Posted on: May 30, 2014

in Blog

Can You Use Predictive Coding To Break Up With Your Current eDiscovery Protocol?


How do you break up with your existing eDiscovery protocol and take up with Predictive Coding? When would a breakup cause a lot of trouble and maybe need a court-approved divorce?

There is a new court decision about predictive coding that promises to be one of the most hotly debated since Da Silva Moore.  In Progressive Casualty Insurance Co v. FDIC, Delaney et al[i] Judge Leen essentially ordered the respondent (Progressive) to adhere to its previously agreed-upon and court-ordered eDiscovery protocol, instead of using predictive coding. We and our clients love predictive coding and we use it all the time. What happened here?

Several writers in the eDiscovery blogosphere have taken Judge Leen to task for not taking the more “progressive” route (pun intended). Before you make up your mind, consider the following important details in this well-considered opinion.

  1. The parties previously agreed on an eDiscovery protocol to cull 1.8 million documents using search terms that they had carefully decided upon, and apparently vetted, in a handful of related cases. The eDiscovery protocol was signed by the Court as an order.
  2. The party now wishing to use predictive coding planned to use it on the pre-culled set of documents, not on the corpus of the collection.
  3. The party now wishing to use predictive coding failed to disclose its use to the party with whom it had a court-ordered eDiscovery protocol, prior to a hearing to discuss it.
  4. The predictive coding protocol itself was flawed. The judge cited the respondent’s own expert, the Da Silva Moore decision by Judge Peck, and several leading treatises[ii], including Judge Peck’s Top Ten Best Practices of Predictive Coding, in order to tease out the facts that there was little transparency to what the respondent was proposing to do, and there was insufficient testing, or disclosure, regarding the adequacy of the process.
  5. The party now wishing to use predictive coding had already under-performed in discovery in several important ways. It had promised a rolling production of ESI but had failed to do so, even after indicating that it had already linearly reviewed 125,000 search hits. It had calculated in an affidavit in support of its use of predictive coding that it would take eight contract attorneys “six or eight months” to complete the linear review, a bad calculation of review metrics that further discloses its poor planning.

Consider these nuggets from the text of the decision:

“Predictive coding has emerged as a far more accurate means of producing responsive ESI in discovery. Studies show it is far more accurate than human review or keyword searches which have their own limitations.”

“[C]ourts give deference to a producing party’s choice of search methodology and procedures in complying with discovery requests… and litigators are loathe to reveal their methodological decisions for various reasons including…work product…discovery about discovery…and the Federal Rules of Civil Procedure only require parties to conduct a reasonable search for responsive documents”.

The Court is a fan of predictive coding and also cognizant of a party’s right to select its own method of review and production. So why did the Court not order predictive coding or agree to allow respondents to continue with it?

The answers are pretty simple.  The parties had agreed on an eDiscovery protocol, and it was a court order. The party wanting to use predictive coding presented a detailed, but still essentially flawed protocol. And that same party was already inexplicably late in producing any ESI and utterly opaque on what it had been doing.

Lessons from Progressive Casualty Insurance Co v. FDIC, Delaney et al for us are clear:

  1. Negotiate the eDiscovery protocol you want. If you have a court-ordered eDiscovery protocol you can no longer live with, re-negotiate it as soon as you can and bring it to the attention of the Court. Don’t wait to spring your new plan for the first time in oral argument.
  2. Predictive Coding is well-accepted by the courts, and the protocols for its execution are now well established. A last-minute hybrid protocol isn’t going to cut it.

Perhaps, if the respondent had been quicker to declare intentions and to produce documents, it could have made the switch. It may even have been allowed to use Predictive Coding to prioritize its review.

[i] District of Nevada, No. 2:11-cv-00678-LRH –PAL, Docket No. 143, filed May 20, 2014

[ii] Reality Bites: Why TAR’s Promises Have Yet to be Fulfilled, William P. Butterfield, Conor R. Crowley & Jeannine Kenney. Maura R. Grossman & GordonV.Cormack, Technology-Assisted Review in E-Discovery Can be More Effective and More Efficient Than Exhaustive Manual Review, Richmond J. L. & Tech. Vo. XVII, No. 3, Article 11 (2011); see also Nicholas M. Pace & Laura Zakaras, RAND Inst. For Civil Justice, Where the Money Goes: Understanding Litigant Expenditures for Producing Electric Discovery 55-58 (2012), available at

The Sedona Conference, Best Practices Commentary on the Use of Search and Information

Retrieval Methods in E-Discovery, 8 Sedona Conf. J. 189, 201-02 (2007), available at; Maura R. Grossman & Terry Sweeney, What Lawyers

Need to Know About Search Tools: The Alternatives to Keyword Searching Include Linguistic and Mathematical Models for Concept Searching, Nat. L. J. (Aug. 23, 2010).   

Discover More Categories

D4 Weekly eDiscovery Outlook

Power your eDiscovery intellect with our weekly newsletter.