Bass, Berry & Sims PLC attorneys Kathryn Walker, Anthony McFarland and Lucas Smith provide The National Law Journal with unique insight on how the recent explosion of technology has created serious financial and time pressures as well as critical liability risks for both clients and their lawyers. Their article, “Technology is the problem. It’s also the solution,” is featured in the Journal’s Electronic Discovery Special Report section in the August 22, 2011 issue. Read below for the article.

Technology is the problem. It’s also the solution
Resistance to change is futile, but tools exist to help lawyers cope with the scale of discovery.
By Kathryn Walker, Anthony McFarland and Lucas Smith
The National Law Journal
August 22, 2011

Modern society craves technology and people are making ever-larger digital footprints. This explosion of technology, however, has placed very serious financial and time pressures and liability risks on both clients and lawyers, especially in the arena of electronic discovery.

The recent Delaware Supreme Court case Genger v. TR Investors LLC, 2011 WL 2802832 (Del. July 18, 2011), illustrates the ever-increasing scope and expense of e-discovery. In Genger, the court affirmed a $3.2 million sanction award against a party for its failure to preserve the “unallocated free space” of computer hard drives — the location in the computer that is used only for temporary storage. With the Delaware court’s ruling, parties now face the prospect of preserving information most users do not even realize resides on their hard drives, in addition to the huge universe of ordinary data.

Faced with this challenge and fearful of both the technology itself and the risks associated with making a mistake, many lawyers resist changing their old ways or, worse, hope to ignore away the elephant in the room. Nevertheless, there is tremendous opportunity to actually embrace technology and use it to fashion creative solutions to the problem of information overload. Stated differently, new technology got us into this mess, and newer technology (such as predictive coding and e-mail threading) may help get us out of it.

Of course, the use of technology to assist with e-discovery is not new. Litigants already widely rely on established technologies such as linear searching to minimize the amount of data that must be reviewed. Courts have largely approved the use of these technologies. These existing solutions, however, possess inherent limitations that render them unsuited to cope with the geometrically escalating amount of discoverable data. For example, linear searching — used to identify documents that contain specific words — is only as reliable as the search terms selected (which terms are normally chosen by attorneys who never participated in the underlying events or communications).

Linear searching therefore often identifies numerous irrelevant documents merely because they contain the chosen search term while simultaneously failing to identify highly relevant documents that do not contain the term. For instance, the search term “football” would identify documents related to European football (i.e., American soccer) in addition to those dealing with American football. This over-inclusive term would not, however, identify documents only containing the acronym “NFL,” a glaring deficiency in a case involving the National Football League.

Moreover, linear searching typically still requires extensive and time-consuming document-by-document review by lawyers. Attorney review of search-term-identified documents is by far the most expensive part of the e-discovery process, often accounting for the majority of the overall discovery costs. Clients (and attorneys) are clamoring for solutions to minimize the number of documents that must be reviewed by attorneys — and e-discovery vendors are racing to meet this demand.

One promising new tool is e-mail threading, a process that groups e-mails from the same conversation or “thread” together so that they all can be read in concert. E-mail threading allows faster review, because earlier e-mails in the same thread can be skipped, and clearly irrelevant e-mail threads can be marked as not responsive simultaneously. Threading also enables reviewers to better understand the context of e-mails within the same conversation, thereby encouraging consistent treatment of all e-mails in the same thread and reducing the chance for attorney coding error.

The efficiencies associated with e-mail threading are substantial. A 2009 survey found that the use of e-mail threading software resulted in a savings of 36% in review time. A. Kershaw & J. Howie, Electronic Discovery Institute Report on Kershaw-Howie Survey of E-Discovery Providers Pertaining to Email Threading (Jan. 5, 2010).

Another important new e-discovery technology is predictive coding, also known as “predictive ranking” or “automated document classification.” Predictive coding is an automated method of identifying responsive documents that begins by having a “super coder” — a seasoned attorney who is familiar with the case — review and code a subset of documents from the larger data collection. The predictive coding software analyzes the super coder’s document classification and thereafter predicts which of the documents in the larger collection will also be responsive or not responsive, based on the attributes of the reviewed documents. This process can be quality-controlled by sampling the results of the initial search and repeating the process as necessary until the attorneys are satisfied the review is complete.

When used properly, predictive coding is an effective tool for reducing litigation cost and expense. Predictive coding can be used to reliably identify and produce responsive documents in a large data collection even though only a small subset of the produced documents actually has ever been reviewed by an attorney. Alternately, predictive coding can be used to weed out documents that are likely irrelevant so that they never have to be reviewed, or to rank documents by predicted relevance for prioritized attorney review.

Because the super-coder attorney essentially teaches the software how to classify the documents, predictive coding does not entirely replace attorney review. Instead, predictive coding uses the judgment and wisdom of a single, seasoned attorney to increase the productivity of all the other attorneys who will be assisting with the review. Indeed, a 2010 survey estimated that predictive coding can save an average of 45% of review time, on top of time savings gained by e-mail threading, de-duplication and removing junk e-mails. Electronic Discovery Institute, Survey on Predictive Coding (Oct. 1, 2010). Via

One major obstacle to the widespread adoption of e-mail threading and predictive coding is mistrust of this unfamiliar technology. For many attorneys, the thought of producing or not producing a document without an attorney having laid eyes on it is cringe-inducing. What if the software fails to produce highly relevant documents? What if privileged documents are unintentionally produced?

These concerns may reflect over-confidence in the nature of comprehensive human review, which itself is subject to significant risk of error. Recent changes in the rules governing the “clawback” of inadvertently produced documents are a recognition of the inevitability of mistakes made even by conscientious attorney reviewers. Nevertheless, given the potentially severe consequences of failing to produce highly relevant documents, and of producing privileged documents, these are not idle questions.

Unfortunately, to date courts have not specifically addressed whether a litigant satisfies its discovery obligations by relying on predictive coding rather than linear review. Judges interpreting the Federal Rules of Civil Procedure speak of the obligation to take “reasonable” steps to search for and produce relevant documents, as demonstrated below. Similarly, Federal Rule of Evidence 502 speaks of the need to take “reasonable steps” to prevent the disclosure of privileged documents.

Unfortunately, courts (some of which are more comfortable with technology than others) take widely varying views of what is reasonable. On the one hand, the U.S. District Court for the Southern District of West Virginia held that a party did not take reasonable steps to prevent the inadvertent disclosure of 377 privileged documents out of a production involving more than 1 million documents, despite the fact that the inadvertent disclosure appeared to be largely the result of a document review software glitch, and despite the fact that the parties had negotiated a clawback agreement. Mt. Hawley Ins. Co. v. Felman Production Inc., 271 F.R.D. 125 (S.D.W. Va., May 18, 2010). (This decision may be explained as an instance of bad facts making bad law, as the inadvertently produced documents arguably described the producing party’s plan to commit insurance fraud.)

On the other hand, the U.S. District Court for the Northern District of California, in Datel Holdings Corp. v. Microsoft Corp., 2011 WL 866993 (March, 11, 2011), held that a party did not waive the privilege through an inadvertent disclosure of documents caused by a software glitch, finding that the defendants’ reliance on the software was reasonable.

The Datel court relied in part on the Advisory Committee Notes to amended Rule 502(b) stating, “Depending on the circumstances, a party that uses advanced analytical software applications and linguistic tools in screening for privilege and work product may be found to have taken ‘reasonable steps’ to prevent inadvertent disclosure.” Id. at *4. Although the Datel decision was not about automated document classification, this portion of the decision supports the argument that reliance on predictive coding software is reasonable.

Many worries about the use of these new discovery technologies could (and should) be addressed by cooperative counsel at the outset of discovery. Especially in cases in which both sides to the dispute face significant discovery burdens, attorneys could save significant effort and expense by agreeing to the parameters of use of technology such as predictive coding, together with reasonable negotiated clawback agreements, to facilitate and expedite document review. Working together, counsel can gain the benefits of the recent advances in discovery technology and reduce the risk that a court will have to decide whether the parties’ reliance on technology is reasonable, without sacrificing their zealous representation of the interests of their respective clients.

Given that the ocean of data daily grows wider and deeper, the problem of managing e-discovery is only getting worse. Human faculties to review data are not developing at a rate that can match the explosion of the creation of that data. Embracing developing technologies is the only feasible method of managing this tremendous volume. Rather than fearing and avoiding these changes, lawyers should recognize that great technology, wielded by great lawyers, is a formidable force for efficiency, efficacy and advocacy.

Kathryn Walker is assistant chairwoman of Bass, Berry & Sims PLC’s litigation department and specializes in data management and e-discovery. Anthony McFarland is a member and Lucas Smith is an associate in the litigation department.