Two recent cases will address important issues in the coming weeks that could help shape the future of predictive coding review technology in electronic discovery (ediscovery). The first case, Da Silva Moore v. Publicis Group et. al., grabbed headlines last week when initial reports erroneously indicated that The Honorable Andrew J. Peck, United States Magistrate Judge for the Southern District of New York, ordered the parties to use predictive coding technology. In reality, the transcript from a February 2012, status conference reveals that the parties agreed to use predictive coding technology, but they struggled significantly to define a mutually agreeable protocol. The challenges surrounding the dispute in Da Silva Moore center on the complexities of attempting to apply a new technological approach to electronic document review that is transparent, accurate, and fair to for all parties.
The second case, Kleen Products LLC v. Packaging Corporation of America, et al., involves alleged antitrust violations for price-fixing in the containerboard products industry. The case is venued in the United States District Court for the Northern District of Illinois with The Honorable Nan R. Nolan presiding over key discovery issues. Kleen Products represents a significant leap from the issues debated in Da Silva Moore because plaintiffs seek a court order requiring defendants, among other things, to use predictive coding technology to respond to plaintiffs’ document requests. The plaintiffs’ position is both novel and controversial considering predictive coding technology is a relatively new approach to electronic document review. Plaintiffs’ position is even more novel considering case law rarely, if ever, provides that one party can dictate whether or what kind of technology tools their opponent must use.
Is Predictive Coding A Substitute For All Predecessor Technologies?
Plaintiffs in Kleen Products take the position that defendants’ use of anything other than what they loosely refer to as “content based advanced analytics” (CBAA) is:
akin to an argument that choosing a horse as a mode of transportation is acceptable because it is the best available horse, even though technology has evolved and a superior form of transportation – the automobile – is now available. (Plaintiffs’ Reply Memorandum of Law For Evidentiary Hearing, at 4, note 5.)
First, one of plaintiffs’ main points of contention is that the court should order defendants to use CBAA in lieu of keyword searches. In response, defendants point out that plaintiffs fail to define the term “analytics.” The defendants go on to explain that, regardless of how plaintiffs define the term, they are using “the best analytical tools” available through their vendor rather than relying exclusively on keyword search technology. (Defendant’s Reply Mem. at 8). Repeated blurring of the lines between the kinds of “analytic” tools, functionality, and results plaintiffs desire, and the capabilities defendants possess, is a common theme that reveals limited knowledge of the range of technology-assisted-review (TAR) tools available to the parties.
Second, plaintiffs’ metaphor comparing different modes of transportation erroneously presumes that new ediscovery technology approaches such as predictive coding should automatically be presumed superior, rather than complementary, to other established TAR tools. This apparent emphasis on one type of TAR tool in favor of all others, is a throw the baby out with the bathwater approach that seems ill advised. To use the plaintiffs’ metaphor, in some situations it is the destination that determines the vehicle of choice. For example, even though automobiles represent a newer mode of transport than horses, a horse would certainly prove to be a more effective means for reaching a high mountain lake without roadways. In other situations, the decision about how to reach a particular destination may depend on personal preference, such as the traveler who chooses to take the train to work instead of a car.
Much like the traveler, the litigator has a range of both new and old vehicles (aka TAR tools) to choose from in order to get the job done. Tech savvy litigators realize that ediscovery technology is not a one-trick-pony – meaning multiple TAR tools such as Boolean based keyword search, find similar, advanced culling, concept search, discussion threading, and other tools are often used together to aid the ediscovery process. The particular tool or tools used for an individual matter may depend on a number of factors including the type of case, deadlines, amount of electronically stored information (ESI), value of the case, budget and many other factors. Predictive coding technology happens to be an extremely promising technology tool that fits underneath the TAR umbrella, but it should not be viewed as a replacement for every TAR tool. Instead, predictive coding is one of many tools that should be included in the litigator’s tool belt.
It’s the Carpenter, Not the Hammer
Two of my favorite quotes come courtesy of an article written by attorney and eDiscovery guru, Craig Ball, involving two companies feuding over whether or not their technology had been “court validated.” In We’re Both Part of the Same Hypocrisy Senator, Ball poignantly states that:
Just because a product is mentioned in passing in a court opinion and the court does not expressly label the product a steaming pile of crap does not render the product ‘court validated.’
He then goes on to state that:
the integrity of the process hinges on the carpenter, not on the hammer.
Even though Ball penned the above article almost three years ago in reference to data collection technology, his words reign true today in the context of predictive coding technology. Just as the operator of a vehicle controls the journey, so too does the technology end user control the ediscovery journey. Technology, regardless of whether it’s a state of the art cruise ship, Boolean keyword search tool, or predictive coding technology, is only as good as the system’s operator. The challenge with today’s predictive coding technology is that the tools are still in their infancy within the legal space, making them difficult for the average attorney operator to use and understand with a sufficient level of confidence. This difficulty and complexity introduces a higher risk of human error and stirs heated debate over proper protocols as evidenced in the Da Silva Moore status conference transcript referenced earlier.
Increased risk of human error combined with concerns over accuracy tends to breed larger concerns about defensibility. These concerns have likely contributed to slow adoption of predictive coding technology tools among attorneys despite the technology’s promise. According to a recent Legal Tech New York survey, (conducted by Symantec) 97% of respondents were familiar with predictive coding, yet only 12% have adopted predictive coding technology. The Catch-22 is that if managed and administered properly, predictive coding technology could ultimately help revolutionize document review by increasing accuracy and decreasing attorney review costs. (See “What Technology-Assisted Electronic Discovery Teaches Us About The Role of Humans in Technology.” Forbes, January 9, 2012). Not surprisingly, the same survey indicates that in addition to more judicial guidance, predictive coding will “go mainstream” if it becomes easier to use, more transparent, and less expensive.