Twitter, Inc. v. Elon R. Musk

COURT OF CHANCERY OF THE STATE OF DELAWARE KATHALEEN ST. JUDE MCCORMICK LEONARD L. WILLIAMS JUSTICE CENTER CHANCELLOR 500 N. KING STREET, SUITE 11400 WILMINGTON, DELAWARE 19801-3734 September 22, 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 decision addresses the Motion to Compel Discovery filed by Defendants Elon R. Musk, X Holdings I, Inc., and X Holdings II, Inc. (“Defendants”) on September 7, 2022, which I refer to as Defendants’ “Fifth Discovery Motion.” 1 Plaintiff Twitter, Inc. (“Plaintiff”) filed its opposition on September 14, 2022. 2 Throughout this letter, I identify specific questions regarding the parties’ respective positions to be addressed during the September 27, 2022 hearing. 1 C.A. No. 2022-0613-KSJM Docket (“Dkt.”) 422 (“Defs.’ Fifth Disc. Mot.”). 2 Dkt. 499 (“Opposition”). C.A. No. 2022-0613-KSJM September 22, 2022 Page 2 of 8 Plaintiff has represented that aspects of Defendants’ Fifth Discovery Motion have been rendered moot. 3 Defendants shall state their position on this representation during oral argument. According to Plaintiff, only the following disputes related to Defendants’ Fifth Discovery Motion remain. Defendants request that I compel Plaintiff to: 1. Produce or log documents containing the terms mDAU, UAM, or stickiness; 4 2. Collect and review 520,000 webpages from an internal policy repository called Confluence; 3. Collect and review custodial emails and Slack threads of Twitter employee Luke Simon; and 4. Collect and review documents stored on Box and Notion. 5 As to the first category, Defendants argue that Plaintiff has failed to produce responsive documents concerning mDAU, UAM, and stickiness. Defendants base this assessment primarily on the following statistics: Plaintiff’s hit counts reflect that its custodians collectively possessed 17,527 documents containing the search term mDAU, but Plaintiff deemed approximately 40% of those documents responsive; 6 Plaintiff’s hit counts reflect that its custodians collectively possessed 5,679 documents containing the 3 Id. at 2 n.1. 4 A note regarding the terms: “mDAU” means monetizable daily active users; “UAM” means user active minutes; and “stickiness” is a metric used by Plaintiff to calculate the number of either DAU or mDAU relative to the corresponding Monthly Active Users (“MAU”) equivalent. 5 Id. at 2. 6 Defs.’ Fifth Disc. Mot. at 2–3. C.A. No. 2022-0613-KSJM September 22, 2022 Page 3 of 8 search term UAM, but Plaintiff deemed approximately 36% of those documents responsive; 7 and Plaintiff has deemed only 20% of documents containing the term stickiness responsive 8 Defendants argue that these responsiveness rates are “highly implausible” and suggest that Plaintiff inappropriately withheld responsive documents containing these terms. 9 As further support, Defendants point to a few instances where Plaintiff redacted portions of a Slack thread concerning UAM and mDAU comparisons. 10 Plaintiff makes several points in response. Plaintiff notes that: Plaintiff did not agree to produce every document containing the search terms at issue; the statistics noted above are slightly off (e.g., Plaintiff produced or logged as privileged 52% of the documents that contained the term mDAU, not 40%); it has re-reviewed the UAM documents; some errant redactions are unsurprising given the pace of this litigation; and it has cured the errant redactions in any event. 11 Plaintiff disputes that “stickiness” was ever identified as a key metric subject to the court’s order. 12 Plaintiff also notified Defendants of its intent to make another production concerning these documents on or about September 3. 13 7 Id. at 8; Opposition at 6. 8 Defs.’ Fifth Disc. Mot. at 6; Opposition at 6. 9 Defs.’ Fifth Disc. Mot. at 4. 10 See id. at 4–6. 11 See generally Opposition at 1–7. 12 Id. at 6. 13 Dkt. 436 (Ex. B to Defs.’ Fifth Disc. Mot.) at 2. C.A. No. 2022-0613-KSJM September 22, 2022 Page 4 of 8 It is unclear whether Plaintiff’s documents production cured the putative deficiency identified in the first category. 14 The parties shall report to the court on the status of this issue during the September 27 hearing. As to the second category, Defendants state that they asked Plaintiff for a list of potential sources of relevant documents at a July 31 meet and confer, including cloud- based platforms such as Google Drive. 15 Plaintiff did not identify Confluence in response to that request. During an August 31 deposition, an employee of Plaintiff identified Confluence, which the employee described as an internal program functioning as a Wikipedia of sorts for technical documents. 16 Defendants asked Plaintiff to search Confluence for responsive documents, and Plaintiff agreed to do so, but only to produce “pages ‘associated with Twitter’s trust and safety function.’” 17 Another employee of Plaintiff, deposed on September 2, then testified that Plaintiff used Confluence for multiple purposes, including to support statements made to Defendants regarding Plaintiff’s spam auditing methodology. 18 Plaintiff then agreed (after Defendants filed this motion) to also produce Confluence pages “related to the quarterly spam estimation process” within a certain group of employees. 19 Through their Fifth Discovery Motion, Defendants state that 14 Defs.’ Fifth Disc. Mot. at 7. 15 Id. at 13. 16 Dkt. 436 (Ex. F to Defs.’ Fifth Disc. Mot.) at 150:20–153:24. 17 Defs.’ Fifth Disc. Mot. at 13 (quoting Ex. I at 2). 18 Dkt. 436 (Ex. J to Defs.’ Fifth Disc. Mot.) at 160:25–167:12. 19 Opposition at 7 (quoting Ex. F. at 2). C.A. No. 2022-0613-KSJM September 22, 2022 Page 5 of 8 they “would be willing to accept the limited production of Confluence pages from Twitter’s data science team(s) and Twitter’s corporate finance and/or sales finance team” hitting on certain search terms. 20 Plaintiff responds by attacking Defendants’ generalized justification for the documents, stating that “Defendants have not identified any reason to believe that other relevant, non-duplicative documents exist on Confluence.” 21 But Defendants have pointed to deposition testimony suggesting that Plaintiff’s use of Confluence was, perhaps, broader than Plaintiff had initially represented or Defendants had understood. 22 Plaintiff also responds that the burden would be too great. Plaintiff argues that collecting and producing documents from Confluence imposes special burdens because Confluence pages cannot be extracted in bulk electronically. Instead, responsive pages must be manually extracted for production after review. [Plaintiff] has more than 520,000 Confluence pages subject to potential extraction. The processing of those pages would present a massive and unjustified burden at this point in the litigation. 23 I am skeptical of Plaintiff’s second argument concerning the Confluence documents, and specifically the representation that manual extraction is truly the only way to mine this repository of information. Perhaps my skepticism is unfounded. I also surmise that the 20 Defs.’ Fifth Disc. Mot. at 14–15. 21 Opposition at 8. 22 See, e.g., Dkt. 436 (Ex. J to Defs.’ Fifth Disc. Mot.) at 161:1–162:16 (describing the range of uses different teams within Plaintiff have for Confluence, including policy documents and company work product). 23 Opposition at 8. C.A. No. 2022-0613-KSJM September 22, 2022 Page 6 of 8 burden could be shifted to a neutral third party, the cost of which the parties could share. During the September 27 hearing, Plaintiff will need to describe the efforts they undertook to investigate how to collect this data and state its position concerning the possibility of relying on a neutral third party to aid the process. As to the third category, Defendants argue that Plaintiff misrepresented its custodians’ level of involvement in analyzing the relationship between mDAU and other metrics, such as advertising revenue and UAM. For example, Plaintiff stated that Julianna Hayes was involved in tracking the relationship between mDAU and advertising revenue. Her deposition testimony revealed that her involvement in such work was limited. 24 Other Twitter employees testified that they deferred to fellow employee Luke Simon on various issues. 25 Given this testimony, Defendants “request that the Court compel [Plaintiff] to add Mr. Simon as a custodian, produce responsive documents, and make him available for a deposition within five days of producing his documents.” 26 24 For instance, although her team was responsible for modeling revenue based on given user assumptions, neither Hayes nor her team were responsible for determining the user behavior information that informed those models. See Dkt. 436 (Ex. M to Defs.’ Fifth Disc. Mot.) 35:8–38:6. And although she used mDAU in her day-to-day activities, see id. 79:22–80:4, she also testified that she has no role in modeling user engagement. Id. 93:2– 14. 25 See, e.g., Dkt. 436 (Ex. C to Defs.’ Fifth Disc. Mot.) 232:2–8 (discussing why a Twitter employee witness would defer to Simon’s knowledge of revenue metrics, given his role at the company). 26 Defs.’ Fifth Disc. Mot. at 20. C.A. No. 2022-0613-KSJM September 22, 2022 Page 7 of 8 Plaintiff responds that Defendants inaccurately portrayed the deposition testimony on which they rely, that Simon’s role was not as significant as Defendants portray, that I previously denied Defendants’ efforts to make Simon a documents custodian, and that my prior ruling should not be unsettled. 27 It is often the case in discovery that depositions reveal new significance to other persons or facts in the case. I have reviewed the deposition testimony on which Defendants rely concerning Simon and believe that their representation is fair. Because Defendants lacked this information when previously requesting Simon as a custodian, I do not view my prior ruling as dispositive on the issue. For this reason, during the September 27 hearing, Plaintiff should be prepared to state the relative burden of collecting and reviewing Simon’s emails. Although I found compelling Plaintiff’s previous assessment of the burden of collecting and reviewing Slack messages, Plaintiff should nonetheless come prepared to state that burden with specificity as to Simon. As to the fourth category, Defendants argue that they learned, during a September 6 deposition of a Twitter employee, that Twitter uses “Box” and “Notion” as additional cloud-based tools, which Plaintiff did not previously identify. 28 Defendants argue that Plaintiff should have to collect and review documents from these repositories for each of their custodians. 29 27 See Opposition at 9–13. 28 Defs.’ Fifth Disc. Mot. at 15. 29 Id. at 16. C.A. No. 2022-0613-KSJM September 22, 2022 Page 8 of 8 Plaintiff responds that the Box and Notion systems are, one, rarely used and, two, duplicative of other repositories that Plaintiff has already searched. 30 Defendants and Plaintiff rely on the September 6 deposition testimony of Manish Chabria. 31 I have reviewed the September 6 deposition testimony on which Defendants rely and agree with Plaintiff’s position. The parties need not present argument on this request; it is denied. IT IS SO ORDERED. Sincerely, /s/ Kathaleen St. Jude McCormick Kathaleen St. Jude McCormick Chancellor cc: All counsel of record (by File & ServeXpress) 30 Opposition at 13–14. 31 Dkt. 436 (Ex. K to Defs.’ Fifth Disc. Mot.).