurora Cooperative Elevator Company v. Aventine Renewable Energy or Independent Living Center of Southern California v. City of Los Angeles, where courts ordered the use of predictive coding after extensive discovery squabbles? Or more recently, Judge Peck declining to order the parties to use TAR in Hyles v. New York City, by which, in that case, he meant predictive coding.

Which illustrates my point: what do we mean when we say TAR?

When it comes to TAR, pretty much everyone agrees with this framing statement made by Maura Grossman and Gordon Cormack in their seminal article, Technology-Assisted Review in E-Discovery Can Be More Effective and More Efficient Than Exhaustive Manual Review, (XVII RICH. J.L. & TECH. 11 (2011):

Overall, the myth that exhaustive manual review is the most effective—and therefore, the most defensible—approach to document review is strongly refuted. Technology-assisted review can (and does) yield more accurate results than exhaustive manual review, with much lower effort.

But then things go south. Why?

First, because it is always unclear if TAR is a synonym for predictive coding as the cases above illustrate.  And as a further example, in a recent post on Ralph Losey’s blog, E-Discovery Team, a sub-title states, “New First Class Added to the TAR Course,” with the first sentence then stating, “We also added a new class on the historical background of the development of predictive coding.”

Second, because any discussion of TAR involves selecting documents using algorithms. Algorithms. Math. Warning. Warning. Danger Will Robinson. Attorneys react to math the way astronaut David Bowman reacted to HAL in 2001: A Space Odyssey – like it’s trying to kill them.

Want a good example? Take a look at Ralph’s most recent blog about his TAR course.  Great course, extremely comprehensive, tremendous insight into TAR.  But here’s a paragraph from the site:

Ralph misspoke in the video at point 8:39 when he said 95% confidence interval, he meant to say 95% confidence level. The random sample of 1,535 documents created a probability of 95% +/- 2.5%, meaning a 95% confidence level subject to an error range or interval of plus or minus 2.5%.

Gain curves, x-axis vs y-axis, HorvitsThompson estimators, recall rates, prevalence ranges and my personal favorite “word-based tf-idf tokenization strategy.”  All this geek talk makes me yearn for the days of Trover and Replevin.

Third, because we’re talking about a process, not a product. The Wall Street Journal made exactly that point in a 2012 article entitled “Why Hire a Lawyer? Computers are Cheaper.”  Ralph calls it the multi-modal approach: a combination of people and computers to get the best result.

Everyone agrees that manual review is inefficient (the lawyer part), but nobody can agree on what software the lawyers should use and how: the geek part.  And when geeks start disagreeing over technology, that’s when things get uncertain.

So where does this leave us? 

The idea behind predictive coding – that technology can help reduce the cost of eDiscovery – is a great one. But figuring out what pieces of technology to apply at what point in the workflow is not so easy, especially when the experts disagree as to the best methodology.

Remember when Judge Facciola said in the O’Keefe case that areas of technical expertise are where even angels fear to tread? Believe me — the angels are taking a BIG detour around this subject.

My advice?

Before you think about using more advanced technology, use basic tools early on: dedupe, denist, cull by dates, sample by custodians and start with basic search terms that are agreed upon by both sides. Then get an expert to perform more advanced analytics who has legal experience and can explain the procedure to you in simple English.

TAR isn’t a piece of software. It’s a process that can include many different steps, several pieces of software, and many decisions by the litigation team. If you and/or your expert can’t quickly and concisely explain that process to the Court as easy as you can order a waffle cone with one scoop of rocky road, one scoop of chocolate pecan fudge, jimmies and some pineapple sauce, then you may find yourself on the receiving end of a document exchange protocol order drafted by the party who could.

To see all the Advanced Discovery blogs, go to http://www.advanceddiscovery.com/blogs/

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.