Predictive Coding

• Author: Aaron Goodman
• Position: The author is with DLA Piper LLP (US), Phoenix.
• Pages: 23-27

Published in Litigation, Volume 43, Number 1, Fall 2016. © 2016 by the American Bar Association. Reproduced with permission. All rights reserved. This information or any portion thereof may not be

copied or disseminated in any form or by any means or stored in an electronic database or retrieval system without the express written consent of the American Bar Association. 23

Predictive Coding

A Better Way to Deal with

Electronically Stored Information

AARON GOODMAN

The author is with DLA Piper LLP (US), Phoenix.

Litigators bemoan the fact t hat the volume of documents and

data produced and requested during d iscovery continues to in-

crease year on year—as doe s the volume of electronically stored

information (ESI) that clients mus t preserve. There is a glut of

ESI arising as a conse quence of the exponential g rowth in the

creation and exchange of electron ic files and email traf fic. The

result is that even in case s with a relatively low monetary value,

there are often sti ll hundreds of thousands or even millions of

files relevant to the ca se, and a significant por tion of the cost of

litigation is driven by t he obligation to preserve, search, review,

and produce this ESI. However, the same computing power that

has provided this overabu ndance of data may also provide the

solution for reviewing, coding, a nd producing it in litigation

with minim al human review.

Computer-assisted review, technology-assisted review, and

“predictive coding ” all refer to the use of computer-generated

algorithm s—based on human modeling—to code docu ments

without the need for document-by-document huma n review.

Predictive coding “pred icts the relevance of discovery docu-

ments [or relative match of any document to the model set] based

on the prior coding of a small sa mple of discovery documents by

an attorney.” Nicholas Barry, Man Versus Machine Review: The

Showdown Between Hordes of Discovery Lawyers and a Computer-

Utilizing Predict ive-Coding Technolog y, 15 V. J. E. & T.

L. 343, 344 (2013). In other words, predictive coding is an itera-

tive process because it depends on the coding f irst applied by

attorneys fami liar with the case and resp onds as the attorneys

continue to intera ct with the docu ments. The computer becomes

an extension of the attor ney reviewer—coding documents in the

review universe based on t he algorithms derived from a model

created by the attorney.

Starti ng Computer-Assisted Revie w

It begins with t he collection and culling of client files a nd data,

which can quickly become a daunt ing task without the support

of a technical specia list. It was not long ago that only the most

significa nt cases would encompass large amounts of ha rd-copy

documents and electronic f iles. In fact, for many years, there

was an inherent proportiona lity between the complexity of the

case and the volume of documents exp ected to be at issue. Today,

however, because each of us and the many devices we use a re

creating data at a near cons tant rate, even small-dollar case s can

have considerable ESI subject to a duty to preserve and dis cov-

ery. Accordingly, it has become commonplace for relevant cus-

todians to posses s hundreds of gigabytes, or even terabytes , of

data sourced solely from emails requ iring preservation, search ,

and review.