1. See Aurora Coop. Elevator Co.
v. Aventine Renewable Energy-Aurora W., LLC, No.
4:12CV230, 2015 WL
10550240, at 1 (D. Neb.
Jan. 6, 2015).
2. Nicholas Barry, Man Versus
Machine Review: The Showdown
Between Hordes of Discovery
Lawyers and A Computer-Utilizing Predictive-Coding
Technology, 15 VAND. J. ENT. &
TECH. L. 343, 344 (2013).
3. Maura R. Grossman & Gordon
Technology-Assisted Review in E-Discovery
Can Be More Effective and
More Efficient Than Exhaustive
Manual Review, 17 RICH. J.L.
& TECH. 11, 2 (2011).
4. Barry, supra note 2, at 354-55.
5. The level of client acrimony
and dissatisfaction with lawyers
and “the system” is a compelling consequence of this process. For once, lawyers are at
a loss for words as they try—
without success—to explain
and justify the need and requirements of ESI production.
6. Estimates of the cost of such
work to our clients are largely a
matter of “oral history.” This is
a remarkable observation in an
article dealing with electronic
information. Unless the
expense is publicly revealed in
a fee application (and in this it
is often buried and rare), the
client’s confidence is preserved
by avoiding public scrutiny of
the monies spent on mere
7. Moore v. Publicis Groupe, 287
F.R.D. 182, 183-84 (S.D.N. Y.
2012) (quoting Andrew Peck,
Search, Forward, L. TECH.
NEWS, Oct. 2011, at 25, 29).
8. Id. at 193.
9. Rio Tinto PLC v. Vale S.A., 306
F.R.D. 125, 127 (S.D.N. Y.
10. Dynamo Holdings Ltd. P’ship v.
C.I.R., 143 T.C. 183, 191-92
11. There are no reported cases—
but expect them to be coming—of parties seeking to
compel, over a court’s denial,
an obligation to use predictive coding. How exactly this
might arise is speculation, but
I presume the court would
have to grant an interlocutory
appeal of a discovery order.
Yes, a long shot. But, ESI was
not in vogue as recently as 15
12. The Sedona Conference Best
Practices Commentary on the
Use of Search and Information
Retrieval Methods in E-Discovery, 8 SEDONA CONF. J. 189,
198-99 (August 2007).
14. See David C. Blair & M. E.
Maron, An Evaluation of
Retrieval Effectiveness for a
System, COMMUNICATIONS OF
THE ACM 289 (1985).
15. Progressive Cas. Ins. Co. v.
Delaney, No. 2:11-CV-
00678-LRH, 2014 WL
3563467, at 8-9 (D. Nev.
July 18, 2014) (citations
omitted); see also Fed. Hous.
Fin. Agency v. HSBC N. Am.
Holdings Inc., No. 11 CIV.
6189 DLC, 2014 WL
584300, at 3 (S.D.N. Y. Feb.
14, 2014) (“The literature
that the Court reviewed at
that time indicated that predictive coding had a better
track record in the production
of responsive documents than
human review, but that both
processes fell well short of
identifying for production all
of the documents the parties
in litigation might wish to
see.”); Malone v. Kantner
Ingredients, Inc., No.
4:12CV3190, 2015 WL
1470334, at 3 n 7 (D. Neb.
Mar. 31, 2015) (“Predictive
coding is now promoted (and
gaining acceptance) as not
only a more efficient and cost
effective method of ESI review,
but a more accurate one.”).
16. FED.R.CIV.P. 26(b)( 1).
17. Sedona Conference, supra note
12, at 198-99; see also Navigat-
ing the Hazards of E-Discovery,
A Manual for Judges in State
Courts Across the Nation,
IAALS (the Institute for the
Advancement of the American
Legal System) (2d ed. 2012).
18. See FED.R.CIV.P. 26(f)( 3)(C).
19. Rule 26, FED.R.CIV.P., Amend-
ment Advisory Committee
Notes (emphasis added).
20. Rule 37(e) and (f) (also
amended in 2015) now specifi-
cally addresses the loss of ESI
and provides for an award of
sanctions required under Rule
26(f), which may provide
further incentive for parties to
agree to a narrowed focus for
preservation, discovery and/or
the use of predictive coding.
21. See FED.R.CIV.P. Rule 37(e)
22. See id. Rule 37, Amendment
Advisory Committee Notes
(“It authorizes and specifies
measures a court may employ
if information that should have
been preserved is lost, and
specifies the findings necessary
to justify these measures.”).
25. Barry, supra note 2, at 345.
narrow, the search will be
more targeted; if it is
broader, there is likely to be a larger, more
Common Client Concerns
Predictive coding is not necessarily a cheap
solution. Rather, it is a more efficient and
often less expensive way to search, code, and
produce documents. It will take many hours
of focused work from a senior attorney to
correctly code each model set. The person
coding the model set must be consistent and
have a good working knowledge of the case.
Clients may understandably fret about
the cost to have an experienced attorney
reviewing documents. Another downside to
predictive coding is that there are no results (i.e., nothing to show for the time
spent by a senior attorney) until the process is complete or substantially complete.
The process could take weeks or months,
• First, the document universe should
Clients are also justifiably concerned
about the inadvertent production of doc-
uments. This is a more difficult prospect—
because there is no guarantee that a privi-
leged or protected document will not be
produced. It is arguably more likely that a
human review will result in an inadvertent
production, but that is difficult for any cli-
ent to stomach. Three steps may help allevi-
ate this common client concern.
be screened for privileged documents
at the outset—that will eliminate most
obviously privileged documents.
• Second, as documents are coded into the
model set, a tag should be created for any
documents identified as work product or
privileged and not identified in the privi-
lege screen. Predictive coding can then
be used to search for these documents.
• Third, there should be a strong claw-back
provision incorporated into an order
entered by the court (e.g., a protective
order, agreement on ESI, or the Rule
26(f) discovery plan) providing for the
automatic return of any such documents.
If a client remains nervous, they can always
elect the greater expense of a document-by-document review. However, they may
change their mind when they review the
This decision point poses the question:
Between two paths, each of which is imperfect, what can a law department offer to its
client that has the least impact on attorney
hours and the most prospect for finding the
needles in the haystack? Computers and predictive coding appear to be the answer.
Predictive Coding and Electronically Stored Information