Harvard Journal of Law & Technology

Volume 31, Special Issue Spring 2018

THE NEW WAVE: COPYRIGHT AND SOFTWARE INTERFACES

IN THE WAKE OF ORACLE V. GOOGLE

Fred von Lohmann*

TABLE OF CONTENTS

I. INTRODUCTION .............................................................................. 517

II. THE NEW WAVE OF SOFTWARE INTERFACE CASES AFTER
ORACLE V. GOOGLE ....................................................................... 519
A. SAS v. WPL............................................................................... 519
B. Synopsys v. ATopTech .............................................................. 523
C. Cisco v. Arista .......................................................................... 524
D. GDC v. Dolby .......................................................................... 525
III. LESSONS FROM THE NEW WAVE OF INTERFACE CASES ............. 527
IV. WHY AFFIRMATIVE DEFENSES ARE NOT ENOUGH .................... 530
V. CONCLUSION ................................................................................ 533

I. INTRODUCTION

Imagine you are interested in starting a car company. Perhaps

your cars will have innovative electric drivetrains, like Tesla’s. Per-
haps you will cater to a niche audience that is not served by main-
stream automobiles, following the example of companies like smart
and Morgan. Perhaps you have something new and entirely novel to
introduce to the automotive market.
Now imagine that, as a condition of entering the market, you first
must design a new interface for your car — something different from
the steering wheel and pedals with which all of your potential custom-
ers are familiar. In order to try out your car, customers would first
have to learn this new “method of operation” (to use the terminology
of 17 U.S.C. § 102(b)). Perhaps you are up to the challenge — per-
haps gestures or a touch screen? But this would be a formidable ob-
stacle, as your potential customers would literally have to relearn how
to drive in order to try your products. If that hurdle was not enough to
put you off the idea of entering the market (entrepreneurs are nothing
if not optimistic!), it might be enough to put off any potential inves-
tors. Certainly, the requirement to create a new interface would be a

* Legal Director, Copyright, Google LLC. The views herein represent the views of the
author, not necessarily those of Google LLC.
518 Harvard Journal of Law & Technology [Vol. 31

substantial barrier to entry for your business and would represent a

significant switching cost for your potential customers.
Fortunately, copyright law has many limiting doctrines that would
make it impossible to assert copyright protection over the steering
wheel and pedals of an automobile. 1 But as Professor Menell’s article
makes clear, while the stakes for software interfaces are just as high as
for automotive interfaces, the copyright law answers are far less set-
tled for software interfaces. The Oracle v. Google litigation — and the
cases that have been brought in its wake — raises the very same ques-
tions posed by the automotive hypothetical: questions about switching
costs, barriers to entry, and network effects. While these kinds of
questions may, at first, seem far afield from copyright law, we can
expect them to recur with increasing frequency in copyright cases in
the years to come. After all, software will likely mediate more and
more of the technologies we depend on. And interfaces are the steer-
ing wheels and pedals by which we operate software.
Those engaged in the copyright debate around interfaces fall
roughly into three camps. First are those who agree with Professor
Menell (I count myself among their number) that interfaces ought not
be protectable by copyright, and who are worried that “[c]ompanies
could use API strategies to lock in consumers and lock out competi-
tors.” 2 On this view, granting copyright protection to software inter-
faces (or “methods of operation” generally), permits the copyright
owner not only to recoup its own investment, but also to unfairly and
inefficiently capture the value of independent investments by its cus-
tomers. The second group lies at the other end of the spectrum, un-
troubled by the “rise of the API copyright dead.” Their reasoning is
that the very purpose of copyright law is to create barriers to entry in
the name of spurring investment and innovation. For them, once an
interface has cleared copyright law’s extremely low threshold of “cre-
ativity,” there is nothing wrong with exclusive rights quelling free-
riding by competitors. The third group takes a middle position, trust-
ing in copyright’s existing affirmative defenses, such as fair use and
scènes à faire, to sort outcomes that enhance social welfare from out-
comes that reduce competition to the net detriment of society.

1. Most importantly, copyright does not protect the design of useful articles, unless the
article’s design includes pictorial, graphical, or sculptural features that are independent of
the utilitarian aspects of the article. See 17 U.S.C. § 101 (2012); Star Athletica LLC v. Var-
sity Brands, Inc., 137 S. Ct. 1002, 1008 (2017).
2. Peter S. Menell, Rise of the API Copyright Dead?: An Updated Epitaph for Copyright
Protection of Network and Functional Features of Computer Software, 31 HARV. J.L. &
TECH. (SPECIAL EDITION) 305 (2018); see also Pamela Samuelson, Functionality and Ex-
pression in Computer Programs: Refining the Tests for Software Copyright Infringement,
31 BERKELEY TECH. L.J. 1215, 1258–59 (2017) (“[f]reedom to reuse APIs, insofar as they
are necessary for interoperability, promoted healthy competition and ongoing innovation in
the software industry.”).
Special Issue] Copyright and Software 519

Professor Menell does a thorough job of explaining why his view

is the best of the three, based on statutory construction, jurisprudential
evolution, and economic policy. Rather than restating those argu-
ments, this comment aims to supplement Professor Menell’s treatment
by reviewing four software interface cases filed in the wake of the
Oracle v. Google litigation. Not only are the “copyright API dead”
rising after decades of relative quiet, they appear to be rising with in-
creasing frequency. Taken together, they lend further support to Pro-
fessor Menell’s arguments and bear out the policy concerns that he
sets out. These four cases also lend further credence to Professor
Menell’s view that affirmative defenses are an inadequate correction
for overbroad software interface copyright protection.

II. THE NEW WAVE OF SOFTWARE INTERFACE CASES AFTER

ORACLE V. GOOGLE

As Professor Menell explains, copyright cases involving software

interfaces were rare in the decades between the Lotus Dev. Corp. v.
Borland Int’l, Inc. 3 and Oracle Am., Inc. v. Google Inc. 4 litigations. In
the wake of Oracle v. Google, however, there have been a spate of
such cases, with their rate of appearance accelerating after the Federal
Circuit’s ruling approving the notion that interfaces can be protectable
by copyright.

A. SAS v. WPL

The earliest in this “new wave” of interface cases is SAS Inst.,

Inc. v. World Programming Ltd., 5 filed in the Eastern District of
North Carolina in 2010. The facts are reminiscent of the early inter-
face cases that Professor Menell discusses — a leading market in-
cumbent sues a new market entrant for studying the interface of an
existing software application, using that knowledge to reimplement
those interfaces, and creating a competing product.
As with most real-world litigations, the case involves a number of
procedural complexities (including a parallel litigation in the United
Kingdom that made its way to the Court of Justice of the European

3. 49 F.3d 807 (1st Cir. 1995), aff’d by an equally divided Court, 516 U.S. 233 (1996).
4. 750 F.3d 1339 (Fed. Cir. 2014).
5. 64 F. Supp. 3d 755 (E.D.N.C. 2014), aff’d in part and vacated in part, 874 F.3d 370
(4th Cir. 2017). This case was filed a few months before Oracle v. Google, but was stayed
for several years as parallel litigation took place in the UK. The appeal drew four dueling
amicus briefs, all of which addressed the copyrightability of software interfaces and the
Federal Circuit’s ruling in Oracle v. Google. Nevertheless, the court of appeals found that
the copyrightability issue need not have been decided because the plaintiff had achieved
complete relief under its breach of contract claim. Accordingly, the court vacated as moot
the district court ruling on the copyrightability issue. 874 F.3d at 389–90.
520 Harvard Journal of Law & Technology [Vol. 31

Union, Europe’s highest court 6) and non-copyright claims (in this

case, breach of contract, on which SAS ultimately prevailed 7). But the
contours are simple enough. SAS is a software company based in
North Carolina that makes the “SAS System,” a suite of software ap-
plications used by enterprises to perform statistical data analysis. 8 The
SAS System is the leading application in this market segment. In or-
der to use the SAS System, its users must first create their own pro-
grams, written in the SAS programming language. 9 Accordingly, SAS
customers face high switching costs if they want to try a different sta-
tistical analysis application, as they have to rewrite programs that they
already wrote for the SAS System, in order to use those programs in
conjunction with a different application.
The defendant, a UK company called World Programming Lim-
ited (“WPL”), identified an opportunity here. If it could independently
develop an application that interoperated with the programs that SAS
customers had already written, it would be able to compete directly
with SAS. In order to create its competing application, WPL studied
publicly available documentation about the SAS programming lan-
guage, the behavior of a legitimately acquired “learning edition” of
the SAS System, and the operation of the SAS System in the hands of
an existing SAS customer. 10 This enabled WPL to create its own in-
teroperable application called the World Programming System, de-
signed to run the programs written by SAS customers — in the words
of SAS, “a cheaper drop-in replacement.” 11
SAS responded by suing for copyright infringement. SAS argued
that WPL had infringed its copyrights “by using certain software lan-
guage functions and by copying the resulting output formats that are
produced when a user runs those language functions through the SAS
System.” 12 In its briefing on appeal, SAS refers to these elements as
the “input and output formats” of the SAS System. 13 With respect to
the “input formats,” SAS characterizes these as “a ‘simple set of . . .

6. SAS, 64 F. Supp. 3d at 760–61; see also Jonathan Band, The Global API Copyright
Conflict, 31 HARV. J.L. & TECH. (SPECIAL EDITION) 615 (2018) (describing the course and
outcome of the European litigation).
7. SAS, 64 F. Supp. 3d at 769–74.
8. Id. at 759, 761.
9. Id. at 762.
10. Id. at 764–67. SAS alleged that WPL nevertheless breached the license agreement
that governed the use of the “learning edition” of the software, but the resolution of that
question was not material to the copyright claim.
11. Redacted Brief of Appellant/Cross-Appellee at 1, SAS Inst. Inc. v. World Program-
ming Ltd., 874 F.3d 370 (4th Cir. 2017) (Nos. 16-1808, 16-1857) [hereinafter SAS Opening
Brief].
12. SAS, 64 F. Supp. 3d at 775.
13. SAS Opening Brief, supra note 11, at 3–4 (“WPL meticulously copied the input and
output formats of the SAS System, which reflect countless hours of creativity on the part of
SAS, its statisticians and programmers.”).
Special Issue] Copyright and Software 521

concise commands’ to request a comprehensive analysis.” 14 With re-

spect to the “output formats,” SAS describes these as “the tables,
graphs, and other forms of output” that are produced by the SAS Sys-
tem from user programs. 15 In other words, the heart of the SAS copy-
right claim is focused on the interface used by SAS customers to
operate the SAS System, as SAS itself admitted:

SAS is unlike commonly used consumer software —

such as a web browser, word processing program, or
videogame — whose users interface with the soft-
ware by clicking a mouse, moving a joystick, or typ-
ing text into a box. In contrast, SAS users interface
by issuing written instructions to the software. Those
instructions are provided by users as text files con-
taining the required instructions and are generally re-
ferred to as “SAS Programs.” SAS Programs are
written in a high-level programming language devel-
oped and maintained by SAS and known as the
“SAS Language.” With a set of commands, a user
can instruct the computer to access and arrange data
and then perform a comprehensive analysis. Differ-
ent from a computer programming language like
FORTRAN or C (the underlying programming lan-
guage used by SAS to write its own software), the
SAS Language allows a user to cause the SAS Sys-
tem to process and analyze data with concise written
instructions that would otherwise require “literally
hundreds of thousands of lines of code” in a low-
level programming language. 16

The district court, however, was unpersuaded. It granted summary

judgment to WPL, holding that all WPL had done here was copy the
“SAS programming language,” and languages are not protectable
copyright subject matter:

In essence, by asking the court to find that defend-

ant’s software infringes its copyright through its pro-
cessing of elements [of] the SAS Language, plaintiff
seeks to copyright the idea of a program which inter-
prets and compiles the SAS Language — a language

14. Id. at 6.
15. Id. at 7.
16. Id. at 7–8 (citations omitted).
522 Harvard Journal of Law & Technology [Vol. 31

anyone may use without a license. However, copy-

right law provides no protection to ideas. 17

On appeal, SAS contended that these “input and output formats”

should be copyrightable because its developers could have chosen a
different structure of command names and output possibilities. 18 In
other words, SAS asserted copyright in the creative choices made in
naming the commands and defining the output possibilities that its
customers must use in order to operate the SAS System.
This sounds reminiscent of Oracle’s copyrightability theory. And,
for that matter, it echoes Lotus’ position regarding its menu com-
mands in Lotus v. Borland. 19 In that case, which Professor Menell
describes in more detail, 20 Lotus contended that the menu command
hierarchy for its Lotus 1-2-3 spreadsheet was protectable by copy-
right. Lotus reasoned that it could have chosen different command
names, and arranged them differently, and thus its competitor, Bor-
land, should be prohibited from copying the names and arrange-
ment. 21 The First Circuit flatly rejected that view: “The fact that Lotus
developers could have designed the Lotus menu command hierarchy
differently is immaterial to the question of whether it is a ‘method of
operation.’” 22
In the end, the Fourth Circuit declined to resolve the question, va-
cating as moot the district court’s copyrightability ruling in SAS v.
WPL. 23 But for purposes of this discussion, the overall contours of the
dispute remain relevant. WPL’s brief on appeal puts the matter most
directly:

Extending copyright protection to the SAS language,

as [SAS] proposes, would impermissibly expand the
scope of [SAS’s] copyright and grant a monopoly
over the SAS language (and indeed over all SAS
language programs written by users), prohibiting
others from developing software that com-
piles/interprets the SAS language, which [SAS] does
not own. 24

17. SAS, 64 F. Supp. 3d at 776 (citing 17 U.S.C. § 102(b) (2012); Feist Publ’ns, Inc. v.
Rural Tel. Serv. Co., 499 U.S. 340, 344–45 (1991)).
18. SAS Opening Brief, supra note 11, at 52–55.
19. See Lotus Dev. Corp. v. Borland Int’l, Inc., 49 F.3d 807, 816–18 (1st Cir. 1995).
20. Menell, supra note 2, at 336–340.
21. Lotus Dev. Corp., 49 F.3d at 816–18.
22. Id. at 816.
23. SAS Inst. Inc. v. World Programming Ltd., 874 F.3d 370 (4th Cir. 2017).
24. SAS Opening Brief at 45–46.
Special Issue] Copyright and Software 523

In short, the story here is fundamentally one of switching costs

for SAS customers, and whether copyright law should force those
customers to rewrite all of their own code as a precondition of trying a
competing statistical analysis system. By failing to resolve the copy-
rightability question, the Fourth Circuit missed a valuable opportunity
to choose between the competing approaches taken by the First Cir-
cuit in Lotus v. Borland and the Federal Circuit in Oracle v. Google.

B. Synopsys v. ATopTech

The second in the “new wave” of interface cases is Synopsys, Inc.

v. ATopTech, Inc., 25 filed on June 26, 2013 in the Northern District of
California. This dispute pitted an incumbent market leader in the field
of electronic design automation (“EDA”) software, Synopsys, against
a smaller competitor, ATopTech. At issue were “place-and-route”
software applications that customers use to plan the layout of a chip
and the electrical connections among its various components. 26
Synopsys’ copyright infringement claim was premised on ATop-
Tech’s copying of input and output formats 27 — essentially the same
theory pressed by SAS in its battle with WPL. With respect to input
formats, Synopsys asserted infringement of “the combination of
names and syntax comprising PrimeTime’s and GoldTime’s input
formats.” 28 Synopsys also claimed copyright over output formats,
“such as including a line of asterisks to demarcate the initial, over-
view information for the report, using a string of dashes to provide
separation before output data, and including flag and reason columns
to label the output.” 29 These interface elements constitute the method
of operating Synposys’ PrimeTime and GoldTime software; a user
cannot operate software without knowing how to issue commands and
what to expect in return. By supporting the same input and output
formats used by Synopsys software, ATopTech promoted interopera-
bility and reduced the switching costs that potential customers other-
wise face. For this, it found itself embroiled in years of expensive
litigation.

25. No. 13-CV-029652965 (N.D. Cal. filed Jun. 26, 2013) (ECF No. 1).
26. Synopsys, Inc. v. ATopTech, Inc., No. 13-CV-02965, 2016 WL 6158216, at *1 (N.D.
Cal. Oct. 24, 2016).
27. See Synopsys, Inc. v. ATopTech, Inc., No. 13-CV-02965, 2016 WL 80549, at *1
(N.D. Cal. Jan. 7, 2016); Defendant ATopTech, Inc.’s Trial Brief in Support of Filtration,
Synopsys, Inc. v. ATopTech, Inc., No. 13-CV-02965 (N.D. Cal. filed Mar. 7, 2016) (ECF
No. 667).
28. Pl. Synopsys, Inc.’s Opp. to ATopTech’s Mot. in Lim. No. 1 at 6, Synopsys, Inc. v.
ATopTech, Inc., No. 13-CV-02965 (N.D. Cal. filed Feb. 2, 2016) (ECF No. 571). The pre-
cise details of the interface, as well as whether ATopTech copied all of the input and output
formats or a subset, are obscured by the redactions made by the parties to protect their pro-
prietary information.
29. Id. at *14.
524 Harvard Journal of Law & Technology [Vol. 31

Over the course of the litigation, the parties faced off across an ar-
ray of claims and counterclaims, including patent infringement,
breach of contract, and antitrust claims. In the end, however, after a
three-week jury trial, the plaintiff prevailed solely on its copyright
claim and was awarded $30.4 million. 30 ATopTech subsequently filed
for bankruptcy protection, resulting in a consent judgment that pre-
cluded further appeal from the copyright verdict. 31

C. Cisco v. Arista

The third in the “new wave” of interface cases after Oracle v.

Google is Cisco Systems Inc. v. Arista Networks, Inc., 32 filed in the
Northern District of California in December 2014. The basic contours
of the dispute once again echo those of previous interface cases. The
plaintiff, an incumbent market leader, brings a copyright claim against
a competitor who tries to mitigate customer switching costs by reim-
plementing the incumbent’s software interface.
In this case, the plaintiff is Cisco, a market leader in Ethernet
switches, routers, and other networking devices. The defendant is
Arista Networks, a competitor founded by former Cisco employees. 33
The dispute centers on the operating system software for the network-
ing hardware sold by both companies. Customers configure and oper-
ate these devices via a text-based command line interface (“CLI”). In
Cisco’s words: “The CLI is the user interface by which users of Cisco
products communicate with the product in order to configure and
manage the product.” 34 For its copyright infringement claim, Cisco
asserted that Arista infringed the user interfaces of the Cisco operating
system software by copying hundreds of multiword commands, as
well as multiword command hierarchies, modes and prompts, com-
mand responses and screen outputs, and help descriptions. 35
While the district court described these interface elements in
some detail, 36 the gist here should be familiar to anyone who has
muddled through a similar command line interface on an early per-
sonal computer (MS-DOS) or a more recent Linux-based computer.
Operating the Cisco switches requires using specific multiword com-

30. Synopsys, Inc., 2016 WL 6158216, at *1. The patent claim was severed for separate
trial.
31. Stipulation and Consent Judgment, Synopsys, Inc. v. ATopTech, Inc., No. 13-CV-
02965 (N.D. Cal. filed Jun. 27, 2017) (ECF No. 967).
32. Complaint for Copyright and Patent Infringement, Cisco Sys. Inc. v. Arista Networks,
Inc., No. 14-CV-053445344 (N.D. Cal. filed Dec. 5, 2014) (ECF No. 1).
33. Second Amended Complaint for Copyright and Patent Infringement at 1, Cisco Sys.,
Inc. Networks, Inc., No. 14-CV-05344 (N.D. Cal. filed July 23, 2015) (ECF No. 64).
34. Id. at 8.
35. Cisco Sys., Inc. v. Arista Networks, Inc., No. 14-CV-05344, Slip Opinion at 2–3
(N.D. Cal. May 10, 2017) (ECF No. 787).
36. See id.
Special Issue] Copyright and Software 525

mands in a particular syntax and format. Examples of multiword

command expressions include “boot system,” “show inventory,” “area
nssa translate type7 always,” and “spanning-tree portfast bpdufilter
default.” 37 In order to configure and operate the Cisco switches, cus-
tomers must employ the correct commands and expect the devices to
return the relevant responses. In other words, this is the method of
operation for Cisco switches.
In its competing products, Arista chose to implement support for
many of the same multiword commands (approximately 500, accord-
ing to Cisco) and other interface elements with which Cisco custom-
ers had become familiar over many years. 38 In various public
statements, Arista specifically described how its support for these
Cisco command line commands would help customers ease adoption
and leverage their long-standing familiarity with Cisco commands. 39
The dispute made its way to a jury trial in December 2016, where
the jury returned a verdict of non-infringement in favor of Arista,
finding the infringement excused by the scènes à faire doctrine. 40 The
matter is, as of this writing, pending before the Court of Appeals for
the Federal Circuit, where it will be heard in 2018 on the heels of Or-
acle’s appeal from the fair use jury verdict in Oracle v. Google. 41
Again, much could be said about the arguments of the parties, the
various theories accepted and rejected by the court, and the details of
the jury verdict. But for purposes of this discussion, the basic factual
contours are most salient. Arista entered the market against an incum-
bent market leader whose customers had made substantial investments
in learning the method for operating the incumbent’s products. Arista
wrote its own original software, but in order to mitigate switching
costs for its potential customers, offered support for many of the same
commands that Cisco’s devices had long used. The similarities with
Lotus v. Borland, Oracle v. Google, and SAS v. WPL, and Synopsys v.
ATopTech are vivid.

D. GDC v. Dolby

The most recent of the new wave of interface cases came in the
form of a declaratory judgment action filed by GDC Technology Ltd.

37. Id.
38. See Second Amended Complaint, supra note 33, at 3.
39. Id. at 2–3, 12–13.
40. Joe Mullin, Arista Beats Cisco’s $335M Copyright Claim with an Unusual Defense,
ARSTECHNICA (Dec. 14, 2016, 4:08 PM), https://arstechnica.com/tech-policy/2016/12/jury-
clears-arista-of-ciscos-335m-copyright-claim/ [https://perma.cc/BD4C-EE8K].
41. The appeal is Cisco Sys., Inc. v. Arista Networks, Inc., No. 17-2145 (Fed. Cir. dock-
eted June 13, 2017). Appellate jurisdiction is exclusive to the Federal Circuit, thanks to a
pendant patent claim in the case.
526 Harvard Journal of Law & Technology [Vol. 31

against Dolby Laboratories, Inc. in April 2016. 42 According to the

complaint, GDC and Dolby are competitors in the digital cinema in-
dustry. 43 GDC sells media servers to theater owners who have transi-
tioned from film to digital cinema systems. 44 These media servers
must interoperate with other systems provided by different vendors,
including projector systems, sound systems, and theater management
systems that coordinate the interoperation of these systems. 45 To or-
chestrate their coordinated functioning, these systems rely on com-
mands that “typically take the form of a four-digit hexadecimal (two
byte) code, embedded in a larger message header that tells the server
that it is about to receive a message.” 46 These command protocols are
widely shared among all players in the industry, and in any event can
easily be observed by reading data transmissions between different
components of the theater system. 47
Dolby’s effort to upend this cooperative state of affairs triggered
the litigation. According to the complaint, shortly after itself entering
the media server market via acquisition, Dolby began asserting pro-
prietary rights over its interconnection codes, telling theater owners
that interconnecting GDC equipment with Dolby equipment would
constitute an infringement of Dolby’s intellectual property rights. 48
Dolby then followed up by sending a cease-and-desist letter to GDC,
explicitly asserting copyright protection over the commands used to
control Dolby devices. 49 GDC then filed the declaratory judgment
action, insisting that “[t]he only element of Dolby’s protocol that
GDC uses is the set of messages/commands and corresponding hexa-
decimal interoperability codes,” 50 and asserting that the commands
were not copyrightable or, in the alternative, that GDC’s continued
use would qualify as a fair use. 51
Like the other “new wave” cases, this dispute bears strong simi-
larities to prior interface cases. Once again, a leading incumbent as-
serted copyright in the commands necessary for the operation of its
software. Prior customer sunk-cost investments (here, investments by
theater owners in systems that include Dolby equipment and software)
create lock-in at the expense of a competitor who seeks to enter the
market. There was no allegation of copying of source code or other

42. Complaint, GDC Tech. Ltd., Inc. v. Dolby Labs., Inc., No. 16-CV-02459 (C.D. Cal.
filed Apr. 11, 2016) (ECF No. 1).
43. Id. at 2.
44. Id. at 3.
45. See id. at 3–4.
46. Id. at 4.
47. Id.
48. Id. at 12–13.
49. Id. at 14.
50. Id. at 7.
51. Id. at 19.
Special Issue] Copyright and Software 527

literal elements of software; rather, the infringement claim was prem-

ised on a competitor creating original software that copies only the
interface — the commands used by customers to operate the software.
Unlike the earlier cases, GDC v. Dolby settled before a judge or
jury could rule on copyrightability or infringement. 52 In a terse press
release, the parties disclosed merely that they “will grant each other
licenses that will allow their respective theatre management systems
to interoperate with the other party’s digital-cinema servers.” 53 This
represents something of an about-face for GDC, which had asserted
that the command protocols were not copyrightable. On that view,
there was nothing for Dolby (and presumably GDC) to license. Never-
theless, in the wake of the settlement, presumably both Dolby and
GDC remain free to assert copyright protection over their command
protocols against other competitors.

III. LESSONS FROM THE NEW WAVE OF INTERFACE CASES

The new wave of software interface cases that have followed on

the heels of Oracle v. Google fit the same pattern that characterized
the previous cases described by Professor Menell. A market leader
asserts a copyright claim over an interface — the method of operating
the software — in an effort to bar a competitor from entering the mar-
ket. The claim is not that the competitor is copying and redistributing
the incumbent’s source or object code. Instead, the allegation focuses
on the competitor creating original software that reimplements some
or all of the interface. The goal in each of these cases is to force com-
petitors to create an entirely original interface as a precondition of
market entry, thereby keeping switching costs high for the incum-
bent’s existing customers. To return to the initial analogy, these are all
examples of copyright being used to prevent a new entrant from using
the equivalent of the steering wheel and pedals as the method for op-
erating a car.
The assertion of copyright to hinder competitive entry is perhaps
most vivid in Lotus v. Borland, SAS v. WPL, and Cisco v. Arista. In
each of those cases, the defendant was entering the market with a di-
rectly competitive product, offering the incumbent’s existing custom-
ers an easy way to bring their existing macros, programs, or scripts
over with them. In Oracle v. Google, Android and Java were not
competing products, but nevertheless the theme of switching costs is
clear. Java developers had independently developed fluency in Java’s

52. Stipulation of Dismissal Pursuant to Fed. R. Civ. P. 41(a), GDC Tech. Ltd. v. Dolby
Labs., Inc., No. 16-CV-02459 (C.D. Cal. Oct. 26, 2016) (ECF No. 26).
53. Film J. Int’l, GDC and Dolby Resolve Litigation (Nov 1, 2016), http://
www.filmjournal.com/news/gdc-and-dolby-resolve-litigation [https://perma.cc/R9RU-
59TF].
528 Harvard Journal of Law & Technology [Vol. 31

API declarations. They also had already written software in Java, code
that would be difficult to repurpose if it had to be entirely rewritten. In
Professor Menell’s words: “Using some of the Java APIs provided a
bridge for the millions of Java programmers.” 54 In this case, the de-
velopers were the customers, and Oracle’s assertion of copyright over
the Java API declarations effectively increased switching costs for
developers who wanted to try a different programming ecosystem.
GDC v. Dolby and Synopsys v. ATopTech also fit the same pat-
tern. In GDC v. Dolby, the assertion of copyright over the command
protocol was a claim limited solely to the interface — Dolby never
suggested that GDC had copied and incorporated Dolby’s software
into GDC’s products. By using copyright law to create artificial in-
compatibility between Dolby and GDC products, customers would be
forced to choose one or the other, thereby increasing switching costs
for customers who wanted to defect from Dolby to GDC, or mix-and-
match products from both vendors. And while some of the facts in
Synopsys v. ATopTech are obscured by redactions to protect the pro-
prietary information of the parties, the story that emerges is also one
of a leading incumbent asserting copyright over an interface — in that
case, input and output formats — in an effort to prevent customers
from defecting to the competitor’s interoperable products.
As Professor Menell explains, it is difficult to see why copyright
law should tolerate these kinds of claims. I will not reiterate the statu-
tory and jurisprudential arguments that Professor Menell thoroughly
covers. On that score, two simple points are enough to persuade me.
First, § 102(b) of the Copyright Act expressly provides that “[i]n no
case does copyright protection for an original work of authorship ex-
tend to any . . . method of operation . . . regardless of the form in
which it is described, explained, illustrated, or embodied in such
work.” 55 Second, as recognized in Lotus v. Borland, this statutory
command applies equally to software. 56 Accordingly, the interface
used to control a software program cannot qualify for copyright pro-
tection, insofar as it serves as the “method of operation” for the soft-
ware. The Federal Circuit was mistaken to the extent it held otherwise
in its copyrightability ruling in Oracle v. Google.
Professor Menell also ably covers the economic and policy ra-
tionales that support his view. Extending copyright protection to soft-
ware interfaces is not necessary to provide adequate incentives for
software developers. 57 Copyright law will continue to protect software

54. Menell, supra note 2, at 470.

55. 17 U.S.C. § 102(b) (2012).
56. Lotus Dev. Corp. v. Borland Int’l, Inc., 49 F.3d 807, 815–18 (1st Cir. 1995), aff’d by
an equally divided Court, 516 U.S. 233 (1996).
57. Menell, supra note 2, at 464 (“[F]unctional features of computer software and ma-
chines fall within the patent system’s domain. The importance of interoperability and com-
Special Issue] Copyright and Software 529

from wholesale, piratical duplication. Patent law may protect interfac-

es, if they qualify under patent law’s more stringent requirements.
And software companies will retain the benefits of trade secret protec-
tion, contract law, and technological protection measures. Together,
these will frequently yield a significant first-mover advantage, as
competitors will be required to study, document, and independently
reimplement interfaces before entering the market. Furthermore, as
explained by Professor Menell, a significant portion of the software
industry has moved to nonproprietary and cloud-based business mod-
els that are protected from much of “the appropriability problem” 58
that copyright is designed to solve. Granting copyright protection to
interfaces, in contrast, conveys a century of legal protection against
new market entrants who want to “build a bridge” 59 for customers
interested in trying their products.
Returning to the fundamental purpose of copyright, moreover,
developers seem to have sufficient incentives to develop interfaces, so
an exclusive right appears unnecessary. After all, the incentive to cre-
ate a method of operating something is inherent in the incentive to
create the thing itself. It would be a strange piece of software indeed
that offered no interface, no method of operation, just as it would be
strange to find a car maker building an automobile with no way for a
driver to operate it. 60
In the end, the question that the new wave of software interface
cases raises is the same one identified by Judge Boudin in Lotus v.
Borland:

But if a better spreadsheet comes along, it is hard to

see why customers who have learned the Lotus menu
and devised macros for it should remain captives of
Lotus because of an investment in learning made by
the users and not by Lotus. 61

patibility bring trademark protection into play. In addition, software can often be protected
through trade secret law. Developers can hide their programming by only releasing object
code versions to the public.”).
58. Id. at 463 (“To a significant extent, platform developers have partially addressed the
appropriability problem associated with software development through the formation of
collaborative clubs.”).
59. Id. at 458 (“Another strategy is to build a convenient bridge over which consumers
can easily migrate to and become accustomed to a new platform.”).
60. What about self-driving cars? To the extent self-driving cars may soon dispense with
the steering wheel and pedals, software will provide the new interface, raising all the con-
cerns discussed here. If consumers become accustomed to a particular set of spoken word
commands to control their self-driving cars, will those commands be subject to copyright
protection, requiring drivers to learn an entirely new command vocabulary in order to try a
different make of self-driving car?
61. 49 F.3d at 821.
530 Harvard Journal of Law & Technology [Vol. 31

Professor Menell restates this in economic terms as the “network

externality dilemma.” 62 It stands to reason that empowering an in-
cumbent market leader in a market characterized by network effects to
recoup not only the value of its initial investment, but also the inde-
pendent investments of its customers, threatens exactly the kind of
monopolistic deadweight loss that Professor Menell describes. 63

IV. WHY AFFIRMATIVE DEFENSES ARE NOT ENOUGH

As discussed earlier, there are those who agree that copyright pro-
tection for software interfaces can have socially detrimental conse-
quences, but who believe that copyright’s existing affirmative
defenses can adequately address these concerns. Professor Menell
addresses this at the end of his article, opining that “the fair use trial
[in Oracle v. Google] was a massive waste of time, party resources,
and judicial resources.” 64 He also emphasizes that sending these cases
to jury verdicts is extremely expensive and does not deliver the legal
clarity that software developers and investors need in order to make
“the difficult, time-sensitive decisions involved in designing products
and platforms.” 65
With respect to Professor Menell’s point regarding expense, skep-
tics may point out that recognizing interfaces as unprotectable “meth-
ods of operation” will not always spare litigants the burden of a trial.
After all, in Oracle v. Google, the district court initially did not rule
on the protectability of the interfaces in question, but instead instruct-
ed the jury to assume copyrightability, and the jury returned a hung
verdict on fair use. 66 Nevertheless, it seems likely that filtering out
methods of operation as unprotectable should, in most cases, result in
earlier (and cheaper) outcomes for litigants. The only relevant inquiry
under § 102(b) should be whether the interface constitutes a “method
of operation.” While the question may not be free from doubt in all
cases, it certainly will not require the kind of broad ranging factual
inquiries required for a fair use or scènes à faire determination. In
many cases, the matter could be resolved at summary judgment, mak-
ing a jury trial unnecessary.
Professor Menell is likewise on solid ground when he notes that
legal clarity suffers if the questions raised by protecting interfaces are
left to copyright’s affirmative defenses. One need only look at the

62. Menell, supra note 2, at 458 (“Intellectual property protection both contributes to and
alleviates the network externality dilemma.”).
63. Id. at 454 (“Monopolistic exploitation distorts market pricing in the short run and can
significantly affect entry and cumulative innovation over longer time horizons.”).
64. Id. at 471.
65. Id. at 472.
66. Final Charge to the Jury (Phase One) and Special Verdict Form, Oracle Am., Inc. v.
Google Inc., 2016 WL 4368346 (N.D. Cal. May 26, 2016) (No. C 10-03561 WHA).
Special Issue] Copyright and Software 531

interface cases where affirmative defenses were litigated to see the

lack of predictability that results. In Oracle v. Google, for example,
the defendant prevailed on a fair use defense (but only after trying the
issue twice), while in Lotus v. Borland and Synopsys v. ATopTech, the
defendants’ fair use defenses were unavailing. 67 In Cisco v. Arista, the
defendant’s fair use arguments failed, but its scènes à faire argument
carried the day. 68 Of course, in every case, the outcome will turn on
the facts. But in cases that turn on affirmative defenses like fair use
and scènes à faire, the outcomes turn on facts that are specific to the
particular product and industry, rather than on any generalizable prin-
ciples regarding the protectability of interfaces that can guide other
software developers.
Some may conclude that this is a feature, rather than a bug, in that
the question of protectability should turn on the particulars of the
product at issue; every interface case is unique. This argument, of
course, reinforces Professor Menell’s point that these cases will then
provide very little legal clarity for others. But those who nevertheless
prefer the close-up, particularized treatment afforded by affirmative
defenses must answer another question: do the affirmative defenses
like fair use and scènes à faire lead courts and juries to ask the right
questions? As discussed by Professor Menell and reinforced by the
recent cases discussed above, the crucial social welfare questions
raised in software interface cases revolve around network effects,
switching costs, and market entry. Are copyright’s existing affirma-
tive defenses the best tools for addressing these questions?
Probably not. In Professor Menell’s words: “The fair use doctrine
is an especially poor vehicle for resolving API copyright disputes.” 69
In some interface disputes, the fair use doctrine may leave room for
the relevant policy considerations. In Oracle v. Google, for example,
the Android operating system was transformative in nature (using a
tiny portion of the Java API interface as part of a much larger, more
ambitious operating system in a different market) and did not compete
with the copyrighted work, the Java SE software. These characteris-
tics not only track the first and fourth fair use factors, 70 but also point
up the switching costs faced by Java developers and the innovation
unlocked thanks to Google “building a bridge” between the Java and

67. See Lotus, 49 F.3d at 812; Jury Verdict at 1, Synopsys, Inc. v. ATopTech, Inc., No.
13-CV-02965 (N.D. Cal. filed Mar. 10, 2016) (ECF No. 687).
68. See Mullin, supra note 40.
69. Menell, supra note 2, at 471.
70. The first fair use factor addresses the “purpose and character of the use” (including
whether the use is transformative) and the fourth fair use factor addresses “the effect of the
use upon the potential market for or value of the copyrighted work” used. 17 U.S.C. § 107
(2012).
532 Harvard Journal of Law & Technology [Vol. 31

Android platforms. 71 But fair use did not carry the day for Arista in its
battle over the use of Cisco’s multiword commands. And what about
cases like SAS v. WPL or GDC v. Dolby, where the defendants are
direct competitors and their products do not easily fit the category of
“transformative”? In these cases, the switching cost and network ef-
fects questions loom equally large, but the fair use factors seem to de-
emphasize these impacts on customers in favor of a focus on potential
harms to the copyright owner.
Scènes à faire also does not seem to reliably grapple with the so-
cial welfare impacts of protection for interfaces. In Cisco v. Arista, it
seemed that the scènes à faire doctrine may have encompassed con-
cerns about switching costs and network effects, insofar as Arista in-
troduced evidence demonstrating that many of the Cisco commands
stemmed from older conventions widely used by networking engi-
neers, leaving the options for alternatives limited. 72 These facts indi-
rectly relate to the notion that Cisco should not be entitled to use
preexisting conventions to block Arista’s entry into the market. Had
the litigation in GDC v. Dolby moved ahead, scènes à faire might
have similarly been adequate, insofar as the commands were short and
largely defined by prior conventions in the theater systems market.
But the relationship between the economic concerns raised in these
cases — switching costs and network effects — and scènes à faire is,
at best, indirect. And cases like SAS v. WPL show how scènes à faire
might fall short in vindicating the economic policies that Professor
Menell identifies. In that case, the SAS input and output formats were
less the product of well-established convention and limited options
(the concerns of scènes à faire), but rather the product of the SAS lan-
guage and the leading position that SAS had established over many
years.
The drawbacks posed by reliance on affirmative defenses to re-
solve copyright disputes over software interfaces, taken together, may
have yet another unfortunate consequence: tilting the playing field
toward big, well-resourced companies. While federal civil litigation is
always expensive, a reliance on affirmative defenses will make it
harder for defendants to prevail at an early stage. As described above,
forcing defendants to endure the entire process of discovery and trial
before a jury will make interface disputes more expensive, drawn out,
and unpredictable. This, in turn, will favor those companies who can
weather the expense and risks of unpredictable, drawn out federal liti-
gation. These realities will likely reduce the number of companies

71. Menell, supra note 2, at 470 (“Using some of the Java APIs provided a bridge for the
millions of Java programmers.”).
72. See Def. Arista Networks, Inc.’s Notice of Mot. and Mot. for Partial Summary Judg-
ment at 9–10, Cisco Sys. Inc. v. Arista Networks, Inc., No. 14-CV-05344 (N.D. Cal. filed
June 30, 2016) (ECF No. 329) (citing expert reports).
Special Issue] Copyright and Software 533

willing to risk disruptive market entry against well-heeled incumbents

with sizeable litigation war chests. Seen in this light, perhaps it is no
surprise that Dolby and GDC settled their dispute with a cross-
licensing arrangement: they have achieved peace while sending a
strong message to any new market entrant.

V. CONCLUSION

In the final analysis, the new wave of interface cases fit into the
historical pattern identified by Professor Menell and bear out the wis-
dom of his ultimate conclusions: “Leaving API design specifications
outside of copyright protection enables entrepreneurs seeking to im-
prove on successful platforms to build bridges for users and pro-
grammers. This avoids excess inertia and accommodates creative
destruction and evolution in those areas where the proprietor of the
standard platform lacks patent protection.” 73

73. Menell, supra note 2, at 468.