RULING REGARDING ELECTRONICALLY STORED INFORMATION ["ESI"]
JOAN GLAZER MARGOLIS, Magistrate Judge.
The factual and procedural history behind this FLSA litigation has been set forth in detail in the multiple rulings issued by U.S. District Judge Janet Bond Arterton and by this Magistrate Judge, namely: Ruling on Plaintiffs' Motion to Compel Production of Documents, filed January 6, 2015 (Dkt. #127), 2015 WL 75884; Ruling on Defendant's Motion to Compel, filed February 10, 2015 (Dkt. #138), 2015 WL 540911; Order Following Telephone Conference, filed February 23, 2015 (Dkt. #143); Ruling Denying Defendant's Motion to Transfer Venue, filed February 23, 2015 (Dkt. #145); Supplemental Ruling Following In Camera Review, filed May 8, 2015 (Dkt. #159); Ruling Granting Plaintiffs' Motion for Conditional Certification, filed June 9, 2015 (Dkt. #168), 2015 WL 3727804; Ruling on Number of Depositions, filed June 10, 2015 (Dkt. #171); Ruling on Objections to Reminder Notice, filed August 14, 2015 (Dkt. #201), 2015 WL 4877173; and Scheduling Order, filed October 20, 2015 (Dkt. #256)["October 2015 Order"].
On November 17, 2014, Judge Arterton first referred this file to this Magistrate Judge for discovery. (Dkt. #103; see also Dkt. #124). There are now 1,047 opt-in plaintiffs in this litigation. (October 2015 Order at 2, ¶ 7). As the multiple rulings and orders indicate, Judge Arterton and this Magistrate Judge have been forced into micro-managing virtually every pretrial aspect of this massive litigation, down to the smallest minutiae. (See also October 2015 Order, at 3, nn. 1-2).
Pursuant to deadlines set in the October 2015 Order (at 2, ¶ 6), on October 30, 2015, plaintiffs and defendant filed letters (Dkts. ##262-63),1 followed by responsive letters on November 6, 2015. (Dkts. ##265-66).
As set forth in plaintiffs' letters (Dkt. #262, at 1-5; Dkt. #265, at 1-4), plaintiffs have proposed 103 search terms from the documents of five custodians, to which defendant objected on the basis of burdensomeness; plaintiffs have proposed three alternatives as a compromise as follows: (1) "sampling and iterative refinement[,]" under which defendant would produce "small samples of documents containing the agreed-on search terms, followed by an iterative analysis of the documents to hone the search terms;" (2) "the virtual warehouse[,]" which constitutes "a quick-peek protocol by which [defendant] provides the entire corpus of documents collected from the custodians," followed by plaintiffs' search "for a limited number of relevant documents, akin to an old-fashioned search through paper files in a warehouse;" or (3) "complete production[,]" described as a "straightforward production of all documents that hit on any search term with the use of a clawback agreement permitting [defendant] to retrieve any inadvertently produced documents."
More specifically, under the first option, "sampling and iterative refinement," plaintiffs propose that if each search string "results in a purportedly unwieldy volume of hits," then defendant can provide only a statistically significant sample of those hits (plaintiffs point to web sites that will provide guidance to counsel for the appropriate sample size), which will include all non-privileged documents, both relevant and irrelevant, so that the parties can then reevaluate and modify the search terms "to focus on the most relevant documents and weed out irrelevant documents, to the extent feasible." (Dkt. #262, at 3 & n.3)(footnote omitted).2 Plaintiffs characterize this as "a limited, reasonable proposal that will result in only a fraction of potentially responsive documents being produced in the first instance." (Id. at 3).
Under the second option, plaintiffs would engage in a "quick peek[,]" where "within an agreed-upon time period," plaintiffs would conduct a "secure, initial examination of all ESI. . . without any search strings being applied to cull documents[,]" plaintiffs would then designate "a limited number of documents" for defendant to formally produce, at which time defendant would have the opportunity to review the designated subset of documents for privilege. (Id. at 4)(emphasis added). Plaintiffs characterize this approach as "a modern equivalent of the old-fashioned method of lawyers going to a warehouse to inspect hard copy documents and tagging documents for production with post-it notes." (Id.).
Lastly, plaintiffs propose "complete production (with clawback)," with defendant producing all documents containing any search terms, along with a qualifying phrase proposed by defendant, and allowing defendant to retrieve irrelevant (and presumably privileged) documents. (Id. at 4-5).
Defendant's letters describe in detail the failed negotiations between counsel with respect to ESI. (Dkt. #263, at 3-6 & Exhs. A-H; Dkt. #266, at 1-4). Defense counsel estimates that one attorney can review 75 to 100 documents per hour. (Dkt. #263, at 4). As represented by defense counsel, plaintiffs' initial proposed search terms against the e-mail accounts of the five custodians3 produced 961,463 documents, or approximately 60% of the 1.5 million documents in the database. (Id. at 3 & Exh. A).4 Indeed, even defendant's proposed search terms currently hit on 61,859 documents, which defendant suggested needs to be further modified to reduce the number of documents to be reviewed. (Id. at 3-4 & Exh. B).5
After several rounds of negotiations, according to defendant, plaintiffs' last proposal was that defendant review all documents where the search terms hit on less than 10,000 documents, as well as review a statistically significant sample of documents where the search terms hit on more than 10,000 documents, and that defendant produce all non-privileged documents, whether relevant or not. (Id. at 5 & Exh. F). According to defendant, its last counterproposal to plaintiffs was to apply plaintiffs' proposed keywords to the e-mails of six custodians, and that for any search term that hit on fewer than 3,000 documents, defendant would review all of the documents and produce any relevant, non-privileged documents, but for any search term that hits on more than 3,000 documents, then defendant would consolidate all of the documents and pull one random sample from the pool of documents, which then would be reviewed and defendant would again produce all relevant, non-privileged documents. (Id. at 5 & Exh. G).
Given that there are 1,047 opt-in plaintiffs, "potentially hundreds more as class members" in the four states of Connecticut, California, North Carolina and Missouri, and a possible verdict in eight or nine digits if plaintiffs are successful here (see Dkt. #265, at 2), defendant's proportionality argument is unavailing. (See Dkt. #263, at 2-3). However, defendant is correct that it should not be obligated to disclose clearly irrelevant documents (id. at 7-8) in an attempt "to [better] hone the search terms[.]" (Dkt. #262, at 1).
In light of all these circumstances, the Magistrate Judge rules as follows: defendant's ESI search shall consist of the search terms proposed by defendant of the files of eight custodians — the six identified in notes 3 and 5 supra, as well as Scott Creasy and Brian Fillebrown (see Dkt. #263, at 3-5 & Exh. B); because these are search terms proposed by defendant, there is a presumption of relevancy, and defendant may remove any documents only if they are clearly and undeniably irrelevant, and of course, if they are privileged; moreover, this conclusion is without prejudice to plaintiffs seeking additional documents, as approved by the court if no agreement is otherwise reached by the parties, in the event that they deem this initial production insufficient. Counsel shall confer with one another as to the date by which defendant shall disclose all relevant documents consistent with this ruling (which may also be on a rolling basis); on or before December 14, 2015, counsel shall notify this Magistrate Judge's Chambers, by letter, the date to which they agree, and if no agreement is reached, the parties' proposed dates.
This is not a Recommended Ruling, but a ruling on a non-dispositive motion, the standard of review of which is specified in 28 U.S.C. § 636; FED. R. CIV. P. 6(a), 6(e) & 72; and Rule 72.2 of the Local Rules for United States Magistrate Judges. As such, it is an order of the Court unless reversed or modified by the District Judge upon timely made objection.
See 28 U.S.C. § 636(b)(written objections to ruling must be filed within fourteen calendar days after service of same); FED. R. CIV. P. 6(a), 6(e) & 72; Rule 72.2 of the Local Rules for United States Magistrate Judges, United States District Court for the District of Connecticut; Small v. Secretary, H&HS, 892 F.2d. 15, 16 (2d Cir. 1989)(failure to file timely objection to Magistrate Judge's recommended ruling may preclude further appeal to Second Circuit).6
