IN RE: ALLERGAN BIOCELL TEXTURED BREAST IMPLANT PRODUCTS LIABILITY LITIGATION This Order Relates to All Actions: In re Allergan Biocell Textured Breast Implant Product Liability Litigation, MCL No. 634 MDL No. 2921 | Civil Action No. 2:19-md-2921 (BRM)(ESK) United States District Court, D. New Jersey Signed October 25, 2022 Counsel Joseph A. Dickson, U.S.M.J. (Ret.), Brittany Manna, CHIESA SHAHINIAN &amp; GIANTOMASI PC, One Boland Drive, West Orange, NJ 07052, (973) 325-1500, Appointed Special Master. Dickson, Joseph A., Special Master SPECIAL MASTER CASE MANAGEMENT ORDER NO. 22 *1 This matter comes before us by way of Defendants’ August 26, 2022 application to implement Defendants’ proposed protocol for the use of technology-assisted review (“TAR”) (or predictive coding) for its document production going forward. Simultaneously, Plaintiffs submitted a brief opposing the use of TAR by Defendants going forward and instead requesting that Defendants be ordered to proceed with its search term and linear review or be ordered to implement the TAR protocol to the full custodial set of documents. On September 9, 2022, the parties each submitted a reply. I. INTRODUCTION The parties are familiar with the facts surrounding the underlying action and claims. Accordingly, we will recite only the relevant procedural and factual background necessary to address the disputes at hand, namely whether the implementation of TAR should apply to the Defendants’ remaining un-reviewed custodial documents after the application of search terms. II. DEFENDANTS’ ARGUMENTS Defendants argue that applying TAR after the application of search terms is standard practice and commonly used to promote efficiency and reduce costs. As the party responding to discovery, Defendants submit that it is in the best position to determine the best review methodology. According to Defendants, the best review methodology, based upon the large number of documents that have been reviewed, is to train the TAR model and implement the TAR protocols to the remaining un-reviewed custodial documents after the application of search terms. Defendants submit that of the sixty custodial files collected to date, there remains approximately 560,000 custodial documents totaling more than 3.5 million pages still to be reviewed by its e-discovery vendor. Defendants assert that their e-discovery vendor estimated that a manual review of those custodial files will take approximately twenty weeks to complete. Therefore, Defendants argue, the application of TAR to those remaining documents will increase the efficiency of the review of the un-reviewed documents and accelerate the completion of discovery. Defendants reject Plaintiffs’ suggestion that Defendants apply the TAR protocols to the full custodial set of documents before search terms are applied citing burden and efficiency concerns. Specifically, Defendants submit that it has collected and indexed 9.371 terabytes of data prior to the application of the parties’ search terms. According to Defendants, under Plaintiffs’ proposal, Defendants would be required to recollect the 9.371 terabytes of data, which is “impractical and wholly unnecessary given the broad search terms already employed.” (See Defs.’ Ltr., Aug. 26, 2022 at 5). Defendants further submit by relying on In re Biomet M2a Magnum Hip Implant Prods. Liab. Litig., Livingston v. City of Chi., Huntsman v. Sw. Airlines Co., and Bridgestone Ams., Inc., that its proposal is consistent with the majority of courts that have addressed whether a producing party may utilize TAR after search term culling. According to Defendants, In re Biomet, Bridgestone Ams. Inc., and Huntsman stand for the proposition that applying search terms culling before predictive coding is permitted and complies with the Federal Rules of Civil Procedure. Similarly, Defendants submit that the court in Livingston rejected the same argument that Plaintiffs attempt to advance here and instead determined that applying TAR after search term culling will achieve the best review in that case. *2 As such, Defendants request that we enter an order permitting it to apply TAR to the remaining un-reviewed custodial documents to which the parties’ search terms have been applied. III. PLAINTIFFS’ ARGUMENT Plaintiffs, on the other hand, submit that the application of TAR to the full corpus of documents, prior to the application of search terms, creates no additional burden on the Defendants and will increase the accuracy of the review. Plaintiffs argue that the application of TAR to the un-reviewed documents after search terms have been applied is prejudicial and unreasonable because it will exclude documents from review and reduce the “efficacy and accuracy” of the review process. (See Pls.’ Ltr., Sept. 9, 2022 at 4). Plaintiffs further argue that Defendants’ have not provided any data to support an estimate of how much time or money would be saved by implementing TAR in the manner Defendants suggest. Plaintiffs advance that any alleged burden on the Defendants must be weighed against the needs of the case, the importance of the issues, and the amount in controversy. Fed. R. Civ. P. 26(b)(1). Additionally, Plaintiffs argue that the cases Defendants relied upon are fact-specific and distinguishable from this case. Plaintiffs assert that I should follow In re Valsartan, Losartan, &amp; Irbesartan Prods. Liab. Litig., 337 F.R.D. 610 (D.N.J. Dec. 2, 2020) instead. As such, Plaintiffs’ request that we deny Defendants’ application and require Defendants to proceed with its linear search term review or, in the alternative, require Defendants to accept a TAR protocol that Plaintiffs propose to be run on the custodial files before the application of search terms. IV. DISCUSSION On September 30, 2022, we heard oral argument from the parties on the TAR issue. As stated, Defendants wish to apply the TAR protocols to the remaining unreviewed documents after search terms have been applied. If we are going to allow Defendants to proceed with TAR, Plaintiffs’ request that Defendants apply the TAR protocols to all of the custodial files before search terms are applied. Defendants suggest that its proposal is “consistent with the majority of courts” that have addressed the TAR/search term issue and seem to have concluded from its review that courts generally have found that search terms are appropriate to cull sets of documents prior to TAR application. (Defs. Ltr. Aug. 26, 2022 at 5). We do not agree with Defendants’ characterization of the case law. There is no such general principle espoused by the courts or the commentators. That is not to say that cases do not exist where parties are permitted to apply TAR after culling by the application of search terms. The courts find solutions to the problems confronting them, but do not settle the question of which method is better. For example, Defendants rely on in In re Biomet M2a Magnum Hip Implant Prods. Liab. Litig., where Biomet used a keyword culling to reduce the universe of documents from 19.5 million to 3.9 million, which was ultimately reduced to 2.5 million after the removal of duplicates. 2013 WL 1729682 at *1 (N.D. Ind., April 18, 2013). Biomet then utilized TAR to identify relevant documents to be produced from the 2.5 million. Id. Plaintiffs there argued that Biomet's use of the key word search in the beginning tainted the process and requested that Biomet be ordered to institute predictive coding to the original 19.5 million documents. Id. at *2. *3 The court then ruled for Biomet after that Biomet provided the court with a cost-benefit analysis that informed the court's decision, including evidence that requiring Biomet to implement TAR to the 19.5 million documents would “entail a cost in the low seven-figures.” Id. at *2. No such analysis is available here. In fact, we are still unclear why data that had been captured in the initial collection and review would now have to be recaptured. Most importantly, the Biomet court specifically stated that it was not deciding whether predictive coding or keyword searching is better; rather the issue was whether Biomet had satisfied its discovery obligations: “[t]he issue before me today isn't whether predictive coding is a better way of doing things than keyword searching prior to predictive coding. I must decide whether Biomet's procedure satisfies its discovery obligations ...” Id. Bridgestone Ams. Ins. Inc. v. IBM is similar to Biomet. The court's opinion was lacking in factual context, and like Biomet, decided eight to nine years ago, which is a lifetime in the world of technological development and advancement of TAR. The Bridgestone court observed that “the uses of predictive coding is a judgment call” and mandated that the parties continue communicate and operate with openness and transparency with respect to the predictive coding process. 2014 WL 4923014 at *1 (M.D. Tenn. July 22. 2014). Here, Defendants provided a Declaration from Paul Ramsey, the Director of Strategic Client Solutions for their e-discovery vendor. (Defs. Ltr. Aug. 26, 2022 at Ex. D ¶ 6). The Declaration set forth the costs Defendants have incurred to-date for their review. The Declaration does not detail how those costs will be increased or decreased based upon the implementation of TAR either before search terms are applied or after search terms are applied. The Defendants, despite being afforded multiple opportunities (in their opening brief, in their reply, and during oral argument on September 30, 2022) to provide a cost-benefit analysis or statistical sampling, have not done so. Defendants fail to explain why they should be permitted to continue using search terms when TAR is generally recognized as more efficient. What is the advantage of using both search terms and TAR? No explanation has been satisfactorily provided. With respect to Defendants’ burden argument, as stated above, we are not convinced that Defendants would have to “recollect[ ] over nine terabytes of data” in order to implement TAR. The question remains: where did the data go? It is customary to host and maintain the full universe of documents on a system. At oral argument on September 30, 2022, Defendants did not sufficiently clarify that point. Clearly the data should be maintained so as to be available in the event of disputes (such as this one). Defendants also did not specify how, if at all, costs would increase or decrease if TAR was implemented in accordance with their preference (after search terms) or in accordance with Plaintiffs’ preference (before search terms). The use of both opens the door for additional discovery disputes related to the accuracy of Defendants’ review. We do know that applying TAR to an already reduced (via search terms) set of documents will only reduce the document pool further and will certainly not reveal documents that the application of search terms has precluded. After reviewing Defendants’ submissions and the Declaration of the Director of Strategic Client Solutions for Defendants’ e-discovery vendor, the question of the benefit of using both search terms and TAR remains unanswered. In re Valsartan, cited by Plaintiffs, is similarly unhelpful as it relies entirely on specific facts that color the decision of that case. Judge Schneider, however, recognized in his lengthy, and clearly irate opinion that TAR requires, “an unprecedented degree of transparency and cooperation among counsel in the review and production of ESI responsive to discovery requests.” 337 F.R.D. 610, 622 (D.N.J. Dec. 2, 2020). This is a principle to which counsel in this matter may not be accustomed. If there is an overriding principle to be taken from all of these cases, this is the one. *4 Also driving our decision is the fact that the parties have not agreed to the application of TAR. More than a year ago, the Court entered Case Management Order No. 15 regarding Electronically Stored Information wherein the parties acknowledged the following General Agreement: The Parties are aware of the importance the Court places on cooperation and commit to continue to consult and cooperate reasonably as discovery proceeds. No Party may seek judicial relief concerning this Order unless it first has conferred with the applicable producing or requesting Party. (ECF No. 194, at * 1, Sept. 4, 2020). The ESI Protocols also address the use of search filtering technology and implies that TAR would be considered, discussed, or addressed at that time, “the parties shall confer on the application, if any, of search or other filtering technologies, including reasonable search terms, file types, date ranges, validation processes, predictive coding, Technology Assisted Review (“TAR”), or other appropriate advanced technology ...” Id. at *4. Further, the ESI Protocols state: The Parties are expected to work in a cooperative and collaborative manner to maximize the efficiency and success of the application of the methodologies proposed at identifying potentially relevant ESI. To the extent the Parties are unable to reach agreement on the application of, or procedures for, any search or filtering processes, the Parties shall raise such issues for resolution by the Court. Following such agreement and/or order of the Court, if either Party believes that revisions to agreed-upon search-term or advanced-technology procedures are necessary to enhance or improve the identification of potentially responsive ESI, the Parties shall promptly meet and confer regarding the proposed revisions prior to implementation. No such revisions shall be permitted absent agreement of the Parties, or order of the Court. Id. (emphasis added). We find that the Defendants have not set forth an adequate basis for ordering the application of TAR after the application of search terms, over the objection of the Plaintiffs. As such, Defendants’ request to apply TAR after search terms have been applied is denied. We have little doubt that the parties knew at the outset the costs of ESI discovery would be high, and the review process would be extensive. The fact is, without testing on an agreed set of documents, no one can predict whether the application of TAR with or without search terms is the more economic and feasible way to proceed. Implementing TAR, at this stage, after the application of search terms, opens the door for potential disputes that may arise related to the accuracy of the review process and will further delay the completion of discovery and drive costs upward. Finally, applying TAR to an already reduced (via search terms) set of documents will reduce further the identified responsive documents and will certainly not reveal documents that the application of search terms has precluded. Because Plaintiffs did not bargain for this at the outset, over a year ago, it is inappropriate to force them to accept it now. CONCLUSION &amp; ORDER For the foregoing reasons, it is on this 25th day of October, 2022, ORDERED that Defendants application to apply TAR after the application of search terms is DENIED. Defendants shall continue with their originally chosen review method: linear, search term review. *5 SO ORDERED. Footnotes  Defendants submit that, at the time of its briefing, it has reviewed more than 2.2 million documents comprising more than 14 million pages. (See Defs.’ Ltr., Aug. 26, 2022 at 3.).  In re Biomet M2a Magnum Hip Implant Prods. Liab. Litig., 2013 WL 1729682 (N.D. Ind. April 18, 2013); Livingston v. City of Chic., 2020 WL 5253848 (N.D. Ill. Sept. 30, 2020); Huntsman v. Sw. Airlines Co., 2021 WL 3504154 (N.D. Cal. Aug. 10, 2021); Bridgestone Ams. Inc. v. Intl. Business Machine Corp., 2014 WL 4923014 (M.D. Tenn. July 22, 2014).