Sharing the Costs of Electronic Discovery
Each year it seems that the cost of discovery in large-and even in not-so-large-cases increases. The pervasive use of computers to create and store communications and data has only exacerbated the problem; vast quantities of electronic data often must be retrieved and reconstructed to respond to a single discovery request. As discovery costs go up, the issue of which party must pay-the party requesting discovery or the party responding to it-becomes more important.
Perhaps surprisingly, the issue of cost shifting was addressed back in 1978 in Oppenheimer Fund, Inc. v. Sanders (437 U.S. 340 (1978)). When the defendant was required to hire a third party to manually sort records, keypunch 300,000 computer cards, and create software programs to extract the names and addresses of class members to enable the plaintiff to give them notice, the U.S. Supreme Court held that it was an abuse of discretion for the lower court to order the defendant to bear the "substantial" expense of producing the information for the benefit of the plaintiff. The expense of hiring the third-party contractor to reconstruct the data, the Court reasoned, would have been no greater for the plaintiff than it was for the defendant, and therefore the plaintiff should bear the financial burden of reconstructing the data.
The "substantial" sum of this endeavor in Oppenheimer Fund amounted to a whopping $16,000. Clearly, the expense of retrieving and reconstructing data has risen dramatically since 1978.
The Problem Today
By contrast, litigants today face ever more complex and costly data-restoration issues. Email alone generates about 400,000 terabytes of new information each year worldwide. To put this in perspective, the amount of information contained in a single terabyte would require 50,000 trees if printed on paper.
Even at a single company, the electronic information maintained on system backup tapes and in central repositories and storage archives can exceed several terabytes. Corporations commonly maintain backup tapes so they can restore files if there is a system failure or disaster such as an earthquake, flood, or fire. Usually, these tapes are not retained for lengthy periods but are periodically overwritten when new backup tapes are generated.
Although litigants sometimes seek to obtain data from backup tapes, doing so is often extremely inefficient and expensive because the data is highly compressed and cannot readily be identified or accessed. In most cases forensic data experts must be hired to restore, retrieve, and reconstruct the data. Then the data must be searched, reviewed, and produced. In the context of discovery, the general rule in both state and federal court is that the responding party must bear the cost of complying with the requests. (Toshiba Am. Elec. Components, Inc. v. Superior Court of Santa Clara Cty., 124 Cal. App. 4th 762, 769 (2004); Oppenheimer Fund, Inc., 437 U.S. at 358). However, as the costs associated with producing electronic documents increase, it is important to consider the possibility of shifting at least part of the financial burden to the demanding party.
In addition to the somewhat dated Oppenheimer Fund case, the issue of cost shifting in electronic discovery has been addressed at both the federal and state levels. But the Southern District of New York Court's decision in Zubulake v. UBS Warburg, LLC (217 F.R.D. 309 (S.D.N.Y. 2003) ("Zubulake I")) has become one of the most important precedents.
Zubulake I sets forth a seven-factor balancing test for analyzing whether shifting costs to the demanding party is appropriate. Under this analysis, cost shifting should be considered only when electronic discovery imposes an "undue burden or expense" on the responding party. The key inquiry is whether the information "is kept in an accessible or inaccessible format (a distinction that corresponds closely to the expense of production)." (Zubulake I, 217 F.R.D. at 318.)
Evaluating whether information is in an accessible format "turns largely on the media on which it is stored." The court identified five categories of data, starting with the most accessible: 1) active, online data (for example, hard drives); 2) near-line data (optical discs); 3) offline storage or archives; 4) backup tapes; and 5) erased, fragmented, or damaged data. The court characterized the first three categories as "accessible." Although the time it takes to access the data in these three formats varies, the information is stored in a readily usable format and no other processing is necessary other than retrieval. Where data are in an accessible format, the general rule is that the responding party bears the cost of producing the data.
By contrast, "inaccessible" data such as that stored on backup tapes and erased, as well as fragmented or damaged data, require significant manipulation and reconstruction to become usable for responding to discovery requests. The cost-shifting inquiry should be entertained only when producing the requested documents requires the responding party to reconstruct inaccessible data.
When the responding party faces incurring substantial expense in reconstructing "inaccessible data," the following seven factors-as articulated by the Zubulake I court-should be considered in determining whether costs should be shifted to the demanding party, weighted in the following order: 1) the extent to which a discovery request is specifically tailored to relevant information; 2) the availability of the requested information from other sources; 3) the total cost of production, compared with the amount in controversy; 4) the total cost of production, compared with the resources available to each party; 5) the relative ability of each party to control the costs and the incentive to do so; 6) the importance of the issue at stake in the litigation; and 7) the relative benefits to the parties of obtaining the information. (Zubulake I, 217 F.R.D. at 322.)
The first two factors are the most important in the Zubulake I balancing test. When the first two factors are met-requests are appropriately tailored and the requested information is not available from another source-courts generally will not favor shifting costs to the demanding party. When it is determined that cost shifting is appropriate, only the costs associated with reconstructing and searching the "inaccessible" data are appropriately shifted. Costs associated with reviewing documents (for privilege and confidentiality) and producing the data once it has been reconstructed (printing, copying, Bates numbering) remain the responsibility of the responding party. (Zubulake v. USB Warburg, LLC, 216 F.R.D. 280, 28991 (S.D.N.Y. 2003) ("Zubulake III"); see also, OpenTV v. Liberate Tech., 219 F.R.D. 474 (N.D. Cal. 2003).)
In the Zubulake case itself, when the court applied its own seven-factor balancing test, it ultimately ordered the demanding party to pay 25 percent of costs associated with restoring backup tapes containing relevant email. (Zubulake III, 216 F.R.D. at 290.)
The Impact of Zubulake I
Zubulake I's influence has been felt far outside the Southern District of New York. For example, a pending proposed amendment to Federal Rule of Civil Procedure 26(b)(2) adopts the principles set forth in Zubulake I. In particular, the amendment reflects the court's emphasis on the delineation between "accessible" and "inaccessible" information-that only the costs associated with restoring inaccessible information may be shifted to the demanding party. The proposed amendment provides that "a party need not produce electronically stored information that is not reasonably accessible because of undue burden or cost." Similar to Zubulake I, the amendment identifies "deleted information," "back-up tape systems," and "legacy data" as examples of inaccessible information.
In addition, as in Zubulake I, under the proposed amendment, the responding party has the burden of showing that the information is not reasonably accessible because of undue burden or cost. However, "[e]ven if the responding party makes this showing, a court may order discovery for good cause and may impose appropriate terms and conditions." Most certainly, "appropriate terms and conditions" encompasses the court's ability to shift costs of reconstruction to the demanding party in cases where the key factors identified in Zubulake I are present.
The proposed amendment to Rule 26(b)(2) was approved by the Judicial Conference on September 20, 2005-and unless Congress rejects, defers, or modifies it, it will become effective as early as December 1, 2006.
Although Zubulake I clearly is influential, other courts have taken a different path. For example, the Sixth District Court of Appeal in California deviated from the seven-factor balancing test in holding that a demanding party should pay the "reasonable expense" for a "necessary translation" of electronic data. (Toshiba Am. Elec. Components, Inc., 124 Cal. App. 4th at 762.)
In Toshiba the court rejected the test set forth in Zubulake I and embraced a statutorily based cost-shifting rule for California state court discovery disputes. In addressing the issue of which party should pay the cost for recovering email stored on 800 computer backup tapes, the court concluded that California Code of Civil Procedure section 2031(g)(1) (currently found at Cal. Code Civ. Proc. Â§ 2031.280(b)) requires the demanding party to pay the "reasonable expense for a necessary translation" of electronic data.
In the underlying action, Lexar Media, Inc. sued Toshiba America Electronic Components, Inc. for misappropriation of trade secrets, breach of fiduciary duty, and unfair competition. Lexar subsequently sought 60 categories of documents from Toshiba. In response, Toshiba produced 20,000 pages of "readily available" responsive documents. However, a dispute arose as to which party should pay for the cost of recovering additional responsive electronic material-the email stored on the backup tapes.
The estimated cost to process, analyze, restore, search, and produce the specified data on the backup tapes was between $1.5 million and $1.9 million. The cost of producing even a relatively small sample-130 tapes surrounding 15 key dates-was estimated at $211,250.
Toshiba asked Lexar to cover some or all of those costs. Lexar refused, arguing that it was unfair for it to bear the cost of production simply because Toshiba had not maintained the data in an easily accessible format. Although the trial court initially granted Lexar's motion to compel, the court of appeal accepted Toshiba's argument that then-section 2031(g)(1) was an automatic cost-shifting provision that required Lexar to cover the reasonable cost of restoring the inaccessible data from Toshiba's backup tapes.
In reaching this decision, the court relied on the statutory language: "If necessary, the responding party at the reasonable expense of the demanding party shall, through detection device, translate any data compilations included in the demand into reasonably usable form." (See Cal. Code Civ. Proc. Â§ 2031.280(b) (formerly Â§ 2031(g)(1).) Although the court acknowledged the general rule that the responding party bears the expense involved in responding to discovery requests, it pointed out that "principles of fundamental fairness require the demanding party to pay any significant 'special attendant' costs beyond those typically involved in responding to routine discovery," and that the California Legislature clearly intended to require cost shifting in some circumstances. (San Diego Unified Port Dist. v. Douglas E. Barnhart, Inc., 95 Cal. App. 4th 1400, 1405 (2002).)
As support for this determination, the court pointed to two state statutes: California Code of Civil Procedure section 2034(i) (currently at Cal. Code Civ. Proc. Â§ 2034.430(b)), a provision that requires a party to pay its opponent's expert witness fees for deposition, and California Code of Civil Procedure section 2034(i) (currently at Cal. Code Civ. Proc. Â§ 2025.510(b)), a provision that requires the party noticing a deposition to bear the cost of transcription. The court concluded that both statutes reflected the Legislature's intent to shift costs to the demanding party in certain circumstances to ensure fairness. The clear language of section 2031(g)(1), the court opined, reflected a similar legislative intent. As such, it held that the expense of translating a data compilation into usable form shifts to the demanding party so long as it is a "reasonable expense for a necessary translation." (Toshiba, 124 Cal. App. 4th at 773.)
In the Toshiba case, the court instructed that in an effort to "inform the analysis of the extent to which [the demanding party] should bear the expenses," the parties should "meet and confer about translating a sample." (Toshiba, 124 Cal. App. 4th at 773.) At the same time, however, the court also concluded that the trial court abused its discretion when it ordered Toshiba to produce its electronic documents without requiring Lexar to pay any portion of the translation costs absent a finding that the translation was not necessary.
Responding to Discovery Requests In addition to reviewing and understanding the federal and state authorities controlling cost shifting for electronic discovery, attorneys should consider the following practical steps when confronted with a case likely to involve electronic discovery issues.
Analyze all standards. Analyze standards from all jurisdictions, such as those set forth in Zubulake and Toshiba, to develop cost-shifting arguments. Both standards give considerable leeway to argue why costs should or should not be shifted. In crafting arguments under the Toshiba standard, the focus should be on the reasonableness of the cost to translate a data compilation and the necessity of the translation. Under the federal standard, remember that cost-shifting arguments will likely be entertained by a court only when the information sought is deemed "inaccessible."
Once "inaccessibility" is established, the two most significant issues will be whether the discovery request is specifically tailored and whether the information is available from alternative sources.
Finally, both the Toshiba and Zubulake standards permit cost shifting only for expenses associated with the retrieval or translation of data. Costs associated with reviewing and producing the material remain, as usual, with the responding party.
Make clear discovery requests. Discovery requests should be clear, specific, and concise. When drafting requests, avoid overbroad, generic references to "all electronic data or documents." Rather, identify particular types and descriptions of electronic data-along with the relevant time period, custodian, and subject matter. This type of tailored request will help reduce the overall costs associated with electronic discovery for all parties involved.
Discuss protocol early. Meet and confer with opposing counsel as early as practicable to discuss an electronic discovery protocol. Early agreements on issues regarding preservation, scope, production of sample data, third-party vendors, and cost sharing can prevent additional expenses associated with litigating these issues later.
Janet H. Kwuon (firstname.lastname@example.org) is a partner at Reed Smith in Los Angeles and a member of the firm's Product Liability Practice Group. Krystal Hauserman (email@example.com) is an associate at Reed Smith in Los Angeles and a member of its Business Trial Practice Group.