Marc J. Goldstein Arbitrator & Mediator NYC
March 05, 2019

TAR: How Shall We Treat This Sticky Subject?

In your Tribunals’ initial case management conferences last month, what discussion took place regarding to the use of Technology Assisted Review (“TAR”) for e-disclosure? Are many of you now scrambling to your web browsers to confirm exactly what is TAR? Is your approach to e-disclosure frozen in time circa 2010 (or perhaps 1995) — even as TAR’s acceptance among judges and magistrates is expanding?  Should questions about the use of TAR in arbitral e-disclosure be left entirely to party agreement? Or is such preliminary delegation  simply a convenient way for arbitrators to steer clear of unfamiliar and challenging terrain?

This post will not educate you about how TAR works or even the details of what TAR is. I am as much of a neophyte about the particulars as many of you may be.

But this much is evidently beyond reasonable dispute: TAR is potentially far more effective and efficient than keyword searching for identifying of relevant and material data in a large universe of data. An arbitral tribunal that claims to be seriously dedicated to an efficient process arguably should steer the parties, at the earliest stages, toward an agreement concerning the use (or not) of TAR for the gathering of data sought by the adverse party.

Here is a modest compilation of evidence supporting the proposition just stated:

  • TAR for Smart People, a 200-page textbook published by the e-discovery consulting firm OpenText (successor by merger as of January 31, 2019 to Catalyst Repository Systems), has recently been issued in its third edition, with a foreword by a just-retired U.S. Magistrate Judge who was a pioneer in the judicial endorsement of  TAR. I have a free download. You may obtain it the same way.
  • There is now a reasonably evolved body of TAR jurisprudence in the US federal courts. This case law arguably should be a foundation for a best-practices approach to e-disclosure among international commercial arbitrators. Some of the accepted principles: (1) TAR is cheaper, more efficient and superior to keyword searching,  (2) TAR will not be imposed upon a party as a method for gathering responsive documents, (3) a party’s decision to use TAR to comply with the adverse party’s requests for production ordinarily should be seen as a reasonable method to achieve compliance, and (4) whereas TAR’s efficacy depends on the sample set of relevant, responsive documents used for the predictive coding, the parties’ early meet-and-confer efforts should focus on compiling a sample set that satisfies the requesting party.
  • In the UK, a Practice Direction issued in August 2018 by the Civil Procedure Rule Committee for the courts of England and Wales makes discussion of the potential use of TAR mandatory in parties’ early conferences concerning disclosure protocol. The Practice Direction requires that the parties justify any decision not to use TAR where disclosure will involve 50,000 documents or more. This Practice Direction is now incorporated a two-year pilot program operative in the Business and Property Courts.

It will take some time for the major provider organizations in international arbitration to come to terms with TAR in their rules, protocols, and guidelines. In the meantime, we as arbitrators have much soul-searching to do, case by case.  I raise some of the questions here.

What is the practical usefulness of traditional written requests for production and written objections thereto, each drafted according to litigation models dating from the pre-digital era? Are arbitrators and counsel unhelpfully channeling their litigator instincts,  carrying on an obsolete custom? If the objective is to provide helpful guidance to the search for documents that the requesting party considers relevant, shouldn’t the guidance conform to the search method?

Let’s suppose a Tribunal takes these questions seriously. And let’s further suppose, to reduce complications in the discussion, that: (1) both parties prefer that document disclosure be obtained before the initial round of written merits submissions, and (2) each side is represented by a multinational law firm based in the US or UK that is known to have (or may reasonably be assumed to have) a well-developed capability for the use of TAR .

Instead of fixing a timetable for court-style requests for production, written responses and objections, perhaps Redfern Schedules, and resolution of ensuing disputes, would you prefer instead that your TAR-sensitive Tribunal (1) require the parties to have their e-disclosure attorneys-in-charge attend the case management conference, and (2) inform the parties that document search method including TAR will be discussed at the conference?

Suppose at this conference both sides indicate a preference to use TAR, reserving the right to assert that perceived shortcomings in the TAR-based production might justify curative application of other methods. If so, isn’t the next question how should each side convey its wishes to the other? If the technology-assisted phase of TAR begins with a very human, attorney-driven, more-or-less “manual” selection of a sample set, then what requesting document will the attorney-samplers use as their reference point to compile the sample set?

Perhaps some e-disclosure attorneys will tell your Tribunal that they prefer old-fashioned court-style requests for production. Or perhaps they will shrug and say they usually do not have a choice, because most of their work is in relation to court litigation, or subpoenas from administrative bodies and grand juries. But perhaps a few of them would say that they would like a document that is like a Proffer, a declaration of what the requesting party expects to prove or disprove on the basis of the responsive documents. Let’s call this, potentially, the TAR Search Template.  It would be required to contain narrative explanations of the relevant and material issues of fact or law to be established by the search, the persons from whom such documents are to be sought (by name or functional profile), and the relevant time frames. (And Template drafters would be reminded to prepare them in compliance with applicable protocols like the ICDR Guidelines and Art. 3.3 of the IBA Evidence Rules).

A provision in Procedural Order No. 1 then might look like this: “TAR Search Templates (the “Templates”) shall be exchanged and delivered to the Tribunal in 30 days, and on [Day 30+7] the Tribunal will hold a telephone conference if necessary to resolve objections to the Templates, the parties having the ability in the +7 period to submit proposed changes and/or to negotiate adjustments to the Templates.”

Ensuing provisions of this TAR-driven PO #1 might require exchange/submission of the sample set, an interval to review/revise the Templates in light of the presence or absence of responsive documents in the sample set, commencement and completion dates for the TAR process, possibly an interim reporting or production requirement so permit TAR’s efficacy to be assessed, and a timetable to request any supplemental search measures (such as keyword searching) after delivery of the documents generated by the TAR process.


I do not know the right answers, but I hope this post poses many of the right questions. I do not think that international arbitrators may continue to be technology trogolodytes who permit the e-disclosure process to be only as efficient as the parties, left mainly to their own initiatives, may or may not succeed in making it. Yet I do not wish to be understood as advocating, at the opposite extreme, for imposition of TAR-based disclosure methods by arbitral fiat.  The effective arbitrator sensibly balances arbitral encouragement and party autonomy.  But this balancing act can’t be done without a fluency in the subject matter. I perceive that there is a fluency problem today. Many of the arbitrators in greatest demand completed their advocacy careers before TAR gained a significant foothold in advocacy practice. Others have had or continue to have only a glancing exposure — optional disengagement from process-oriented details being a privilege of stature.

Here is a vote for universal fluency. Let’s all read some of the leading TAR decisions of US federal judges. And download your copy of TAR for Smart People today. (I have no connection to the publisher or the authors, and derive no remuneration from advocating that you read it).

One Response to “TAR: How Shall We Treat This Sticky Subject?”

  1. Philip Lacovara says:

    Extremely useful . . . but would have been even more informative with a thumbnail sketch of the key differences between TAR and standard key-word searches.

Leave a Reply