So far, the issues under this amendment have arisen largely in cases involving unindexed, hard-to-search backup tapes, “deleted” information, and “legacy” information that was created and can only be accessed by obsolete hardware or software. The issues will change as the technology changes; what is hard to access today may be readily accessible tomorrow, while changes to systems may make information readily accessible today very difficult to produce in a few years. The case law already shows that judges understand that these issues require detailed court management.

One new requirement is that the responding party must disclose the existence of electronically stored information located on sources that are not reasonably accessible, even if the party has not searched it. Although the limits of this identification requirement are unclear, a party need not identify everything it is not searching but only sources that are likely to contain relevant information.

Some judges have taken a creative approach in using an “identification as not-reasonably-accessible log,” similar to a privilege log, as a way to address the identification requirement. But there are important differences between such a log and a privilege log that should be kept in mind. The producing party knows the content of what it lists on a privilege log. A party may not know what information is contained on a source that is not reasonably accessible, including whether it is responsive or relevant. The responding party may not be able to find the answers to this basic question without the type of expensive and time-consuming restoration steps that Rule 26(b)(2)(B) is intended to avoid unless a good-cause showing is made and appropriate terms and conditions imposed. Requiring a privilege type of log to meet the identification requirement may make that requirement itself a source of cost and delay. For many lawyers and litigants, privilege logs epitomize the worst features of discovery: they are expensive; they take forever; and when finished they are rarely used. A privilege type of log may also obscure the difference between identifying the unsearched sources of potentially responsive electronically stored information by category or type, as the Committee Notes direct, and identifying the information itself, which may require knowledge about the contents. The task for judges is to manage this part of electronic discovery under a rule designed to reduce recurring problems and associated burdens and costs, without introducing new and potentially onerous additional requirements.

The uncertainty about the content of electronically stored information that is not reasonably accessible will also continue to complicate preservation decisions. A party must determine whether not-reasonably-accessible sources might contain relevant information and must evaluate the risks of allowing that information to be overwritten, deleted, or otherwise become increasingly more difficult to access. A primary factor to consider when evaluating those risks is whether the information likely to be found on those sources is also available on other, reasonably accessible sources. In evaluating preservation choices that parties made in the past, a judge will need to analyze—without the distortion of hindsight—the reasonableness of the party’s litigation hold, including any conclusion that information stored on not-reasonably-accessible sources was either irrelevant or available elsewhere and did not need to be preserved.

One question that is emerging is whether a data producer that has met its preservation duty by saving information available on sources that are not reasonably accessible may shift costs under Rule 26(b)(2)(B). The line drawn in one case addressing the question again demonstrates the close judicial involvement required and the need to balance effective discovery supervision with an understanding of the challenges facing the parties. In Quinby v. Westle AG, an employment dispute reminiscent of Zubulake v. UBS Warburg L.L.C., the plaintiff had asked for e-mails covering a broad range of the defendants’ employees, topics, and time. The defendant/employer had followed a consistent practice, existing long before the litigation was anticipated or filed, of deleting former employees’ e-mails from the accessible database and maintaining them on backup tapes. The plaintiff sought e-mails from a number of individuals, including former employees. The defendant had saved the backup tapes containing the former employees’ e-mails and sought to shift to the plaintiff the costs it incurred restoring and searching these tapes for e-mails responsive to the plaintiff’s discovery requests. The issue was not preservation. This defendant had understood that once it reasonably anticipated litigation, it was under a duty to preserve information it knew or reasonably should have known would be relevant to the litigation.

The Quinby court recognized that the defendant could choose how to meet its preservation obligation and should not be sanctioned for selecting one format over another, not even for selecting an inaccessible format over an accessible format. In reaching this result, the judge disagreed with the result reached in Treppel v. Biovail Corp., a case decided by another judge in the same district. But although the Quinby court refrained from sanctioning the defendant, the court also held that the defendant could not shift the costs of accessing the e-mails stored on the backup tapes because it had placed the e-mails on backup tapes after it should have “reasonably foreseen” that the e-mails “would be discoverable material."

This approach to Rule 26(b) requires a data producer faced with anticipated or filed litigation to draw a very fine line. That data producer must, of course, preserve electronically stored information it knows or reasonably should know is relevant to anticipated or filed litigation. It may choose to secure this information in formats that are not reasonably accessible. But to recover costs, it must secure electronically stored information that it should reasonably foresee would be discoverable only on accessible sources. And it may have to make this distinction between information “reasonably-foreseeable-as-relevant-to-the-litigation” and “reasonably-foreseeable-as-discoverable” early in the process. The Quinby case aptly demonstrates how hard it is for judges to draw distinctions that will provide predictable guidelines for discovery, in the detail necessary to provide effective discovery control and supervision, without imposing standards that may be clearer in hindsight than when the litigants made the preservation decisions.

The increase in costs that has accompanied electronic discovery and the explicit identification of the judge’s cost-allocation authority in the Rule 26(b)(2)(B) Committee Notes are likely to lead parties to be more creative and aggressive in seeking to shift costs. For example, parties may ask to shift the costs of putting information in a specific form, of removing information that is confidential or private, or of reviewing information for relevance or privilege. Courts are likely to see these and similar requests under Rule 26(b)(2)(B), and deciding such requests will require careful and detailed judicial management of the discovery process. Even before electronic discovery and Rule 26(b)(2)(B), courts had the authority to allocate all or part of the reasonable discovery costs. The Committee Notes that accompany the electronic discovery amendments make clear that all discovery is subject to the proportionality limits of Rule 26(b)(2)(C). The amended rule does not say that judges may only consider cost allocation if the subject of the discovery is electronically stored information, if the electronically stored information is not reasonably accessible, or if the costs that the producing party is seeking to shift go beyond the costs of forensic work necessary to make certain information accessible. Nor does the amended rule preclude producing parties from seeking to shift costs of producing electronically stored information that is reasonably accessible or from seeking to shift other costs of producing not reasonably accessible information.

Requesting parties may argue that if they are willing to pay the costs of restoring and retrieving information that is not reasonably accessible, then that information must be produced. But whether to require the production of information that is not reasonably accessible is a separate question from deciding who will bear the costs. The Committee Notes specifically state that a requesting party’s willingness to share or bear the costs of accessing information that is not reasonably accessible is one factor in determining good cause, but not the only factor. A court is also to consider the producing party’s burdens and costs in reviewing the information after it is accessed, for relevance and privilege, as a factor weighing against permitting the required discovery. The rules do not limit a court’s consideration to the forensic costs of accessing electronically stored information. Instead, a court assessing whether the benefits from production amount to good cause may also consider the human costs of reviewing the information that is accessed for relevance and for privilege. What costs to consider in determining whether to compel production of the “second tier” and whether or to what extent those costs should be shifted are decisions that can only be made with a detailed understanding of the requested discovery.

Rule 26(b)(2)(B) does not create new authority for judges to limit discovery or to allocate the costs of that discovery. But the increased costs and burdens of electronic discovery are likely to generate new arguments as to the limits of this authority, with a corresponding need for judges to analyze and define those limits without drawing lines so fine that they cannot readily be understood by litigants and lawyers looking to those lines for future guidance.

Lee H. Rosenthal is a United States District Judge for the Southern District of Texas. Rebecca Bolin, Yale Law School ‘06, provided helpful assistance and suggestions.

Preferred Citation: Lee H. Rosenthal, A Few Thoughts on Electronic Discovery After December 1, 2006, 116 Yale L.J. Pocket Part 167 (2006), http://yalelawjournal.org/forum/not-reasonably-accessible-information-and-allocating-discovery-costs.

A Few Thoughts on Electronic Discovery After December 1, 2006

• Installment 1: An Overview of the E-Discovery Rules Amendments
• Installment 2: Meeting and Conferring
• Installment 3: Not Reasonably Accessible Information and Allocating Discovery Costs
• Installment 4: Privilege Review
• Installment 5: Metadata and Issues Relating to the Form of Production
• Installment 6: Sanctions
• Installment 7: Conclusion