C.A. No. 2022-0613-KSJM.

Re: Twitter, Inc., v. Elon R. Musk et al.,

Court of Chancery of Delaware.

August 25, 2022.

Peter J. Walsh Jr., Esquire, Edward B. Micheletti, Esquire Kevin R. Shannon, Esquire Lauren N. Rosenello, Esquire Christopher N. Kelly, Esquire Skadden, Arps, Slate, Meagher & Flom LLP Mathew A. Golden, Esquire 920 N. King Street, 7th Floor Callan R. Jackson, Esquire P.O. Box 636 Potter Anderson & Corroon LLP Wilmington, DE 19899-0636 1313 N. Market Street Hercules Plaza, 6th Floor Wilmington, DE 19801 Brad D. Sorrels, Esquire Wilson Sonsini Goodrich & Rosati, P.C. 222 Delaware Avenue, Suite 800 Wilmington, DE 19801.

Re: Twitter, Inc. v. Elon R. Musk et al., C.A. No. 2022-0613-KSJM.

Dear Counsel:

This letter resolves issues raised in the August 15, 2022 letter to the court from Twitter, Inc. ("Plaintiff"), which this decision refers to as Plaintiff's "Second Discovery Motion." The motion seeks relief in connection with discovery directed to Defendants Elon R. Musk, X Holdings I, Inc., and X Holdings II, Inc. ("Defendants") concerning Defendants' data science analysts (the "Data Scientists").1

Through this action, Plaintiff seeks to specifically enforce an Agreement and Plan of Merger (the "Merger Agreement") dated April 25, 2022, under which Defendants agreed to acquire Plaintiff.2 Section 6.4 of the Merger Agreement grants Defendants the right to information from Plaintiff (subject to certain conditions, restrictions, and exceptions) "for any reasonable business purpose related to the consummation of the transactions contemplated by this [Merger] Agreement."3 In May 2022, Defendants demanded information pursuant to Section 6.4 concerning Plaintiff's methods of calculating monetizable daily active usage or users ("mDAU").4 In response, Plaintiff provided its "firehose" data—i.e., a live feed of data concerning public accounts on Plaintiff's platform. Defendants then provided Plaintiff's firehose data to the Data Scientists, who conducted a "preliminary analysis" of that data.5 Based expressly in part on that preliminary analysis, Defendants terminated the Merger Agreement on July 8, 2022. Plaintiff filed suit, and Defendants responded with counterclaims that reference analyses performed by the Data Scientists (with the preliminary analysis, the "Analyses") at least eight times.6

Defendants have advanced a variety of positions to shield from discovery the Data Scientists' information, including the Analyses. Although they abandoned most of those positions, they maintain that the Data Scientists' information is insulated from discovery by the non-testifying expert protection and the work product doctrine. Through its Second Discovery Motion, Plaintiff seeks a declaration that no blanket protection or privilege applies to this information.

Non-Testifying Expert Protection

Court of Chancery Rule 26(b)(4)(B) restricts discovery from persons retained as non-testifying experts in anticipation of litigation.7 Rule 26(b)(4)(B) tracks the language of its federal counterpart, and this court has looked to federal authorities when applying it.8 The restriction stems from the concern that "[a]llowing routine discovery as to [non-testifying experts] would . . . deter thorough preparation of the case and reward those whose adversaries were most enterprising."9 The restriction prevents the "additional drawback" that, after learning of experts that a party retained but elected not to have testify, "the opponent might later seek to call them as witnesses to attest to views that the opponent found congenial."10 The theory is that "[i]f the price for seeking expert consultation is possibly to live or die by the unknown opinion of the expert first consulted, there will be even more reluctance to consult."11

Rule 26(b)(4)(B) does not provide unqualified immunity. A party may obtain discovery from a non-testifying expert "upon a showing of exceptional circumstances under which it is impracticable for the party seeking discovery to obtain facts or opinions on the same subject by other means."12 Rule 26(b)(3), which protects work product, contains a similar exception and permits discovery of work product "only upon a showing that the party seeking discovery has substantial need of the materials in the preparation of the party's case and that the party is unable without undue hardship to obtain the substantial equivalent of the materials by other means."13 The exception for non-testifying experts in Rule 26(b)(4)(B) has been described as "more exacting" than the exception for work product in Rule 26(b)(3).14

Rule 26(b)(4)(B) does not "shield testimony by a natural fact witness" from discovery.15 As the advisory note to the rule's federal counterpart explains, the rule does not protect "the expert whose information was not acquired in preparation for trial but rather because he was an actor or viewer with respect to transactions or occurrences that are part of the subject matter of the lawsuit. Such an expert should be treated as an ordinary witness."16 When a party invokes Rule 26(b)(4)(B) to shield discovery of a fact witness, this court is rightly skeptical.17

In this case, Defendants insist that the Data Scientists were retained by counsel in anticipation of litigation and to assist counsel in rendering legal services as non-testifying experts. The language of the Data Scientists' engagement letters provides support for this assertion, as does the timing of their retention.18

Yet, the Data Scientists were also "actor[s] or viewer[s] with respect to transactions or occurrences that are part of the subject matter of [this] lawsuit."19 Defendants expressly relied on the Data Scientists' preliminary analysis to justify making additional information requests and later terminating the Merger Agreement.20 Defendants also rely on the Analyses throughout their counterclaims.21 In these circumstances, it is clear that the Data Scientists did not act solely as non-testifying experts. Rather, at least with respect to the Analyses, they are fact witnesses.

The Data Scientists' dual-natured role forces the following question: To what degree do the protections of Rule 26(b)(4)(B) extend to persons who played a dual role as fact witnesses and non-testifying experts?

Plaintiff answers this question by citing Hexion, where this court declined to recognize a party's designation of a fact witness with knowledge about issues at the heart of a case as a non-testifying expert.22 There, the plaintiff-buyer terminated a merger agreement with the defendant-seller, claiming that the seller had suffered a material adverse effect and that the merger could not close because the buyer's financing sources could not be provided with a "satisfactory solvency opinion[.]"23 The seller purported to retain its financial advisor for the merger, Merrill Lynch, as a litigation consultant and attempted to shield all of Merrill Lynch's financial advice in connection with the merger from discovery under Rule 26(b)(4)(B).

The Hexion court rejected the seller's argument, holding that "Rule 26(b)(4)(B) does not apply to Merrill Lynch in its role as [the seller's] financial advisor because in that role Merrill Lynch has acquired and continues to acquire information as an `actor' in or `viewer' of the [merger] transaction, which lies at the heart of this lawsuit."24 Furthermore, because Merrill Lynch did not clearly distinguish between work it performed as a financial advisor and as an ostensible litigation consultant, the court declined to recognize Merrill Lynch as a non-testifying expert for any purpose whatsoever.

The logic of Hexion favors Plaintiff. Rule 26(b)(4)(B) does not protect the Analyses that lie at the heart of this lawsuit. Because Rule 26(b)(4)(B) does not protect the Analyses, it does not protect documents or communications relating to those Analyses. And if Defendants failed to distinguish between the Data Scientists' work performed for those ends and any work performed in an expert capacity, Rule 26(b)(4)(B) does not protect any of it.

Plaintiff argues for the same result under the "exceptional circumstances" standard. As to the Analyses, Plaintiff has demonstrated "exceptional circumstances under which it is impracticable for the party seeking discovery to obtain facts or opinions on the same subject by other means."25 Defendants relied on the Analyses as a basis for termination and in support of their counterclaims. To probe the veracity of Defendants' assertions, Plaintiff requires the Analyses. Plaintiff cannot obtain those facts through any other means.

Invoking case law interpreting the work product doctrine, Defendants argue that "dual-purpose" documents generated by an alleged expert should be protected, even though the materials also had a nonlitigation purpose, if the documents were prepared for litigation.26 Defendants urge the court to apply the "because of" test rather than following Hexion. Under the "because of" test, the court asks in light of "the totality of the circumstances" why the document was prepared.27 "If a document was generated `because of litigation,' then it is likely privileged," but "[i]f the document was created for some other reason, such as a business purpose, then it is likely not protected."28

Defendants cite to cases declining to compel the production of documents under the "because of" dual-purpose test.29 Each of the dual-purpose cases on which Defendants rely analyzes the dual-purpose issue with some reference to the substantial-need standard of Rule 26(b)(3).30 As discussed above, the substantial-need standard is less exacting and easier to meet than the exceptional-circumstances standard. It follows that if a court denied protection under the lower standard, then the court also should deny protection under the higher standard.

Defendants' reliance on Rohm & Haas illustrates why their reliance on the because-of case law is misplaced. The discovery dispute there concerned a dynamic financial model that the defendant used for corporation decision-making. The plaintiff argued that the model was critical to its ability to understand the defendant's representations to banks, rating agencies, and its board. The defendant produced multiple versions of the model but claimed that a recently updated version was protected work product because it was created at the request of the defendant's head of litigation for "litigation support and settlement analysis."31 The court agreed with the defendant, observing that it was "hard pressed to think of any information that warrants greater protection under attorney work product doctrine than potential settlement strategies prepared at the direction of counsel."32 Applying the because-of test, the court observed that "the litigation purpose of the [model] sufficiently permeates the business purpose of the model to warrant work product protection."33 The court then factored that holding into the governing substantial-need standard and concluded that the plaintiff failed to meet its burden.

Defendants parrot the "permeates the business purpose" quote of Rohm & Haas in support of their position, claiming here that the litigation purpose of the Data Scientists' Analyses sufficiently permeates their business purpose as to render them immune from discovery. The court accepts Defendants' representation to this effect. Even so, that is not sufficient to shield the Analyses from discovery. Unlike in Rohm, Plaintiff would win under a work product analysis because it can demonstrate substantial need for the Analyses. The plaintiff in Rohm & Haas could not demonstrate substantial need because the plaintiff already had access to multiple versions of the model at issue, and the model did not form the basis for any allegations at the heart of the case. Here, Plaintiff does not have access to any versions of the Analyses, which are central to various aspects of this case.

Plaintiff's ability to demonstrate substantial need for the Analyses renders Defendants' because-of cases inapposite on the application of Rule 26(b)(4)(B). Plaintiff has shown the exceptional circumstances necessary to warrant production of Analyses, even if they would otherwise be protected under Rule 26(b)(4)(B).

Work Product Protection

In addition to using work product cases to bolster their invocation of the protection for non-testifying experts, Defendants invoke the work product doctrine independently. To that end, they contend that any otherwise-discoverable material in the Data Scientists' possession was generated at least in part "in anticipation of litigation" and is thus shielded by Rule 26(b)(3).

Rule 26(b)(3) generally protects "documents, electronically stored information, and tangible things" that have been "prepared in anticipation of litigation or for trial by or for another party or by or for that other party's representative (including the other party's attorney, consultant, surety, indemnitor, insurer, or agent)."34

Rule 26(b)(3) does not provide an unqualified protection. As discussed above, the rule permits discovery of work product "upon a showing that the party seeking discovery has substantial need of the materials in the preparation of the party's case and that the party is unable without undue hardship to obtain the substantial equivalent of the materials by other means."35 This decision has already explained that Plaintiff meets the substantial-need test as to the Analyses.

Equally important, Rule 26(b)(3) only protects the work product of the attorney. "At its core, the work-product doctrine shelters the mental processes of the attorney."36 The doctrine does not shelter the mental processes of the expert, which are subject to a separate legal analysis and framework discussed above.37 A non-testifying expert's information "is not itself work product."38 Thus, to the extent Defendants object to production of all the Data Scientists' documents on the grounds that they categorically constitute protected work product, that objection is overruled.

That said, "as a collaborator in the development of pretrial strategy, a non-testifying expert may become a unique repository of insights into counsel's opinion work product[.]"39 And while Plaintiff has demonstrated exceptional circumstances and thus a substantial need for the Analyses, it is possible that communications with counsel concerning the Analyses reflect work product. It is difficult to assess that possibility at this stage because the documents at issue have not been logged. In any event, to the extent that Plaintiff seeks a declaration that Defendants may not assert work product protection over any of the Data Scientists' documents, that request is denied.


Summing it up, to the extent that Defendants have lodged a blanket objection to producing all of the Data Scientists' documents as non-testimony expert materials or work product, that objection is overruled. At a minimum, Defendants must produce the Analyses.

That leaves documents and communications and drafts concerning the Analyses. Rule 26(b)(4)(B) does not apply to those materials. To the extent Defendants claim that any document contains work product, Defendants must identify that document on a privilege log. At that point, Plaintiff can seek production of specific documents.40 Plaintiff's Second Discovery Motion is therefore granted in part.

This decision also has not reached the thorny question of whether Defendants may assert privilege over the Data Scientists' documents and communications that post-date the Analyses. To be clear, this decision neither endorses nor rejects any arguments concerning that discovery. Given the dual role of the Data Scientists, however, Defendants should log any such communications withheld on privilege grounds.



1. C.A. No. 2022-0613-KSJM, Docket ("Dkt.") 160 ("Pl.'s Second Disc. Mot."); see also Dkt. 168 ("Defs.' Response to Pl.'s Second Disc. Mot."); Dkt. 196 (Pl.'s Reply to Pl.'s Second Disc. Mot.).
2. See generally Dkt. 1 ("Compl.").
3. Compl. Ex. 1 (Merger Agreement) § 6.4.
4. Pl.'s Second Disc. Mot. at 4.
5. Id. Ex. A at 6.
6. Dkt. 42, Countercls. ¶¶ 13 ("[P]reliminary expert estimates of the false or spam accounts in Twitter's mDAU population, based on the data Twitter has provided and using a publicly available machine learning algorithm, yield findings that are shocking," including that "one-third of visible accounts may have been false or spam accounts[.]"), 17 ("The Musk Parties' preliminary analysis shed light as to why Twitter has stonewalled—Twitter did not want the Musk Parties (or the market) to discover that Twitter has been misleading investors regarding its `key metric.'"), 116 ("[C]ontrary to Twitter's representations that its business was minimally affected by false or spam accounts, the Musk Parties' preliminary estimates show otherwise."), 117 ("An analysis of Firehose data from the first week of July, including processing visible accounts using a publicly-available machine-learning spam detection model, shows that, during that timeframe, false or spam accounts accounted for 33% of visible accounts."), 118 ("[E]ven assuming that every single one of the invisible accounts is a legitimate user, and not a false or spam account (an assumption as conservative as mathematically possible), these preliminary findings indicate a floor for the prevalence of false or spam accounts among Twitter's mDAU of 10%, rendering Twitter's statements that less than 5% of mDAU is comprised of false or spam accounts materially misleading.") (emphasis in original), 119 ("[C]ontrary to the implication in the 2021 10-K that fewer than 10.85 million mDAU were false or spam accounts, preliminary findings suggest that more than 20 million mDAU were false or spam accounts."), 120 ("Not only does preliminary analysis reveal that Twitter's false or spam accounts exceed 10% of mDAU, the Musk Parties estimate that false and spam accounts make up an even more significant portion of the mDAU that actually see ads based on Twitter's own data regarding ad engagement among its userbase."), 150 ("In reality, as discussed above, preliminary estimates based on only the 30% of mDAU visible in the Twitter Firehose already indicate that one third of visible accounts and 10% of the mDAU count may be made up of false or spam accounts.").
7. Ct. Ch. R. 26(b)(4)(B).
8. See, e.g., Hexion Specialty Chems., Inc. v. Huntsman Corp., 959 A.2d 47, 50-51 (Del. Ch. 2008); In re Speedway Motorsports, Inc., 2001 WL 818169, at *1 (Del. Ch. July 10, 2001); see also State v. Figg Bridge Eng'rs., Inc., 2013 WL 5293549, at *4-5 (Del. Super. Sept. 19, 2013); Pfizer v. Advanced Monbloc Corp., 1999 WL 743868, at *5 (Del. Super. Sept. 20, 1999); Winchester v. Hertrich, 658 A.2d 1016, 1020-21 (Del. Super. 1995); Harris v. Parmi Tool Co., Inc., 1990 WL 9495, at *4 (Del. Super. Jan. 3, 1990).
9. 8A Charles A. Wright & Arthur R. Miller, Fed. Prac. & Proc. § 2032 at 94 (3d ed.) [hereinafter "Wright & Miller"].
10. Id. at 97.
11. Id. at 100.
12. Ct. Ch. R. 26(b)(4)(B).
13. Ct. Ch. R. 26(b)(3).
14. People ex rel. Wheeler v. S. P. Transp. Co., 1993 WL 816066, at *8 (E.D. Cal. Sept. 2, 1993) ("[S]ince the exceptional circumstances condition to preclude application of subdivision (b)(4)(B) of Rule 26 is more exacting than the substantial need requirement found in subdivision (b)(3), the court will analyze the exception under this more stringent standard."); see also MeadWestvaco Corp. v. Rexam, PLC, 2011 WL 2938456, at *6 (E.D. Va. July 18, 2011) ("Plaintiff cannot meet the standard of `exceptional circumstances' because it cannot even meet the lower threshold of `substantial need.'"); U.S. v. 22.80 Acres of Land, 107 F.R.D. 20, 23 (N.D. Cal. 1985) (describing the exceptional-circumstances standard as "more demanding" than the substantial-need standard).
15. Hexion, 959 A.2d at 51.
16. Fed. R. Civ. P. 26(b)(4)(B) advisory committee's note (1970) (quoted in Hexion, 959 A.2d at 50).
17. See Hexion, 959 A.2d at 51.
18. See Defs.' Response to Pl.'s Second Disc. Mot. Ex. A at 1 (retaining one of the Data Scientists' companies "to serve as an outside expert and to support Skadden's provision of legal advice to Client," to "work at Skadden's direction and report directly to Skadden," and to provide "services of a character and quality which would necessarily be adjunct to Skadden's services as attorneys"); id. Ex. B at 1 (same); id. Ex. C. at 1 (same).
19. Fed. R. Civ. P. 26(b)(4)(B) advisory committee's note (1970) (quoted in Hexion, 959 A.2d at 50).
20. Pl.'s Second Disc. Mot. Ex. M at 2.
21. Dkt. 42, Countercls. ¶¶ 13, 17, 116-20, 150.
22. Hexion, 959 A.2d at 52.
23. Id. at 49.
24. Id. at 50.
25. Ct. Ch. R. 26(b)(4)(B).
26. See Defs.' Response to Pl.'s Second Disc. Mot. at 24 (referring back to Defendants' work-product argument); see also 8 Wright & Miller § 2024 at 512-14 (discussing the dual-purposes scenario).
27. Pfizer, 1999 WL 743868, at *5.
28. JPMorgan Chase & Co. v. Am. Century Cos., Inc., 2013 WL 1668393, at *3 (Del. Ch. Apr. 18, 2013).
29. Defs.' Response to Pl.'s Second Disc. Mot. at 19 (citing Rohm & Haas Co. v. Dow Chem. Co., 2009 WL 537195 (Del. Ch. Feb. 26, 2009); In re Grand Jury Subpoena (Mark Torf/Torf Env't Mgmt.), 357 F.3d 900 (9th Cir. 2004) ("Torf"); U.S. v. Adlman, 134 F.3d 1194 (2d Cir. 1998); U.S. Inspection Servs., Inc. v. NL Engineered Solutions, LLC, 268 F.R.D. 614 (N.D. Cal. 2010); and Hollinger Int'l Inc. v. Hollinger Inc., 230 F.R.D. 508 (N.D. Ill. 2005)).
30. See Rohm & Haas, 2009 WL 537195, at *2 (holding that "[g]iven the information to which it already has access, Rohm and Haas has not demonstrated a substantial need for the Litigation Support Model"); Adlman, 134 F.3d at 1204 (rejecting the argument that "[t]he IRS contends that even if the Memorandum qualifies as work product, it has made a sufficient showing of substantial need and unavailability so as to overcome the qualified protection accorded by Rule 26(b)(3)"); U.S. Inspection Servs., 268 F.R.D. at 626 (holding that "Fulks-Graham has not contended and cannot demonstrate that a substantial need exists for this document"); see also Torf, 357 F.3d at 904 (declining "to consider the government's contention made in its brief filed with this court, but not presented to the district court, that it has a substantial need for the withheld documents and that it would incur undue hardship in obtaining substantially equivalent information"); Hollinger, 230 F.R.D. at 511 n.5 (observing that "[d]efendants have not argued that they have a substantial need for the undisclosed material").
31. Rohm & Haas, 2009 WL 537195, at *1.
32. Id.
33. Id. at *2.
34. Ct. Ct. R. 26(b)(3).
35. Id.
36. Tackett v. State Farm Fire & Cas. Ins. Co., 653 A.2d 254, 261 (Del. 1995).
37. See 8A Wright & Miller § 2029 at 17-18 (explaining that "[t]he knowledge of an expert is not privileged" and "is not part of the work product"); see also Fed. R. Civ. P. 26(b)(4)(B) advisory committee's note (1970) (noting that the provisions of subdivision (b)(4) "repudiate the few decisions that have held an expert's information privileged simply because of his status as an expert" and "reject as ill-considered the decisions which have sought to bring expert information within the work-product doctrine").
38. 8A Wright & Miller § 2032 at 96.
39. Id.
40. As a word of caution, Defendants are reminded that a large volume of logged entries can raise a red flag for the court. See, e.g., Mechel Bluestone, Inc. v. James C. Justice Cos., Inc., 2014 WL 7011195, at *6 (Del. Ch. Dec. 12, 2014).


1000 Characters Remaining reserves the right to edit or remove comments but is under no obligation to do so, or to explain individual moderation decisions.

User Comments

Listed below are the cases that are cited in this Featured Case. Click the citation to see the full text of the cited case. Citations are also linked in the body of the Featured Case.

Cited Cases

  • No Cases Found

Listed below are those cases in which this Featured Case is cited. Click on the case name to see the full text of the citing case.

Citing Cases