Who’s a Custodian, What’s Protected, and What Counts as Reasonable? Judge Sallie Kim’s Opinion on Shenwick v. Twitter

Magistrate Judge Sallie Kim issued a discovery opinion that should be included in ediscovery chapters of Civil Procedure textbooks. In a security class action against Twitter, Judge Kim methodically issued orders covering whether to include custodians in the scope of discovery, application of the Stored Communications Act, and proximity search terms.1

Should the Co-Founder and CEO Be Included as a Custodian?

The parties had agreed on 25 custodians; however, there was disagreement on whether to include Jack Dorsey, the CEO and co-founder of Twitter. Dorsey was the CEO from 2007 to 2008, was Chair of the Board of Directors, and CEO again in 2015. Dorsey was the individual who “came clean” on Twitter’s metrics that went to the issues of misleading investors.

The Defendants argued that searching Dorsey’s email was premature, which the Court rejected. Moreover, the Court also rejected the argument that Dorsey had limited involvement, since he was the one who admitted Twitter’s actual metrics. Even though other custodians might have responsive information, it would not justify eliminating Dorsey as a custodian. As such, the Court ordered the inclusion of Dorsey’s files to be searched.

Direct Messages Are Protected by the Stored Communication Act

The Court found that the Stored Communication Act (SCA) prohibited the production of Direct Messages sent within Twitter, other than between the two individuals who were parties to the litigation. The Court treated Twitter as a separate entity from the individuals who had Direct Messages within Twitter, because Twitter did not require its employees to use Direct Messages. The SCA prohibits the production of third-party electronic communications from electronic communication service providers (ECS). As Twitter is an ECS under the SCA, the Court adjudicated Twitter separately from users with privacy rights under the SCA.

Wildcard and Proximity Searches

The parties had one disagreement over a search term: “engag*”. This is a “wildcard” search in many review applications that would generate hits for the every word beginning with “engage.” The Defendants warned this term could have false-positive results that would include anything from “engagement parties” to political engagement.

The Defendants argued the term “engag*” needed to be within the proximity of five other keywords; the Plaintiffs countered that was too limiting. The Court acknowledged there was not a perfect solution and ordered the term “engag*” be searched within 10 words of other search terms.

Bow Tie Thoughts

The facts of a lawsuit matter in determining which custodians need to be included in the scope of discovery, reasonable search methodologies, and whether a party has actual control of data in the case. This case logically addressed and resolved all of these issues.

Proximity searches should be discussed  at Rule 26(f) conferences between the two parties. Parties frequently list hundreds of search terms for use, not taking into consideration that hundreds of “OR” searches could generate anywhere from tens of thousands to millions of hits. Searching for these terms within a certain distance of other terms, or used by specific individuals, can focus search hits on those that are reasonable for review.

Consider the following example: An “OR” search for the terms “gas,” “interest,” and “earnings.”  There are 276,597 hits.

This is an excessive amount of records to review from OR searches. However, if the search is refined to finding records that contain “gas” within twenty words of the phrase “interest earnings,” the number of hits drops to two.

Any party responding to discovery requests must demonstrate whether they had a reasonable process to find responsive data. Are two search hits reasonable? Yes, if the two hits prove the case. In most cases, however, two hits are not reasonable, and the proximity search should be expanded out to validate if there are any other possibly responsive hits.

Search terms should be tested, experimented with, and validated. This requires leveraging proximity searches and other advanced search strategies to find responsive data, instead of merely relying on lengthy “OR” searches. Thinking in big-picture terms can help legal teams avoid swimming in thousands of search hits and focus on finding the discovery that supports their case.

1Shenwick v. Twitter, Inc., 2018 U.S. Dist. LEXIS 22676 (N.D. Cal. Feb. 7, 2018).