As predictive coding technology began garnering attention, one of the earliest and most common questions among attorneys paying attention to e-Discovery issues – probably the second most common question, right after “does it work” – was whether judges would accept it if opposing counsel challenged its use. As one writer put it in August 2011, attorneys were “left to wonder if predictive coding meets the standard of ‘reasonableness’ …” Pepper, Robot Review: Will Predictive Coding Win the Trust of the Courts?, Law Technology News, August 2011. A few months later, Magistrate Peck, writing in the same publication, offered his personal opinion as a support for attorneys considering predictive coding but reluctant to make the leap without judicial blessing:
“Until there is a judicial opinion approving (or even critiquing) the use of predictive coding, counsel will just have to rely on this article as a sign of judicial approval. In my opinion, computer-assisted coding should be used in those cases where it will help ‘secure the just, speedy, and inexpensive’ determination of cases in our e-Discovery world.”
- Peck, Search, Forward: Will Manual Document Review and Keyword Searches be Replaced by Computer Assisted Coding?, Law Technology News, October 2011.
A few months later, Judge Peck authored the first judicial opinion endorsing the use of predictive coding. Da Silva Moore v. Publicis Groupe, 287 F.R.D. 102 (S.D.N.Y. 2012). The Da Silva Moore decision, however, widely discussed as it was, still left counsel with defensibility concerns for two reasons. First, Judge Peck was (and is) well known to be in the e-Discovery vanguard, and was already well acquainted with the concept of predictive coding. Helpful as his decision was, the question remained whether other, potentially less tech-savvy judges would be as quick to accept computer review as a reasonable substitute for attorney review of each potentially responsive document. And second, the protocol agreed to by the parties and endorsed by Judge Peck in Da Silva Moore treated predictive coding as a unique species of discovery requiring special safeguards and atypical access to what would typically be the work-product process of making relevance decisions: attorneys for both parties would be provided access to the “seed sets” used to train the predictive coding engine, including documents excluded from production as non-responsive. 287 F.R.D. at 187.
Recent decisions – both reported and unreported – are providing comforting evidence that predictive coding is gaining wide judicial acceptance as a tool available to counsel on appropriate cases.
In April 2012, a Virginia state judge, James Chamblin, endorsed the use of predictive coding over the vigorous protest of opposing counsel. Global Aero. Inc. v. Landow Aviation, L.P., No CL 61040, 2012 Va. Cir. LEXIS 50 (Va. Cir. Ct., Apr. 23, 2012). In a July 2012 decision addressing the sufficiency of a FOIA search, Judge Schiendlin – herself known for her attention to e-Discovery issues – noted that “parties can (and frequently should) rely on … ‘computer assisted’ or ‘predictive’ coding” to identify responsive documents. Nat’l Day Laborer Org. Network v. United States Immigration & Customs Enforcement Agency, 877 F. Supp. 2d 87, 109 (S.D.N.Y. 2012). In January 2013, a Delaware Vice Chancellor issued a sua sponte order directing the parties in a litigation before him to either use predictive coding or show cause why they should not be required to do so. EORHB, Inc. v. HOA Holdings, Inc., No. 7409-VCL (Del. Ch. Ct. Oct. 15, 2012). And just a few weeks ago, two Judges of the Southern District of New York relied in part on the availability of predictive coding in rejecting a burdensomeness objection to a subpoena. See Chevron Corp. v. Donziger, No. 11 Civ. 0691, 2013 U.S. Dist. LEXIS 36353 (S.D.N.Y. Mar. 15, 2013); Harris v. Subcontracting Concepts, LLC, No. 1:12-MC-82, 2013 U.S. Dist. LEXIS 33593 (S.D.N.Y. Mar. 11, 2013).
And, in a December 2012 hearing in which opposing counsel challenged Foley’s use of predictive coding and demanded, at the very least, access to samples of the documents excluded as non-responsive (as in the Da Silva Moore protocol), Judge Andrews of the District of Delaware recognized that there was no more reason to provide such access where documents were excluded by predictive coding than there would be to grant access to a sample of documents deemed nonresponsive as a result of linear review:
why isn’t that something — you know, you answered their discovery however you answered it — why isn’t it something where they answer your discovery however they choose to answer it, complying with their professional obligations? How do you get to be involved in the seed batch?
Robocast, Inc. v. Apple, Inc., No. 11-235 (D. Del.) 12/5/2012 Transcript at 16:4-8.
Of course, this is a relatively small sample, so it is too soon to say that the defensibility of predictive coding can be taken as much for granted as, say, keyword search. But the fact that every reported case involving predictive coding has authorized its use suggests that day may not be too far off.