Not Reasonably Accessible ESI

Not Reasonably Accessible ESI

Rule 26(b)(2)(B) gives electronically stored information a limit of its own: a party need not provide discovery of ESI from sources it identifies as not reasonably accessible because of undue burden or cost.¹ The rule then runs in three moves. The responding party identifies the source. If the requester moves to compel, or the responder moves for a protective order, the responder must show that the source is not reasonably accessible because of undue burden or cost. If the responder makes that showing, the court may still order discovery from the source when the requesting party shows good cause, considering the limitations of Rule 26(b)(2)(C), and the court may set conditions for the discovery.²

This limit sits beside proportionality, which since 2015 has been part of the definition of scope in Rule 26(b)(1).³ The two exist side by side because electronic sources differ so widely in what they cost to use. Active data on a hard drive is relatively easy to reach, while archival backup tapes may need time-consuming and costly restoration before anyone can review them. When retrieval and review climb into the tens of thousands of dollars or beyond, it is fair to ask whether the expense is proportional to the need for the information and its potential value.⁴ The 2006 amendment answered by creating, in effect, a per se category of unduly burdensome or expensive discovery: ESI from sources determined to be not reasonably accessible. A court may allow discovery from that category only on a showing of good cause, with restrictions that guard against undue burden or expense. That structure opens a second way to control the cost of e-discovery, cost shifting or cost sharing, which the next section takes up.⁵

Two cautions about the case law apply to everything below. The accessibility vocabulary predates the rule: in Zubulake IV, one of the pioneering e-discovery decisions, the court drew the line between accessible and inaccessible ESI, and the rulemakers adopted that distinction in 2006.⁶ And the 2015 amendments dropped the phrase “reasonably calculated to lead to the discovery of admissible evidence” from the scope provision, so older opinions may well come out differently under the current rules.⁷ Take the test from the rule, and read the older opinions for their reasoning about cost and burden. The three subsections follow the rule's order: what makes a source inaccessible, how the responding party identifies and proves it, and what the requester must show to compel discovery anyway.

What Counts as Inaccessible

The rule's label has two parts that must travel together: a source, meaning a particular system, medium, or repository of data, and a cause, which is undue burden or cost.⁸ The rule lists no qualifying media. The sorting scheme comes from the Zubulake decisions, whose distinction between accessible and inaccessible ESI the rulemakers adopted in 2006.⁹

For electronic data, the starting point is that search engines make any data retained in machine-readable form typically accessible.¹⁰ Accessibility therefore turns largely on the media where the data sits, and the literature on electronic storage describes five categories, ranked from most to least accessible:¹¹

1. Active, online data, such as a hard drive, where files are created, received, and processed and where access is very fast, measured in milliseconds.¹²
2. Near-line data, such as optical disks in a robotic library that locates and reads the media automatically, with access times running from milliseconds up to 10 to 30 seconds for optical disk and 20 to 120 seconds for sequentially searched media such as magnetic tape.¹³
3. Offline storage and archives, meaning removable optical disks or magnetic tape kept on a shelf or rack, traditionally as disaster copies or archival records unlikely to be retrieved. Reaching them takes manual intervention and may take minutes, hours, or even days.¹⁴
4. Backup tapes, which are sequential-access devices: reading any block of data means reading every block before it, and the data is organized to mirror the computer's structure, not to retrieve individual files.¹⁵
5. Erased or fragmented data, which must be reconstructed or de-fragmented before anyone can use it.¹⁶

The first three categories are typically treated as accessible and the last two as inaccessible.¹⁷ The line falls at usability. Accessible data sits in a readily usable format and needs no restoration or other manipulation, even when reaching it takes anywhere from milliseconds to days. Inaccessible data must first be restored, de-fragmented, or reconstructed.¹⁸ That is why an offline archive that takes days to retrieve still ranks above a backup tape: once the archive's media arrive, their contents are ready to use, while a tape's contents must still be restored.

The Zubulake court applied the line to its own defendant. UBS's active mail files were online, and its e-mails on optical disks had to be located and read with the correct hardware but could be searched simply and automatically once found. For both sources, the court said, even considering cost shifting would be wholly inappropriate. UBS kept the data in an accessible, usable format and could respond cheaply and quickly, so, as with most discovery requests, the producing party bears the cost.¹⁹

Backup tapes are the standard inaccessible source because of how they are built. A tape is read sequentially and arranged to mirror the computer's structure and not the structure of human record keeping, so it is not organized for retrieving individual documents.²⁰ Tapes also typically use data compression, which lets more data fit on each tape but makes restoration more time-consuming and expensive, especially since no uniform standard governs how the compression is done.²¹ Purpose matters alongside medium. Zubulake IV treated backup tapes held solely for disaster recovery as inaccessible and tapes accessed in the regular course of business as accessible.²²

The label also shapes preservation. Zubulake IV held that a party need not suspend its normal overwriting of disaster-recovery tapes even after the duty to preserve is triggered, unless it knows or should know that the tapes hold potentially relevant data that duplicates nothing on accessible sources.²³ A later court restated the general rule in similar terms: a litigation hold does not ordinarily reach inaccessible backup tapes, which may keep being recycled on the schedule in the company's policy.²⁴

The burden of restoring tapes can be counted. In Cache La Poudre Feeds LLC v. Land O'Lakes, Inc., the court recounted that finding a document containing a particular word on about 400 backup tapes from the last four years meant restoring each tape first and then searching it. Restoration took roughly eight hours per tape, in serial fashion because only a limited number of restoration servers existed, so restoring the tapes alone would take over 3,200 hours.²⁵ The court added that, as of December 2006, a responding party must identify but need not produce ESI that is not reasonably accessible because of undue burden or cost, and that backup tapes containing archived data might be one such source.²⁶ The passage comes from a discussion of what the producing party had to preserve, so it supplies the arithmetic of a tape burden and does not decide a motion to compel.

The Zubulake categories predate the rule, and the rule asks a different question: whether burden or cost is undue. The medium is a strong clue, and the evidence decides. Because “undue” is relative, a cost is measured against what the discovery is worth. In Best Buy Stores, L.P. v. Developers Diversified Realty Corp., a magistrate judge found the Odom database reasonably accessible at a cost of $124,000, reasoning that the figure was reasonable given potential contract damages above $800,000, potential enhanced damages on fiduciary-duty and fraud claims, and the potential long-term economic impact of the litigation on all parties.²⁷ That reasoning shows how the stakes of a case enter the question whether a cost is reasonable. The defendants, for their part, argued that the database's accessibility should be determined from a particular point in time.²⁸ Both points show that the label is argued case by case and is not read off the medium.

Courts accept a showing that the cost or burden is excessive in relation to the value of the discovery. In one 2017 decision, a declaration established that extracting the requested data would take six months, cost at least $120,000, and require new software before the data was usable, and that the work would divert staff from other duties. The court weighed that burden against the limited relevance of the data as requested, found it not proportional to the needs of the case at that time, and found the burden undue.²⁹ The tempting exam shortcut is to answer that data on backup tapes is inaccessible because it sits on backup tapes. That answer skips the proof the rule demands: the medium points toward the conclusion, and the evidence of burden or cost reaches it.

Identifying and Proving Inaccessibility

The rule fixes who acts first and who must prove what. A party need not provide discovery from sources it identifies as not reasonably accessible, and when the requester moves to compel or the responder moves for a protective order, the party from whom discovery is sought must show that the source is not reasonably accessible because of undue burden or cost.³¹ One court called this a two-tiered approach. The producing party first declines to search the source it has identified. If the requester then moves to compel, the court weighs the bona fides of the producing party's representations together with the considerations in Rule 26(b)(2)(C), such as whether the discovery is unreasonably cumulative or duplicative and whether its burden or expense outweighs its likely benefit.³²

The first tier belongs to the responding party. Rule 26(b)(2)(B) lets it withhold relevant, responsive ESI held in not reasonably accessible formats on its own judgment about accessibility.³³ What it must do is identify the source, since a responding party must identify but need not produce such ESI.³⁴ Identification gives the requester a choice: accept the gap, negotiate a narrower search, or move to compel. Either side can bring the claim before the court, because the responder may move for a protective order, which requires a good-faith effort to resolve the dispute with the requester and then a showing of good cause.³⁵ Once the requester moves to compel, the responding party must demonstrate inaccessibility based on undue burden and cost.³⁶ The burden stays with the responder because the facts about hours, systems, and dollars sit in its own files, and a court asked to accept a conclusion with nothing behind it would be guessing.

The proof at the third step is evidence. A party objecting that a request is unduly burdensome must submit affidavits or other evidence to substantiate the objection.³⁷ A bare assertion that requested discovery is unduly burdensome is ordinarily insufficient to bar production, and one court rejected such an objection where the objecting party gave no estimate of the hours or money compliance would take.³⁸ Another court called a standalone claim of “undue burden” wholly inadequate, citing authority that the objecting party must particularize its objection because generalized assertions are inadequate.³⁹ The reason is practical: a court cannot weigh numbers it has not been given.

A showing that does the job names the sources, the work per source, and the total. The Cache La Poudre record has that shape: about 400 backup tapes, each restored and then searched, roughly eight hours per tape in serial order because restoration servers were limited, and more than 3,200 hours in all.⁴⁰ Each figure is a fact the court can test and the requester can dispute. The court recited those numbers in a dispute over preservation, so they show what such proof contains and do not record a ruling on a motion to compel.

Aguilar v. Immigration and Customs Enforcement shows the procedure working for the responding party. The defendants made a persuasive showing at a conference that restoring and reviewing the backup tapes generated by their decentralized e-mail system would be extremely burdensome. The court then weighed the Rule 26(b)(2)(C) considerations and noted that the plaintiffs had not shown a likelihood of recovering important information not previously disclosed. The cost was unquestionably high and the likely benefit low, so the defendants did not have to review and produce any data regarding e-mails on the tapes.⁴¹ The opinion quotes the cost-benefit limit in its pre-2015 form. That limit has since moved: proportionality now sits in Rule 26(b)(1), and the limit in Rule 26(b)(2)(C) now reaches discovery outside the scope that Rule 26(b)(1) permits.⁴² The order of steps in Aguilar is still the rule's order: the responder proves the burden first, and the requester's showing of likely benefit comes second.

A claim of burden can also fail because the party's own conduct contradicts it. In LaBrier, the producing party said it could not search for payment data with data inquiries because the data sat in a picture format reachable only through a system called ECS, and that calculating a labor deduction for any claim would mean downloading estimates one at a time and recalculating each manually. Yet its own witness had used estimating software at the start of the lawsuit to identify the putative class members and to estimate the damages alleged. The court found incredible the suggestion that no cost-effective way existed to match one database against another, observed that data sorting is what computers do in very short amounts of time, and held that even claim-by-claim matching did not justify preventing access to critical information.⁴³

Equal Employment Opportunity Commission v. SVT, LLC shows how an early statement frames a later cost dispute. Before serving its requests, the Commission told SVT in writing how it wanted ESI produced: documents in near-native TIFF format, and spreadsheets and databases in native format. SVT initially communicated that it could comply, and the Commission's requests then sought applications, screening assessments, interview guides, spreadsheets about applicants and employees, and pay rate information. The parties did not dispute that this ESI was relevant.⁴⁴ With relevance conceded, any later claim that the requested format is too costly has to be reconciled with SVT's first answer.

The burden to proceed with care runs to the requester and its counsel as well. An objection to discovery should come only after a lawyer has paused and considered whether, on a reasonable inquiry, there is a factual basis for it, and boilerplate objections that a request is overbroad and unduly burdensome are improper unless based on particularized facts.⁴⁵ The same rule makes a lawyer's signature on a discovery request a certification that the request is neither unreasonable nor unduly burdensome or expensive, considering the needs of the case, prior discovery, the amount in controversy, and the importance of the issues.⁴⁶ In the Mazda litigation, a finding that Mazda violated Rule 26(g) stood, yet the trial court's sanction was wildly excessive given the plaintiffs' overbroad discovery requests and the trial court's inaction.⁴⁷ Later sections take up objections and certification in detail.

Good Cause to Compel Anyway

Rule 26(b)(2)(B) does not end with the responding party's proof of inaccessibility. Once that proof is in, the court “may nonetheless order discovery from such sources if the requesting party shows good cause, considering the limitations of Rule 26(b)(2)(C),” and it “may specify conditions for the discovery.”⁴⁹ The burden changes hands at that point. The responder carried the first showing by proving that undue burden or cost put the source out of reach, and the requester now carries the second by showing that its need justifies reaching into the source anyway.⁵⁰

The two showings answer different questions. The first asks what it costs to reach the source, and the facts sit in the responding party's own systems. The second asks whether what the source probably holds is worth that cost, and the requesting party is the one who knows what it is trying to prove. If inaccessibility ended the inquiry, important evidence could drop out of a case merely because it sat on awkward media. If the label counted for nothing, a requester could demand costly restoration out of habit and leave the responder with the bill. Placing the second burden on the requester, and tying it to Rule 26(b)(2)(C), keeps the inquiry on the same cost-and-benefit concerns that govern every other discovery request.

The Rule 26(b)(2)(C) limitations are the mandatory limits discussed earlier in this chapter. The court asks whether the discovery sought is unreasonably cumulative or duplicative, or obtainable from a source that is more convenient, less burdensome, or less expensive;⁵¹ whether the requester has had ample opportunity to obtain the information by discovery in the action;⁵² and whether the proposed discovery is outside the scope permitted by Rule 26(b)(1).⁵³ The court in Best Buy described the cross-reference as a balancing of costs and potential benefits that applies to all discovery requests.⁵⁴ That opinion predates the 2015 amendments. It quotes the third limit as itself weighing the burden or expense of the proposed discovery against its likely benefit, taking into account the needs of the case, the amount in controversy, the parties' resources, the importance of the issues, and the importance of the proposed discovery.⁵⁵ Proportionality now sits in Rule 26(b)(1), and the third limit points there, so the court performs the same balance under a different address.⁵⁶

The rule does not define good cause. The Advisory Committee's notes supply seven factors for the inquiry, which Best Buy sets out:⁵⁷

• the specificity of the discovery request;
• the quantity of information available from other and more easily accessed sources;
• the failure to produce relevant information that seems likely to have existed but is no longer available on more easily accessed sources;
• the likelihood of finding relevant, responsive information that cannot be obtained from other, more easily accessed sources;
• predictions as to the importance and usefulness of the further information;
• the importance of the issues at stake in the litigation; and
• the parties' resources.

The seven divide into two groups that ask different things. The first four ask whether the requester needs this particular source, and they turn on substitutes: if the same facts sit in an active file or on paper, restoring the inaccessible source buys nothing. The third is the easiest to see in practice, since a gap in the accessible record where relevant material should be suggests that the material survives, if at all, in the harder source. The last three ask whether the information is worth the effort, and they echo the proportionality considerations in Rule 26(b)(1). A requester who argues only relevance has therefore not yet addressed good cause, because relevance is what carried the request through the scope definition before anyone raised inaccessibility.

Zubulake I sits at the favorable end. It came before Rule 26(b)(2)(B), so the court spoke of cost shifting and not of good cause, but the same facts would likely satisfy the good-cause inquiry today.⁵⁸ The plaintiff had found one e-mail urging that she be fired “ASAP” after her EEOC complaint, to deprive her of year-end bonuses, and she had retained 450 relevant e-mails from work, which suggested strongly that more such messages sat on the backup tapes.⁵⁹ Those facts line up with the factors on likelihood and substitutes: relevant e-mail seemed likely to exist and was unlikely to be found elsewhere. The court did not demand that she first prove what the tapes held. It ordered UBS to produce, at its own expense, responsive e-mails from any five backup tapes she selected, and to file an affidavit on the results and on the time and money spent, after which the court would conduct the cost-shifting analysis.⁶⁰ A sample of that kind replaces guesswork with tangible evidence of what the tapes hold and what restoring them costs.⁶¹

Aguilar sits at the other end. The defendants made a persuasive showing that restoring and reviewing the tapes would be extremely burdensome, and the plaintiffs had not shown a likelihood of recovering important information not previously disclosed. With cost unquestionably high and likely benefit low, the court did not require review and production of any e-mail data from the tapes.⁶² The opinion framed the question through the Rule 26(b)(2)(C) balance and did not list the seven factors by name, though its focus on new and important information tracks the fourth and fifth. A requester can lose the same way on the need side: if the information it wants sits in an easier source, it has shown no good cause for the harder one.

A finding of good cause does not require the court to order everything or nothing, because the rule lets it specify conditions for the discovery.⁶³ The sample ordered in Zubulake I is one such condition. The court may also allocate the expense of discovery, since Rule 26(c)(1)(B) authorizes an order specifying terms for disclosure or discovery, including the allocation of expenses.⁶⁴ The next sections take up those tools. For now the point is that good cause opens the door to an inaccessible source, and the court's conditions decide how far through the door the requester goes and who pays for the trip. The usual exam error is to treat the label “not reasonably accessible” as the end of the analysis, when the label only shifts the burden to the requester to show good cause.