2005 UCLA J.L. & Tech. 2

Electronic Discovery and the Challenge Posed by the Sarbanes-Oxley Act
Daniel B. Garrie & Matthew J. Armstrong

Contributions by:
Ervin Adler, William R. Burdett, & Thomas J. Routt

Disclaimer:
The views expressed in this article do not necessarily represent the views of the U.S. Department of Justice or the United States and Boeing Corporation and NS8.1

INTRODUCTION

[¶1] In 2002, the United States Congress passed the Sarbanes-Oxley Act, a complex set of provisions that imposes additional federal requirements2 on all publicly traded corporations.3 Sarbanes-Oxley influences the business operations of public corporations because its provisions mandate changes pertaining to financial document storage4 and the financial transparency of public companies.5 Furthermore, these changes impact current legal rules and practices. One specific legal area Sarbanes-Oxley impacts is electronic discovery (“e-discovery”), and more specifically, e-discovery cost-shifting.6

[¶2] Currently, courts determine whether to grant cost-shifting relief in e-discovery disputes by applying a seven-part test, or some variation thereof,7 established in Zubulake v. UBS Warburg.8 While the Zubulake cost-shifting test is a considerable improvement over prior cost-shifting approaches, the test is not a panacea.9 Applications of the Zubulake cost-shifting test can occasionally produce harmful and unjust outcomes in three ways: first, applications of the Zubulake test encourage litigants to use e-discovery blackmail10 by creating financial incentives for wealthy producing parties to settle claims rather than comply with expensive e-discovery requests. Second, the Zubulake test fosters e-discovery evasion11 because producing parties can use discovery-inefficient storage systems12 to prevent production of otherwise discoverable digital documents by making production costs too expensive to justify the request. Third, the Zubulake test will not generate equitable results under Sarbanes-Oxley because public companies will, on average, be compelled to divulge more digital documents in e-discovery disputes than their private counterparts.13 These three negative scenarios are not a result of poor judicial construction, but rather arise from the difficulty of extrapolating traditional discovery rules to the rapidly changing digital medium.

A BRIEF HISTORY OF E-DISCOVERY COST-SHIFTING

[¶3] Today, businesses are awash in electronic information. One study found that approximately 547.5 billion e-mail messages were transmitted in the United States alone in 2003.14 Furthermore, a single large Fortune 500 company can generate and receive several million e-mail messages and digital documents each day.15 Finally, in 2001, 93% of all newly recorded information was stored on magnetic media.16 These facts demonstrate society’s growing dependence upon digital information.17 In light of these statistics, one would expect e-discovery to be the norm and paper discovery to be the exception.18 However, statistics suggest that litigators continue to underutilize e-discovery techniques in litigation.19 Recent research indicates that most companies are unaware that civil discovery can result in the impoundment or copying of computer hard drives, e-mails, instant messages, or servers.20 Despite these oversights, litigants are held to demanding e-discovery standards even though the judiciary system lacks a mechanical cost-allocation test that consistently delivers fair results.21

A. The History of E-Discovery

[¶4] In 1970, the FederalRules of Civil Procedure (“Federal Rules”) were amended in an attempt to clarify the issue of e-discovery.22 The Advisory Committee Notes for the 1970 amendments to the Federal Rules of Civil Procedure revised the description of "documents" in Rule 34(a) to accord with changing technology.23 “It makes clear that Rule 34 applies to electronic data compilations from which information can be obtained only with the use of detection devices, and that when the data can as a practical matter be made usable by the discovering party only through respondent's devices, respondent may be required to use his devices to translate the data into usable form.”24 Corporations have been ordered to produce, sometimes at considerable expense, computerized information, including e-mail messages, support systems, software, voice mail systems, computer storage media and backup tapes and telephone records.25 The language of the Federal Rules, as well as recent case law, both affirm that e-discovery requests are controlled by the traditional discovery rules provided by Federal Rules 26 and 34.26

[¶5] Over the past thirty years, courts have struggled to integrate digital data production’s highly variable cost structure into the Federal Rules’27 traditional discovery principles.28 During the past decade, the federal courts have attempted to align e-discovery with technological advances in such cases as McPeek v. Ashcroft,29 Rowe Entertainment, Inc. v. The William Morris Agency, Inc.,30 and Zubulake v. UBS Warburg, LLC.31 In each case, the respective court read and applied Rule 26(b)(2) with the intent of protecting producing parties from undue burden or expense.32 All three courts sought to resolve the digital discovery conundrum by: (1) clarifying the situations where courts should consider cost shifting; and (2) developing a test derived from the Federal Rules to determine which party should bear data production costs.33

[¶6] The federal district court in McPeek established a rudimentary e-discovery cost-shifting test to determine whether to grant cost-shifting relief.34 The McPeek court applied a marginal utility approach, 35 stating that “[t]he more likely it is that the backup tape contains information that is relevant to a claim or defense, the fairer it is that the [producing party] search at its own expense.”36 This holding represents an early attempt to develop a cost-shifting test for e-discovery disputes.37

[¶7] Courts applied McPeek, or some variation thereof, until the Rowe decision created an eight-factor e-discovery cost-shifting test that incorporated McPeek’s marginal utility test.38 The Rowe court used these eight factors to emphasize that courts should focus on data’s marginal utility when evaluating e-discovery disputes.39 By recognizing a multitude of relevant factors, the Rowe decision represented a significant step for courts in creating a more precise e-discovery cost-shifting test.40

[¶8] Courts considered the Rowe test to be “the gold standard,”41 until Zubulake augmented its factors to reflect those listed in FRCP 26(b)(2)(iii).42 The Zubulake court consolidated the Rowe test into seven factors that courts should consider when deciding whether to grant cost-shifting relief.43 The seven Zubulake factors are: (1) the extent to which the request is specifically tailored to discover relevant information; (2) the availability of such information from other sources; (3) the total cost of production, compared to the amount in controversy; (4) the total cost of production, compared to the resources available to each party; (5) the relative ability of each party to control costs and its incentive to do so; (6) the importance of the issues at stake in the litigation; and (7) the relative benefits to the parties of obtaining the information.44 The Zubulake court stated that the first six factors of the seven-factor test correspond to the three explicit considerations of Rule 26(b)(2)(iii).45 The Zubulake court held that these changes were necessary because the Rowe test generally favored cost-shifting, undercutting the presumption46 that the producing party bears document production costs.47 Thus, Zubulake represents a step towards creating a cost-shifting test that relies on Rule 26(b)(2)’s proportionality factors48 to maintain consistency with the Federal Rules’ presumption that the producing party bears the cost of production.49 Presently, courts rely upon the seven-factor Zubulake test, or some modification thereof, to resolve these e-discovery cost-shifting disputes.50

[¶9] In summary, the Zubulake, Rowe, and McPeek courts sought to resolve the digital discovery conundrum by: (1) clarifying the situations where courts should consider cost shifting; and (2) providing a test derived from the Federal Rules that determines which party should bear data production costs.51 While each of these cases represents notable milestones in the application of discovery principles to technology, the reality is that courts will never be a step ahead of technology.52 Consequently, courts should develop an e-discovery cost-shifting test of broad application, such that the legal test is independent of the particular characteristics of the underlying technology at issue.

CRITICISMS OF THE ZUBULAKE TEST

[¶10] Although courts currently apply the Zubulake test in e-discovery disputes, it does not always generate just results53 because it allows technology to dictate judicial outcomes.54 The Zubulake test has three significant problems that must be corrected in order to ensure an equitable forum for litigants. First, the Zubulake test facilitates e-discovery blackmail by plaintiffs who force wealthy defendants to settle cases by constructing narrowly tailored yet expensive e-discovery requests that cause production costs to economically prohibit litigation.55 Second, the Zubulake test fosters e-discovery evasion because it rewards parties who use discovery-inefficient legacy systems, while penalizing those who choose, or are required by law, to use efficient electronic record systems.56 Third, because the Zubulake test relies heavily on computer system costs in e-discovery disputes, the Zubulake test places an unfair burden upon companies that are legally required to retain data.57 The courts must either modify the Zubulake test, or adopt a new e-discovery cost-shifting solution that addresses these three deficiencies.

A. The Zubulake Test Permits E-Discovery Blackmail

[¶11] First, the Zubulake test does not always generate equitable results because it is unable to prevent the opportunistic use of e-discovery blackmail.58 This scenario arises when a discovering party constructs a narrowly tailored, yet expensive e-discovery request such that the total cost of production approaches, or in some cases exceeds the cost of settling a claim.59 Thus, e-discovery blackmail enables discovering parties to compel wealthy producing parties to settle claims60 that they would otherwise contest.61

[¶12] Although the third Zubulake factor does in fact weigh the total cost of production against the amount in controversy, such a consideration is just one of seven factors in the test,62 and often is overcome by the other six factors.63 Additionally, the fourth Zubulake factor considers the financial resources available to each litigant relative to the total cost of production,64 which inhibits wealthy corporate defendants from receiving equitable cost-shifting relief65 upon receiving an expensive e-discovery request.66 Corporations with large fiscal resources are thereby placed at a serious disadvantage because they can only avoid e-discovery blackmail when they can effectively file a 12(b)(6) motion to dismiss before discovery takes place.67 E-discovery blackmail further harms litigants by effectively denying them a forum for their grievances68 if the e-discovery request is not rejected at the outset, on grounds of being unduly burdensome or expensive.69 Courts must therefore modify the Zubulake test to eliminate the opportunistic use of e-discovery blackmail in the discovery process.70

B. The Zubulake Test Permits E-Discovery Evasion

[¶13] Second, the Zubulake test fosters e-discovery evasion by producing parties because it fails to recognize that data can be stored on legacy storage systems to prevent discovery of otherwise discoverable data, thereby causing the combined costs of finding and producing data to exceed the amount in controversy.71 In such situations, courts applying Zubulake are likely to grant cost-shifting relief to alleviate the alleged burden placed upon the producing party, especially when the producing party has limited financial resources.72 Once a court decides to shift costs, it becomes unlikely that a producing party will be compelled to produce digital documents73 because the discovering party has to pay costs that approach or exceed the value of the case, making discovery economically impractical.74 Therefore, courts applying Zubulake effectively reward parties for using discovery-inefficient legacy storage systems, while punishing parties that choose, or are required by law, to use efficient electronic record systems.75 Zubulake thus creates a perverse disincentive that prevents companies from investing in more efficient data storage technologies, because parties with efficient storage systems are generally forced to produce more digital documents than parties using legacy storage systems.76 Although companies eventually may determine that the need for a newer storage system exceeds the risks posed by broad electronic discovery,77 litigants should not be forced to weigh potential adverse legal consequences against the benefits that could be realized by investing in appropriate systems for their business needs.

[¶14] Zubulake itself offers a prime example of e-discovery evasion.78 In Zubulake, the court used its discretion to construct a narrower e-discovery request than originally requested, because the respondent’s documents were stored on 94 inaccessible backup tapes that required extremely costly restoration.79 The Zubulake court ordered the discovering party to pick five of the tapes to be initially restored, while restoration of the remaining tapes remained contingent upon a successful initial search.80 The producing party, by following its corporate e-mail policy, prevented the discovering party from discovering all of the requested e-mails by saving them on backup disks instead of a more accessible medium.81 Zubulake’s holding implies that by maintaining a complex disaster and data recovery system, litigants can sidestep e-discovery requests for data residing in their data warehouses. The Zubulake court’s failure to differentiate between acceptable and unacceptable electronic information repositories in creating its cost-shifting test encourages e-discovery evasion.82 If this shortcoming remains unaddressed by the judicial system, e-discovery evasion will likely become a more frequently used means for abusing the e-discovery process.

C. The Zubulake Test Will Not Yield Equitable Judicial Outcomes Once Provisions of the Sarbanes-Oxley Act Become Effective

[¶15] Third, the Zubulake test will be unlikely to generate equitable judicial decisions, because Sarbanes-Oxley requires public companies to store and access financial data differently than private companies.83 Prior to Sarbanes-Oxley most public and private companies in industries other than financial services and healthcare did not have to comply with burdensome legally mandated data retention policies.84 Under Sarbanes-Oxley, however, public companies are distinguished from their private counterparts in that they must retain financial data in order to comply with the legislation.85 Not only are public companies forced to retain more data than private companies, but public companies are now required to maintain the data in an easily accessible manner.86 In e-discovery disputes governed by Zubulake, therefore, courts are more likely to compel public companies to produce digital documents than their private counterparts because Sarbanes-Oxley requires public companies to maintain efficient storage systems. Public companies’ efficient storage systems reduce their data production costs, and consequently, their chances of receiving cost-shifting relief.87 Thus, under Zubulake, publicly traded companies are subject to much greater litigation risks and costs than their private counterparts, giving private corporate competitors an advantage in lawsuits involving e-discovery.88

[¶16]For example, Bill Teuber, the CFO of the information-storage giant EMC, “spent more than $1 million and thousands of man-hours complying with two of the main statutes in the Sarbanes-Oxley Act of 2002 - Section 404, relating to internal controls; and Section 302, mandating CEO and CFO certifications of quarterly financial statements.”89 The actual cost of full compliance with Sarbanes-Oxley is uncertain and the resulting burden on public corporations is expected to exist for the foreseeable future because of the ongoing testing and disclosure requirements.90 EMC’s private corporate competitors are not incurring these costs. Thus, if EMC becomes involved in litigation, under Zubulake, it will likely incur further costs in the form of an e-discovery dispute ― because its data production costs will be lower than its private counterparts, thereby increasing the scope and likelihood of e-discovery. Courts must address the imbalance between public and private companies arising in e-discovery cost-shifting disputes because such imbalance unfairly burdens public companies, thereby creating inequitable judicial outcomes.

AN ELECTRONIC DISCOVERY COST-SHIFTING SOLUTION

[¶17] The previous section of this paper discussed three significant problems with the Zubulake test: (1) it permits e-discovery blackmail; (2) it permits e-discovery evasion; and (3) it unfairly burdens companies that must comply with digital records legislation. Any solution must be designed to alleviate these problems, account for a multitude of e-discovery situations, and be rigid enough to provide a mechanical rule that gives guidance to future litigants.91 This is a difficult task because traditional discovery rules were designed to function with paper documents, not digital ones. We propose the Benchmark Cost-Shifting Test (“BCST”)92 as an alternative e-discovery dispute resolution system to Zubulake and its progeny. The BCST adequately addresses these problems by separating the e-discovery dispute resolution process into three distinct parts: (1) a data discoverability test, (2) a system-normalized cost analysis, and (3) a query relevancy assessment. The BCST’s three-step process provides a mechanical and equitable test that gives adequate guidance to future litigants while addressing Zubulake’s deficiencies by normalizing production costs and accounting for technological change.

[¶18] Although courts have extrapolated traditional discovery principles from paper documents to digital ones, courts have also been challenged by production cost differences between paper and digital documents.93 The costs of digital data production have the potential to exceed the cost of ordinary paper discovery94 because of the potential size of the data set95 as well as the complex steps of finding96 and viewing digital documents. Both technological complexities and poorly designed information systems continue to frustrate judicial efforts to formulate a unified test that governs all e-discovery requests. Courts, when faced with expensive e-discovery requests, thus generally rely on cost-shifting to insulate producing parties from enormous costs.97 While cost-shifting remains a viable option in some e-discovery disputes, technological advancements have eliminated the need to use cost-shifting in the majority of e-discovery disputes.

[¶19] The BCST solves these problems by applying traditional discovery rules to e-discovery disputes while limiting cost-shifting to the most unduly burdensome production requests.98 The BCST is composed of three parts: (1) a data discoverability test, (2) a system-normalized cost analysis, and (3) a query relevancy assessment. It also contains two exceptions that account for situations where there is: (i) a significant public interest;99 or (ii) a request for deleted data.100 The BCST is designed to grant cost-shifting relief only in the most unduly burdensome situations where production costs economically prohibit litigation. The BCST also addresses the wide spectrum of e-discovery production costs by evaluating all production requests against a benchmark standard. Thus, the BCST creates a mechanical rule that is both responsive to technological change and capable of providing equal treatment for all litigants.

A. The Data Discoverability Test

[¶20] In the BCST’s first step courts apply a data discoverability test to determine whether requested data is discoverable under Federal Rules of Civil Procedure 34(a) and 26(b)(1),(2).101 Generally, all documents are discoverable under the Federal Rules of Civil Procedure unless: (1) the documents are privileged; (2) the discovery sought is unreasonably cumulative or duplicative; (3) the discovery sought is obtainable from some other source that is more convenient, less burdensome, or less expensive; (4) the discovering party already had ample opportunity to obtain the information sought in discovery; or (5) the burden or expense of the proposed discovery outweighs its likely benefit.102 In close cases where an e-discovery request’s expense approaches its likely benefit,103 courts should generally favor discoverability because production costs can be shifted or minimized under the second and third elements of the BCST.104 If, however, it is obvious that a discovering party purposefully abused the discovery process by requesting an extremely broad or expansive search, a court should deny the discovery request.105 By stipulating that courts must initially apply the data discoverability test, the BCST ensures that courts evaluate digital document requests in the same way as paper document requests.106

B. The System-Normalized Cost Analysis

[¶21] In the BCST’s second step, courts determine whether to shift costs by applying the system-normalized cost analysis test and then evaluating its results.107 Courts apply this test when a producing party argues that an e-discovery request is unduly burdensome, thereby warranting cost-shifting relief.108 Courts perform this analysis in four-sequential steps by: (1) determining the amount in controversy; (2) computing the adjusted amount in controversy; (3) computing the system-normalized cost of production; and (4) comparing the adjusted amount in controversy against the system-normalized cost of production.

[¶22] First, a court determines the amount in controversy by calculating the expected return of the claim. To do so, a court evaluates the factual record and subjectively assigns different probabilities to different possible outcomes based on the strength of the case. The court then multiplies the probabilities of recovery against the respective projected amounts of recovery and adds the results to determine the expected return of the claim.109 For additional guidance, courts may allow both the producing party and the responding party to provide the court with their own calculations of the claim’s expected return. For instance, if a claim has a 25% chance of receiving a positive jury verdict yielding $1 million, a 50% chance of obtaining a $600,000 settlement, and a 25% chance of receiving a negative jury verdict yielding nothing [$110, the claim’s expected return would be $550,000. A court then uses the figure representing the claim’s expected return, here $550,000, as the amount in controversy when evaluating the claim under the remainder of the BCST.

[¶23] Second, a court computes the adjusted amount in controversy by multiplying the amount in controversy by a diminishing factor that represents the importance of the data to the discovering party.111 Courts evaluate each case individually and apply a diminishing factor between 30 and 80 percent, where 30 percent represents data that is helpful but not crucial to the discovering party, and 80 percent represents data that forms the basis of the discovering party’s claim. If the data’s value to the discovering party is somewhere in-between, courts can use a factor between 30 and 80 percent that adequately represents that value. The 30 to 80 percent factor determined by a court is then multiplied against the amount in controversy, yielding the adjusted amount in controversy. For instance, in our hypothetical case, if a court determines that the requested data is critical to the discovering party’s claim, it will multiply the amount in controversy of $550,000 by a diminishing factor of 80 percent, yielding an adjusted amount in controversy of $440,000.

[¶24] Courts should augment the amount in controversy prior to weighing it against the system-normalized cost of production because data production costs have the potential to economically prohibit litigation before the total cost of production equals the amount in controversy. For instance, the upper end of the range for the diminishing factor is set at 80 percent of the amount in controversy because producing parties face other costs - such as attorney fees - that at times can account for up to 20 percent of the amount in controversy. If the amount in controversy is not multiplied by 80 percent, producing parties could potentially face costs in excess of 100 percent of the amount in controversy.112 This outcome would be counterproductive because it would economically prohibit a producing party from litigating claims filed against it. Without this protective buffer zone, discovering parties could compel producing parties to settle claims that they would otherwise contest, effectively denying them a forum for their grievances. The lower end of the range utilizes a 30 percent reducing factor to limit production costs imposed upon producing parties for relevant, yet unnecessary, discoverable data. The 30 percent factor gives discovering parties access to a reasonable amount of relevant data while still providing producing parties with the possibility of obtaining a better outcome through litigation than can be obtained through settlement.

[¶25] For the third step in the analysis, a court calculates the system-normalized cost of discovery by utilizing the System-Normalized E-Discovery Cost Curve.113 This graph represents the average production costs based on the size of a data storage system and the size of a data request. The curve is determined by an evolutionary algorithm114 that calculates data production costs for different systems of all types and sizes.115 To use the graph, a court must first estimate the size of the data request116 and the size of the of the producing party’s system.117 Then, the court calculates the system-normalized cost of production for the e-discovery request by imputing these two values. For example, to find the cost of discovering 10-megabytes of information on a 1,500-gigabyte system, a judge looks up 10-megabytes on the X-axis, which represents the size of the data request. Next, the judge looks up 1,500-gigabytes on the Z-axis, which represents the size of the producing party’s system. Finally, the judge reads off the corresponding value on the Y-axis of $100,000, which represents the system-normalized cost of production for the e-discovery request.

Figure 1: System-Normalized E-Discovery Cost Curve

X-axis = The size of the data request in Megabytes.

Y-axis = The system-normalized cost of production in Dollars.

Z-axis = The size of the producing party’s system in Gigabytes.

[¶26] Step four of the analysis requires a court to compare the adjusted amount in controversy to the system-normalized cost of production. Cost-shifting relief is granted whenever the system-normalized cost of production exceeds the adjusted amount in controversy.118 Otherwise, no cost-shifting relief is granted and the producing party bears the production costs.119 Thus, the BCST’s second step grants producing parties cost-shifting relief whenever the system normalized cost of production would economically prohibit them from contesting a claim in court. For example, in the hypothetical case discussed above, a court comparing the system-normalized cost of production value of $100,000 against the adjusted amount in controversy of $440,000 would not grant cost-shifting relief. In this situation, the data production costs do not economically prohibit the producing party from engaging in litigation, so, according to the presumption, the producing party must bear the cost of production.120 Consequently, the BCST’s system-normalized cost analysis test ensures that neither litigant can gain an unfair advantage by using different data storage systems to manipulate judicial outcomes.121

C. Query Relevancy Assessment

[¶27] In the BCST’s third step, a court applies a query relevancy assessment122 to modify unduly burdensome or expensive discovery requests.123 This BCST element allows courts to exercise judicial discretion to minimize discovery costs borne by producing parties by formulating more appropriate search parameters.124 In making this assessment, a court considers: (1) the relatedness of the query terms to the material in dispute; (2) the practicality of the resultant data set; (3) the formulation of the search terms; and (4) the degree of expertise required to draft an appropriate query that discovers the desired documents.125 Courts may elect to alter overly broad discovery requests by forcing the discovering party to limit its search to certain keywords in specified locations.126 Courts should seek to construct fair solutions by employing query relevancy assessments in most e-discovery disputes.127 Only where it is evident that a discovering party purposefully abused the discovery process by requesting an extremely broad search should courts take a firm stance and deny the discovery request altogether.128

D. Exceptions to the Rule

[¶28] Although the BCST is a versatile test that courts can effectively apply in many situations, its application could create undesirable results when there is either: (1) a significant public interest; or (2) a request for deleted data.129 In these situations, courts should refrain from making a normal BCST application, and instead should apply rules developed specifically to address these two exceptions.

[¶29] First, the public interest exception permits courts to compel digital production in situations where the total cost of production exceeds the amount in controversy when the issues at stake are of significance to the public.130 In such situations, a court may permit e-discovery to whatever extent is necessary to discover the truth.131 This exception grants a court the discretion to ascertain whether a matter of public interest is sufficiently important to justify unfairly burdening the producing party, which must either settle the case or incur high production costs. This exception should only be applied when an extraordinary public interest exists because its application is manifestly unjust to producing parties.

[¶30] Second, the spoliation exception gives courts greater leniency when addressing cost-shifting disputes involving deleted data. When a litigant seeks to discover previously deleted files, courts should apply different standards, depending upon whether the producing party deleted the data negligently, maliciously, or in the normal course of business.132 If a producing party deleted data negligently or maliciously, that party should be required to bear the cost of producing the requested data; otherwise, the producing party would be unjustly rewarded.133 If, however, the producing party deleted the data in the normal course of business, the discovering party should bear the costs of restoring the data.134 Under current discovery rules, no party has a duty to preserve electronic data unless it knows or should know that it is likely to be requested in a future litigation, or if retention is specifically mandated by law.135 In cases involving deleted data, producing parties already suffer the adverse affects of the resilient nature of electronic data.136 It is therefore unjust for a court to require a producing party to bear the additional cost of finding and restoring legally-deleted data, especially if the discovery of that deleted data proves to be against the producing party’s interests.137

BCST: A MORE RELIABLE TOOL FOR JUDICIAL GUIDANCE

[¶31] The BCST introduces five innovative ideas that enable the judicial system to resolve e-discovery cost-shifting disputes in an equitable manner. First, the BCST clarifies e-discovery dispute resolution by dividing it into three distinct steps so that courts can more readily determine: (1) whether discovery is permissible; (2) whether cost-shifting is warranted; and (3) whether costs can be reduced by performing a query relevancy assessment. Second, the BCST adds a system-normalized component to the balancing test that greatly reduces e-discovery evasion and ensures equitable treatment of all litigants. Third, the BCST introduces a query relevancy assessment that enables courts to minimize e-discovery production costs by changing search parameters. Fourth, the BCST further ensures equitable treatment of all litigants by removing the resources available to each party as a factor in the cost-shifting balancing test. Fifth, the BCST establishes a common-law digital document standard by encouraging all potential litigants to utilize industry standard data storage systems.

A. E-Discovery Disputes are Analyzed in Three Distinct Steps

[¶32] The BCST divides the e-discovery process into three distinct steps: (1) whether discovery is permissible, (2) whether cost-shifting is warranted, and (3) whether costs can be reduced by performing a query relevancy assessment.138 By separating e-discovery evaluations into three steps, courts can address issues of data discoverability and cost-shifting with greater clarity than before. By applying mechanical rules at each step, courts can eliminate the confusion that potentially can result from multi-factor balancing tests. This solution benefits both courts and litigants because it is easier to apply and provides better guidance to litigants about how courts resolve e-discovery disputes.

[¶33] In the BCST’s first step, a court determines whether requested materials are discoverable under the Federal Rules. Byisolating this step from cost-shifting considerations, the BCST ensures that courts apply similar discovery rules to digital documents and paper documents. In the BCST’s second step, a court performs a system-normalized cost analysis to determine whether cost-shifting is warranted. Here a court evaluates an e-discovery dispute based on a benchmark standard cost of data production derived from the System-Normalized Digital Discovery Cost Curve, rather than basing its decision on the highly variable and potentially inflated cost of production on the producing party’s system. The BCST approach thus fixes many of Zubulake’s problems by using an industry standard data production cost to evaluate all e-discovery requests, regardless of the financial status of the litigants. Finally, in the BCST’s third step, a court performs a query relevancy assessment to minimize data production costs by revising litigants’ search parameters. This step enables courts to minimize wasteful discovery while still providing litigants with complete access to discoverable data.

B. A System-Normalized Component is Incorporated into E-Discovery Cost-Shifting Disputes

[¶34] The BCST’s primary innovation is its use of a benchmark standard to determine litigants’ data production costs instead of arbitrary and speculative “actual” costs of production. This approach limits a producing party’s ability to use e-discovery evasion because the “actual” cost of data production plays no notable role in the cost-shifting assessment. Thus, under the BCST, producing parties are unable to avoid producing digital documents by storing troublesome data on discovery-inefficient legacy storage systems. The BCST brings e-discovery cost-shifting in line with the treatment of traditional “paper” discovery requests because cost-shifting is limited to situations where production costs economically prohibit litigation.

[¶35] Additionally, the BCST’s system-normalized component yields two ancillary benefits. First, it eliminates the need for courts to differentiate between different levels of data accessibility by creating a single e-discovery test that applies to data stored on many types of storage systems. Second, the BCST promotes economic efficiency by encouraging litigants to use storage systems that comport with industry standards, instead of rewarding litigants for using discovery-inefficient systems that ultimately frustrate discovery requests.

C. A Query Relevancy Assessment is Incorporated into E-Discovery Cost-Shifting Disputes

[¶36] The BCST’s query relevancy assessment incorporates a new element into e-discovery cost-shifting analyses that enables courts to reduce wasteful discovery searches by limiting their search parameters. When making a query relevancy assessment, courts should utilize their discretion in order to formulate narrowly-tailored searches that minimize costs, while still providing the parties access to discoverable data. Courts applying the BCST are therefore able to take a more active role in minimizing wasteful e-discovery costs caused by inefficient discovery requests that substantially burden plaintiffs, yet do not rise to the level of being unduly burdensome.139 This ability also encourages courts to apply traditional discovery rules to e-discovery disputes, because it gives courts the ability to mitigate the anti-cost-shifting inclination that is built-in to a system-normalized cost analysis.

D. The Parties’ Resources Are Removed From the Factor List

[¶37] The BCST encourages equitable treatment of all litigants by removing the parties’ resources from consideration in cost-shifting analyses. Unlike the Rowe140 and Zubulake141 tests, which confuse data discoverability factors with cost-shifting factors, the BCST separates the two different steps so courts can address each item with more clarity. The BCST does not incorporate the parties’ resources into the cost-shifting step of the analysis because the court already has considered the parties’ resources when determining whether data is discoverable.142

[¶38] Furthermore, the BCST does not use the parties’ resources as a factor because such information should play no role in courts’ assessments of whether data production costs economically prohibit litigation. The BCST also addresses the tendency of courts to predicate their cost-shifting decisions upon the parties’ respective financial resources whenever courts are allowed to consider economic resources as a factor.143 In an area of law as widely used as e-discovery, it is preferable to have a mechanical test that applies universally to all litigants, irrespective of the underlying technology used by the litigants.

E. The BCST Creates a Common Law Digital Document Standard

[¶39] Perhaps most importantly, the BCST establishes a common law digital document standard that steps in where technology has fallen short, as seen by the hundreds if not thousands of document formats and storage mediums existing in business today. The BCST provides a mechanism that enables courts to ignore proprietary computer systems, complex networks, and other technological instruments when evaluating e-discovery disputes. Thus, the BCST provides courts with a test capable of evolving with technology, while simultaneously rectifying the problems that plagued previous judicial tests.

A REAL WORLD EXAMPLE

[¶40] As noted in the criticism section, the Zubulake seven-factor balancing test can create inequitable outcomes in e-discovery cost-shifting disputes. In November of 2003, for example, the Zubulake test led to an unfair result in OpenTV v. Liberate Technologies.144 While neither litigant attempted to abuse the Zubulake test’s weaknesses by engaging in e-discovery blackmail or e-discovery evasion,145 the OpenTV court nevertheless reached a holding that complied neither with traditional discovery rules nor common law precedent.146 Instead, the OpenTV court exercised its judicial discretion under Zubulake’s balancing test to decide the case.147 The producing party should not have received cost-shifting relief, while the discovering party shouldered the burden of paying for the costs of discovery. The court’s unbridled use of judicial discretion in applying the Zubulake test highlights the balancing test’s inability to generate consistent results.

A. Applications of the Zubulake Test Can Create Unfair Results

[¶41] In OpenTV the litigants disagreed about who should bear the costs of finding and producing the source code stored in the producing party’s database.148 The producing party offered the discovering party unrestricted use of its office database to extract the source code.149 The discovering party rejected the offer, however, arguing that it improperly shifted the cost of production because it required the discovering party to find and extract data from the database before it could even be reviewed.150 The producing party’s offer, therefore, violated the traditional discovery rule that the producing party must bear the cost of finding and producing documents.151 In resolving this dispute, the OpenTV court looked to persuasive authority and thereby applied the Zubulake test.152

[¶42] The OpenTV court began its analysis by stating that e-discovery cost-shifting is only considered when inaccessible data is sought.153 Thereafter, the OpenTVcourt determined that the producing party’s source code was relatively inaccessible,154 warranting an application of the Zubulake cost-shifting test.155 Next, the court applied each of the seven Zubulake factors separately,156 finding that factors one, two, three, and five weighed against cost-shifting, while factors four and seven favored cost-shifting, and factor six was neutral.157 As in Zubulake, the OpenTV court determined that while the list of factors is instructive, “it is not merely a matter of counting and adding.”158 The court then applied its discretion to create an equitable solution by granting disproportionate weight to factor four: “the cost of production, compared to the resources available to each party.”159 Even though the Zubulake factors collectively weighed against cost shifting, the OpenTV court nevertheless ordered the two litigants to split the data extraction costs because extracting the source code imposed an undue burden and expense upon the producing party and both litigants had similar fiscal resources.160

[¶43] The OpenTV court made two errors in its application of the Zubulake test. First, it disregarded the balancing test’s defined order by applying disproportional weight to factor four.161 Second, the court seemingly based its entire decision upon the financial status of both parties, a factor that should not dictate the amount one pays in judicial actions.162 Under the Zubulake test, factors one and two receive the most weight, while factors three, four, and five receive less weight, and factors six and seven are given variable weightings based on their relevance to the particular case.163 Since factor four falls into the second group of factors,164 the OpenTV court should not have given it greater weight than factors one and two, which weighed heavily against cost-shifting.165

[¶44] Thus, the OpenTV court incorrectly exercised its discretion to shift costs based solely upon the financial strength of the parties, notwithstanding a preponderance of factors that weighed against cost-shifting.166 While the holding arguably achieved a just result under the circumstances, it did not comport with traditional discovery rules.167 The OpenTV holding illustrates the difficulty of applying the seven-factor Zubulake test in a consistent manner. Furthermore, OpenTV demonstrates how extensive use of judicial discretion in e-discovery can result in inconsistent legal applications that provide little useful guidance for future litigants. In an area of law as widely used as e-discovery, courts would be better off to apply a mechanical test such as the BCST that regularly generates fair and legally consistent cost-shifting determinations.

B. OpenTV Under the BCST

[¶45] The OpenTV court would have had the opposite outcome had it applied the BCST. Instead of splitting data extraction costs between the parties,168 the producing party would have borne all of the data production costs. However, the court could have reduced the producing party’s costs by altering the search terms through the BCST’s query relevancy assessment.

[¶46] A court applying the BCST to OpenTV would first determine whether the requested materials are discoverable under Federal Rules 34 and 26(b)(1), (2).169 Generally, all documents, both physical and digital,170 relevant to the claim or defense of a party are discoverable unless: (1) the documents are privileged; (2) the request is unreasonably cumulative or duplicative; (3) the discovery sought is obtainable from some other source that is more convenient, less burdensome, or less expensive; (4) the discovering party already had ample opportunity to obtain the information sought in discovery; and (5) the burden or expense outweighs the likely benefit.171 Here a court should hold that the producing party’s source code is discoverable because the data is not privileged, is not unreasonably duplicative, is not obtainable from another source or another method in discovery, and it probably does not create a burden that outweighs its likely benefit. The only potential problem with the request is that the burden or expense could arguably be found to outweigh the likely benefit since the discovering party requested multiple versions of similar programs.172 In this case, the discovering party requires the majority of the files to support its case because it needs to analyze the progression of compatibility between its program and the producing party’s programs.173 Furthermore, courts should generally favor discoverability of data in questionable e-discovery disputes when applying the BCST, because courts are still able to shift or minimize costs after determining whether the data is discoverable. Thus, a court applying the BCST to OpenTV should permit discovery, because the benefit gained by the discovering party would likely outweigh the burden placed upon the responding party.

[¶47] Once a court establishes that the disputed source code is discoverable, it then must determine whether to consider shifting costs when a producing party claims that an e-discovery request is unduly burdensome or expensive. Under the BCST’s system-normalized cost analysis, cost-shifting relief is granted when data production costs on a benchmark system either economically prohibit litigation or exceed the benefit to the discovering party.

[¶48] The first step of a system-normalized cost analysis requires a court to determine the amount in controversy by finding the expected value of the claim. In OpenTV, the discovering party had filed a patent infringement action requesting specific performance in the form of injunctive relief.174 Although the opinion does not provide a monetary estimation of the claim’s value or a probability of its success, the expected value appears to be well in excess of $200,000.175 Thus, a court applying the BCST to OpenTV could use $200,000 as the amount in controversy.

[¶49] The second step of the BCST’s system-normalized cost analysis requires a court to calculate the adjusted amount in controversy. A court performs this action by applying a diminishing factor between 30 and 80 percent that represents the data’s importance to the discovering party’s claim. In OpenTV, the data sought by the discovering party is necessary to prove its case, so a court applying the BCST to OpenTV would use the upper end of the cost-shifting range and multiply the amount in controversy by 80 percent.176 Thus, a court applying the BCST to OpenTV would use $160,000177 as the adjusted amount in controversy.

[¶50] The third step of the BCST’s system-normalized cost analysis requires a court to determine the system-normalized data production cost from the system-normalized e-discovery cost curve. To do so, a court uses as inputs the size of the data set requested and the size of the producing party’s system into the equation. In OpenTV, the discovering party has requested 100 versions of source code,178 and there was no mention of the size of the producing party’s system.179 Even on a large benchmark system, this search would not cost much more than $10,000.180 Therefore, a court applying the BCST to OpenTV would likely use $10,000 as the system-normalized data production cost.

[¶51] In the fourth step of the BCST’s system-normalized cost analysis a court weighs the adjusted amount in controversy against the system-normalized data production cost to determine whether to grant cost-shifting relief. Under the BCST, it is recommended that courts shift costs when the system-normalized data production cost exceeds the amount in controversy. Here, a court would not likely grant cost-shifting relief because the system-normalized data production cost of $10,000 is much less than the adjusted amount in controversy of at least $160,000. Thus, a court applying the BCST to OpenTV would find that cost-shifting would not be warranted, because the benchmark data production cost is only about 6.25% of the amount in controversy.181

[¶52] As a final point, a court applying the BCST to OpenTV could, at its discretion, make a query relevancy assessment to minimize the responding party’s data production costs. In performing a query relevancy assessment, a court considers: (1) the relatedness of the query terms to the material in dispute; (2) the practicality of the resulting data set; (3) how the search terms are formulated; and (4) the degree of expertise required to draft a query to discover the desired documents. In OpenTV, the query terms bear a high correlation to the disputed material because they are names of specific versions of source code; the resulting data set is impractical because multiple versions of similar programs have been requested; the search terms are names of specific programs that cannot be modified by the court and; the degree of expertise required to draft the search is minimal because each search consists of specific program names. A court would modify the requesting party’s search parameters to increase the practicality of the resultant data set because the redundant program versions create additional, unnecessary production costs.182 Such modification of search parameters minimizes the costs incurred by the producing party while still allowing the discovering party to prove its case by tracking the changes between different program versions.

[¶53] In sum, a court applying the BCST to OpenTV would likely reach an opposite result than that reached in the actual case. Instead of crafting a holding based upon the financial similarities of the litigants as was done in OpenTV, a court applying the BCST should determine whether to shift costs based upon the benchmark cost of production. In addition, the court, at its discretion, could modify the query to increase the likelihood of discovering the requested data. Thus, a court applying the BCST is able to use a methodical approach that evaluates all parties evenhandedly, irrespective of the extent of each party’s data storage systems and financial resources.

CONCLUSION

[¶54] The BCST replaces the Zubulake test and eliminates its unintentional consequences by notably reducing the effects of e-discovery blackmail, e-discovery evasion, and legislative disparities between public and private companies.183 The BCST severely limits e-discovery blackmail because it does not include the litigants’ resources as a factor in the cost-shifting analysis. The BCST ensures that cost-shifting relief is granted whenever the benchmark production costs exceed the adjusted amount in controversy. Thus, plaintiffs will be unable to force wealthy defendants into settlement simply by making expensive discovery requests that exceed the cost of settling a claim. E-discovery blackmail could, however, still occur in a few situations where producing parties use legacy systems that are more costly to use than the benchmark standard. Some may argue that the BCST unfairly burdens litigants by forcing them to comply with the benchmark system standard. Such argument is unfounded because the BCST merely provides notice to litigants that e-discovery disputes will be evaluated by an impartial standard that accounts for technological change.

[¶55] The BCST eliminates e-discovery evasion because the cost of production is derived from a benchmark system, rather than the producing party’s system.184 Thus, digital documents will only remain undiscovered under the BCST when the benchmark cost of production economically prohibits litigation by exceeding the adjusted amount in controversy.185 Litigants will be unable to avoid digital document production requests by asserting that it is too costly to find and recover data on their own storage systems. Instead, courts will calculate litigants’ e-discovery costs from the BCST’s system-normalized e-discovery cost curve. Thus, the BCST eliminates the e-discovery practice of hiding digital documents on legacy systems while retaining them for private use.

[¶56] Applying the BCST also minimizes cost-shifting inequalities created by disparities in legislative treatment between public and private companies because it treats all litigants similarly when calculating e-discovery production costs. Instead of considering the actual cost of production on a producing party’s system, the BCST uses the benchmark cost of production derived from the system-normalized e-discovery cost curve. The BCST thereby avoids imposing greater costs on companies that are required by law to retain data on easily accessible storage systems.186 In addition, the BCST encourages equitable treatment of all corporate litigants without forcing private companies to retain electronic data. Therefore, it does not extend Sarbanes-Oxley to private companies, but rather levels the playing field in e-discovery disputes.

[¶57] While the BCST eliminates the most troublesome abuses of the e-discovery process, it does not alleviate all of the problems caused by applying traditional discovery rules to digital documents. For example, courts must still address the costs of finding and producing digital documents in accessible formats, a step not required in paper discovery.187 Unfortunately, no common law e-discovery test can completely remove these additional cost factors because traditional discovery rules were not created with digital documents in mind.188 The BCST does, however, attempt to account for these additional costs by acting as a common law digital document standard. Thus, the BCST reduces a notable number of flaws and ambiguities created by the various e-discovery cost-shifting tests utilized by the judiciary today.

APPENDIX: TECHNICAL DATA STORAGE AND RETRIEVAL SOLUTIONS

[¶58] Despite the prevalence of legal compliance departments in most large corporations, e-discovery issues typically go unaddressed, due to the complexity of creating and implementing uniform data management and retention policies on decentralized storage systems. Currently within most large enterprises, Electronic Case Filing Management (“ECF”), the process of implementing and modifying existing technology to ensure compliance with e-discovery laws, is largely decentralized and generally receives little legal guidance.189 ECF has two distinct components: (1) document management – receiving, transmitting, and processing of legal documents, and (2) records management – storing and archiving legal documents over time. These two components cannot be maintained by merely introducing new technology, but rather require a change from the current state of fragmented case file management policies to uniform policies implemented across an entire enterprise.190

[¶59] An ideal enterprise ECF legal solution would rely on XML architecture running Google-like services.191 This system must be designed to be reliable, highly available, fault-tolerant, and scalable while performing at acceptable levels. Currently there are three major companies in the market that deliver enterprise ECF legal solutions with these capabilities: (1) Sun/BEA Systems, (2) IBM, and (3) Microsoft.192 In addition to having their own solutions, these companies support a wide range of vendors that deliver legal solutions that are compatible with these enterprise platforms.193 IBM, for example, has 166 different solution providers that deliver a wide range of legal solutions, such as records management, case file management, billing, integration, document processing, and imaging.194 This wide range of solutions within the IBM framework guarantees that IBM can meet the needs of each individual component on a common enterprise platform. IBM and the other two previously mentioned vendors all have experience delivering these types of solutions. It is wise for small, medium, and large companies to retain an independent consultant for assistance in making an initial platform/vendor selection because each enterprise provider seeks to promote its own products. The three aforesaid vendors rely on either J2EE 195 (the Java 2 Platform, Enterprise Edition) or .NET 196 (Microsoft.NET), which supply the “plumbing” that goes into building web services, such as XML interoperability, transactions, and load balancing.197

[¶60] Any practical technical solution must work within the existing corporate infrastructure. Each of the three previously mentioned companies and their respective vendors has implemented enterprise ECF systems.198 The viability of these solutions, however, hinges upon open collaboration between technologists, attorneys, and executives in drafting a document management solution. While e-discovery law is complex and constantly evolving, it is possible to design a digital ECF enterprise system that meets both the functional and legal requirements of the enterprise.

Footnotes