Discovering A New Approach

When Dylan went electric, everything changed, but electronic discovery still has the same pitfalls as its paper cousin. A leading discovery consultant suggests a better—and somewhat radical—way. 

By Adam S. Bendell

DISCOVERY MAY HAVE GONE ELECTRONIC, BUT THE MANNER OF IDENTIFYING WHICH DOCUMENTS ARE produced to the other side is largely unchanged. Party A crafts a broadly worded document request and relies upon the good faith and internal ethical standards of Party B to interpret it. Party B’s lawyers supervise the collection of data and theoretically produce all responsive, nonprivileged documents. Absent the extraordinary involvement of a discovery referee, Party B’s lawyers determine what is responsive and what is privileged.

All of the endless, mind-numbing discovery battles between parties relate to the subset of documents that Party B’s lawyers have identified as potentially responsive. Although the logistics of document collection vary, all current approaches require unsupervised judgment calls by Party B. Sometimes clients identify and collect potentially responsive documents; other times paralegals look through files; often lawyers go on-site to collect documents. But in all of these approaches documents are only collected from the file cabinets and desk drawers identified by the client. Even when lawyers go onsite, they do not rummage through every file drawer in the company. So if Party B’s lawyers never go and look in a particular place for documents, unless the proponent can find out about the existence of the document through other means, such as deposition testimony, the document will never be considered for production. The identification of potentially responsive documents is an extraordinarily imprecise lawyers’ art, implemented inconsistently, without meaningful oversight, because Party B almost always has better information about its documents than Party A. The process is costly, frustrating, and subject to cynical gamesmanship. There has to be a better way.

Up to now, lawyers and the courts have largely applied the “unsupervised discretion” model of civil discovery developed for the paper world to the current age of electronic data storage, with erratic results. Today more than 90 percent of business documents exist in digital form: in e-mail, word processing documents, spreadsheets, presentations, intranets, public Web sites, and enterprise databases. These files are stored on PC hard drives, the flash memory of handheld computers, file servers, Web servers, CDs, DVDs, removable disks, and thumb drives. The files might be relatively well organized, a total bleeping mess, or, most likely, somewhere in between.

The volume of business documents retained by litigants has increased dramatically because of electronic storage, and by some measures is more than doubling every year. The low cost of electronic storage, the lack of centralized management of that storage, and the ubiquity of e-mail have all contributed to overwhelming quantities of documents. Yet the overall approach to electronic discovery (as distinct from the processing mechanics) is largely unchanged from the paper world. The requesting party still crafts broadly worded document requests and relies upon the recipient counsel’s good faith and internal ethical standards to interpret them. To the requester, discovery is an opaque black box. But now the recipient faces such a mountain of data that it cannot possibly review every document, so the arbitrariness inherent in a party’s opponent choosing which documents to collect for initial review has skyrocketed.

That breakdown manifests in one of two ways: missing relevant evidence and getting buried in junk. Broad requests yield gigabytes of data, so much that recipients cannot possibly review every document. Overly narrow requests allow the respondent to hide the ball. The model of “honorable compliance” has broken down under the volume of electronic data storage.

Conceptual search providers such as Cataphora, Inc., Attenex Corporation, and Engenium Corp. are taking a first and critical step by providing tools to bypass review of 100 percent of collected documents. These vendors offer search engines and processes that look for conceptually related documents rather than simply those that contain certain keywords. But they face sharp questions from litigators more comfortable with judgments made by lawyers, however imperfect, than by machines. And even cutting-edge tools cannot solve the core problem: that the respondent makes all the judgment calls.

The solution is “discovery transparency.” Imagine, to go back to the original example, if Party A was able to search through Party B’s documents using a negotiated set of search terms and parameters. In a single stroke, Party B would forgo costly page-by-page document review, and Party A would no longer be subject to the discretionary judgment of its adversary. The disadvantage of electronic data storage—the cheap and therefore perpetual accumulation of digital data—is offset by the potential for automated searching.

Here is how it might work: Party A requests that all available electronic data of a number of identified witnesses be loaded into a searchable system and searched using search terms negotiated between the two parties. In addition to negotiating the search terms, the parties would also negotiate a maximum number of pages or documents to be produced. If the search exceeds that number, the search would be narrowed. Otherwise, the document set would be searched to exclude privileged documents using attorney names and words like “work product.”

Documents meeting the privileged search criteria would be provisionally excluded. Ideally, Party A would agree not to assert waiver of privilege by reason of inadvertent production, so that privileged documents that slipped through the cracks would be returned. Because document searching is an imperfect art, Party A would be allowed to review the search results and run refined searches a few times to eliminate false positives and capture false negatives. Of course, at each iteration, provisionally privileged items would be excluded before Party A saw the results. After an agreed number of iterations (say three), Party B’s counsel would review the results of the privileged search, make definitive privilege calls, and prepare a privilege log.

At that stage, the nonprivileged documents produced by the search would constitute the entire universe of documents to be produced in the litigation. A costly page-by-page review of all documents by Party B’s lawyers would be avoided—they could review only produced documents, and could do so later so as not to delay production.

ELECTRONIC DISCOVERY COULD CONCEIVABLY BE COMPLETED IN WEEKS, NOT MONTHS OR YEARS

Discretionary judgment would be eliminated, instead being reflected in the arm’s length negotiation over search terms between the parties. Party B would have saved money, and Party A would not have the lingering sense that it was hoodwinked. Discovery could be completed in weeks, not months or years.

There are logistical hurdles to making the foregoing example practical. The parties would need to decide whether to search native files or to first convert to a “common denominator” image format, such as TIFF or PDF, and accept the ramifications of each approach. Identifying whether a particular electronic document “belongs” to a particular witness is tricky, because in a corporation there are often shared electronic storage areas for which there is no single clear owner. The mere harvesting of certain data, such as content from company intranets and enterprise databases, is complex and potentially expensive. The parties must address the questions of whether to review backup tapes, which are usually filled with numerous copies of duplicate data but may have some data not otherwise available. The choice of search engine may impact which documents are produced, because each search engine works differently and may return different result sets for the same query. Both sides will need to retain technical experts to avoid technical gamesmanship.

The Federal Rules of Civil Procedure and state law equivalents set forth a broad framework governing the discovery process. The discovery transparency proposed here generally does not require any changes in the rules. Parties can agree to the procedures outlined here and, as necessary, seek protective or other court orders that endorse the process.

By far the greater challenge is the change in mind-set required of participants in such a process. Most litigators will shudder at the thought of such a cooperative process and can unleash a flurry of reasons why it will never work. But this process is already working in some cases. Negotiated search terms are a growing practice among private litigants, typically as an adjunct to, rather than a complete replacement for, traditional discovery. Even the federal bureaucracy has seen the future. Last year the Federal Trade Commission suggested in guidelines for merger investigations that “[a] thorough and well-executed term search of electronic files may be an efficient way to respond to a [Hart-Scott-Rodino] second request.”

Given the blossoming burden and expense of page-by-page review of the slag heap of electronically stored data, it’s only a matter of time until practicality and economics leads to fundamental change in the practice of civil discovery in the United States.

Adam S. Bendell is the president of Strategic Discovery, Inc., and former chief technology counsel of Gibson, Dunn & Crutcher. E-mail: abendell@strategicdiscovery.com