Florida State University Law Review Florida State University Law Review
Volume 39 Issue 3 Article 3
2012
Anticompetitive Product Design in the New Economy Anticompetitive Product Design in the New Economy
John M. Newman
Follow this and additional works at: https://ir.law.fsu.edu/lr
Part of the Law Commons
Recommended Citation Recommended Citation
John M. Newman,
Anticompetitive Product Design in the New Economy
, 39 Fla. St. U. L. Rev. (2012) .
https://ir.law.fsu.edu/lr/vol39/iss3/3
This Article is brought to you for free and open access by Scholarship Repository. It has been accepted for
inclusion in Florida State University Law Review by an authorized editor of Scholarship Repository. For more
information, please contact efarrell@law.fsu.edu.
FLORIDA STATE UNIVERSITY
L
AW REVIEW
A
NTICOMPETITIVE PRODUCT DESIGN IN THE NEW ECONOMY
John M. Newman
V
OLUME 39
S
PRING 2012
N
UMBER 3
Recommended citation: John M. Newman, Anticompetitive Product Design in the New
Economy,
39 FLA. ST. U. L. REV. 681 (2012).
ANTICOMPETITIVE PRODUCT DESIGN IN THE NEW
ECONOMY
JOHN M. NEWMAN
ABSTRACT
Claims alleging anticompetitive product design and redesign lie at the very core of one
of antitrust law’s most challenging dilemmas: the intersection between innovation and regu-
lation, invention and intervention. For over three decades, courts and scholars have strug-
gled to determine the proper analytical framework within which to address such cases.
Meanwhile, the very industries in which challenged conduct occurs have been undergoing
fundamental changes.
As demonstrated by the ongoing and recent antitrust litigation involving high-
technology firms Apple, Intel, and Microsoft, distinctive features characterize most product
markets in what has been called the “New Economy”—and what increasingly has become
simply “the economy.” Many of these features not only uniquely incentivize anticompetitive,
design-related conduct but also render such conduct uniquely susceptible to antitrust scru-
tiny. Accordingly, this Article both supplies a proper understanding of code-based product
markets and, perhaps more importantly, provides a structured, efficient, and rational meth-
od for analyzing design-related conduct in those markets.
I.
I
NTRODUCTION ................................................................................................ 682
II. TECHNOLOGY-INTENSIVE AND CODE-BASED PRODUCT MARKETS ................... 685
A. The Operation of the Markets: Stable Network Markets,
Innovation-Based Rivalry, or Both? ........................................................... 688
1. Network Markets and Effects ............................................................... 688
(a) Positive Network Externalities ..................................................... 688
(b) The “Side Effects” of Network Markets ......................................... 689
2. Innovation-Based Rivalry .................................................................... 691
B. Antitrust Harm in Code-Based Product Markets ...................................... 693
III. ANTICOMPETITIVE PRODUCT (RE)DESIGN IN CODE-BASED
PRODUCT MARKETS ............................................................................................. 695
A. Design and Redesign: Computer Programming and Software Updating . 695
1. The Nature of Computer Code and the Products It Underlies ............. 695
2. Redesigning Code-Based Products ....................................................... 697
3. The Apple iPod iTunes Antitrust Litigation ........................................ 697
(a) The Audio-File Market ................................................................. 698
(b) Apple Closes the Network ............................................................. 699
B. Software Updates as a Uniquely Attractive Method of Using Product
Redesign to Foreclose Rivals ...................................................................... 703
1. Relatively Low Design-Related Costs to Firms: Lower C
i
Incentivizes
Anticompetitive Product Redesigns ..................................................... 705
(a) Low Development Costs Relative to Physical Product Redesign .. 705
(b) Speed, Ease, and Low Costs to Firms of Online
Software-Update Distribution ....................................................... 706
2. Relatively Low Risk That Consumers Will Reject Redesigns: How
Lower R Further Incentivizes Anticompetitive Product Redesigns ..... 707
(a) Relatively Low Cost to Consumers Lowers R ............................... 708
(b) The Trend Toward Automatic Software Updates ........................ 708
Trial Attorney, U.S. Department of Justice, Antitrust Division. The views ex-
pressed herein are solely those of the author and do not necessarily reflect those of the
Department of Justice. I owe an enormous debt of gratitude to Professor Herbert
Hovenkamp for his invaluable expertise and insight. Finally, for participating in what
must have seemed like innumerable discussions of this paper, I am grateful to Andrew
Tran, Bryan Sullivan, and Rachel Howard.
682 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
3. Relatively Low Losses Incurred if Product Redesigns Fail: How
Lower L Further Incentivizes Anticompetitive Redesigns ................... 710
C. Code-Based Product Redesign as Uniquely Susceptible
to Antitrust Scrutiny ................................................................................. 711
1. Software Updates Are Conceptually Separable from the
Base Product ........................................................................................ 711
2. Lines of Code Are Separable from a Software Redesign as a Whole ... 712
3. Code Can Be Written Different Ways to Perform the Same Function.. 713
IV. PRODUCT DESIGN AND TECHNOLOGICAL TIES:EARLY ANALYTICAL
APPROACHES THROUGH THE DIGITAL AGE........................................................... 714
A. The IBM Plug-Compatible Peripherals Litigation .................................... 715
B. Berkey Photo, Inc. v. Eastman Kodak Co. ................................................ 717
C. C.R. Bard, Inc. v. M3 Systems, Inc............................................................ 718
D. United States v. Microsoft Corp. ............................................................... 720
E. Allied Orthopedic Appliances, Inc. v. Tyco Health Care Group LP .......... 722
F. In re Intel ................................................................................................... 723
V. LOOKING FORWARD:ASUGGESTED ANALYTICAL FRAMEWORK ....................... 725
A. Defendant-Friendly Approaches Are Inappropriate .................................. 725
B. “Actual Improvement” and “Reasonableness”............................................ 728
1. “Actual Improvement”: Unfocused ....................................................... 728
2. Reasonableness: Overly Vague ............................................................. 729
C. Intent: Relevant, but Not Dispositive ......................................................... 729
D. Consumer Acceptance, Not Preference: A Constrained
and Uninformed “Choice” .......................................................................... 730
E. Shifting Burdens, Yet Avoiding Both “Balancing” and a
Defendant-Friendly Default Rule .............................................................. 731
VI. CONCLUSION .................................................................................................. 733
I. INTRODUCTION
“Innovation” has come to be broadly regarded as one of the prima-
ry factors in achieving increased consumer and societal wealth.
1
As
such, it has also become an increasingly prized goal of antitrust law.
2
And, consequently, antitrust courts have traditionally been hesitant
to condemn unilateral conduct related to product design and redesign
for fear that such decisions will place an undue chill on innovation in
the marketplace.
3
1. See J. Gregory Sidak & David J. Teece, Dynamic Competition in Antitrust Law, 5
J. C
OMPETITION L. & ECON. 581, 603 (2009) (“Industry after industry can demonstrate
gains from dynamic (innovation-driven) competition that overshadow the gains when com-
petition is present but innovation is absent.”); Gaia Bernstein, In the Shadow of Innova-
tion, 31 C
ARDOZO L. REV. 2257, 2258 & n.3 (2010) (“We put brilliance, creativity, and inno-
vation up on a pedestal, often tying these qualities to progress and overall improvement in
human welfare.”).
2. See, e.g., U.S. D
EP’T OF JUSTICE &FED.TRADE COMM’N, HORIZONTAL MERGER
GUIDELINES 23-24 (2010), available at http://www.ftc.gov/os/2010/08/100819hmg.pdf (de-
voting an entire subsection to “Innovation and Product Variety” and stating that enforce-
ment agencies account for effects on innovation in conducting merger reviews).
3. See, e.g., Berkey Photo, Inc. v. Eastman Kodak Co., 603 F.2d 263 (2d Cir. 1979); In
re IBM Peripheral EDP Devices Antitrust Litig., 481 F. Supp. 965, 1003 (N.D. Cal. 1979),
aff’d sub nom. Transamerica Computer Co. v. IBM Corp., 698 F.2d 1377 (9th Cir. 1983).
2012] ANTICOMPETITIVE PRODUCT DESIGN 683
The archetypical design-conduct challenge alleges that a firm,
dominant in one product market, designed a new version of that
product so as to maximize interoperability with its own complemen-
tary product(s), essentially requiring customers to buy the two to-
gether. Thus, the theory goes, the defendant either engaged in “fore-
closure” (excluding rivals who make interoperable complementary
goods from the market), “leveraging” (extracting overcharges from
customers in the complementary-good market), or both. Such conduct
is often referred to as “technological tying” because of its conceptual sim-
ilarity to contractual tying.
4
And as with contractual tying, a healthy
debate surrounds the viability of claims that it is anticompetitive.
5
Yet, perhaps because the early cases consisted of similar fact pat-
terns, too much emphasis has been placed on this resemblance to
traditional tying, which tends to focus on the controversial “leverag-
ing” theory of antitrust harm. The possibilities for firms to engage in
anticompetitive product design and redesign are broader than tech-
nological tie-ins, particularly in markets for digital, or computer
code-based, products. Take, for instance, a dominant manufacturer of
central processing units (CPUs) for personal computers that also cre-
ates “compilers,” software programs that expert computer program-
mers use in developing new software.
6
If the firm were to design or
redesign its compilers such that the resulting software runs slower
on personal computers that use rivals’ CPUs, its conduct would have
the effect of prolonging its CPU monopoly by foreclosing those rivals
from the market. And if the dominant firm did so with no procompet-
itive justification, its behavior was anticompetitive.
Thus, this Article looks beyond conduct involving technologically
tied products and more generally toward product design and rede-
sign
7
in the product markets that characterize what has been called
4. See, e.g., Charles M. Gastle & Susan Boughs, Microsoft III and the Metes and
Bounds of Software Design and Technological Tying Doctrine, 6 V
A.J.L.&TECH. 7 (2001).
5. Compare Einer Elhauge, Tying, Bundled Discounts, and the Death of the Single
Monopoly Profit Theory, 123 H
ARV.L.REV. 397, 399 (2009) (“Even without a substantial
foreclosure share, tying by a firm with market power generally increases monopoly profits
and harms consumer and total welfare, absent offsetting efficiencies.”), with Erik
Hovenkamp & Herbert Hovenkamp, Tying Arrangements and Antitrust Harm, 52 A
RIZ.L.
R
EV. 925, 927 (2010) (“The great majority of ties are beneficial or at least benign, measured
by either welfare standard, and this is true without even considering production efficien-
cies that many ties produce.”).
6. This fact pattern is broadly drawn from In re Intel Corp., No. 9341 (F.T.C.), but it
is not intended to pass judgment on the defendant’s behavior in that case.
7. Another term frequently used to describe anticompetitive, design-related conduct
is “predatory innovation.” See, e.g., Joseph Gregory Sidak, Debunking Predatory Innova-
tion, 83 C
OLUM.L.REV. 1121 (1983). As a label, however, it is (to at least some extent)
conclusory, “[a]nd conclusory labels fall prey to hopeless circularity.” Hon. Richard D.
Cudahy & Alan Devlin, Anticompetitive Effect, 95 M
INN.L.REV. 59, 59 (2010). More specif-
ically, the term itself begs the question by assuming that behavior can be at once both
“predatory” and truly “innovative.” Yet, as argued infra, in the product markets under
discussion herein, this is rarely—and indeed, likely never—the case.
684 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
the “New Economy”
8
—and what increasingly has become simply “the
economy.”
9
Here, the focus is on the destruction of interoperability
between a dominant firm’s products and those of its rivals—
foreclosure, rather than leveraging.
10
The product markets that dom-
inate the new economy are technology intensive and based on intan-
gible computer code, unlike the physical-based product markets tra-
ditionally subject to antitrust scrutiny.
11
And, I argue, these new
markets have distinctive features that not only uniquely incentivize
anticompetitive, design-related conduct but also render such conduct
uniquely susceptible to antitrust scrutiny.
Part II of this Article addresses the functioning of code-based
product markets, focusing on their tendency to be characterized by
positive network externalities. It contends that this tendency, com-
bined with a number of other factors, serves to refute the argument
that antitrust scrutiny is wholly inappropriate in these markets. Part
II concludes with a discussion of some general principles of anticom-
petitive harm in new-economy product markets.
In Part III, I begin by first discussing the mechanics of designing
and redesigning code-based software products. Next, I introduce the
ongoing Apple FairPlay antitrust litigation as an example of how
software updates, the principal method of redesigning software pro-
grams, can be used by a dominant firm to exclude rivals from a closed
network market. Part III concludes by demonstrating that software
updates are a uniquely attractive method of using product redesign
to foreclose rivals and that code-based product redesign is uniquely
susceptible to antitrust scrutiny.
Part IV of this Article begins by assessing the landmark antitrust
product-design decisions, underlining the various approaches used by
courts in addressing design claims, and foreshadowing the ineptitude
of the various analytical frameworks used to date for addressing such
8. See, e.g., THE NEW ECONOMY AND BEYOND:PAST,PRESENT AND FUTURE (Dennis
W. Jansen ed., 2006);
James Crowe, Regulation and Free Markets: How To Regulate the
Telecommunications Industry in the New Economy, 2 J.
ON TELECOMM.&HIGH TECH. 429
(2003); Richard A. Posner, Antitrust in the New Economy, 68 A
NTITRUST L.J. 925 (2001).
9. See Hon. Orrin G. Hatch, Antitrust in the Digital Age, in C
OMPETITION,INNOVA-
TION AND THE
MICROSOFT MONOPOLY:ANTITRUST IN THE DIGITAL MARKETPLACE 19, 22
(Jeffrey A. Eisenach & Thomas M. Lenard eds., 1999) (“High technology is the single larg-
est industry in the United States, leading all other sectors in terms of sales, employment,
exports, and research and development.”).
10. For an excellent discussion of foreclosure strategies engaged in by intellectual
property owners, as well as the harms to competition and innovation that may be caused
thereby, see Christina Bohannan, IP Misuse as Foreclosure, 96 I
OWA L. REV. 475 (2011).
11. See Dan Hunter, Culture War, 83 T
EX.L.REV. 1105, 1107 (2005) (“As the modern
era advanced, the importance of industrial production waned. No longer was heavy ma-
chinery and physical plant the predominant means of production; no longer was physical
inventory the most important asset to industry.”).
2012] ANTICOMPETITIVE PRODUCT DESIGN 685
conduct. It also serves to reemphasize the burgeoning potential for anti-
competitive product design and redesign in code-based product markets.
Finally, Part V examines and rejects the various standards em-
ployed by courts and advanced by scholars as inappropriate for as-
sessing design-related claims involving code-based products. In their
stead, it offers a structured, rational, and efficient analytical frame-
work for assessing design-related claims. Crucially, this method of
analysis allows the avoidance of both an overly defendant-friendly
default rule and the inherent uncertainty of “balancing” anticompeti-
tive effects against innovative justifications.
II. T
ECHNOLOGY-INTENSIVE AND CODE-BASED PRODUCT MARKETS
“Whatever faults this world has, it at least cannot be charged with
being petrified; on the contrary it transforms itself unceasingly.”
--Charles Gide
12
Much has been said about the “New Economy,” the paradigm
shift
13
away from an economy based on physical goods and networks
to digital ones.
14
What, exactly, is meant by this phrase tends to dif-
fer depending on who is using it. Generally speaking, it refers to a
trend toward an information-based economy and away from the
heavy focus on physical manufacturing and industrialization that
characterized much of the early- to mid-twentieth century.
15
It en-
compasses primarily digital technological innovations that have al-
lowed for sizable increases in productivity
16
and an accompanying
12. LAURENCE GRONLUND,THE NEW ECONOMY:APEACEABLE SOLUTION OF THE SO-
CIAL
PROBLEM 13 (1898).
13. The term “paradigm shift” originated in Thomas Kuhn’s seminal The Structure of
Scientific Revolutions and described a dramatic change in the paradigms underlying a
prevailing theory of scientific study. See T
HOMAS S. KUHN,THE STRUCTURE OF SCIENTIFIC
REVOLUTIONS 85 (3d ed. 1996). It has, however, subsequently entered the popular lexicon
as a phrase more loosely describing any major change in political, social, artistic, or com-
mercial structures, and it is in this latter sense that I employ the term.
14. See generally Turgut Ayhan Beydogan, Interoperability-Centric Problems: New
Challenges and Legal Solutions, 18 I
NT’L J. L. & INFO.TECH. 301 (2010) (arguing that en-
suring interoperability is necessary to encourage innovation in emerging information-
technology markets); Geoffrey A. Manne & Joshua D. Wright, Google and the Limits of
Antitrust: The Case Against the Case Against Google, 34 H
ARV.J.L.&PUB.POL’Y 171, 171
(2011) (discussing antitrust enforcers’ “increased focus on innovative companies in high-
tech industries”); Posner, supra note 8, at 925 (using the phrase to refer to computer-
software manufacturing, Internet-based businesses, and information technology providers);
Philip J. Weiser, Innovation, Entrepreneurship, and the Information Age,
9J.
ON TELE-
COMM
.&HIGH TECH. L. 1 (2011) (addressing the role of antitrust enforcers in the “Infor-
mation Age”).
15. See Hunter, supra note 11, at 1107.
16. See J
ANSEN, supra note 8, at 2-3. The annual growth rate of average labor produc-
tivity declined precipitously from 1974 to 1995, averaging 1.44% per year during that peri-
od. Yet from 1995 to 2000, the growth rate averaged 2.36% per year, an increase attributa-
ble largely to advances in information technology. Id.
686 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
shift in the number and type of occupations available and sought af-
ter in the workplace.
17
And it identifies a change in the way everyday
consumers purchase—or decline to purchase—products, a change
spurred on by the speed and ease of the transmission of information
allowed by these new digital technologies. Witness, for example, the
rapid rise of online retailing, a means of commerce entirely unknown
to firms and consumers only a few short decades ago.
18
Yet the degree to which this economy is really “new,” and the im-
plications the answer to that question holds for antitrust law and pol-
icy, is unclear. Some argue that the concept of a “new” economy is
merely, or at least largely, the product of semantics.
19
And it is true
the phrase itself is hardly new, having appeared in literature at least
as far back as the late 1800s.
20
“But it is safe to say that the economy
has changed in the last decade or so, and that these changes are of
sufficient importance to deserve the title ‘New Economy.’ ”
21
And the
changes that have occurred are due primarily to changes in the ways
in which individuals in society access, process, store, and communi-
cate information.
22
Those changes have been facilitated by two pri-
mary technological developments: (1) the advent of affordable, per-
sonal, digital computing devices and (2) the rise of inexpensive, high-
speed, readily available access to the Internet
23
utilizing those devices.
17. See Hatch, supra note 9, at 22.
18. In 2008, e-commerce in retail-trade industries totaled $141 billion. U.S. C
ENSUS
BUREAU,STATISTICAL ABSTRACT OF THE UNITED STATES: 2011, at 662 tbl.1054 (2011),
available at http:/www.census.gov/compendia/statab/2011/tables/11s1054.pdf. At only 3.6%
of total sales, id., online retailing has certainly not yet come to dominate these markets.
Yet considering that, in recent memory, retail revenues from e-commerce totaled $0.00, the
$141 billion figure is substantial.
19. See J
EAN GADREY,NEW ECONOMY,NEW MYTH 1 (2003). As Professor Gadrey
memorably stated:
We have entered a new age, we are told. Its characteristics are ill-defined, the
issues at stake uncertain, but we are told it is new and that we have to go along
with it. The myths put about to stir people into action are flourishing, but little
serious thought is being given to the general nature of this new age or to the
social risks we might have to guard against. Trenchant turns of phrase and
portmanteau words seem to be sufficient. However, these new words and the
old ones have many points in common.
Id.
20. Attorney (and ardent socialist) Laurence Gronlund used the phrase in 1898 as the
title of his major work, a cry that “[s]omething . . . must be done!” about the emergence of
an unrivalled threat to workers: the Trust. G
RONLUND, supra note 12, at 1, 15.
21. J
ANSEN, supra note 8, at 2.
22. Cf. Posner, supra note 8, at 925 (“I shall use the term ‘new economy’ to denote . . .
the manufacture of computer software[,] . . . Internet-based businesses[,] . . . [and] commu-
nications services and equipment designed to support the first two markets.”).
23. The increase in the number of Americans with ready Internet access has been
unbelievably precipitous—take, for example, the second half of the year 2000 when “the
number of American adults with Internet access grew from about 88 million to more than
104 million.” L
EE RAINIE &DAN PACKEL,PEW INTERNET PROJECT,MORE ONLINE,DOING
2012] ANTICOMPETITIVE PRODUCT DESIGN 687
These two developments have given rise to a wealth of new indus-
tries that differ widely from the paradigmatic firms around which
antitrust doctrine grew and developed. Traditional antitrust targets
were industries that manufactured or were based upon physical
products: oil, steel, railroads, tobacco, etc.
24
These industries tended
to be relatively stable; entry and exit (and even substantial market
share shifts) were rare.
25
Entry entailed relatively high sunk costs,
and economies of scale offered existing market participants substan-
tial advantages over potential entrants.
26
These firms also incurred
both fixed and variable costs throughout their production cycle, with
average variable cost rising even at relatively low levels of output.
On the whole, these tended not to be characterized by positive net-
work externalities (aside from the railroads and—of course—the tele-
communications industry, the archetypical network industry). And
the pace of innovation was modest.
27
The nascent industries that have been built around digital com-
puters and the Internet, on the other hand, use intangible—instead
of physical—means as the dominant means of production.
28
The
products and services they offer increasingly consist largely (or even
solely) of computer code. As a result, intellectual property protections
have become an increasingly crucial incentive to firms.
29
These mar-
kets frequently exhibit positive network externalities
30
and are char-
acterized by high fixed costs and very low variable costs (often ap-
proaching zero). The barriers to creating entirely new products tend
to be lower,
31
and innovations—perhaps as a result of this—appear to
be more frequent.
MORE 2 (2001), available at http://www.pewinternet.org/~/media//Files/Reports/2001/
PIP_Changing_Population.pdf.
24. Cf. Alan Devlin, A Neo-Chicago Perspective on the Law of Product Tying, 44 A
M.
B
US. L.J. 521, 560-61 (2007) (“[T]raditional antitrust jurisprudence has been aimed at
those industries with eventual rising average costs in production.”).
25. See Posner, supra note 8, at 926.
26. Chris Butts, Note, The Microsoft Case 10 Years Later: Antitrust and New Leading
“New Economy” Firms, 8 N
W.J.TECH.&INTELL.PROP. 275, 275 (2010).
27. Posner, supra note 8, at 926.
28. See Bernstein, supra note 1, at 2262.
29. Id. (“As the dominant means of production changed from the physical to the in-
tangible, the determination of the appropriate scope of intellectual property rights became
more crucial.”). Of course, U.S. intellectual property regimes do not perfectly incentivize
innovation in high-technology (or any other) industries. The disclosure requirement of pa-
tent law, for instance, can allow competitors to “design around” innovative firms’ patented
products. “Moreover, . . . competitors may actually gain an advantage by not being the first
to introduce a product since they can then free-ride on the research-and-development in-
vestments of the original firm to produce a better product.” Damon C. Andrews, Why Pa-
tentees Litigate, 12 C
OLUM.SCI.&TECH.L.REV. 219, 230 (2011).
30. See Posner, supra note 8, at 926 (referring to positive network externalities as
“economies of scale in consumption”).
31. See L
AWRENCE LESSIG,THE FUTURE OF IDEAS:THE FATE OF THE COMMONS IN A
CONNECTED WORLD 121 (2001) (“The barriers of cyberspace in its natural state are radical-
ly different from the barriers in real space.”).
688 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
A. The Operation of the Markets: Stable Network Markets,
Innovation-Based Rivalry, or Both?
As noted above, the product markets that have emerged in the
new economy are both dynamic and different from the industries that
dominated the economy in the past. Multiple explanations have
emerged for this dynamism; perhaps the two foremost (and compet-
ing) theories advanced are the impact of network effects and innova-
tion-based, or “Schumpeterian,” rivalry. Both have a part to play in
determining antitrust’s role in evaluating product design and rede-
sign in these markets.
1. Network Markets and Effects
Network markets are those markets in which one user of a good or
service impacts, through her use, the value of that good or service to
other users. As an initial matter, a network may be either “closed” or
“open.” A “closed” network is controlled by one firm, which limits full
access to the network.
32
An “open” network, on the other hand, may
be fully accessed by all who wish to do so. All networks, however, are
characterized by “network effects,” a term that is sometimes used in
place of “positive network externalities” (described infra) but more
properly includes all effects one user can have on the value of a prod-
uct to other users in a network market.
Network effects can be direct or indirect. “Direct” network effects
occur where an additional use or user causes a direct impact on the
value of the product to other users. “Indirect” network effects, on the
other hand, occur where an increase in usage incentivizes firms to
produce additional complementary goods, increasing the value of the
original product to all users.
(a) Positive Network Externalities
Network markets are frequently characterized by positive network
effects, or “externalities”—“benefits to society that accrue as the size
of a network grows.”
33
In a network market, the value to each indi-
vidual consumer increases with the number of other individuals who
use the same network.
34
Take, for example, a telephone network. If
very few telephones existed, I would not value a telephone very high-
ly, or at least not nearly as highly as I do today. The more individuals
who use the telephone network, the more individuals I am able to
32. It should be noted, however, that a closed network does not necessarily dominate
or constitute a product market.
33. Howard A. Shelanski & J. Gregory Sidak, Antitrust Divestiture in Network Indus-
tries, 68 U. C
HI.L.REV. 1, 7-8 (2001).
34. Another term sometimes used to describe this phenomenon is “economies of scale
in consumption.” See, e.g., R
ICHARD A. POSNER,ANTITRUST LAW 246 (2d ed. 2001).
2012] ANTICOMPETITIVE PRODUCT DESIGN 689
reach with my own telephone—and the more valuable my phone be-
comes to me. Thus, telephone networks allow for direct, positive net-
work externalities.
Many of the industries that have arisen from the advent of digital
computers and broadly available Internet access exhibit similar posi-
tive network externalities, both direct and indirect.
35
Online social
networking sites like Facebook allow for direct, positive effects (much
like the telephone industry). And a broad range of product markets
allow for indirect, positive network externalities. Take, for example,
Apple’s iPod: as more individuals purchase iPods, Apple is increas-
ingly incentivized to invest in adding songs to its iTunes Music Store,
thereby increasing the value of the iPods. The current market for
Blu-ray players operates similarly,
36
as does the market for Mi-
crosoft’s Windows operating system (OS). These positive network ex-
ternalities that tend to characterize new, digital industries can, how-
ever, have less benign implications.
(b) The “Side Effects” of Network Markets
If its product is not interoperable with rivals’ products, a first-
mover or firm with a relatively large market share at an early stage
in a network market can gain a decisive advantage over other firms
in the market simply by virtue of its early lead. This is so “because
consumers are attracted to the good or service that offers the largest
network benefit, and that benefit then only becomes larger and more
attractive to later consumers.”
37
Thus, a network owner is faced with
a choice. First, it can choose to allow interoperability with competi-
tors’ products (that is, choose an open network), thereby increasing
the number of users in the network and increasing the network’s
value to all users. On the other hand, it may refuse to interconnect
(i.e., choose a closed network), which “would lead to a race to become
the dominant player.”
38
Absent interoperability, consumers will de-
sire to be a part of the largest network, which naturally offers the
most value. Less risk-averse consumers will buy into the network
early on, predicting that more consumers will follow (increasing the
35. See id. at 245-46; Arun Sundararajan, Network Effects, IOIT,
http://oz.stern.nyu.edu/io/network.html (last visited Feb. 13, 2012) (“Today, [network ef-
fects] are widely recognized as a critical aspect of the industrial organization of IT indus-
tries, and are prevalent in a wide variety of sectors, including software, microprocessors,
telecommunications, e-commerce and electronic marketplaces.”).
36. It is perhaps worth venturing to predict that, as consumers move further and fur-
ther into the “cloud,” such physical-disk-based technology will eventually go the way of the
dinosaur and the VCR, demonstrating the disruptive effect that innovation can sometimes
play—even in network markets.
37. Shelanski & Sidak, supra note 33, at 8.
38. Daniel F. Spulber & Christopher S. Yoo, Mandating Access to Telecom and the
Internet: The Hidden Side of Trinko, 107 C
OLUM.L.REV. 1822, 1890 (2007).
690 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
value of the early consumers’ investment). At some point, consumers
will perceive that a certain product will become dominant and accord-
ingly flock to that product—a phenomenon known as the market’s
“tipping point.”
39
And “[o]nce the market reaches its tipping point, the
value of the network of the dominant player will so far outstrip that of
its competitors that the market collapses into a natural monopoly.”
40
The possibility of charging higher-than-marginal-cost prices acts
as a powerful incentive for firms to avoid interoperability and instead
attempt to become the dominant firm in a closed network market.
41
Thus, in network markets, the good or service that constitutes the
market
42
is frequently produced by only one firm and is frequently
not interoperable with competing goods or services. Consumers who
have bought the dominant firm’s product (that is, joined its network)
in order to obtain the larger value conferred by the larger network
would have to incur switching costs in order to purchase or begin us-
ing a competitor’s product.
43
And as a result, “consumers can become
‘locked in’ to a particular network,”
44
creating a barrier to potential
39. See, e.g., William J. Baer & David A. Balto, New Myths and Old Realities: Recent
Developments in Antitrust Enforcement, 1999 C
OLUM.BUS.L.REV. 207, 221 n.47 (“Particu-
larly in markets characterized by networks, an industry may reach a ‘tipping’ point where
manufacturers of complementary products adopt the dominant technical standard of those
products with which interface is necessary, or consumers adopt the product that is most
favored by other consumers.”).
40. Spulber & Yoo, supra note 38, at 1889. A relatively recent example of this phe-
nomenon was the high-definition optical-disc format war that took place from around 2006
to 2008. The HD DVD and Blu-ray formats struggled for dominance until the market
reached a tipping point early in 2008, with several major manufacturers announcing that
they were exclusively adopting the Blu-ray format. Very shortly thereafter, Toshiba issued
a press release indicating that it planned to discontinue the manufacture and sale of its
HD DVD players, effectively ceding the market to Blu-ray. See Press Release, Toshiba,
Toshiba Announces Discontinuation of HD DVD Businesses (Feb. 19, 2008), available at
http://www.toshiba.co.jp/about/press/2008_02/pr1903.htm.
41. Daniel F. Spulber & Christopher S. Yoo, Access to Networks: Economic and Consti-
tutional Connections, 88 C
ORNELL L. REV. 885, 922 (2003) (“[A] growing number of scholars
have raised the concern that network owners may be able to use interconnection (or, more
accurately, the refusal to interconnect) as an anticompetitive weapon.”).
42. More precisely, the output of a network is “the use of the network’s services.” Id.
at 894.
43. These costs would include any financial outlays, as well as time spent learning to
operate the new technology.
44. Shelanski & Sidak, supra note 33, at 9. Take, for example, Facebook’s social-
networking website. The value of the code-based product is entirely dependent on the
number of users in the network—and as of this writing, there are more than 900 million
active users. Fact Sheet,
FACEBOOK, http://newsroom.fb.com/content/
default.aspx?NewsAreaId=22 (last visited Feb. 13, 2012). Given that many users, the net-
work’s value is very substantial, as reflected in a recent valuation (completed in prepara-
tion for Facebook’s planned 2012 IPO) placing the Facebook corporation’s value at an esti-
mated $50 billion. Anupreeta Das, Geoffrey A. Fowler & Liz Rappaport, Facebook Sets
Stage for IPO Next Year, W
ALL ST. J. (Jan. 6, 2011), http://online.wsj.com/article/
SB10001424052748703730704576066162770600234.html. A potential rival, on the other
hand, would naturally start with zero users and would offer relatively negligible value to
potential users. Since Facebook is currently not interoperable with any other social-
2012] ANTICOMPETITIVE PRODUCT DESIGN 691
entrants. Consequently, the dominant firm may become entrenched.
Further, that firm’s design choices will determine the market’s tech-
nological standard—and later inventions will have to follow the
“path” set by the market leader. This holds true even if that path is
not the highest quality or most efficient possible.
45
2. Innovation-Based Rivalry
As noted above, another characteristic of the new economy seems
to be a more rapid pace of innovation than was seen in many of the
industries that traditionally fell under antitrust scrutiny. This has
given rise to scholars and commentators proclaiming that digital
markets are driven primarily by innovation, rather than static price
competition.
46
Their vision, which traces its roots directly to the writ-
ings of Joseph Schumpeter,
47
is of a marketplace periodically domi-
nated by innovative firms, which are in turn displaced by firms that
market the next wave of technology.
48
The core of the concept is that
“the pursuit of market power is a creative and dynamic force that ‘in-
cessantly revolutionizes the economic structure from within, inces-
santly destroying the old one, incessantly creating a new one.’ ”
49
Un-
der this view, though a firm might temporarily gain monopoly power
in a market, subsequent innovation by competitors will ensure that
its ability to charge supracompetitive prices will be short-lived.
50
And
networking websites, rivals would need some massive advantage to offer users in order to
enter the market. One potential rival, “PeopleString,” actually pays users; its market share
remains dwarfed by Facebook’s. See P
EOPLESTRING, http://peoplestring.com (last visited
Feb. 13, 2012).
45. See Shelanski & Sidak, supra note 33, at 9.
46. See, e.g., Weiser, supra note 14, at 3 (“The dynamics of today’s information age
have pushed economists further away from the classic, static focus on prices—which re-
mains an important part of economics, to be sure—to a greater appreciation for the impact
of innovation.”).
47. See generally J
OSEPH A. SCHUMPETER,CAPITALISM,SOCIALISM AND DEMOCRACY
(1942).
48. See, e.g., Robert W. Crandall & Clifford Winston, Does Antitrust Policy Improve
Consumer Welfare? Assessing the Evidence, 17 J.
E
CON.PERSP. 3, 23 (2003) (arguing that
there is no persuasive evidence that antitrust enforcement actually benefits consumers, in
part because it has difficulty addressing the “new economy characterized by dynamic com-
petition, rapid technological change and important intellectual property”).
49. Shelanski & Sidak, supra note 33, at 11 (quoting J
OSEPH A. SCHUMPETER,CAPI-
TALISM
,SOCIALISM AND DEMOCRACY 83 (1942)).
50. Schumpeter also famously suggested “that large firms and monopolists may be
more innovative than firms in competitive markets.” Jonathan B. Baker, Beyond Schum-
peter vs. Arrow: How Antitrust Fosters Innovation, 74 A
NTITRUST L.J. 575, 578 (2007). Per-
haps the primary argument in favor of this proposition is that large firms and monopolists
are more capable of funding research and development projects than are smaller firms. See
id. On the other hand, as Judge Learned Hand memorably stated in the seminal Alcoa
decision: “Many people believe that possession of unchallenged economic power deadens
initiative, discourages thrift and depresses energy; that immunity from competition is a
narcotic, and rivalry is a stimulant, to industrial progress; that the spur of constant stress
is necessary to counteract an inevitable disposition to let well enough alone.” United States
v. Aluminum Co. of America, 148 F.2d 416, 427 (2d Cir. 1945). Indeed, some have argued
692 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
as a result, adherents call for a reduced role of antitrust enforcement
in markets that seem to thrive on innovation-based rivalry.
51
It is true that the advent and continuing improvement of the digi-
tal computer and the now widespread availability of high-speed, mo-
bile Internet access has spawned many new advances—including
“products that could not, or would not, have been built before the
Net. Among these we could include the dynamically generated maps
with driving directions[,] massive translation engines . . . , and on-
line dictionaries covering hundreds of languages that otherwise
would not be available . . . .”
52
Computing power has historically in-
creased at such a rapid pace that a term, “Moore’s Law,” has been
coined to refer to its rate of development.
53
And the Internet has ex-
ponentially increased, and continues to increase, the speed with
which individuals can transfer large amounts of information.
54
Yet
many of the goods and services that may seem “new” in fact are not—
take, for example, e-mail, which is generally perceived as a product of
the 1990s, when in fact it was developed over fifty years ago.
55
Many
are merely digital versions of physical-based services, such as the
online retailing offered by big-box department stores like Nordstrom
and Macy’s. Such “new” markets are generally not characterized by
rapid rates of market-shifting innovation;
56
their business models are
not much different from the mail order businesses that advertise
through physical catalogs and have existed for over a century.
Further, it is dangerous to assume, based on past innovation in a
market, that that market will continue to be characterized by rapid
innovation. Take the telephone as an example: although it was rapid-
ly improved and implemented immediately following its invention in
1876, the basic, underlying technology then remained static for dec-
ades. And it is, of course, never certain that even innovative rivals
will unseat a dominant firm in a network market—giants like Face-
book, for example, have yet to be dethroned.
57
That is particularly
that, rather than market structure driving innovation, it is instead innovation that drives
structure. Sidak & Teece, supra note 1, at 596–98.
51. See, e.g., Crandall & Winston, supra note 48.
52. L
ESSIG, supra note 31, at 122 (citations omitted).
53. I
NTEL,MOORE’S LAW 1 (2005), available at ftp://download.intel.com/museum/
Moores_Law/Printed_Materials/Moores_Law_2pg.pdf.
54. See Christopher S. Yoo, The Changing Patterns of Internet Usage, 63 F
ED.COMM.
L.J. 67, 68 (2010) (“[The Internet] has revolutionized the way people communicate with one
another and obtain information . . . .”).
55. See Tom Van Vleck, The History of Electronic Mail, http://www.multicians.org/
thvv/mail-history.html (last updated Dec. 20, 2010).
56. But cf. Alan J. Meese, Price Theory, Competition, and the Rule of Reason, 2003 U.
I
LL.L.REV. 77, 139-40 (“Innovation includes the discovery of new organizational forms and
the application of old forms to new contexts.”).
57. Indeed, some predict that, given the extremely strong positive network externali-
ties that characterize social-networking websites and the fact that Facebook has hit a “crit-
ical mass” of users, “Facebook isn’t going to go away anytime soon.” Farhad Manjoo, Is
2012] ANTICOMPETITIVE PRODUCT DESIGN 693
true where the market leader refuses to allow interoperability, for an
innovation may not be enough to tempt users away from the largest
existing network (and the firm that controls it). Finally, “the existing
body of theoretical and empirical literature on the relationship be-
tween competition and innovation ‘fails to provide general support for
the Schumpeterian hypothesis that monopoly promotes either in-
vestment in research and development or the output of innovation.’
”
58
Given all of this, it is not at all certain that the monopolies created
in digital product markets will be fleeting or quickly destroyed by
subsequent innovations. Thus, Schumpeterian tenets alone—which
some contend would call for antitrust enforcement to play a limited
role in markets characterized by relatively rapid innovation—cannot,
and should not, dictate the degree of antitrust’s involvement in as-
sessing product design and redesign in new-economy markets.
59
B. Antitrust Harm in Code-Based Product Markets
Though the new digital product markets on the whole appear to be
characterized by a relatively rapid rate of innovation, they still allow
the potential for—and sometimes even uniquely incentivize—
anticompetitive behavior. As noted above, digital product markets
tend to be characterized by substantial positive network externali-
ties, whether direct or indirect.
60
The potential to capture monopoly
profits by becoming the dominant firm in a network market may en-
gender a “race to gain and to maintain dominance”—one that pro-
vides “motive[] to engage in anticompetitive conduct.”
61
And those
monopoly profits may be more substantial and enduring than would be
possible in other industries, given the barrier to entry created by a large,
locked-in network of users. Indeed, a “trend to monopoly occurs as a re-
sult of the ‘network effects’ present in most high technology markets.”
62
High-technology firms also tend to face different cost and profit
structures that may encourage them to pursue monopolization
Facebook a Fad? What Social Networks Will Look Like in Five Years, SLATE (Mar. 31,
2011), http://www.slate.com/id/2290039.
58. Joshua D. Wright, Antitrust, Multidimensional Competition, and Innovation: Do
We Have an Antitrust-Relevant Theory of Competition Now?, in C
OMPETITION POLICY AND
PATENT LAW UNDER UNCERTAINTY:REGULATING INNOVATION 228, 244 (Geoffrey A. Manne
& Joshua D. Wright eds., 2011) (quoting Richard J. Gilbert, Competition and Innovation, in
1 I
SSUES IN COMPETITION LAW AND POLICY 600 (W. Dale Collins ed., 2008)).
59. Cf. Thomas M. Lenard, Introduction and Overview, in C
OMPETITION,INNOVATION
AND THE MICROSOFT MONOPOLY:ANTITRUST IN THE DIGITAL MARKETPLACE, supra note 9,
at 3 (“The notion that rapid innovation, per se, makes the market for software immune
from anticompetitive conduct and/or monopolization is simplistic and misleading.”).
60. See supra notes 33-35 and accompanying text.
61. Shelanski & Sidak, supra note 33, at 9; see also 3B P
HILLIP E. AREEDA &HERBERT
HOVENKAMP, ANTITRUST LAW ¶ 776c, at 297 (3d ed. 2006) (“[N]etwork status may increase
both the incentive and the opportunities for certain kinds of anticompetitive behavior.”).
62. Thomas A. Piraino, Jr., A Proposed Antitrust Approach to High Technology Com-
petition, 44 W
M.&MARY L. REV. 65, 77 (2002).
694 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
through anticompetitive conduct. Digital products can be reproduced
extremely cheaply, often with marginal costs approaching zero.
Firms’ fixed costs, on the other hand, tend to be relatively high.
63
Thus, firms that compete in digital product markets “have a substan-
tial incentive to engage in conduct that will artificially extend the
duration of their market power.”
64
As to the harm that can be inflicted by anticompetitive behavior in
digital, code-based product markets, the governing principles are
much the same as in other industries. A dominant firm might, for
example, engage in tying practices, which can cause harm through
either foreclosure of rivals
65
or “leveraging”—the extraction of over-
charges from customers of complementary products.
66
Or it might en-
gage in predatory pricing, again foreclosing rivals (and ultimately
extracting overcharges, though in the short run consumers “benefit”
from the artificially low prices). And (most relevantly for purposes of
this Article) “[b]ecause the users of networks place a premium on
compatibility, a dominant firm can exclude rivals anticompetitively
by engineering incompatibilities between the dominant product and
the product offered by rivals.”
67
Thus, firms may be uniquely incen-
tivized to create a closed network (and reap monopoly profits if that
network dominates the market) by denying interoperability through
design changes—as Microsoft attempted to do by steering users of its
OS away from Netscape’s Navigator browser.
68
While the governing principles of harm are similar, the means
that predatory firms have at their disposal have changed. The possi-
bilities offered by digital product design open up a new range of anti-
competitive strategies, as demonstrated by the allegations in the
Federal Trade Commission’s complaint against Intel.
69
As digital
technology continues to evolve, new avenues for anticompetitive, de-
sign-related behavior will continue to develop.
63. See Posner, supra note 8, at 926-27 (“Intellectual property is characterized by
heavy fixed costs relative to marginal costs. It is often very expensive to create, but once it
is created, the cost of making additional copies is low, dramatically so in the case of soft-
ware . . . .”).
64. Piraino, supra note 62, at 95-96.
65. See Hovenkamp & Hovenkamp, supra note 5, at 926. “Foreclosure occurs when a
tie ousts or unreasonably limits the opportunities of rivals, typically in the tied product
market.” Id. at 927.
66. Id. (“The second type of antitrust harm is extraction, which involves overcharges
that purchasers of tied packages are forced to pay.”).
67. A
REEDA &HOVENKAMP, supra note 61, ¶ 776c, at 297.
68. United States v. Microsoft Corp. (Microsoft III), 253 F.3d 34, 63-64 (D.C. Cir.
2001).
69. Complaint of Petitioner at 2-5, In re Intel Corp., No. 9341 (F.T.C. Dec. 16, 2009),
available at http://www.ftc.gov/os/adjpro/d9341/091216intelcompt.pdf.
2012] ANTICOMPETITIVE PRODUCT DESIGN 695
III. ANTICOMPETITIVE PRODUCT (RE)DESIGN IN CODE-BASED
PRODUCT MARKETS
As noted above, the advent of the digital computer and the wide-
spread availability of Internet access has given rise to new indus-
tries—what some have called the “New Economy.”
70
Importantly, the
primary product created by the participants in these industries is
constructed from computer code, instead of the physical materials
that characterized more traditional industries.
71
Even companies sell-
ing physical goods online, like the big-box department stores men-
tioned above, offer a product that consists at least in part of the ease
of use and speed of transactions allowed by code-based programs.
72
The variety of code-based products offered today is mind-boggling.
From personal-use entertainment (for example, digitized music, film,
and literature, as well as video games) to educational software to
productivity-enhancing programs like word processors and spread-
sheet programs to social-networking websites—all are intangible,
code-based products.
73
Thus, understanding the implications of code-
based product design and redesign is crucial to understanding these
new industries and markets—and the ways in which dominant mar-
ket participants can engage in anticompetitive design changes.
A. Design and Redesign: Computer Programming and
Software Updating
1. The Nature of Computer Code and the Products It Underlies
Computer code can be broadly divided into “source code” and “ob-
ject code.” “Source code” generally refers to the text produced by a
human computer programmer, written in a programming language.
74
“Object code” (or “machine code”) refers to the binary data that a
computer actually executes; it is unreadable by humans.
75
Program-
ming languages vary in their level of abstraction from machine
70. See supra notes 7-10, 13-31 and accompanying text.
71. See Posner, supra note 8, at 926 (“A key to understanding these industries is that,
with the partial exception of communications equipment, their principal output is intellec-
tual property, namely computer code, rather than physical goods.”).
72. Cf. id. (stating that “Internet-based businesses, even those that sell physical
products, such as groceries and books,” are offering a service that “is a function to a large
extent both of the sophistication of the business’s computer software and of its trademarks
and copyrights”).
73. See generally J.B. D
IXIT,COMPUTER PROGRAMMING 26-27 (2d ed. 2007) (describing
some of the different types of software available).
74. See Source Code, T
ECHTERMS.COM, http://www.techterms.com/definition/sourcecode
(last updated Oct. 11, 2007).
75. Thus, digital computers process “digital” data—that is, binary code. See D
IXIT,
supra note 73, at 2.
696 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
code.
76
Thus, at low levels of abstraction, code is much closer to ma-
chine code than to human language; conversely, high-level languages
tend to be easier to understand and use.
77
As a result, the more ab-
stract (that is, high-level) a language is, the fewer technical details
the programmer must remember and input—but the less efficiently
the resulting program tends to run. Low-level languages, on the oth-
er hand, tend to allow efficient programming, a necessity where com-
putational ability and speed are of high importance.
78
The source code that underlies computer programs is written line
by line, not necessarily as a cohesive whole—what looks to the un-
trained eye like a paragraph-length section may actually contain
multiple sections of code that do not relate to each other.
79
Frequent-
ly, programmers will begin by setting out definitions of variables and
functions that will be used throughout the program. The remainder
of the program will consist of sections of code that perform certain
tasks,
80
referring back to the internally defined variables and func-
tions (and sometimes to each other). Thus, a programmer may be able
to look at a single section of code and explain its immediate function,
though it may require a more holistic understanding of the program to
explain that function fully if that section is highly interrelated to other
sections. On the other end of the spectrum, however, the purpose and
functionality of a single section of code that is relatively less interre-
lated to other sections may be much more readily apparent.
Source code by itself cannot be read by a computer.
81
Thus, once
written by a programmer, high-level source code must be run
through a compiler, a software program (itself consisting of code) that
translates an entire program written in human-readable source code
into a program written in machine-readable object code.
82
The result-
76. See generally Programming Languages, RZERO, http://alternatives.rzero.com/
lang.html (last visited Feb. 13, 2012) (explaining and comparing various programming
languages).
77. The lowest level programming language is known as “assembly language.” See id.
78. Id. (“[I]f you need speed and/or compactness above all else, Assembly is the solu-
tion.”).
79. In fact, though it is not common, a single program may be written in different
programming languages.
80. While a smaller program may draw from a single source code file, most larger
programs reference hundreds or thousands of separate source code files. See Source Code,
supra note 74.
81. See G
EORGES IFRAH,THE UNIVERSAL HISTORY OF COMPUTING:FROM THE ABACUS
TO THE
QUANTUM COMPUTER 309 (E.F. Harding trans., John Wiley & Sons, Inc. 2001)
(1994) (“The written programme is then interpreted by a reader connected to the computer,
its contents being translated into the machine’s internal language . . . . The computer may
then carry out the programme . . . .”).
82. Cf. id. (using the term “reader” in place of “compiler”).
2012] ANTICOMPETITIVE PRODUCT DESIGN 697
ing product is a file executable by computers.
83
Consequently, code-
based products are generally distributed to users solely in the form of
object, rather than source, code. Although the primary method of dis-
tributing these code-based products to users was once physical disks,
distribution via the Internet has become increasingly prevalent. Vid-
eo games, for instance, were once sold primarily in brick-and-mortar
stores like GameStop on physical disks. The same games, however,
are now sold (by some of the same firms, including GameStop) via
websites that allow users to download the files directly to their com-
puters’ hard drives.
84
2. Redesigning Code-Based Products
Like any other product, a software program is frequently not per-
fect, particularly in its early iterations.
85
Software programs, howev-
er, can also be retroactively “redesigned” through software updates.
These updates are typically distributed to users in the form of exe-
cutable files written in object code that modify the software program
itself.
86
Software can be, and is, redesigned via updates for myriad
reasons—from adding new features, to fixing “bugs” in the software
that cause it to run poorly or inefficiently, to ensuring security from
malicious hackers and viruses, to external factors like an upgrade to
a new OS.
87
As the In re Intel Corp.
88
and United States v. Microsoft
Corp. (Microsoft III)
89
cases demonstrate, however, software design
can also be anticompetitive
90
—and software redesign through updat-
ing is no exception.
3. The Apple iPod iTunes Antitrust Litigation
On January 3, 2005, a class action suit was filed against Apple
Computer, Inc. alleging (among other things)
91
that Apple had violat-
83. An “interpreter” is a program (somewhat similar to a compiler) that instead trans-
lates source code into object code one statement at a time, then immediately executes that
statement. D
IXIT, supra note 73, at 28.
84. See, e.g., Downloads, G
AMESTOP, http://www.gamestop.com/downloads (last visit-
ed Feb. 13, 2012). And the future may be in the “cloud”—already, the technology exists for
(among other things) playing video games on remote servers, making the physical platform
on which a game is played largely irrelevant to individual users’ enjoyment of it.
85. See The Irrepressible Reasons for Upgrading Software, I
TOMIC (Mar. 12, 2010),
http://www.itomic.com.au/article/The-Irrepressible-Reasons-for-Upgrading-Software (“The
need to update software is almost as inevitable as death and taxes.”).
86. Updates do so in one of two ways: they either (1) completely replace part of the
object code underlying the program or (2) simply insert additional object code into the pro-
gram’s object code.
87. See id.(discussing the primary reasons software updates are issued).
88. See infra Part IV.F.
89. See infra Part IV.D.
90. See supra notes 68-69, and accompanying text.
91. The plaintiffs also initially included a section 1 claim, which was subsequently
dismissed. See The Apple iPod iTunes Antitrust Litigation, C 07-6507 JW, 2010 WL
698 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
ed section 2 of the Sherman Act.
92
The plaintiffs’ theory—refined over
the course of multiple amended complaints (and several years)—
came to rest on product-design allegations. In 2001, Apple introduced
a portable, digital music player, the soon ubiquitous “iPod.” It alleg-
edly subsequently obtained monopoly power in the portable digital
audio player market.
93
Apple also introduced “iTunes,” a software
program (freely downloadable from Apple’s website and compatible
with the iPod) that allowed computer users to catalog and organize
their digital music libraries.
94
The iPod initially could play only “un-
protected” song files in .mp3 format.
95
“Unprotected” formats contain
no restrictions on customers’ use of the files; “protected” formats, on
the other hand, contain digital rights management (DRM) software
that restricts end uses. Subsequently, however, Apple launched the
“iTunes Music Store,” an online music store that allows customers to
purchase and download a wide variety of songs and albums.
96
With
the iTunes Music Store, Apple was entering a new market—the mar-
ket for audio-file downloads.
97
At least according to the plaintiffs, Ap-
ple also obtained monopoly power in this market.
98
(a) The Audio-File Market
The audio-file market is complex and, at least to some degree, a
network market exhibiting both direct and indirect positive network
externalities. Taking into account the seemingly inevitable copyright
infringement that accompanies downloading audio files in unprotect-
2629907, at *1 (N.D. Cal. June 29, 2010) (“In its October 30 Order, the Court found that
Plaintiffs failed to state a tying claim pursuant to Section 1 of the Sherman Act . . . on the
ground that Plaintiffs’ allegations of a technological interrelationship between iTunes and
iPods did not amount to a coercive tie.”).
92. Class Action Complaint at 3, Slattery v. Apple Computer, Inc., No. C05-00037 JW
(N.D. Cal. Jan. 3, 2005).
93. Amended Complaint at 23, The Apple iPod iTunes Antitrust Litigation, C 07-6507
JW (N.D. Cal. July 9, 2010) (“Apple has monopoly power in the Portable Digital Player
Market.”).
94. Id. at 10.
95. The “.mp3” format uses data compression to reduce the size of CD-quality audio
files, facilitating their storage and transmission. The trade-off for the more convenient file
size is a (generally imperceptible) loss in audio quality.
96. See id.
97. This product market might more properly include only non-copyright infringing
audio-file-downloads—a robust market for pirated music (and films) existed then as it does
now, though there may be relatively low cross-elasticity of demand for the two types of
files, given the potential liability entailed by downloading pirated files. See generally Who
Music Theft Hurts, RIAA, http://www.riaa.com/physicalpiracy.php (last visited Feb. 13,
2012) (discussing what the Recording Industry Association of America terms “music theft”).
At least according to the RIAA (far from an unbiased source), during a five-year span from
2004 to 2009, “approximately 30 billion songs were illegally downloaded on file-sharing
networks.” For Students Doing Reports, RIAA, http://www.riaa.com/faq.php (last visited
Feb. 13, 2012).
98. See Amended Complaint, supra note 93, at 11-12 (stating that Apple obtained a
92% market share in the “Audio Download Market” as defined by the plaintiffs).
2012] ANTICOMPETITIVE PRODUCT DESIGN 699
ed file formats, the audio-file-download market is characterized by
direct positive network externalities. This is so because when song
files are sold to users in unprotected file formats, those users can
then share audio files with other users, whether as a gift or in ex-
change for song files or other remuneration from such users—albeit
in violation of the Copyright Act.
99
Thus, increased usage by others
directly increases the value of the original product to the user. The
indirect positive network effects, on the other hand, arise because
increased usage incentivizes sellers to increase the number of titles
offered through their online stores, which in turn increases the value
of audio-file players to users.
Further, the network may be open or closed depending in part on
whether unprotected or protected file formats are used. If audio files
are distributed in an unprotected format, all firms that wish to offer
files for download are generally free to do so using that format. The
files themselves can be played on a variety of software programs and
portable digital audio players, and users can even download song
files from each other via person-to-person file-sharing networks
(again, frequently in violation of U.S. copyright laws). If, however, a
firm encodes its audio files with its proprietary DRM software, it can
effectively restrict consumers from playing the files on other firms’
audio players or uploading files to or downloading them from other
users’ music libraries. Thus, a firm that produces both DRM-
protected song files and digital audio playback programs (whether
installed on personal computers, portable digital audio players, or
both) can effectively create a closed market with which other firms’
products are not interoperable.
(b) Apple Closes the Network
Initially, Apple intended to compete in an open network—in fact,
it “touted the iPod’s interoperability as a selling point, and made no
efforts to restrict the music that could be played on the device. Simi-
larly, iTunes users could sync their [digital music] libraries with non-
Apple devices.”
100
According to the complaint, however, at the same
time Apple launched the iTunes Music Store, it “secretly ‘updated’
the iTunes software so that iTunes and its iPod were compatible with
FairPlay, Apple’s proprietary DRM.”
101
Apple at all times refused to
99. See generally 17 U.S.C. § 106(1) (2006) (granting to copyright owners the exclusive
right “to reproduce the copyrighted work in copies or phonorecords”). There is (almost)
universal agreement among courts and scholars that naked copying of media files consti-
tutes copyright infringement. See John M. Newman, Raising the Bar and the Public Inter-
est: On Prior Restraints, “Traditional Contours,” and Constitutionalizing Preliminary In-
junctions in Copyright Law, 10 V
A.SPORTS &ENT. L.J. 323, 363 & n.247 (2011).
100. Amended Complaint, supra note 93, at 4.
101. Id. at 11.
700 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
license FairPlay to other firms that offered digital audio down-
loads. And, with all of that, the pieces were then in place for Apple
to close its network and strive for continued dominance in the au-
dio download market.
Concurrently with the introduction of the iTunes Music Store, Ap-
ple allegedly updated the iTunes software and issued a new genera-
tion of its iPod music players so that (1) the iTunes music library
program could export audio files only to iPods (not competing porta-
ble media players) and (2) iPods would accept files only from iTunes
(not competing software).
102
It also began selling song files encoded
with FairPlay; such files could be played only via iTunes (on comput-
ers) and on iPod portable music players. Thus, Apple had effectively
created a closed network—it offered song files for purchase and
download through the iTunes Music Store that could be played only
via its iTunes and iPod products. Meanwhile, iTunes and iPods could
play only Apple’s proprietary, DRM-encoded files (not those offered
by other firms).
103
Though further factual development would be needed for a more
definite explanation, it seems likely that Apple made the choice to
switch to a closed network, purposely prompting the race for domi-
nance that such decisions tend to cause.
104
Such a move is a gamble—
on the one hand, Apple might have chosen to instead maximize in-
teroperability with rival music players in order to maximize sales of
digital audio files from the iTunes Music Store. But as noted above,
if a closed market “tips” in favor of one firm, the resulting monopo-
ly will be protected by the natural barrier to entry constituted by
the size of the network itself, allowing the dominant firm to charge
monopoly prices.
105
With a closed network in place, one that (at least according to the
plaintiffs) dominated the market, Apple could rationally begin setting
prices above marginal cost. In such a setting, it would be natural for
the resulting profits to attract new entrants. And, indeed, RealNet-
102. Id. at 4 (“Unbeknownst to consumers, Apple released a new generation of iPods
and altered the iTunes software so that iTunes and iPods were only interoperable with
each other.”).
103. This network, it should be noted, was closed more specifically only as to DRM-
encoded song files—iTunes and iPods remained capable of playing unprotected .mp3-
format audio files.
104. See supra notes 38-40 and accompanying text. Alternatively, it could be argued
that Apple’s using FairPlay was an attempt to comply with the wishes of record companies,
who have a strong interest in preventing music piracy. See Defendant Apple Inc.’s Answer
and Defenses to Plaintiffs’ Consolidated Complaint at 7-8, The Apple iPod iTunes Antitrust
Litig., No. C 05-00037 JW (N.D. Cal. June 6, 2007). Yet, as the plaintiffs’ amended com-
plaint points out, “[t]hese companies . . . did not require Apple to limit Audio Downloads to
use with Apple products. It was Apple’s choice to encode the Audio Downloads sold through
iTS with its own proprietary DRM, FairPlay.” Amended Complaint, supra note 93, at 11.
105. See supra notes 39-45 and accompanying text.
2012] ANTICOMPETITIVE PRODUCT DESIGN 701
works, a competitor in the audio file download market, announced in
January 2003 that its software-based music player had become capa-
ble of playing songs bought from the iTunes Music Store.
106
Subse-
quently, in July 2004, RealNetworks announced the release of its
“Harmony” software, which allowed songs purchased from RealNet-
works’ online store to play on iPods.
107
Likely of greatest concern to
Apple, however, was the fact that RealNetworks was offering legal
music downloads—in a file format that was fully compatible with
Apple products—at prices as low as $0.49 per song, substantially un-
dercutting Apple’s traditional price of $0.99 per song.
108
Apple responded quickly. In a press release, it stated: “We are
stunned that RealNetworks has adopted the tactics and ethics of a
hacker to break into the iPod, and we are investigating the implica-
tions of their actions under the DMCA and other laws.”
109
Beyond
merely investigating the copyright law implications of RealNetworks’
actions, however, Apple also issued a veiled threat: “We strongly cau-
tion Real and their customers that when we update our [iPod] soft-
ware from time to time it is highly likely that Real’s Harmony tech-
nology will cease to work with current and future [iPods].”
110
And Ap-
ple allegedly made good on this threat, issuing software updates to
its iTunes music library program that blocked RealNetworks and its
Harmony software from any interoperability with Apple products.
111
According to the plaintiffs, this pattern repeated itself multiple
times. “For example, in or about the beginning of 2005, a
software program known as JHymn . . . was developed . . . so that
Audio Downloads purchased from [the iTunes Store] could be played
on . . . Apple’s iPod or any non-Apple device.”
112
Apple, however,
“immediately began issuing software updates to thwart JHymn.”
113
It
106. John Borland, RealNetworks Breaks Apple’s Hold on iPod, CNET (July 26, 2004),
http://news.cnet.com/RealNetworks-breaks-Apple's-hold-on-iPod/2100-1027_3-5282063.html.
107. Id.
108. At least according to RealNetworks, its prices were incredibly successful in at-
tracting customers away from Apple’s iTunes Store, even over a short period of time. “In its
first three weeks of selling iPod-compatible music files, RealNetworks sold approximately
three million music files,” which equated to about 10% of the market shifting from Apple to
RealNetworks. Amended Complaint, supra note 93, at 14, 15.
109. J
AMES BOYLE,THE PUBLIC DOMAIN:ENCLOSING THE COMMONS OF THE MIND 110
(2008) (quoting Apple’s press release).
110. Jim Dalrymple, Apple Accuses RealNetworks of Hacking, PCW
ORLD (July 30, 2004),
http://www.pcworld.com/article/117183/apple_accuses_realnetworks_of_hacking.html (quoting
the same press statement).
111. See Amended Complaint, supra note 93, at 15 (“True to its threat, beginning in
October 2004, Apple ‘updated’ its iPod and iTunes software to prevent songs downloaded
from RealNetworks’ music store from being played on iPods.”). All iTunes users were alleg-
edly required to install the new software updates if they wished to continue to use Apple’s
iTunes Store. Id. Further, Apple allegedly failed to disclose to consumers that the updates
would block interoperability with RealNetworks’ products. See id.
112. Id. at 16.
113. Id. at 17.
702 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
also used software updates to block similar programs from achieving
interoperability with Apple products.
114
By 2009, Apple found itself in
a game of cat-and-mouse with Palm, which had programmed one of
its smartphones to accept digital music files uploaded from Apple’s
iTunes music library program. For months, the two companies issued
warring software updates to their respective devices—Palm would
issue an update reopening connectivity and Apple would respond
with an update re-blocking it.
115
Importantly, the plaintiffs alleged that Apple’s software updates
served no procompetitive, innovative purpose.
116
On the strength of
those allegations, the district court denied Apple’s motion to dismiss
the product-design section 2 claims.
117
In a subsequent amended an-
swer, on the other hand, Apple admitted only “that it released up-
dates to its FairPlay technology to, among other things, stop JHymn
and other hacks that circumvented the content protection required by
the [major record] labels on music purchased from the iTunes Store .
. . .”
118
The inclusion of “among other things” raises—or at least does
not eliminate—the possibility that Apple will allege some procompet-
itive justification other than merely foreclosing rivals. Thus, at least
at the pleading stage, no clear answer has emerged as to whether the
updates were pro- or anticompetitive (unlike, for example, Microsoft
III, where the defendant failed to identify any procompetitive justifi-
cation for certain of its design-related actions).
119
Yet if the plaintiffs’
114. See id. (identifying QTFairUse and PlayFair as additional programs targeted by
Apple’s software updates).
115. See id. (referring to three separate Apple updates and two Palm updates over a
three-month period, each of which either reestablished or blocked interoperability between
Apple and Palm products).
116. Id. (“Apple continually redesigned its software even though it admitted that doing
so served no genuine antipiracy purpose.”).
117. The Apple iPod iTunes Antitrust Litigation, Nos. C 05-0037 JW, C 07-06507 JW,
2010 WL 2629907, at *4 (N.D. Cal. June 29, 2010) (“The Court finds that Plaintiffs has [sic]
alleged facts, which if proven true, would suffice to hold Defendant liable under Section 2
of the Sherman Act.”).
118. Defendant Apple Inc.’s First Amended Answer and Defenses to Plaintiffs’ Amend-
ed Consolidated Complaint ¶ 65, at 10, The Apple iPod iTunes Antitrust Litigation,
No. C 05-00037 JW (N.D. Cal. Sept. 24, 2010), 2010 WL 6323163.
119. Microsoft III, 253 F.3d 34, 67 (D.C. Cir. 2001). Recently, the district court denied
Apple’s motion for a protective order and allowed the plaintiffs to depose Apple’s CEO,
Steve Jobs. The order, however, included the requirement that the deposition be limited to
the subject of Apple’s anticompetitive actions toward RealNetworks. See The Apple iPod
iTunes Antitrust Litigation, No. C 05-0037 JW, 2011 WL 976942, at *3 (N.D. Cal. Mar. 21,
2011) (“[T]he deposition shall be limited to the topics of (a) the July 26, 2004 RealNetworks
announcement [regarding Harmony], (b) the July 29, 2004 Apple Announcement in re-
sponse thereto, and (c) Apple’s software updates in October 2004 that rendered the Real-
Networks’s digital music files once again inoperable with iPods.”); see also Philip Elmer-
DeWitt, Judge Orders Steve Jobs to Appear, F
ORTUNE (Mar. 22, 2011),
http://tech.fortune.cnn.com/2011/03/22/judge-orders-steve-jobs-to-appear/ (detailing the
events leading up to the court order, including a continuance granted so that Jobs could
“focus on his health”).
2012] ANTICOMPETITIVE PRODUCT DESIGN 703
allegations are correct—and if they are susceptible of proof at tri-
al
120
—Apple may very well be (eventually) found guilty of monopoli-
zation through product redesign.
121
B. Software Updates as a Uniquely Attractive Method of Using
Product Redesign to Foreclose Rivals
The facts of the Apple iPod iTunes Antitrust Litigation present
issues unique to antitrust product design jurisprudence. In particu-
lar, the use of software updates to quickly, cheaply, and easily—yet
effectively—redesign code-based products with an eye toward closing
a network, destroying interoperability to preserve a closed-network
monopoly, or otherwise excluding rivals from the market poses a dis-
tinctive threat to competition and innovation. And it is a threat that
does not implicate, or implicates less heavily, the traditional concerns
with condemning product redesigns under the antitrust laws.
One of the primary concerns with condemning product redesigns
as anticompetitive is that “in most cases if the conduct is excessively
exclusionary it is also self-deterring.”
122
More specifically, it is
thought that “a product innovation is extremely costly and time con-
suming to develop, design, manufacture, and place on the market.”
123
Further, redesigning goods merely to make them incompatible with
competitors’ products will likely cause customers to react negative-
ly.
124
And there are “sunk” costs associated with innovation, costs
that cannot be recovered once invested.
125
If negative customer re-
sponse forces the firm to pull the new product, the firm will forfeit
these sunk costs; it will also incur reversal losses by switching back
to its former product.
120. The plaintiffs may also face difficulty in showing that Apple has monopoly power
in the relevant markets given the somewhat unfamiliar product markets the plaintiffs
will need to define for their monopolization claims to succeed. See generally
Philip Elmer-DeWitt, Does Apple Have a Monopoly?, F
ORTUNE (Feb. 16, 2011),
http://tech.fortune.cnn.com/2011/02/16/does-apple-have-a-monopoly/ (“The answer to . . .
whether Apple has sufficient control over a particular market for regulators to even ask
whether its behavior has been abusive or coercive depends on how you define the market.”).
121. Certainly, the Apple litigation bears some resemblance to the landmark Microsoft
antitrust litigation. See id. (“[I]n the months ahead . . . you will hear Apple using many of
the same legal arguments Microsoft made . . . .”). Both are high-technology firms that ap-
pear dominant in their core competencies—code-based product markets characterized by
positive network effects—and both are unquestionably innovative.
122. H
ERBERT HOVENKAMP,THE ANTITRUST ENTERPRISE:PRINCIPLE AND EXECUTION
275 (2005).
123. A
REEDA &HOVENKAMP, supra note 61, ¶ 775c, at 284.
124. See H
OVENKAMP, supra note 122, at 274-76.
125. See Spulber & Yoo, supra note 41, at 919-21 (discussing sunk costs generally and
in relation to the telecommunications industry).
704 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
Thus, if C
i
> P
m
– LR,
126
a rational firm will not engage in the an-
ticompetitive, design-related behavior. Redesigning physical (as op-
posed to intangible) products is generally fairly expensive, so C
i
in
such markets is relatively high, making true anticompetitive product
redesigns less common. Since innovations in such markets are
thought to be “extremely costly” and most of those costs are unrecov-
erable in the face of a failed product launch, L is also considered to be
high. Additionally, firms can expect to incur substantial reversal
losses, making L still higher. Finally, expected customer resistance to
product changes that do not represent true innovations also makes R
relatively high in these traditional markets, further dissuading such
behavior. Therefore, the argument goes, the market itself tends to
discipline potential, anticompetitive product-design decisions. The
left side of the formula tends to outweigh the right.
As an initial matter, the potential rewards of anticompetitive,
code-based product redesign through software updates may frequent-
ly be greater than those of physical product design, increasing the
value of P
m
in the formula above. This is so because, as noted supra,
the relevant markets tend to be characterized by positive network
externalities that form a natural barrier to entry.
127
Given that, a
firm that dominates a closed network market may enjoy a more sta-
ble monopoly than would be possible in a non-network market.
128
But more importantly, and more certainly, other factors in the
formula are different in the realm of digital product markets and
software updates. Specifically, I contend first that C
i
tends to be low-
er, reducing the value that the right side of the formula must over-
come in order to make an anticompetitive product redesign ration-
al.
129
Next, I argue that R also tends to be lower, reducing the amount
that usually offsets the potential of monopoly profits.
130
Finally, I also
contend that L tends to be lower. Each of these features tend to make
anticompetitive product redesign through software updates uniquely
attractive to firms.
131
126. In this formula, Ci is the costs a firm incurs by redesigning the product and bring-
ing the replacement to market; Pm is the potential monopoly profits; L represents the po-
tential losses from unrecoverable sunk costs, reversal costs, or damages awards in anti-
trust suits; and R represents the risk—the probability that the design will fail due to cus-
tomers rejecting the new product or the design being condemned under antitrust laws.
127. See supra notes 33-45 and accompanying text.
128. This is subject, of course, to the possibility that it will be “leapfrogged” by firms
innovating around its network. See supra notes 13-31 and accompanying text (discussing
the argument that the New Economy tends to be characterized by higher rates of innova-
tion).
129. See infra Part III.B.1.
130. See infra Part III.B.2.
131. It should be noted that, if Ci < Pm – LR, a rational, profit-maximizing firm still
would not necessarily engage in the anticompetitive design-related behavior. Instead, at
the point that Pm – LR exceeds Ci, the firm would be faced with deciding whether Pm – LR
2012] ANTICOMPETITIVE PRODUCT DESIGN 705
1. Relatively Low Design-Related Costs to Firms: Lower C
i
Incentivizes Anticompetitive Product Redesigns
As the costs to firms of developing and distributing a product re-
design decrease, the attractiveness of an anticompetitive product re-
design increases. The following two subparts discuss two reasons
redesigning code-based products through software updates is inex-
pensive relative to redesigning physical products. Part III.B.1.a ex-
plains that developing a software update tends to entail lower sunk
costs than developing a new physical good, while Part III.B.1.b ex-
plores the low distribution costs associated with issuing software
updates via the Internet. Importantly, both factors tend to lower C
i
and, as a result, make anticompetitive product redesign more attrac-
tive to firms.
(a) Low Development Costs Relative to Physical Product Redesign
Redesigning code-based products through software updates gener-
ally entails low development costs relative to redesigning physical
products, making it an inviting avenue for anticompetitive design-
related conduct. Software can be effectively redesigned through an
update—which may affect only a small part of the software pro-
gram—relatively cheaply. Take, for example, the product designs at
issue in some of the seminal product-design antitrust cases discussed
in the following Part: a new computer system (the IBM plug-
compatible cases);
132
an instant camera and redesigned, complemen-
tary film (Berkey Photo, Inc. v. Eastman Kodak Co.);
133
or a skin-graft
gun and redesigned replacement needles (C.R. Bard., Inc. v. M3 Sys-
tems, Inc.).
134
Each required a team of engineers, changes to produc-
tion facilities, and substantial marketing costs. A software update, on
the other hand, can be designed and produced by a single computer
programmer working at a single computer; at most, it requires only a
team of programmers.
135
Thus, the development costs of software up-
dates tend to be lower than those entailed by redesigning physical
products, lowering the value of C
i
in the formula above. Yet software
updates can be used just as effectively to exclude rivals from a mar-
ket as a physical redesign—as Apple allegedly did by issuing soft-
is also greater than Pc, where Pc represents the profits that could be obtained through
either making a procompetitive design change or declining to make any change. Since the
analysis that follows is unaffected by its omission, the remainder of the Article declines to
mention this consideration for purposes of simplicity and readability.
132. In re IBM Peripheral EDP Devices Antitrust Litig., 481 F. Supp. 965, 972-73 (N.D.
Cal. 1979), aff’d sub nom. Transamerica Computer Co. v. IBM Corp., 698 F.2d 1377 (9th
Cir. 1983).
133. Berkey Photo, Inc. v. Eastman Kodak Co., 603 F.2d 263, 268-71 (2d Cir. 1979).
134. C.R. Bard., Inc. v. M3 Systems, Inc., 157 F.3d 1340, 1346-48 (Fed. Cir. 1998).
135. Cf. P
ETER SEIBEL,CODERS AT WORK:REFLECTIONS ON THE CRAFT OF PROGRAM-
MING
154-56 (2009) (discussing writing and rewriting computer code).
706 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
ware updates to block interoperability with RealNetworks’ digital
audio files.
136
So, while the potential rewards of code-based product design
through software updates are generally greater than or similar to
those of physical product design, the sunk costs of development are
relatively lower. Thus, the overall risk usually associated with a
purely exclusionary physical product design is relatively higher than
the risk of issuing an anticompetitive product update. And as a re-
sult, it is more likely that C
i
< P
m
– LR, making anticompetitive de-
sign decisions more likely rational—and the market less able to act
as a check on exclusionary design conduct in this area.
(b) Speed, Ease, and Low Costs to Firms of Online
Software-Update Distribution
Like the initial distribution of software,
137
updating software was
once a costly, time-intensive project. It generally required design and
subsequent distribution of code-based program files on physical disks
to consumers. Given that files typically needed to be stored on such
disks to allow for transportation to end users, distribution channels
were necessarily physical, and, as a result, distribution was relatively
slow and costly. Immediate updates for the purpose of repairing unfore-
seen problems with software—or for any other reason—were impossible.
Installing the updated files, at least in the early days of digital
computing, also required expertise on the part of consumers or, al-
ternatively, required consumers to hire expert installers. In these
ways, updating code-based software was fairly similar to updating
the design of other physical products—take, for example, a car. Fre-
quently, automobile manufacturers design, develop, and ultimately
sell a particular make and model to consumers and then discover a
defect in the design months (or even years) down the road—perhaps
a flaw in the vehicle’s airbag system. In order to “update” the design,
the manufacturer must issue a recall announcement asking consum-
ers to bring their vehicles to a dealer for replacement or repair of the
faulty system. The manufacturer must also physically distribute the
relevant replacement parts to its dealers. Consumers would then
have to bring their vehicles to the dealer in order to complete the in-
stallation of the update, making the entire process quite lengthy—
months or even years—and costly to manufacturers.
138
136. See supra notes 106-11 and accompanying text (noting that Apple admitted to
doing so).
137. See L
ESSIG, supra note 31, at 7 (noting the “high costs of production” and “the
extraordinarily high costs of distribution” present in real-space markets).
138. The scale of such recalls can be massive—currently, the largest auto recall in his-
tory included 14.1 million Ford vehicles that had been built with faulty cruise-control
switches. The model years included were 1993-2004; Ford announced the recall in two
2012] ANTICOMPETITIVE PRODUCT DESIGN 707
With the advent of nearly universal access to the Internet, howev-
er, firms marketing products based on computer code were given an
unprecedented, even revolutionary, distribution tool. Programmers
can now write update files that a firm can distribute to users almost
instantaneously, eliminating the lion’s share of the time lag inherent
in physical distribution systems. Likewise, the Internet has made
costs of update distribution to software firms relatively negligible—
unlike, for example, the costs an auto manufacturer would incur by
shipping fourteen million cruise-control switches via physical distri-
bution systems.
139
As a direct result, updating software through files
distributed via the Internet has become an industry norm.
140
The speed and ease of distribution of software updates can also
allow a firm to be more nimble in anticompetitively redesigning its
products. Consider again the rapid series of updates issued by Apple
to repeatedly block interoperability with Palm’s products.
141
Firms
wishing to foreclose competition by redesigning physical products
were hampered by the length of time required for the new product to
reach consumers—in the meantime, competitors could enjoy interop-
erability and access to the market. Redesigning software through up-
dates, however, allows firms to respond much more rapidly to threats
by rivals.
And the relatively low distribution costs associated with issuing
software updates contributes to the rebalancing of the risk–reward
analysis discussed above, further incentivizing anticompetitive be-
havior. Given low development and distribution costs, and holding
constant—or, in a network market, even increasing—the potential
reward of monopoly overcharges, exclusionary product redesign be-
comes yet more attractive to firms. Or, put another way, it is even
more likely that C
i
< P
m
– LR. Again, the market acts as much less of
a check in such a scenario, mitigating one primary concern with regu-
lating anticompetitive, design-related behavior.
2. Relatively Low Risk That Consumers Will Reject Redesigns:
How Lower R Further Incentivizes Anticompetitive Product Redesigns
As the risk that customers will reject product redesigns decreases,
the likelihood that a firm can offset development and distribution
costs by reaping monopoly profits increases. As a direct result, lower
separate announcements, one in 2008 and the other in 2009, years after the defective
vehicles were sold. See Vincent Aubé, Ten Biggest Auto Recalls, MSN A
UTOS,
http://autos.ca.msn.com/photos/gallery.aspx?cp-documentid=22338216 (last visited Feb. 13,
2012).
139. See generally id. (describing such a recall).
140. UpdateShield 1.0: Product Overview, S
CIENSOFT, http://www.sciensoft.com/
products/updateshield (last visited Feb. 13, 2012) (“Automatic software update is becoming
a standard feature in many software products today.”).
141. Amended Complaint, supra note 93, at 17.
708 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
R further incentivizes anticompetitive product redesigns. The next
two subparts demonstrate that redesigning code-based products
through software updates does, in fact, involve relatively low risk
that customers will react negatively to product redesigns.
(a) Relatively Low Cost to Consumers Lowers R
As shown above, redesigning code-based products through soft-
ware updates entails relatively low costs to firms. Firms can and do
take advantage of these low costs by offering low-cost or free updates
to software users, yet ultimately make a profit. This can be done for
legitimate purposes—for example, Facebook’s constantly updated, yet
free, online social-networking service. Facebook’s profit mechanism is
selling advertising space to other firms; it can afford to update its
software at no extra charge to users because software updates are
relatively inexpensive and it is profiting from selling advertising
space. But such design changes can also be done for exclusionary
purposes. Take, for example, Apple issuing free updates to its iTunes
music library program and iPod portable digital audio players in or-
der to exclude RealNetworks from the market for music downloads.
Assuming Apple’s iTunes Store dominated the market for legal music
downloads, Apple could have issued updates for free yet made a prof-
it because software updates are relatively inexpensive and it would
prolong its monopoly by excluding RealNetworks from the market.
Where C
i
is high, firms must either absorb the costs or pass them on
to consumers; where it is relatively low, firms can charge consumers
less for the product redesign. And that, in turn, lowers R—an ele-
ment of which is the risk that consumers will reject the redesign.
Another element of cost consumers take into account in deciding
whether to accept a redesigned product is the time required to do so.
As noted above, the time required of customers to receive and install
software updates has been reduced considerably by the advent of
widespread Internet access and the corresponding elimination of re-
liance on physical distribution processes.
142
As compared to physical
goods, consumers are required to expend minimal time accepting
software redesigned through updates delivered via the Internet. This
further lowers their costs, further lowering R and further increasing the
probability that an anticompetitive product redesign will be profitable.
(b) The Trend Toward Automatic Software Updates
Even further mitigating this concern is one final aspect of soft-
ware updating in the context of ubiquitous, high-speed Internet ac-
cess: the trend toward automatic software updates. As noted supra,
142. See supra Part III.B.1.b.
2012] ANTICOMPETITIVE PRODUCT DESIGN 709
one of the arguments against antitrust oversight of product redesigns
is that customers will reject new products that are not really im-
provements over the old.
143
This can hold true where customers are
given a meaningful choice whether to keep the old product or switch
to the new. Where, for example, a firm begins selling a new instant
camera (and new complementary film), customers can make a rela-
tively informed choice whether to keep their old cameras and buy in-
teroperable film from rivals or switch to the new camera and film. If
the design changes represent true innovation that offers substantial-
ly more value to consumers, then customers will switch over; if the
changes were merely exclusionary, customers will (so the theory
goes) rationally reject the new products.
In the new era of code-based product markets and widespread,
high-speed Internet access, however, “automatic” software updates
are becoming increasingly prevalent.
144
Automatic software updates
are triggered by code written into software programs themselves.
This code directs the users’ computers to check a certain server via
the Internet for any available updates, then download and install any
and all updates that are available. So long as the users’ computers
connect to the Internet at some point, no further action is needed on
the part of users to install the updates. There are plausible procom-
petitive advantages to automatic, versus manual, software updates:
users need not oversee the updating process, they allow for more ef-
fective and rapid virus protection, problems with software can be
fixed or “patched” more rapidly (perhaps even before the user ever
notices a problem), etc.
145
But automatic software updates also eliminate customers’ ability
to make an informed ex ante choice as to whether they should accept
the redesigned product. Customers may not have any information
available to them regarding what, exactly, the software update does.
In fact, they may not even be aware that an automatic software up-
date has occurred.
146
By allowing firms to count on the virtual cer-
tainty that a product redesign will be accepted by customers,
automatic software updates slash R dramatically, further increasing
the attractiveness of an anticompetitive product redesign.
Thus, one of the primary concerns with condemning product-design
143. See supra notes 124-27 and accompanying text.
144. UpdateShield 1.0: Product Overview, supra note 140; Tamara Wilhite, Automatic
Software Updates: The Good, the Bad, and the Ugly, IIE B
LOGS (Jan. 14, 2008, 12:00 AM),
http://www.iienet2.org/Blogger_comment.aspx?id=11216&blogid=612 (“Automatic software
updates have a number of advantages, which is why they are so frequently used in the IT
world.”).
145. See Wilhite, supra note 144 (noting four distinct advantages of automatic software
updates).
146. And even when software updates are not entirely automatic, frequently all a cus-
tomer sees is a small pop-up window asking whether to or not to “install new updates.”
710 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
decisions—that excessively exclusionary design conduct is generally
deterred by potential customer resistance—does not apply with
much force (if at all) to code-based product redesigned through
software updates.
3. Relatively Low Losses Incurred if Product Redesigns Fail: How
Lower L Further Incentivizes Anticompetitive Redesigns
As discussed above, C
i
tends to be lower for code-based product
redesign through software updates, especially for updates distributed
via the Internet. Since C
i
takes into account sunk costs, L (the total
costs incurred by a firm that undertakes a failed product redesign),
which similarly accounts for sunk costs, also tends to be lower in these
markets. Yet redesign of software programs also carries another unique
feature that lowers L even further—relatively low reversal costs.
If a firm redesigns and brings to market a new physical product, it
has likely altered its assembly and packaging lines, marketing, etc.,
in ways that would entail significant reversal costs in the face of a
failed product launch. Take, for example, the new Instamatic camer-
as at issue in Berkey Photo.
147
To launch the Instamatic, Kodak pre-
sumably needed to alter its manufacturing facilities, place orders for
component parts and materials, develop and begin assembling new
packaging, mount a marketing campaign, and actually ship thou-
sands of the physical cameras to brick-and-mortar stores. If the In-
stamatic had been a failure, Kodak would have needed to revert its
production facilities to the status quo, dump its new packaging, pull
promotional materials and mount a new advertising campaign tout-
ing the old product, and (depending on contractual arrangements)
perhaps account for the mass of unwanted product in the hands of
retailers. All of this would have added up to significant reversal
costs, making L high.
Conversely, imagine that Apple’s exclusionary software updates
had been installed by customers who, upon realizing the updates
foreclosed interoperability with RealNetworks, began to substitute
away from Apple products. Apple could quickly, easily, and above all
cheaply reverse the design by simply issuing another update that
contained commands undoing the exclusionary code functions. It
would not need to revert production facilities, change packaging, al-
ter its advertising campaign, or account for unwanted inventory at
its retail stores. This scenario would hold true for almost all code-
based product redesigns through online software updates. Thus, in
addition to the reductive effect of relatively low sunk costs, the lower
reversal costs further reduce L. This, in turn, increases the value of
147. A more detailed discussion of Berkey Photo’s facts—and especially the Second
Circuit’s reasoning behind its decision in that case—follows. See infra Part IV.B.
2012] ANTICOMPETITIVE PRODUCT DESIGN 711
the right side of the formula, again making anticompetitive product
redesigns more attractive.
C. Code-Based Product Redesign as Uniquely Susceptible to
Antitrust Scrutiny
The other primary concern regarding antitrust scrutiny of product
design decisions is that, “[a]s a general matter, antitrust courts are
not competent to second-guess decisions about product design.”
148
Multiple unique features of modern code-based product redesign,
however, render it uniquely susceptible to antitrust scrutiny. First,
redesign through software updates allows firms to make targeted de-
sign changes to products; thus, these redesigns are much less com-
prehensive than typical physical product redesigns—making anti-
competitive elements more conceptually separable from procompeti-
tive innovations. Second, and similarly, sections of code perform spe-
cific functions and are separable from surrounding sections, again
facilitating the ability of courts to discern between exclusionary and
innovative design elements. Finally, since computer code can be con-
figured multiple ways to perform the same function, a design change
to a code-based product that limits interoperability is more likely mo-
tivated by the desire to foreclose rivals from the market than physi-
cal product changes.
1. Software Updates Are Conceptually Separable from the
Base Product
The stereotypical, allegedly anticompetitive product design (like
the design decisions challenged in, for example, the IBM plug-
compatible cases or Berkey Photo) consists of introducing to the mar-
ket a significantly redesigned product that renders competitors’ com-
plementary products noninteroperable. In such situations, courts are
wisely reluctant to condemn the design change—given the conceptual
difficulty of separating potentially exclusionary design elements from
procompetitive innovations, courts recognize the potential that using
antitrust law to condemn such changes will frequently chill innova-
tion.
149
Yet Apple was (allegedly) able to use software updates to re-
design small segments of products already in the hands of consumers
and was able to do so with an ease and frequency that was previously
unheard of (witness the cat-and-mouse interoperability battle with
Palm).
150
By delivering targeted product redesigns almost instanta-
148. HOVENKAMP, supra note 122, at 275.
149. See, e.g., In re IBM Peripheral EDP Devices Antitrust Litig., 481 F. Supp. 965,
1003 (N.D. Cal. 1979), aff’d sub nom. Transamerica Computer Co. v. IBM Corp., 698 F.2d
1377 (9th Cir. 1983).
150. See supra notes 106-15 and accompanying text (detailing a series of redesigns
made possible by online software updates that occurred over a short period of time).
712 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
neously via the Internet, Apple’s behavior highlighted the possibili-
ties for anticompetitive product design available to firms competing
in the new, code-based product markets. Yet it also highlighted one
reason such redesigns are uniquely susceptible to antitrust scrutiny.
As noted above, software updates—the dominant method of rede-
signing software programs—do not constitute a total redesign; in-
stead, they contain a relatively small amount of object code that al-
ters certain functions of the base software program’s underlying code.
The update code interacts with, but exists separately from, the previ-
ous code underlying the program. Not so with the redesigned Insta-
matic camera and accompanying film in Berkey Photo—there, the
court was faced with trying to separate the design elements that led
to an improved product
151
from those that made Berkey Photo’s film
incompatible. Yet the very same elements performed both func-
tions, making conceptual separation of the two a Sisyphean task
that the court quite rightly declined to attempt.
152
Since the ele-
ments and functionality of a software update are relatively easily
conceived of as separate from the elements of the base software pro-
gram affected by the update, courts are more competent to address
their effects on competition than the same courts would be in the ste-
reotypical product-design case.
153
2. Lines of Code Are Separable from a Software Redesign
as a Whole
The objection might be raised that a software update itself may
contain both innovative and anticompetitive aspects. Thus, while an
update may be conceptually separable from the base software pro-
gram, it does not necessarily follow that anticompetitive elements are
likewise capable of being isolated. Yet, recall that a software program
consists of a series of functions expressed in object code.
154
Each line
or section of code performs a specific, finite function. Thus, within the
code that composes a software update, specific functions are separa-
ble. If a software update serves to foreclose interoperability with a
rival’s products, that noninteroperability can be traced to those cer-
tain functions. And as a result, even if a software update contains
multiple design changes, the lines of code that dictate functions with-
151. And the Instamatic (along with Kodacolor II film) was indeed perceived as such by
consumers—they “went on to become two of the great successes of amateur photography.”
H
OVENKAMP, supra note 122, at 275.
152. See Berkey Photo, Inc. v. Eastman Kodak Co., 603 F.2d 263, 286-87 (2d Cir. 1979).
153. It should be emphasized at this point that I do not take issue with the usual con-
tention that “courts and juries are generally incapable of addressing the technical merits or
anticompetitive effects of innovation.” A
REEDA &HOVENKAMP, supra note 61, ¶ 775c, at
284 (emphasis added); see also Sidak, supra note 7, at 1143-48 (arguing that a rule of per
se legality should apply to product innovations).
154. See supra Part III.A.1.
2012] ANTICOMPETITIVE PRODUCT DESIGN 713
in the update are separable, allowing direct analysis of what those
respective functions are.
3. Code Can Be Written Different Ways to Perform the
Same Function
In writing the source code for a software program, a programmer
first decides what functions are needed for the program to perform its
intended role. She next organizes these functions into an organiza-
tional flow chart, rearranging as necessary to achieve secondary
goals, like allowing for efficient computing. Finally, the programmer
writes the source code itself.
155
“This stage in the process is compara-
ble to the novelist fleshing out the broad outline of [her] plot by craft-
ing from words and sentences the paragraphs that convey the
idea.”
156
Thus, programmers constantly make choices that are fre-
quently arbitrary—even in crafting simple calculations and formu-
las.
157
Indeed, “in the computer field ‘[t]here exists a virtually unlim-
ited number of instruction sequences that would enable a program-
mer to construct a program which performs even the more basic algo-
rithmic or mathematical procedures.’ ”
158
Given that computer code can be written multiple ways to perform
the same function, it is more likely that a function that destroys in-
teroperability was programmed and distributed in order to foreclose
rivals. Alleged innovative justifications traditionally raise the possi-
bility that a redesign was procompetitive, implicating the concern
that “courts are not competent to second-guess decisions about prod-
uct design”
159
without unduly chilling innovation. With physical
products, a plausible claim can usually be made that design choices
were dictated by either functional, cost-related, or aesthetic concerns,
all of which may be valued by customers.
160
But with the actual writ-
155. See Marc T. Kretschmer, Note, Copyright Protection for Software Architecture:
Just Say No!, 1988 C
OLUM.BUS.L.REV. 823, 825-27 (describing the general process of
software development).
156. Id. at 826.
157. SAS Inst., Inc. v. S & H Computer Sys., Inc., 605 F. Supp. 816, 825 (M.D. Tenn.
1985) (“At every level, the process is characterized by choice, often made arbitrarily, and
only occasionally dictated by necessity. Even in the case of simple statistical calculations,
there is room for variation . . . .”).
158. M. Kramer Mfg. Co. v. Andrews, 783 F.2d 421, 436 (4th Cir. 1986) (alteration in
original) (quoting Anderson L. Baldy III, Note, Computer Copyright Law: An Emerging
Form of Protection for Object Code Software After Apple v. Franklin, 5 C
OMPUTER L.J. 233,
251 (1984)).
159. H
OVENKAMP, supra note 122, at 275.
160. Though courts tend to be somewhat more suspicious of claims that a redesign was
a purely aesthetic improvement, it is not clear that aesthetic improvements should thus be
disfavored (or at least discounted) by antitrust law. “[C]onsumer preferences are driven by
aesthetic concerns as much as by concerns for lower cost, increased speed, decreased size,
lightness, flexibility, and other factors.” A
REEDA &HOVENKAMP, supra note 61, ¶ 776a, at
287.
714 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
ing of lines of source code, functional concerns less frequently dictate
the particular expression chosen by the programmer.
161
And aesthetic
concerns drive decisions only to the extent that the programmer her-
self prefers a certain design for its aesthetic value,
162
for consumers
almost never see the object (let alone the source) code underlying the
programs they use.
163
Thus, alleged innovative justifications are
much more capable of judicial scrutiny in code-based product mar-
kets than in traditional, physical product markets.
Take, for example, the Microsoft III court’s careful dissection of
the files containing the commingled Internet Explorer and Windows
OS code (discussed in greater detail below).
164
In Microsoft III, “the
defendant took advantage of the fact that computer code can readily
be configured in practically any fashion that programmers wish.”
165
Microsoft could offer no innovative or procompetitive justification for
its choices.
166
The fact that computer code can easily, quickly, and
cheaply be configured—or reconfigured—in multiple ways further
alleviates the traditional concern that courts should not second-guess
product-design decisions.
IV.
PRODUCT DESIGN AND TECHNOLOGICAL TIES:EARLY ANALYTICAL
APPROACHES THROUGH THE DIGITAL AGE
Forging a cohesive, efficient analytical framework for addressing
anticompetitive product design in code-based product markets re-
quires first examining the somewhat tangled skein of past design-
related antitrust cases. For over three decades, district and appellate
courts have wrestled with how to approach claims that firms have
engaged in anticompetitive, design-related behavior. A wide-ranging
variety of analyses have been used, though due to the fear of stifling
innovation noted above, courts have on the whole tended to be quite
deferential to defendants.
167
The prevailing school of thought runs
161. See SAS Inst. Inc., 605 F. Supp. at 825. Certainly, at a certain level of abstraction,
generally the organizational flow-charting stage, programmers make choices based primar-
ily on efficiency. But at the very lowest level of abstraction, the writing of lines of source
code itself, multiple ways of expressing the same idea exist.
162. Cf. Anthony L. Clapes et al., Silicon Epics and Binary Bards: Determining the
Proper Scope of Copyright Protection for Computer Programs, 34 UCLA L. R
EV. 1493, 1535
(1987) (“Not surprisingly, each programmer develops her own style of expression.”).
163. See supra note 84 and accompanying text.
164. See Microsoft III, 253 F.3d 34, 66 (D.C. Cir. 2001) (“Another Government expert
likewise testified that one library file, SHDOCVW.DLL, ‘is really a bundle of separate
functions. It contains some functions that have to do specifically with Web browsing, and it
contains some general user interface functions as well.’ ” (quoting the trial testimony of
Edward Felten)).
165. H
OVENKAMP, supra note 122, at 276.
166. See Microsoft III, 253 F.3d at 67.
167. Naturally, this defendant-friendly standard employed by most antitrust courts
lowers one component of R in the formula explored above—the potential costs entailed by
defending or losing an antitrust lawsuit or both.
2012] ANTICOMPETITIVE PRODUCT DESIGN 715
that because (1) the market itself generally disciplines anticompeti-
tive, design-related conduct and (2) antitrust courts are generally not
competent to second-guess design decisions, condemning such con-
duct runs a high risk of producing false positives. Yet, antitrust
courts have generally been called upon to consider design conduct
involving more traditional, physical products, as in the first three
cases discussed below.
The seminal Microsoft III marked the first major antitrust deci-
sion involving design of a computer code-based product; it also
marked a departure from the general mode of analysis used by previ-
ous antitrust courts in design-related cases.
168
Yet default rules favor-
ing defendants are still very much en vogue, as demonstrated by Al-
lied Orthopedic v. Tyco,
169
a 2010 Ninth Circuit Court of Appeals de-
cision also discussed here.
170
This Part concludes with a brief discus-
sion of the In re Intel consent decree,
171
also issued in 2010, as an ex-
ample of the ever-expanding avenues of anticompetitive product de-
sign available to high-technology firms.
A. The IBM Plug-Compatible Peripherals Litigation
In 1964, IBM introduced its extremely successful third-generation
computer systems to the nascent computer system market.
172
IBM’s
resulting profits naturally attracted competitors, and in the late
1960s, “several companies began marketing copies of IBM’s tapes,
disks and printers, which were ‘plug-compatible’ with IBM CPUs.”
173
These plug-compatible devices were not only perfect substitutes for
IBM peripherals, they were much cheaper.
174
In response, IBM intro-
duced the “370” system, which used a “new and improved peripher-
al[]” design that was incompatible with the plug-compatible manu-
facturers’ existing designs.
175
Several of the plug-compatible manufacturers (PCMs) sued IBM
under sections 1 and 2 of the Sherman Act, alleging (among other
168. 253 F.3d 34.
169. Allied Orthopedic Appliances, Inc. v. Tyco Health Care Grp., LP, 592 F.3d 991,
994-95 (9th Cir. 2010).
170. See infra Part IV.E.
171. Agreement Containing Consent Order, In re Intel Corp., No. 9341 (F.T.C.
July 28, 2010) (non-public), available at http://download.intel.com/pressroom/legal/ftc/
FTC_Final_Executed_Agreement.pdf.
172. In re IBM Peripheral EDP Devices Antitrust Litig., 481 F. Supp. 965, 972 (N.D.
Cal. 1979), aff’d sub nom. Transamerica Computer Co. v. IBM Corp., 698 F.2d 1377 (9th
Cir. 1983).
173. Id. at 973.
174. Id. (“A user could simply unplug the IBM peripheral, substitute the cheaper copy,
and plug it into the IBM CPU.”).
175. Id. at 972.
716 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
theories) that IBM’s new product design was anticompetitive.
176
The
theory was fairly straightforward: “[A] monopolist could utilize the
design of its own product to maintain market control or to gain a
competitive advantage.”
177
More specifically, it was essentially a
“technological tie-in” theory—IBM was forcing customers of its com-
puter systems to also buy its peripherals.
The courts addressing these cases set forth some of the founda-
tional principles that have generally been reflected in subsequent
antitrust jurisprudence considering the legality of product design.
Primary among these was the concern for preserving incentives to
innovate. As the district court deciding the suit brought by
Transamerica put it, “[E]quipment design can have pro-competitive
as well as anticompetitive aspects. Truly new and innovative prod-
ucts are to be encouraged, and are an important part of the competi-
tive process.”
178
Taken to the extreme, this concern can result in a
standard like the one employed by the ILC Peripherals Leasing Corp.
court, which reasoned that if expert testimony so much as differs on a
design’s legitimacy, the design is irrebuttably presumed legal.
179
Yet
deciding on a more appropriate standard proved difficult.
180
The op-
posite extreme, per se illegality for product bundling or redesigns
that have the effect of foreclosing rivals’ interoperability, holds the
obvious likelihood of unduly chilling competition through innovation.
The courts addressing these cases generally disposed of the sec-
tion 1 “tying” claims by holding either that the separate-products re-
quirement was not met or that the design yielded “facially plausible
benefits.”
181
As to the section 2 monopolization claims, the courts
generally found that the redesigned product was technologically su-
176. Id. at 971; Telex Corp. v. IBM Corp., 367 F. Supp. 258, 267 (N.D. Okla. 1973),
rev’d, 510 F.2d 894 (10th Cir. 1975).
177. In re IBM Peripheral EDP Devices Antitrust Litig., 481 F. Supp. at 1002.
178. Id. at 1003. Similarly, the district court in Telex Corp. v. IBM Corp. stated, “Nor
should the court unnecessarily involve itself in the task of the administration of . . . product
designs and technological applications or other functions . . . .” 367 F. Supp. 258, 356 (N.D.
Okla. 1973), rev’d, 510 F.2d 894 (10th Cir. 1975). Notably, that court issued a decree en-
joining IBM to release upon request “at the time of first customer shipment of an IBM CPU
or its channel, information describing the design of the electronic interface for such prod-
uct.” Id.
179. ILC Peripherals Leasing Corp. v. IBM Corp., 458 F. Supp. 423, 439 (N.D. Cal.
1978) (“Where there is a difference of opinion as to the advantages of two alternatives
which can both be defended from an engineering standpoint, the court will not allow itself
to be enmeshed ‘in a technical inquiry into the justifiability of product innovations.’ ”)
(quoting Response of Carolina, Inc. v. Leasco Response, Inc., 537 F.2d 1307, 1330 (5th Cir.
1976)).
180. In re IBM Peripheral EDP Devices Antitrust Litig., 481 F. Supp. at 1003 (“It is
more difficult to formulate a legal standard for design conduct than it is to imagine clearly
illegal situations.”).
181. David A. Heiner, Assessing Tying Claims in the Context of Software Integration: A
Suggested Framework for Applying the Rule of Reason Analysis, 72 U. C
HI.L.REV. 123, 125
(2005).
2012] ANTICOMPETITIVE PRODUCT DESIGN 717
perior, and thus it was not “unreasonably restrictive of competi-
tion.”
182
Some imprecisely formulated the standard as “reasonable-
ness”—as the Ninth Circuit put it, the “reasonableness” of IBM’s be-
havior precluded finding an antitrust violation.
183
B. Berkey Photo, Inc. v. Eastman Kodak Co.
Around the same time the protracted IBM antitrust litigation was
wending its way through the court system, Berkey Photo brought a
somewhat similar claim against Eastman Kodak under section 2 of
the Sherman Act.
184
Both companies manufactured still cameras tar-
geted at amateur photographers, although Kodak’s market share far
outstripped Berkey Photo’s.
185
A giant in the field at the time, Kodak
had gained its position through aggressive innovations that were ex-
ceedingly well received by consumers.
186
Berkey Photo manufactured
cameras and film that were interoperable with Kodak’s products.
187
Kodak, however, unveiled a new system, the 110 Instamatic camera
and Kodacolor II film. The camera was smaller and more convenient
than its predecessors and was, by all measures, a “dramatic suc-
cess.”
188
It was also incompatible with Berkey Photo’s products.
Berkey’s antitrust claims included a product-design theory—that
Kodak hastily launched the new products together in order to exclude
Berkey Photo from the camera market.
189
At the trial court level, Ko-
182. In re IBM Peripheral EDP Devices Antitrust Litig., 481 F. Supp. at 1003; see also
Cal. Computer Prods., Inc. v. IBM Corp., 613 F.2d 727, 744 (9th Cir. 1979) (“The reasona-
bleness of IBM’s conduct in this regard did not present a jury issue.”). But see ILC Periph-
erals Leasing Corp., 458 F. Supp. at 430-31 (granting IBM’s motion for a directed verdict as
to the monopolization charges on the grounds that the plaintiff, Memorex, had failed to
show that IBM had monopoly power in the relevant product markets).
183. Cal. Computer Prods., Inc., 613 F.2d at 744; accord Transamerica Computer. Co.
v. IBM, 698 F.2d 1377, 1383 (9th Cir. 1983) (“[W]e affirm the district court’s ruling that the
interface changes were not unreasonable.”).
184. See Berkey Photo, Inc. v. Eastman Kodak Co., 457 F. Supp. 404, 419 (S.D.N.Y.
1978), rev’d, 603 F.2d 263 (2d Cir. 1979).
185. Berkey Photo, Inc., 603 F.2d at 269-70.
186. See generally Vrinda Kadiyali, Eastman Kodak in the Photographic Film Industry:
Picture Imperfect?, in M
ARKET DOMINANCE:HOW FIRMS GAIN,HOLD, OR LOSE IT AND THE
IMPACT ON ECONOMIC PERFORMANCE 89, 91-99 (David I. Rosenbaum ed., 1998) (describing
the competitive strategies that contributed to the rise and prolonged dominance of East-
man Kodak).
187. Berkey Photo, Inc., 603 F.2d at 269 (“Other camera manufacturers, including
Berkey, copied both these inventions but for several months after each introduction anyone
desiring to purchase a camera in the new format was perforce remitted to Kodak.”).
188. Id. at 278.
189. See id. at 285. Berkey Photo also premised its claim to some degree on Kodak’s
failure to give advance notice of the new film’s altered technical specifications—a claim
that has been properly identified as having anti-innovative implications. See Geoffrey A.
Manne & Joshua D. Wright, Innovation and the Limits of Antitrust, 6 J. C
OMPETITION L. &
E
CON. 153, 173-74 (2010). If firms were required to disclose novel innovations to competi-
tors before introducing them to the market, antitrust law would essentially be “turning a
novel innovation into an old hat.” Id. at 174. The first-mover advantage that accrues to
718 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
dak contended that introducing new products must “ ‘as a matter of
law’ be immune from attack under the antitrust laws.”
190
The trial
court rightly rejected this argument, reasoning that product intro-
ductions, like other ordinarily legal business decisions, may in the
right circumstances be anticompetitive.
191
Yet it declined to enumer-
ate the proper analysis. On appeal, the Second Circuit held that even
though there was evidence that Kodacolor II was inferior to Kodacol-
or X in several ways, there was also evidence that it was an im-
provement in at least one regard.
192
The court thus held that, as a
matter of law, the design and introduction were not anticompetitive,
recognizing that “[p]reference is a matter of individual taste.”
193
The analysis employed by the Second Circuit was more focused
than the “reasonableness” standard identified by the Ninth Circuit.
194
It was also very deferential to the possibility of harming innovation.
Whereas the IBM decisions generally rested on a finding that the
new product was on the whole superior to the old, the Second Circuit
required only evidence of some innovation or improvement—even in
the face of proven disadvantages—for the product design to be held
lawful.
195
Notably, the Berkey Photo court also identified customer
acceptance of the redesign as relevant to its inquiry; while such evi-
dence may be relevant in traditional product markets,
196
for reasons
discussed subsequently, it should bear much less heavily in design
cases involving code-based products.
197
C. C.R. Bard, Inc. v. M3 Systems, Inc.
A less defendant-friendly analysis was used by the Federal Circuit
in deciding C.R. Bard, Inc. v. M3 Systems, Inc.
198
Like Berkey Photo
and the IBM plug-compatible peripherals litigation, C.R. Bard in-
volved an antitrust challenge to a dominant firm’s redesign of physi-
cal products. In C.R. Bard, the eponymous firm manufactured an au-
tomated, reusable biopsy gun, as well as replacement needles for the
successful innovators is a crucial element in incentivizing innovative (and thus procompeti-
tive) behavior.
190. Berkey Photo, Inc. v. Eastman Kodak Co., 457 F. Supp. 404, 411 (S.D.N.Y. 1978),
rev’d, 603 F.2d 263 (2d Cir. 1979).
191. Id. But cf. Sidak, supra note 7, at 1143-48 (arguing that a rule of per se legality
should apply to product innovations).
192. Berkey Photo, Inc., 603 F.2d at 286.
193. Id. at 287.
194. See Cal. Computer Prods., Inc. v. IBM Corp., 613 F.2d 727 (9th Cir. 1979).
195. Berkey Photo, Inc., 603 F.2d at 286-87.
196. See A
REEDA &HOVENKAMP, supra note 61, ¶ 777b, at 306 (“Clearly, antitrust
should not intervene when the invention pleases consumers . . . .”).
197. See supra Part III.B.2. (demonstrating that, in such markets, there is a relatively
low risk that consumers will reject even anticompetitive redesigns).
198. 157 F.3d 1340 (Fed. Cir. 1998).
2012] ANTICOMPETITIVE PRODUCT DESIGN 719
gun.
199
M3 Systems, a competitor, manufactured identical replace-
ment needles that were substitutes for C.R. Bard’s needles.
200
Bard
responded by redesigning its biopsy-sampling gun so that M3 Sys-
tems’ needles were no longer interchangeable (and thus, no longer
usable as substitutes for Bard’s needles).
201
M3 Systems then sued
Bard under section 2, basing its claims on a technological tie-in theo-
ry—that Bard had “leveraged its monopoly power in the guns to ob-
tain a competitive advantage in replacement needles by modifying its
gun to accept only Bard needles.”
202
On appeal, the Federal Circuit opened its analysis of the jury’s
antitrust verdicts by identifying a standard that seemed to hinge on
intent. As Judge Bryson put it, “M3 was required to prove that Bard
made a change in its Biopty gun for predatory reasons, i.e., for the
purpose of injuring competitors in the replacement needle market,
rather than for improving the operation of the gun.”
203
To support
this, the Federal Circuit decision cited one of the IBM plug-
compatible decisions discussed above.
204
Yet, there, the district court
had merely identified two hypothetical situations in which a monopo-
list redesigned a product with the sole “purpose and effect” of fore-
closing rivals,
205
emphasizing the point that it was not thereby at-
tempting to announce an analytical framework.
206
The heavy focus on
intent, then, represented a novel approach to product-design analysis.
En route to ultimately affirming the jury’s finding of a section 2
violation, the appellate decision admitted that at least some evidence
had been adduced at trial to the effect that Bard had modified its gun
to facilitate loading and unloading of needles—a procompetitive in-
novation.
207
Yet the Federal Circuit pointed out that “there was sub-
stantial evidence that Bard’s real reasons for modifying the gun were
to raise the cost of entry to potential makers of replacement needles, to
make doctors apprehensive about using non-Bard needles, and to pre-
199. Id. at 1346.
200. See id. at 1382.
201. Id.
202. Id. at 1367.
203. Id. at 1382 (citing In re IBM Peripheral EDP Devices Antitrust Litig., 481 F.
Supp. 965, 1002 (N.D. Cal. 1979), aff’d sub nom. Transamerica Computer Co. v. IBM Corp.,
698 F.2d 1377 (9th Cir. 1983)).
204. Id.
205. In re IBM Peripheral EDP Devices Antitrust Litig., 481 F. Supp. at 1002-03.
206. Id. at 1003 (“It is more difficult to formulate a legal standard for design conduct
than it is to imagine clearly illegal situations.”).
207. C.R. Bard, Inc., 157 F.3d at 1382 (“Bard contended at trial that it modified its
Biopty gun to make it easier to load and unload . . . .”).
720 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
clude the use of ‘copycat’ needles.”
208
Thus, the court held that a jury
could reasonably have found that the redesign was anticompetitive.
209
Like the record before the Second Circuit in Berkey Photo, the rec-
ord in C.R. Bard contained conflicting evidence as to whether the
product design constituted an overall innovation. Yet the Second Cir-
cuit in Berkey Photo held that in such cases the design is not anticom-
petitive as a matter of law. The Federal Circuit in C.R. Bard, on the
other hand, held that such cases may, at the very least, go to the jury.
D. United States v. Microsoft Corp.
In the D.C. Circuit’s 2001 decision in the Microsoft antitrust liti-
gation (often referred to as “Microsoft III”), one area of Microsoft’s
challenged conduct involved the integration of Internet Explorer and
the Windows OS.
210
Microsoft primarily achieved this integration via
three means: (1) excluding the Internet Explorer programs from the
Windows OS’s “Add/Remove Programs” function; (2) programming
Windows such that, when users chose to set Internet browsers other
than Internet Explorer as their default browsers, Windows some-
times overrode that choice; and (3) commingling Internet Explorer’s
code with Windows code “so that any attempt to delete the files con-
taining IE would, at the same time, cripple the operating system.”
211
These tactics, it was alleged, were designed to deter Original Equip-
ment Manufacturers (OEMs) and consumers from using different In-
ternet browsers, particularly Netscape Navigator,
212
which could per-
form basic OS functions (thus threatening Microsoft’s OS monopoly).
Thus, while it bore some resemblance to more traditional technolog-
ical-tying claims, the United States’ allegations were more complex
208. Id.
209. It has been urged that this was an example of the court using (if implicitly) a “bal-
ancing test” to determine whether a product redesign was, on the whole, anticompetitive.
See Jonathan Jacobson, Scott Sher & Edward Holman, Predatory Innovation: An Analysis
of Allied Orthopedic v. Tyco in the Context of Section 2 Jurisprudence, 23 L
OY.CONSUMER
L. REV. 1, 13 (2010) (“Although the [C.R.] Bard court did not articulate a ‘balancing test’ for
determining whether such conduct violated the antitrust laws, the decision squarely fo-
cused on whether, on balance, Bard’s conduct would harm customers, even though it was
undisputed that the ‘innovation’ did marginally improve the Bard product.”). It must be
remembered, however, that the C.R. Bard court was merely deciding whether there was
sufficient evidence to support the jury’s finding that the antitrust defendant had engaged
in exclusionary conduct—a decision that falls under the “clearly erroneous” standard of
review. Rather than balancing, then, the Federal Circuit was concerned merely with
whether the jury’s decision was clearly incorrect. Further, recall that the Federal Circuit’s
analysis began with the premise that product-design liability turns on a firm’s purpose. See
supra note 203-206 and accompanying text. As a result, if the court was engaged in any
balancing, it was not looking to the redesign’s anticompetitive or procompetitive effects;
rather, it was considering Bard’s purpose in redesigning the gun and replacement needles.
210. Microsoft III, 253 F.3d 34, 45 (D.C. Cir. 2001).
211. Id. at 64-65.
212. Id. at 63.
2012] ANTICOMPETITIVE PRODUCT DESIGN 721
than a simple tie-in theory. The district court found them convinc-
ing and held that Windows had thereby violated section 2 of the
Sherman Act.
213
Like the Federal Circuit in C.R. Bard, the D.C. Circuit’s Microsoft
III decision was relatively less deferential to the possibility of deter-
ring innovation.
214
Rather, it applied a more traditional section 2
analysis that used burden-shifting and (somewhat controversially) a
“balancing” inquiry. In analyzing the three instances of conduct set
forth above, the D.C. Circuit stated that the initial burden rests on
the plaintiff to show that the defendant’s conduct caused anticompet-
itive effects.
215
The court found that, in this case, the United States
had “plainly” made a prima facie case of anticompetitive effect.
216
The
court’s analysis then shifted the burden to Microsoft to show some
procompetitive justification for its behavior.
217
As to (1) and (3), it
held that Microsoft failed to show that the conduct served any pur-
pose other than maintaining its monopoly on the OS market.
218
As to
(2), the court held that Microsoft alleged “valid technical reasons” for
its actions.
219
As a result, the burden shifted back to the United
States to rebut the justification and show “that the anticompetitive
effect of the challenged action outweighs it.”
220
And since no rebuttal
was offered, the D.C. Circuit found this aspect of Microsoft’s design-
related conduct to be legal.
221
Notably, the Microsoft III product-design claims dealt with soft-
ware products, making the D.C. Circuit’s holding the first major deci-
sion on potentially anticompetitive software design.
222
Microsoft III,
then, presented the first example of an antitrust court examining ex-
pert testimony on computer-code-based products, and the D.C. Cir-
213. United States v. Microsoft Corp., 87 F. Supp. 2d 30, 44 (D.D.C. 2000) (“[T]he pred-
atory nature of [Microsoft’s] conduct compels the Court to hold Microsoft liable under § 2 of
the Sherman Act.”).
214. The D.C. Circuit began by recognizing the concern for avoiding undue deterrence
to innovation that animates much of the product-design jurisprudence. 253 F.3d at 65 (“In
a competitive market, firms routinely innovate in the hope of appealing to consumers,
sometimes in the process making their products incompatible with those of rivals; the im-
position of liability when a monopolist does the same thing will inevitably deter a certain
amount of innovation.”). It quickly went on to state, however, that “[j]udicial deference to
product innovation . . . does not mean that a monopolist’s product design decisions are per
se lawful.” Id. at 65.
215. Id. at 59.
216. Id. at 67.
217. Id. at 59.
218. Id. at 67.
219. Id. (internal quotation marks omitted). The accepted justification was that some of
Windows’ Internet-based functions depended on controls and formats not supported by
competitors’ Internet browsers.
220. Id.
221. Id.
222. Conversely, “[t]he IBM cases of the 1970s all concerned computer hardware.” Hei-
ner, supra note 181, at 126.
722 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
cuit’s seminal opinion reflects some of the unique aspects of such
products.
223
Unlike, for example, the Kodacolor II film under discus-
sion in Berkey Photo, the code underlying Internet Explorer was capa-
ble of dissection into and analysis of separate functions,
224
allowing the
court more freedom to analyze anticompetitive aspects of its design.
E. Allied Orthopedic Appliances, Inc. v. Tyco Health Care Group LP
In Allied Orthopedic, the defendant, Tyco Healthcare Group LP,
was a dominant manufacturer of its patented pulse-oximetry sys-
tems.
225
Anticipating that the expiration of its patent would allow for
a flood of generic entries, Tyco drew up a new design for its monitor-
and-sensor system that moved the system’s digital memory chip from
the monitor to the sensor, making generic sensors incompatible.
226
A
group of competitors, including four rivals that had planned to begin
manufacturing compatible generic sensors, sued Tyco on a theory re-
sembling a technology-tying claim.
227
The design decision, however,
also “allowed Tyco to add new features” to the sensors
228
and reduced
the switching costs incurred when customers decided to adopt a new
type of sensor—which, the Ninth Circuit reasoned, in turn facilitated
innovation in the sensor product market.
229
In weighing the merits of the plaintiff’s claim, the court empha-
sized that, “[a]s a general rule, courts are properly very skeptical
about claims that competition has been harmed by a dominant firm’s
product design changes.”
230
It then rightly noted that product-design
changes are not categorically immune from scrutiny under antitrust
laws.
231
Strongly departing from the D.C. Circuit’s Microsoft III deci-
sion, the Allied Orthopedic court eliminated the third step of the D.C.
Circuit’s analysis, the “balancing” of procompetitive benefits and an-
ticompetitive harm. As the Ninth Circuit put it:
223. The Microsoft III decision has been called “the seminal decision with regard to
claims of anticompetitive product redesign.” See Jacobson et al., supra note 209, at 21.
While this perhaps overstates the case’s significance, the decision has been heavily dis-
cussed by other courts, antitrust enforcers, and legal scholars—it has been cited over 3000
times.
224. See Microsoft III, 253 F.3d at 66 (“Another Government expert likewise testified
that one library file, SHDOCVW.DLL, ‘is really a bundle of separate functions. It contains
some functions that have to do specifically with Web browsing, and it contains some gen-
eral user interface functions as well.’ ”) (quoting the trial testimony of Edward Felten).
225. Allied Orthopedic Appliances Inc. v. Tyco Health Care Group LP, 592 F.3d 991,
994 (9th Cir. 2010).
226. Id.
227. Allied Orthopedic Appliances, Inc. v. Tyco Health Care Group L.P., No. CV 05-
06419 MRP (AJWx), 2008 WL 7346921, at *1-2 (C.D. Cal. July 9, 2008).
228. Allied Orthopedic Appliances Inc., 592 F.3d at 994.
229. Id. at 994-95.
230. Id. at 998 (alteration in original) (quoting Microsoft III, 253 F.3d 34, 65 (D.C. Cir.
2001)) (internal quotation marks omitted).
231. Id.
2012] ANTICOMPETITIVE PRODUCT DESIGN 723
There is no room in this analysis for balancing the benefits or
worth of a product improvement against its anticompetitive ef-
fects. If a monopolist’s design change is an improvement, it is “nec-
essarily tolerated by the antitrust laws” . . . .
. . .
To weigh the benefits of an improved product design against
the resulting injuries to competitors is not just unwise, it is unad-
ministrable. There are no criteria that courts can use to calculate
the “right” amount of innovation, which would maximize social
gains and minimize competitive injury.
232
Given that it was “undisputed” that Tyco’s redesigned sensor allowed
for some improvements, the court’s rubric naturally dictated that the
product redesign did not violate the Sherman Act.
233
This forceful
holding led some scholars to characterize it as having established a
per se rule shielding product redesign from antitrust scrutiny.
234
While this is likely overstating the Ninth Circuit’s position to some
degree, it is not too far off the mark. The Allied Orthopedic analysis
would presumably still allow a court to find an antitrust violation in
a situation like, for example, Microsoft III, where the defendant could
offer no procompetitive justification whatsoever for its behavior. The
analysis would, however, likely immunize the behavior condemned in
C.R. Bard, where the defendant produced evidence at trial that its
redesign created at least some product improvements.
235
F. In re Intel
On December 16, 2009, the Federal Trade Commission (FTC) filed
a complaint against Intel Corporation, alleging that it had violated
section 5 of the Federal Trade Commission Act,
236
which bans
“[u]nfair methods of competition in or affecting commerce.”
237
In the
Intel complaint, the FTC charged Intel with anticompetitively rede-
232. Id. at 1000 (citation omitted).
233. Id. at 1001-02.
234. See Jacobson et al., supra note 209, at 4 (“[T]he Ninth Circuit’s decision in Allied
Orthopedic v. Tyco—which established a per se rule protecting redesign—was a significant
departure from the normal standards employed to determine whether a product redesign is
predatory . . . .”).
235. C.R. Bard, Inc. v. M3 Systems, Inc., 157 F.3d 1340, 1382 (Fed. Cir. 1998).
236. See Complaint of Petitioner, supra note 69, at 1.
237. 15 U.S.C. § 45(a)(1) (2006). This language has been interpreted as granting the
FTC jurisdiction over anticompetitive practices. See generally Appendix 1—Laws Enforced
by the FTC, F
ED.TRADE COMM’N (last modified June 25, 2007), http://www.ftc.gov/opp/
gpra/append1.shtm (referring to the FTC’s “antitrust law enforcement responsibilities”).
There is some debate over whether, and to what extent, the scope of section 5 of the FTC
Act extends beyond that of the Sherman Act. Herbert Hovenkamp, The Federal Trade
Commission and the Sherman Act, 62 F
LA.L.REV. 871, 873 (2010) (“[T]he power to con-
demn undefined practices that have been held not to violate the Sherman Act is hardly
uncontroversial.”).
724 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
signing and selling its “compiler” software.
238
“Compilers” consist of
code-based software that serves a sort of translating function—they
convert source code (written by human programmers) into object code
(binary that can be run through a “linker” utility that combines it
into an executable program).
239
The FTC alleged that Intel deliberate-
ly programmed its compilers so that they output software programs
that ran more slowly on computers with non-Intel CPUs than on
those with Intel CPUs.
240
This, in turn, would cause consumers to
perceive non-Intel CPUs (such as those produced by AMD, Intel’s
main rival) as inferior to Intel CPUs. According to the FTC, this con-
duct had no procompetitive justification.
241
The complaint also al-
leged that Intel initially encouraged manufacturers of graphics pro-
cessing units (GPUs) to innovate based on Intel’s CPU platform.
242
This interoperability was made possible by Intel’s use of open inter-
faces between its CPUs and chipsets. Yet after perceiving increasing-
ly sophisticated GPUs as a threat to its dominant position in the
CPU market, Intel “selectively cut off or hindered” interoperability.
243
Intel agreed to a consent decree on July 28, 2010;
244
and the FTC
issued its decree on October 29, 2010.
245
As to the GPU-
interoperability issue specifically, the decree required that Intel re-
frain from designing any of its interfaces so as to intentionally make
any competitors’ GPUs noninteroperable with Intel’s CPUs.
246
And,
as to the product-design claims generally, the consent decree re-
quired that Intel “not make any engineering or design change” to its
products if the change “(1) degrades the performance of a . . .
[p]roduct sold by a competitor of [Intel] and (2) does not provide an
actual benefit” to the Intel product.
247
The decree further put the
burden of showing such a benefit on Intel.
248
Interestingly, the FTC
explained in a Notice accompanying the proposed consent agreement
238. Complaint of Petitioner, supra note 69, at 10.
239. Definition of: Compiler, PCM
AG.COM, http://www.pcmag.com/encyclopedia_term/
0,2542,t=compiler&i=40105,00.asp (last visited Feb. 13, 2012).
240. Complaint of Petitioner, supra note 69, at 10. The FTC further alleged that “Intel
represented to [users] that programs inherently performed better on Intel CPUs than on
competing CPUs.” Id.
241. Id. at 15.
242. Id. at 14.
243. Id. at 14-15.
244. Agreement Containing Consent Order, In re Intel Corp., No. 9341 (F.T.C.
July 28, 2010) (non-public), available at http://download.intel.com/pressroom/legal/ftc/
FTC_Final_Executed_Agreement.pdf.
245. Decision and Order at 21, In re Intel Corp., No. 9341 (F.T.C. Oct. 29, 2010), avail-
able at http://www.ftc.gov/os/adjpro/d0341/101102inteldo.pdf.
246. Id. at 6.
247. Id. at 13.
248. Id.
2012] ANTICOMPETITIVE PRODUCT DESIGN 725
that merely showing “actual benefits” would be sufficient.
249
Thus,
the decree did not contemplate any sort of “balancing” of anticompeti-
tive harms and procompetitive benefits.
250
Yet the Notice went on to
state that “[a] balancing test would be appropriate in a legal chal-
lenge to an Intel design change under Section 5 of the FTC Act or
Section 2 of the Sherman Act.”
251
Like Microsoft III (and unlike the other product-design cases dis-
cussed supra), In re Intel dealt with software design. And software is
essentially a code-based, intangible product rather than physical
hardware like camera film, biopsy-gun replacement needles, or even
the peripherals in the IBM plug-compatible cases. While it would be
overreaching to infer too much from Intel’s entering into a consent
decree rather than having been found guilty of an antitrust violation,
In re Intel at the very least demonstrates the potential that code-
based products offer for anticompetitive product redesign, a potential
that presents unique problems for antitrust law and its enforcers.
V. L
OOKING FORWARD:ASUGGESTED ANALYTICAL FRAMEWORK
The Supreme Court has never decided a section 2 claim based on a
theory of anticompetitive product redesign, leaving lower courts to
employ the broad spectrum of analytical approaches in cases like
those discussed above. As a baseline principle, all courts have recog-
nized that “product improvement without more is protected and be-
yond antitrust challenge.”
252
Beyond that, however, antitrust courts
and scholars have differed on the proper approach to product-
redesign liability. This Part seeks both to demonstrate the flaws of
past analyses as applied to design-related conduct in code-based
product markets and to set forth a structured, efficient, and usable
method for analyzing such conduct in the future.
A. Defendant-Friendly Approaches Are Inappropriate
On one end of the spectrum, some have argued that a rule of per
se legality should apply to all product innovations.
253
No court has yet
gone so far, though some have come close. The Allied Orthopedic v.
Tyco opinion, for one, has been characterized as creating such a
249. Analysis of Proposed Consent Order to Aid Public Comment, 75 Fed. Reg. 48338,
48345 (F.T.C. Aug. 10, 2010), available at http://www.ftc.gov/os/fedreg/2010/august/
100810intelconsent.pdf.
250. Id.
251. Id.
252. A
REEDA &HOVENKAMP, supra note 61, ¶ 776a, at 286 (citing Transamerica Com-
puter Co. v. IBM, 481 F. Supp. 965 (N.D. Cal. 1979), aff’d, 698 F.2d 1377 (9th Cir. 1983)).
253. See Sidak, supra note 7, at 1148 (“Courts should advance from their strong pre-
sumptions of legality for technological tie-ins and acknowledge that marketing strategies
for product innovations should be per se legal.”).
726 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
rule.
254
Even Allied Orthopedic, however, required that the defendant
demonstrate at least some procompetitive justification for the rede-
sign; if any justification is shown, the Ninth Circuit reasoned, the
conduct would be legal.
255
The court in ILC Peripherals Leasing Corp.
used a similar analysis, holding that if expert testimony so much as
differs on a design’s legitimacy, the design is irrebuttably presumed
legal.
256
And the Second Circuit in Berkey Photo used a somewhat
similar approach, holding that in the face of conflicting evidence as to
whether the new product is an improvement, the conduct is legal (at
least given demonstrated consumer preference for the new product).
257
What all these approaches have in common is that they place a
thumb on the scale in favor of defendants, at least as compared to the
generally used section 2 exclusionary-conduct inquiry,
258
essentially a
rule-of-reason analysis. The D.C. Circuit in Microsoft III set forth the
general method of analysis, complete with allocations of the burden
of proof. First, the burden is on the plaintiff to make a prima facie
case that the defendant has engaged in monopolistic conduct (proper-
ly defined).
259
If the plaintiff does so, the burden then shifts to the
defendant to show a procompetitive justification for the redesign.
260
If
the defendant fails to do so, the conduct is exclusionary.
261
If, howev-
er, the defendant shows some plausible justification, the burden
shifts back to the plaintiff to rebut that justification.
262
If the plaintiff
fails to do so, then the plaintiff must show that the anticompetitive
harm outweighs the procompetitive justification.
263
The leading trea-
254. See Jacobson et al., supra note 209, at 4 (arguing that “the Ninth Circuit’s decision
in Allied Orthopedic v. Tyco . . . established a per se rule protecting redesign”).
255. Allied Orthopedic Appliances Inc. v. Tyco Health Care Grp, 592 F.3d 991, 1000-02
(9th Cir. 2010).
256. ILC Peripherals Leasing Corp. v. IBM Corp., 458 F. Supp. 423, 439 (N.D. Cal.
1978) (“Where there is a difference of opinion as to the advantages of two alternatives
which can both be defended from an engineering standpoint, the court will not allow itself
to be enmeshed ‘in a technical inquiry into the justifiability of product innovations.’ ”)
(quoting Response of Carolina, Inc. v. Leasco Response, Inc., 537 F.2d 1307, 1330 (5th Cir.
1976)).
257. Berkey Photo, Inc. v. Eastman Kodak Co., 603 F.2d 263, 286-87 (2d Cir. 1979).
258. Broadly speaking, in the realm of section 2 jurisprudence, “a structured approach
to the issue of exclusionary conduct has emerged.” 2 J
OSEPH P. BAUER &WILLIAM H. PAGE,
K
INTER FEDERAL ANTITRUST LAW ¶ 14.15, at 353 (2002) (describing a burden-shifting
analysis).
259. Microsoft III, 253 F.3d 34, 58-59 (D.C. Cir. 2001). There are, of course, myriad
definitions of what constitutes “monopolistic conduct”; which (if any) is “correct” may vary
depending on the particular facts of a case and is, at any rate, far beyond the scope of the
present discussion. See generally 3 P
HILLIP E. AREEDA &HERBERT HOVENKAMP,ANTITRUST
LAW:AN ANALYSIS OF ANTITRUST PRINCIPLES AND THEIR APPLICATION ¶ 651, at 96-109 (3d
ed. 2006) (describing alternative definitions, their various advantages and disadvantages,
and the degree to which they have been used by courts).
260. Microsoft III, 253 F.3d at 59.
261. Id.
262. Id.
263. Id.
2012] ANTICOMPETITIVE PRODUCT DESIGN 727
tise takes issue with the last step, at least insofar as it seems to call
for courts to engage in “balancing” of close cases—advocating instead
a burden-shifting analysis that, while perhaps somewhat less de-
fendant-friendly than the above approaches, calls for “resolv[ing]
close cases in favor of the defendant.”
264
The various approaches de-
scribed above, however, end the analysis and dismiss the claim as
soon as the defendant shows any plausible justification for its behav-
ior. This favorable treatment traditionally accorded to defendants in
this area is due largely to the concerns noted above—the fear that,
because (1) the markets themselves act as a check on exclusionary
product redesigns (making them quite rare) and (2) antitrust courts
are generally not competent to second-guess design changes, con-
demning product redesigns will tend to unduly stifle innovation.
Yet, as shown above, these concerns largely dissipate in the types
of markets under discussion. As to the first, the nature of code-based
products and the widespread availability of high-speed Internet ac-
cess have combined to make the now standard method of redesigning
these products—software updates—a uniquely attractive method of
foreclosing rivals. This is so for three primary reasons: (1) low devel-
opment and distribution costs,
265
(2) low risk that consumers will re-
ject redesigns,
266
and (3) low losses incurred if these product rede-
signs fail.
267
Additionally, new-economy markets tend to be character-
ized by strong positive network externalities, which may further in-
centivize monopolistic behavior.
268
Given the confluence of these fac-
tors, it is much more likely that C
i
> P
m
– LR in these markets.
And with regard to the second concern, as shown above, the inher-
ent and unique nature of code-based product redesign makes it
uniquely susceptible to antitrust scrutiny.
269
Given that such rede-
signs are more easily analyzed than traditional, physical product re-
designs, it should come as no surprise that firms may be able to offer
no justification for their conduct (as occurred in Microsoft III). Alter-
natively, they may simply settle out of court or enter into consent de-
crees (as may have occurred in In re Intel). At any rate, the point is
that antitrust courts no longer need to simply throw up their hands
and find for defendants in design-related cases.
Since these concerns largely dissipate in these markets, the need
to place a thumb on the scale in favor of defendants—that is, the
need for the inquiry to end as soon as the defendant makes any plau-
264. AREEDA &HOVENKAMP, supra note 259, ¶ 651e2, at 122.
265. See supra Part III.B.1.
266. See supra Part III.B.2.
267. See supra Part III.B.3. Further, if the relevant market is a network market, as these
markets tend to be, the incentive to engage in anticompetitive behavior even higher.
268. See supra notes 60-69 and accompanying text.
269. See supra Part III.C.
728 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
sible claim of a procompetitive benefit—dissipates as well. And in the
formula expressed above, a defendant-friendly approach lowers R by
reducing the risk of antitrust liability for engaging in exclusionary,
design-related conduct. Absent the usual check of market forces, such
an approach even further incentivizes such conduct. Firms can and
almost certainly do engage in anticompetitive design in these mar-
kets; witness Microsoft’s commingling of code,
270
the FTC’s theory in
In re Intel,
271
or Apple’s allegedly exclusionary software updates.
272
While courts are rightly reluctant to review antitrust challenges to
physical product design changes, code-based product markets exhibit
unique features that obviate the need for an overly defendant-
friendly analysis.
B. “Actual Improvement” and “Reasonableness”
Alternative approaches used by courts in evaluating challenged
product redesigns include whether the redesign constituted an “actu-
al improvement”
273
or whether the defendant’s conduct was “reasona-
ble.”
274
Both are inappropriate for somewhat similar reasons—
primarily that they provide little helpful guidance to courts consider-
ing code-based product-redesign claims. Further, the first would tend
to misdirect focus in such cases; the second is simply too vague to be
of much use.
1. “Actual Improvement”: Unfocused
Inquiring whether a product redesign constitutes an “actual im-
provement” in the realm of code-based products involves too high a
level of abstraction in analyzing defendants’ conduct. As an initial
matter, it is geared toward scrutinizing redesigned physical products,
which (as noted above) may contain inextricably interwoven design
elements that serve to both improve the product and destroy interop-
erability with rivals’ complementary products. Thus, if the rede-
signed product as a whole was an improvement, it was deemed legal
because courts were cautious (and likely incapable) of attempting to
dissect these elements.
275
Yet the coded functions that underlie soft-
270. Microsoft III, 253 F.3d 34, 64-65 (D.C. Cir. 2001).
271. Complaint of Petitioner, supra note 69, at 5.
272. The Apple iPod iTunes Antitrust Litigation, Nos. C 05-00037 JW, C 07-06507 JW,
2010 WL 2629907, at *3 (N.D. Cal. June 29, 2010).
273. In re IBM Peripheral EDP Devices Antitrust Litig., 481 F. Supp. 965, 1003 (N.D.
Cal. 1979) aff’d sub nom. Transamerica Computer Co. v. IBM Corp., 698 F.2d 1377 (9th
Cir. 1983).
274. See Cal. Computer Prods., Inc. v. IBM Corp., 613 F.2d 727, 744 (9th Cir. 1979)
(“The reasonableness of IBM’s conduct in this regard did not present a jury issue.”); accord
Transamerica Computer Co. v. IBM Corp., 698 F.2d 1377, 1383 (9th Cir. 1983) (“[W]e af-
firm the district court’s ruling that the interface changes were not unreasonable.”).
275. See In re IBM Peripheral EDP Devices Antitrust Litig., 481 F. Supp. at 1003.
2012] ANTICOMPETITIVE PRODUCT DESIGN 729
ware products are uniquely separable from the whole product,
276
and
a particular line of code is much less likely to constitute both an “ac-
tual improvement” and an exclusionary element, especially given the
fact that code can be configured multiple ways to perform the same
function.
277
Thus, courts are less bound to analyze products as a
whole and are more free to directly scrutinize particularized instanc-
es of design-related conduct. Additionally, code-based products are
today almost exclusively redesigned through software updates deliv-
ered via the Internet; as noted above, these updates are thus capable
of isolation and separate analysis from the base product.
278
2. Reasonableness: Overly Vague
Insofar as the “reasonableness” standard is simply shorthand for
antitrust’s rule-of-reason analysis, it does little harm, for “[t]he law of
exclusionary practices lies at the center of antitrust’s rule of rea-
son.”
279
But given the complex issues at stake, the substantial harm
that can be caused by exclusionary design practices (particularly in
network markets),
280
and the potential for chilling innovation that is
present in all product-design and product-redesign cases, reasona-
bleness is too broad a standard to serve as a standalone approach.
Whether a change is “reasonable” is an inquiry that offers little value
to antitrust courts
281
and, perhaps worse, provides little to no real
predictability for firms considering innovation.
C. Intent: Relevant, but Not Dispositive
While “intent” should be relevant to some degree in addressing
product-design and -redesign claims in code-based product markets,
it should not alone be dispositive. Recall that in C.R. Bard, the court
characterized the inquiry as whether the plaintiff had shown that the
defendant had engaged in conduct “for predatory reasons, i.e., for the
purpose of injuring competitors in the replacement needle market,
rather than for improving the operation of the gun.”
282
Thus, its ap-
276. See supra Part III.C.2.
277. See supra Part III.C.3.
278. See supra Part III.C.1.
279. A
REEDA &HOVENKAMP, supra note 259, ¶ 651e3, at 121.
280. See H
OVENKAMP, supra note 122, at 281-82 (highlighting the increased threat of
anticompetitive practices in network markets). But cf. P
OSNER, supra note 34, at 249 (“The
gale of creative destruction that Schumpeter described, in which a sequence of temporary
monopolies operates to maximize innovation that confers social benefits far in excess of the
social costs of the short-lived monopoly prices that the process also gives rise to, may be the
reality of the new economy.”).
281. Cf. Cudahy & Devlin, supra note 7, at 59 (“Inherently nebulous terms such as
‘reasonable,’ ‘sufficient,’ ‘foreseeable,’ and ‘justified’ pervade the law.”).
282. C.R. Bard, Inc. v. M3 Systems, Inc., 157 F.3d 1340, 1382 (Fed. Cir. 1998) (citing In
re IBM Peripheral EDP Devices Antitrust Litig., 481 F. Supp. 965, 1002 (N.D. Cal. 1979),
aff’d sub nom. Transamerica Computer Co. v. IBM Corp., 698 F.2d 1377 (9th Cir. 1983)).
730 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
proach seemed to hinge on subjective intent, rather than actual or
likely anticompetitive effects. Other courts, including the Supreme
Court, have been less than clear on the import of “intent” in monopo-
lization cases.
283
Yet the core concern of antitrust law is with prevent-
ing anticompetitive effects. Consequently, “[a]ntitrust appraisal of
conduct depends on an understanding of its likely anticompetitive
consequences as well as its possible social benefits.”
284
The focus
should be primarily directed toward effects, not intent.
And this is doubly true in the realm of product-redesign jurispru-
dence. A firm that redesigns its product generally has the goal of
harming rivals by causing customers to substitute away from their
products and to the firm’s. Yet this goal is not in itself an antitrust
concern
285
—a firm that is wildly innovative may be motivated solely
by the desire to drive its rivals out of the market, yet directly and
greatly increase both consumer and societal wealth through its inno-
vations. As a result, it is frequently difficult to distinguish competi-
tive intent from monopolistic intent.
286
Allowing intent inquiries to
inform antitrust review of design-related conduct poses a high risk of
confusing the pertinent issues. Again, antitrust’s focus must be on
effects; thus, subjective intent should be deemed relevant only in am-
biguous cases where it can potentially shed light on such conse-
quences or benefits.
287
C.R. Bard may have been such a case—as not-
ed above, in addressing the comprehensive redesign of a physical
product, it is difficult (if not impossible) to conceptually separate an-
ticompetitive from beneficial design elements. If so, the court proper-
ly used intent as a means of discerning likely anticompetitive effects.
But, as demonstrated supra, code-based product redesigns will rarely
present a truly “ambiguous” case. Thus, in such markets, intent should
rarely be necessary, and intent alone should never be dispositive.
D. Consumer Acceptance, Not Preference: A Constrained
and Uninformed “Choice”
The court in Berkey Photo relied at least to some degree on evi-
dence of consumer preference to find that the new camera-and-film
283. Compare United States v. Aluminum Co. of Am., 148 F.2d 416, 431 (2d Cir. 1945)
(“We disregard any question of ‘intent.’ ”), with United States v. Grinnell Corp., 384 U.S.
563, 570-71 (1966) (stating that “willful acquisition or maintenance” of monopoly power is
an element of the monopolization offense).
284. A
REEDA &HOVENKAMP, supra note 259, ¶ 651c, at 112.
285. See B
AUER &PAGE, supra note 258, ¶ 14.16, at 354 (“Nor is it unlawful, by itself,
for a firm to harm its rivals . . . .”).
286. See id. ¶ 14.17, at 357 (“In competitive markets, managers often use warlike lan-
guage in referring to their rivals, even if they only plan to compete vigorously.”).
287. See Aspen Skiing Co. v. Aspen Highlands Skiing Corp., 472 U.S. 585, 602 (1985)
(“In [monopolization offenses] evidence of intent is merely relevant to the question whether
the challenged conduct is fairly characterized as ‘exclusionary’ or ‘anticompetitive’ . . . .”).
2012] ANTICOMPETITIVE PRODUCT DESIGN 731
combination was technically superior to preceding products. This log-
ic holds true, however, only where consumer choice was uncon-
strained and at least somewhat informed. As noted supra, those
conditions are rarely present in instances of software redesign. In-
stead of expressing that they “prefer” such redesigns, users generally
“accept” them.
Consumer choice in this area is often constrained by software
companies requiring users to accept and install software updates in
order to continue using the program or all of its features. Witness, for
example, Apple’s requirement that iTunes users download its alleg-
edly exclusionary software updates in order to access the iTunes Mu-
sic Store.
288
Unlike in Berkey Photo, the choice here was not between
buying a competitor’s product and buying the dominant firm’s prod-
uct. Rather, the choice that modern consumers must increasingly
face is between being able to continue to use a familiar, already
owned product and either losing that ability or being forced to con-
tinue using a less functional version of the same product. Given
switching costs, particularly in closed network markets, consumer
acceptance can hardly be said to be “unconstrained” in such scenarios.
Further, as noted above, the dominant trend toward automatic
software updating means that consumers are not allowed to make a
meaningful, informed choice as to whether to accept an update.
289
And even before automatic software updating became the dominant
method of code-based product redesign, little information was gener-
ally available to users. Since customer “choice” is neither unconstrained
nor informed, antitrust courts must be extremely cautious in considering
it as evidence of a technical improvement to a code-based product.
E. Shifting Burdens, Yet Avoiding Both “Balancing” and a
Defendant-Friendly Default Rule
A burden-shifting analysis offers a rational, useful, and efficient
way for courts to approach claims of anticompetitive conduct, particu-
larly design-related claims in the product markets under discussion
here. As the party that bears the ultimate burden of proving its case,
the plaintiff must first make a prima facie showing that the defend-
ant has engaged in exclusionary conduct. The burden then shifts to
the defendant to show a procompetitive justification for its actions. If
the defendant fails to do so, the inquiry ends. And under the analyses
employed in many product-design cases, if the defendant succeeds in
doing so, the inquiry also ends. But, as shown above, the rationale for
288. The Apple iPod iTunes Antitrust Litigation, No. C 05-00037 JW, C 07-06507 JW,
2010 WL 2629907, at *3-4 (N.D. Cal. June 29, 2010).
289. See supra Part III.B.2.b (discussing the trend toward automatic software updates
and its implications for consumers).
732 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
such defendant-friendly analyses does not hold in code-based product
markets (and especially markets characterized by positive network
externalities). Thus, the burden should shift back to the plaintiff, al-
lowing the opportunity to rebut that justification. If the plaintiff can
do so, the conduct is rightly deemed unlawful.
But where the plaintiff fails to rebut the justification, courts
would seem to be forced into an unpleasant choice between (1) at-
tempting to “balance” the anticompetitive effects against the procom-
petitive justification
290
or (2) adopting a “tie-goes-to-the-runner” de-
fault rule favoring the defendant.
291
The problems that inhere in a
balancing test have been extensively documented
292
and may even be
exacerbated in the realm of product-design conduct, where such a
test would require courts to “balance” anticompetitive effects against
the benefits of innovation—a task that may well be workable in clear
cases
293
but could be so fraught with uncertainty as to preclude any
meaningful analysis of close cases. Yet a default rule favoring defend-
ants is, as noted above, generally inappropriate in addressing code-
based product redesign claims (particularly in network markets).
294
In the realm of code-based product redesign, however, this dilem-
ma is largely illusory. Given the unique separability of coded func-
tions from a code file as a whole and the fact that code can be config-
ured multiple ways to perform the same function, design-related con-
duct that truly causes both anticompetitive effects and constitutes
procompetitive innovation will be at most extremely rare, and likely
nonexistent. Thus, as in Microsoft III (and quite possibly In re Intel),
many defendants that have engaged in anticompetitive conduct will
likely not be able to even show a plausible procompetitive justifica-
tion. In the cases in which a plausible justification is shown, close
examination at a low level of abstraction will generally reveal the
effect of each coded function.
295
Finally, antitrust courts should keep
in mind the fact that code can be configured multiple ways to perform
290. See Microsoft III, 253 F.3d 34, 59 (D.C. Cir. 2001); Decision and Order, supra
note 245.
291. A
REEDA &HOVENKAMP, supra note 259, ¶ 651e3, at 121-23. At least in the realm
of unilateral conduct, such a rule favoring plaintiffs has neither been adopted nor urged,
given the difficulties that inhere in achieving and maintaining monopoly power.
292. See, e.g., Allied Orthopedic Appliances, Inc. v. Tyco Health Care Grp. LP, 592 F.3d
991, 1000 (9th Cir. 2010) (“To weigh the benefits of an improved product design against the
resulting injuries to competitors is not just unwise, it is unadministrable. There are no
criteria that courts can use to calculate the ‘right’ amount of innovation, which would max-
imize social gains and minimize competitive injury. A seemingly minor technological im-
provement today can lead to much greater advances in the future.”).
293. One need not use a very precise scale to determine that an elephant weighs more
than a chicken.
294. See supra notes 252-74 and accompanying text.
295. This is particularly true of programs written in relatively higher-level program-
ming languages. See supra notes 74-78 and accompanying text.
2012] ANTICOMPETITIVE PRODUCT DESIGN 733
the same function, making truly ambiguous cases very rare. If a
court finds itself seemingly unable to determine whether a defend-
ant’s conduct was anticompetitive or innovative, it should go back
and take a hard look at the alternative methods of designing the code
that were, ex ante, available to the defendant. To the extent that in-
tent can be ascertained, it could also be considered at this point.
Thus, courts employing a burden-shifting analysis should be enabled
to identify anticompetitive conduct with an acceptably low rate of
false positives, all while avoiding both the uncertainty of balancing
close cases and an inapposite default rule favoring defendants.
VI.
CONCLUSION
The new economy has brought with it a host of innovations that
have vastly benefited consumers and society. Yet it has also opened
novel avenues for firms to engage in exclusionary conduct, creating
the potential for untold harm to both consumer and total wealth. As
shown above, the distinctive aspects of these markets and the prod-
ucts they comprise uniquely incentivize firms to engage in anticom-
petitive, design-related behavior—here, perhaps more frequently
than in any other markets, C
i
> P
m
– LR. Fortunately, these aspects
also render code-based product redesign uniquely susceptible to anti-
trust scrutiny. They also allow courts to apply a burden-shifting
analysis that avoids balancing. Perhaps more importantly, they allow
courts to properly avoid employing an overly defendant-friendly de-
fault rule that would reduce R, further tipping the balance toward
incentivizing harmful exclusionary conduct.
After decades of debate over the principles and ideology of anti-
trust law, some consensus has emerged regarding core antitrust val-
ues. The resulting “antitrust doctrine is supple enough, and its com-
mitment to economic rationality strong enough, to take in stride the
competitive issues presented by the new economy.”
296
Despite this
measure of doctrinal accord, however, some continue to criticize anti-
trust’s institutions as unwieldy or its rules as unfocused.
297
This Arti-
cle seeks to allay those concerns. Given a proper understanding of
code-based product markets and a structured, efficient, and rational
method of analyzing design-related conduct in those markets, anti-
trust’s institutions will also be well-equipped to protect competition
in such markets—and can do so without stifling true innovation.
296. Posner, supra note 8, at 925.
297. See, e.g., id. (“The real problem lies on the institutional side: the enforcement
agencies and the courts do not have adequate technical resources, and do not move fast
enough, to cope effectively with a very complex business sector that changes very rapidly.”).
734 FLORIDA STATE UNIVERSITY LAW REVIEW [Vol. 39:681
