Harvard Journal of Law & Technology Occasional Paper Series — July 2013*

GOOGLE AND SEARCH-ENGINE MARKET

POWER
By Mark R. Patterson

*
The Harvard Journal of Law & Technology Occasional Paper Series is an online-only component intended to
showcase thoughtful and innovative writings on a wide variety of subjects related to the law and technology field.
These pieces are reformatted for consistency, but otherwise undergo no editing or cite-checking by the members of
the Harvard Journal of Law & Technology.
 Harvard Journal of Law & Technology

GOOGLE AND SEARCH-ENGINE MARKET POWER

Mark R. Patterson*

TABLE OF CONTENTS

I. INTRODUCTION .................................................................................. 1
II. SEARCH ENGINE MARKETS .............................................................. 6
A. The Irrelevance of Market Share.................................................. 6
B. The Importance of Quality ........................................................... 7
C. The Google-Bing Comparison ..................................................... 9
III. SEARCH-RESULT QUALITY AND INFORMATION COSTS ................ 11
A. Search, Experience, and Credence Goods ................................. 11
B. The Effects of Ignorance ............................................................ 12
C. Search Engine Optimization ...................................................... 14
IV. SEARCHERS AND WEB SITES ........................................................ 15
A. Two-Sided Markets..................................................................... 16
B. Distortion as Abuse and Power .................................................. 18
C. Competitive Responses ............................................................... 19
D. The Power to Manipulate Search Results .................................. 20
V. THE OBJECTIVITY BASELINE ......................................................... 22
VI. CONCLUSION ................................................................................ 23

I. INTRODUCTION

Allegations of anticompetitive conduct by Google have

prompted investigations by antitrust agencies on both sides of the At-
lantic,1 as well as several unsuccessful private suits. Although the al-

* Professor of Law, Fordham University School of Law. I am grateful for comments re-
ceived from Aditi Bagchi, James Grimmelmann, Martin Hansen, Giorgio Monti, Frank
Pasquale, Steven Thel, and participants at presentations at a Columbia Law and Economics
Workshop and at the November 2011 Mediadem conference on “Pluralism and Competition
in the Regulation of New Media” at the European University Institute, for which an earlier
version of this paper was prepared. I would also like to acknowledge the organizers of that
conference, Fabrizio Cafaggi and Pier Luigi Parcu. I received valuable research assistance
from Maxwell Meadows. This paper is part of a larger project addressing the competition
law implications of information intermediaries like search engines.
1. Michael Liedtke & Joelle Tessler, Google Confirms FTC Antitrust Investigation,
HUFFINGTON POST, June 24, 2011, http://www.huffingtonpost.com/2011/06/24/google-
confirms-ftc-antitrust-investigation_n_883951.html; Press Release, European Commission,
Antitrust: Commission probes allegations of antitrust violations by Google (Nov. 30, 2010),
http://europa.eu/rapid/pressReleasesAction.do?reference=IP/10/1624; Claire Cain Miller,
Texas Probes Google on Ranking of Search Results, N.Y. TIMES, Sept. 3, 2010, at B3,
available at http://www.nytimes.com/2010/09/04/technology/04google.html?.
2 Harvard Journal of Law & Technology

legations are of several kinds, the primary one is that Google has ma-
nipulated its search results — either its so-called “organic”2 search
results or the results provided by its AdWords sponsored-advertising
program — in order to disadvantage competitors or potential competi-
tors. Generally speaking, the claims are that Google has artificially
pushed competitors down in its search results,3 making it more diffi-
cult for them to reach searchers seeking the sorts of services that those
competitors provide. That is, the claims are that Google does not de-
liver the results that would best serve consumers, but instead alters
those results to serve its own competitive interests.
The competitors that Google is alleged to have disadvantaged
are of two kinds: information providers like Yelp, and so-called “ver-
tical search engines.”4 Web sites like Yelp allege both that Google has
misappropriated their content and that Google demotes them in its
organic results in order to give an advantage to Google products like
Google Places and Zagat.5 “Vertical search engines” are search en-
gines like Foundem6 or TradeComet7 that are aimed at those seeking
particular sorts of information, such as medical information or infor-
mation about the prices of consumer electronics goods. These vertical
search engines allege that Google has manipulated the pricing of its
AdWords program to disadvantage them as potential competitors of
Google.8 Because it may be feasible to develop a successful search

2. Google uses “organic” to describe its main search results, i.e., the list of results that are
not sponsored advertising. See The Power of Google: Serving Consumers or Threatening
Competition?: Hearing Before the Subcomm. On Antitrust, Competition Policy, & Consum-
er Rights of the S. Comm. on the Judiciary, 112th Cong. 5 (2011) (testimony of Eric
Schmidt, Executive Chairman, Google Inc.) [hereinafter Schmidt Testimony], available at
http://www.judiciary.senate.gov/pdf/11-9-21SchmidtTestimony.pdf (“Google was one of
the first search engines to clearly distinguish advertisements from our organic search re-
sults.”).
3. For AdWords, the effectively equivalent claim is that Google artificially charges its
competitors more for placement in the AdWords results.
4. In fact, the definition of “vertical search engine” is broad enough and vague enough
that it could encompass information providers like Yelp.
5. The Power of Google: Serving Consumers or Threatening Competition: Hearing Be-
fore the Subcomm. On Antitrust, Competition Policy, & Consumer Rights of the S. Comm.
on the Judiciary, 112th Cong. (2011) (testimony of Jeremy Stoppelman, Cofounder and
CEO, Yelp! Inc.), available at http://www.judiciary.senate.
gov/hearings/testimony.cfm?id=3d9031b47812de2592c3baeba64d93cb&wit_id=3d9031b47
812de2592c3baeba64d93cb-5-1; see also Matthew Ingram, Google’s Zagat Buy Could Give
Search Critics More Ammo, GIGAOM, Sept. 9, 2011, http://gigaom.
com/2011/09/09/googles-zagat-buy-could-give-search-critics-more-ammo/.
6. See Adam and Shivaun Raff, Background to EU Formal Investigation,
SEARCHNEUTRALITY.ORG (Nov. 30, 2010 8:01 PM), http://www.searchneutrality.
org/foundem-google-story/eu-launches-formal-investigation.
7. See Complaint, TradeComet.com LLC v. Google Inc., No. 09-CIV-1400 (S.D.N.Y.
Feb. 17, 2009).
8. Id.; see also Jeff Bliss and Sara Forden, Google Ad Rate for Microsoft Said to Be In-
vestigated by U.S., BLOOMBERG NEWS, Sept. 21, 2011, http://www.
businessweek.com/news/2011-09-21/google-ad-rate-for-microsoft-said-to-be-investigated-
 Google and Search Engine Market Power 3

engine in a limited area at relatively low cost, a vertical search engine

could gain a foothold in a particular area and pose a competitive threat
to Google and other general search engines, at the very least taking
away advertising revenue for searches in that area.
The theory of an antitrust claim against Google would be that
the manipulation of search results excludes competitors — vertical
search engines, say — from effectively competing to serve those seek-
ing information, either generally or through Google. Foreclosure from
access to those Google users, even if only through being pushed down
in Google’s results, could effectively exclude competitors from the
search market. Moreover, this sort of exclusion would also harm con-
sumers if consumers value the services that those vertical search en-
gines provide, and would prefer that they be included, or appear
higher, in Google’s results. The U.S. antitrust statute applicable to
such a claim is Sherman Act § 2, under which a monopolization viola-
tion can be established by proof of two elements: possession of mo-
nopoly power, and willful acquisition or maintenance of that power.
A significant and growing body of commentary considers wheth-
er possible manipulation of search results by Google could satisfy the
second, conduct-focused element of monopolization.9 Some argue that
deviations from some “objective” or “unbiased” standard for provid-
ing search results can constitute exclusionary and anticompetitive
conduct. Others argue that such conduct should not be the basis for an
antitrust violation, either because Google and other search engines
should be free to provide whatever results they like or because it is too
difficult to distinguish procompetitive from anticompetitive conduct
in this area.
Surprisingly, though, little serious attention has been paid to
whether Google satisfies the power element of a monopolization
claim.10 Under Sherman Act § 2 (or its European analogue, Article

by-u-s-.html (“U.S. antitrust enforcers are investigating whether Google Inc. illegally in-
creased advertising rates 50-fold for rival Microsoft Corp.”).
9. For examples of the allegations against Google, see the sources cited in notes 5–8 su-
pra. Those who appear to advocate or accept the possibility of such liability include, e.g.,
Oren Bracha & Frank Pasquale, Federal Search Commission? Access, Fairness, and Ac-
countability in the Law of Search, 93 CORNELL L. REV. 1149 (2008); Mark R. Patterson,
Non-Network Barriers to Network Neutrality, 78 FORDHAM L. REV. 2843 (2010). Those
who reject it, either on factual or legal grounds, are James Grimmelmann, Some Skepticism
About Search Neutrality, in THE NEXT DIGITAL DECADE: ESSAYS ON THE FUTURE OF THE
INTERNET (Berin Szoka & Adam Marcus eds., 2010); Daniel Crane, Search Neutrality as an
Antitrust Principle (Univ. of Michigan L. School, Pub. L. Working Paper Series, No. 256);
Geoffrey A. Manne & Joshua D. Wright, Google and the Limits of Antitrust: The Case
Against the Case Against Google, 34 HARV. J.L. & PUB. POL’Y 171 (2011); Joshua D.
Wright, Defining and Measuring Search Bias: Some Preliminary Evidence (Geo. Mason L.
& Econ. Research Paper No. 12–14, 2011), available at http://papers.ssrn.com
/sol3/papers.cfm?abstract_id=2004649.
10. Some blogs have briefly addressed the issue. See, e.g., Amit Runchal, Let’s Stop Say-
ing Google Has a Monopoly, INTERACTIONED (Mar. 2, 2013, 10:43 PM), http://www.
interactioned.com/post/15743869510/lets-stop-saying-google-has-a-monopoly
4 Harvard Journal of Law & Technology

102 TFEU), a plaintiff, whether private party or government, must

show that the defendant possesses monopoly power (or, in Europe,
dominance), which is a large degree of market power. Those who fa-
vor antitrust scrutiny of Google generally cite its large market share,
from which they infer or assume its power.11 Those who are skeptical
of competition law’s role in regulating search, on the other hand, usu-
ally cite Google’s “competition is only a click away” mantra to sug-
gest that Google’s market position is precarious.12 In fact, the issue of
Google’s power is more complicated and interesting than either of
these approaches suggests.
A fundamental problem is that for information intermediaries like
Google, the most commonly used measure of monopoly power, mar-
ket share, is not a valid one.13 To that extent, the “competition is a
click away” story is an accurate one. Consumers can indeed easily
find and switch to alternatives. But the ease of clicking to another
search site does not mean that Google has no power. For the ease of
clicking away to constrain Google, it must also be the case both that
users do not suffer other costs in switching and that they can deter-
mine just when it is advantageous to click away. That is, it is im-
portant that both switching costs and information costs are low.
Although some of Google’s practices seem designed to increase
switching costs, such costs probably do remain low.14 Information

None of the monopoly definitions that people have been throwing

around are very clearly defined, which may be part of the problem
when we start arguing about this stuff. But however you want to de-
fine it, you can’t argue that Google has any ‘real’ monopolistic pow-
ers, even if they controlled 90% of the search market.
And an article co-authored by former F.T.C. Commissioner Pamela Jones Harbour ad-
dressed the issue of defining markets in “new-technology markets,” discussing Google but
not focusing particularly on it. Pamela Jones Harbour & Tara Isa Koslov, Section 2 in a Web
2.0 World: An Expanded Vision of Relevant Product Markets, 76 ANTITRUST 769 (2010).
11. The Power of Google: Serving Consumers or Threatening Competition?: Hearing
Before the Subcomm. on Antitrust, Competition Policy, & Consumer Rights of the S. Comm.
on the Judiciary, 112th Cong. (2011) (question of Sen. Herb Kohl to Eric Schmidt, Execu-
tive Chairman, Google Inc.), available at http://www.gpo.gov/fdsys/pkg/CHRG-
112shrg71471/pdf/CHRG-112shrg71471.pdf (“But you do recognize that in the words that
are used in antitrust kind of oversight, your market share constitutes monopoly, dominant —
special power, dominant firm, monopoly firm?”).
12. Miguel Helft, Google Makes a Case That It Isn’t So Big, N.Y. TIMES, June 28, 2009,
at B1 (quoting Dana Wagner, Google senior competition counsel, as saying “Competition is
a click away”); Schmidt Testimony, supra note 2, at 7 (“[I]f consumers don’t like what one
website is providing them, they can switch to another website with just one click.”).
13. See infra part II.A.
14. That is true at least for those who use a search engine’s home page. For search
through other pages, Google has paid considerable sums to be the default search engine,
suggesting a belief that there is value in being the first search engine users see and that users
will not necessarily switch to the engine that serves them best. See Kara Swisher, Google
Will Pay Mozilla Almost $300M Per Year in Search Deal, Besting Microsoft and Yahoo,
ALL THINGS D, (Dec. 22, 2011), http://allthingsd.com/20111222/google-will-pay-mozilla-
almost-300m-per-year-in-search-deal-besting-microsoft-and-yahoo/; see also Matthew
Panzarino, The FTC subpoena of Apple could spell the end of Google’s default search status
 Google and Search Engine Market Power 5

costs, however, can be significant. This essay argues that it could be

quite difficult for search engine users to determine whether the results
they are receiving from a particular search engine justify switching to
another.
The analysis of power for Google or other information pro-
viders is also complicated by the multifaceted role played by infor-
mation. The commentary on Google has not focused on information
as a product and generally has not considered the ways in which it
differs from other products. A key feature of information is described
by Arrow’s paradox regarding information: “its value for the purchas-
er is not known until he knows the information, but then he has in
effect acquired it without cost.”15 In many instances of search, a con-
sumer will be seeking information in circumstances in which she will
be unable to evaluate the quality of the information she receives. As
will be discussed in more detail below, this lack of transparency in
quality can give an information provider market power, just as can an
absence of transparency in price for other products. The significance
of this effect, however, is difficult to evaluate.
This paper proceeds in several steps. It begins in Part II with
a brief discussion of factors that make assessment of Google’s power
more difficult than for many other sellers. This discussion emphasizes
the particular characteristics of markets for providing information, and
explains how those characteristics serve to make market share a poor
measure of power. Part III then specifically examines the task of
measuring informational market power, arguing that a focus on the
difficulty of assessing the quality of information is critical. Part IV
tentatively suggests an approach to measuring Google’s power that
relies on its position as a two-sided market platform, and uses the
pricing of Google’s paid AdWords placements as a means of measur-
ing its power. Lastly, Part V discusses a particular informational issue,
the objectivity of Google’s search results, and its implications for
market power. Part VI concludes.

on the iPhone, THE NEXT WEB, (Mar. 13, 2012), http://thenextweb.com/

apple/2012/03/13/the-ftc-subpoena-of-apple-could-spell-the-end-of-googles-default-search-
status-on-the-iphone/.
15. Paper by Kenneth J. Arrow, Econ. Div., RAND Corp., Economic Welfare and the Al-
location of Resources and Invention 10 (Dec. 15, 1959) , available at http://www.rand.org/
pubs/papers/2006/P1856.pdf.
6 Harvard Journal of Law & Technology

II. SEARCH ENGINE MARKETS

A. The Irrelevance of Market Share

In cases under Sherman Act § 2 and Article 102 TFEU, the usual
measure of market power is market share.16 Competition law uses
market share as proxy for power because it often reflects the ability of
a firm to act without regard to competition. Generally speaking, if a
firm with a large market share seeks to act anticompetitively, smaller
competitors will be unable to compensate for the anticompetitive acts
by meeting the unmet demand themselves, because their response will
be limited by their relatively smaller size. Although antitrust typically
uses sales to measure market share, it is actually capacity that is the
relevant measure.17 That is so because it is unused or expanded capac-
ity that allows competitors to respond to anticompetitive demand by a
monopolist, not current sales. In most instances, however, at least in
traditional product markets, capacity and sales are closely related, so
the distinction is unimportant.
Where the product is information, however, firms may be able
quickly to expand output. Consider a search engine that competes
with Google, for example. If Google were to act anticompetitively, its
competitor would easily be able to “produce” products to meet the
demand of those who were unsatisfied with Google’s products. After
all, the products at issue are search results, and the algorithm for pro-
ducing them is already available, so the only obstacle to producing
more of them is the availability of server capacity to deliver the re-
sults to customers. Although expanding server capacity imposes some
costs and takes some time, those limitations are small compared with,
say, expansion of capacity in the production of the archetypal widget.
Hence, market share is a relatively poor proxy for power when the
product at issue is information.18
It should be emphasized, however, that the focus on capacity here
is a narrow one, referring only to the capacity to deliver search results
to users. There is another, perhaps more important element related to
the volume of search results delivered: the advantage a search engine

16. See, e.g., United States v. Aluminum Corp. of America, 148 F.2d 416 (2d Cir. 1945);
Case 85/76, Hoffmann-La Roche & Co. AG v Commission, 1979 ECR 461 ¶ 39 (“The
existence of a dominant position may derive from several factors which, taken separately,
are not necessarily determinative but among these factors a highly important one is the
existence of very large market shares.”).
17. See U.S. DEP’T OF JUSTICE & FEDERAL TRADE COMM’N, HORIZONTAL MERGER
GUIDELINES (1992, revised 1997).
18. The value of approaches focused on market definition and market share is a subject of
current debate. Cf. Louis Kaplow, Why (Ever) Define Markets?, 124 HARV. L. REV. 437
(2010) with Gregory J. Werden, Why (Ever) Define Markets? An Answer to Professor
Kaplow (Feb. 13, 2012) (unpublished manuscript) (on file with U.S. Dep’t. of Just., Anti-
trust Div.), available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2004655.
 Google and Search Engine Market Power 7

gains from the information gathered from searches. A search engine

delivering a larger volume of search results gains valuable infor-
mation from its users’ searches and thus is able to deliver better
search results. This advantage is akin to learning effects, and it could
be viewed as lowering the cost of delivering high-quality results. The
advantage is not, however, so much an effect of current market share
as one of the cumulative number of searches delivered. For that rea-
son, it does not support using current market share as a measure of
power. It is better treated, perhaps, as privileged access to a valuable
input, in the sense that prior searches are the raw material from which
search results are in part derived.19 The focus of this paper is on de-
mand-side effects, but supply-side effects such as this one may also be
important.

B. The Importance of Quality

If market share is not the source of power in the search-engine

market, then what is? Market power can derive from the inability of
consumers to evaluate the quality of the product they are receiving. If
the quality of a search engine’s results is difficult for consumers to
assess, the result can be the diminished competition that provides
market power. The issue can be illustrated by reference to George
Stigler’s classic description of the problem of price dispersion in the
market for consumer purchase of automobiles:

Price dispersion is a manifestation — and, indeed, it

is the measure — of ignorance in the market. Disper-
sion is a biased measure of ignorance because there
is never absolute homogeneity in the commodity if
we include the terms of sale within the concept of the
commodity. Thus, some automobile dealers might
perform more service, or carry a larger range of vari-
eties in stock, and a portion of the observed disper-
sion is presumably attributable to such differences.
But it would be metaphysical, and fruitless, to assert
that all dispersion is due to heterogeneity.20

In the case of Google, it is quality, not price, that is at issue. Price is

homogeneous because search engines provide search results for free,
but quality of course differs among search engines.21

19. It has been argued that this raw material should be shared among search engines.
20. George J. Stigler, The Economics of Information, 69 J. POL. ECON. 213, 214 (1961).
21. Some factors, like search response time and the ease of use of web sites, could be
placed either in the price or quality category. That is, they might be viewed as part of the
quality that one receives from a search engine, or they might be viewed as factors in the
price that one must pay to receive search results.
8 Harvard Journal of Law & Technology

Whether the issue is price or quality, the basic insight still holds.
Just as prices for homogeneous products like automobiles differ (and
Stigler provides evidence that they do), the quality of homogeneously
priced search results will differ. And one reason is the same: consum-
er ignorance. As will be described further below, it is difficult for
consumers to assess the quality of at least some of the search results
they receive, which means that search engines need not provide the
highest quality of search results to succeed. They cannot provide very
poor results, of course, but the difficulty and cost to users of assessing
search results means that there is some freedom to provide less than
the very best quality, just as the cost (once higher) of price compari-
sons means that automobile dealers need not always sell at the very
best price.
The task of assessing Google’s power is also complicated be-
cause there is no established baseline for comparison. For price, cost
is at least in theory a measure against which price can be compared,
because competition tends to drive price to marginal cost. Analogous-
ly, if there were some clear measure of the quality of search infor-
mation that should be provided for free, then perhaps we could say
that providing search information of lower quality would be an exer-
cise of power. In fact, as will be discussed below,22 Google has at
times claimed that it provides un-manipulated search results, so that
could be a standard applied to assess its power. It is not clear, howev-
er, that this is a proper standard. For example, should a finding of
power turn on whether a search engine states or denies that it manipu-
lates its search results?23
In any event, if competition does not force search engines to pro-
vide high-quality search results (and the next part of this essay argues
that it does not), that freedom will give them some degree of market
power. That is, Google’s ability, if any, to provide less-than-optimal
search results would be evidence of market power. But the distortion
of search results is also the anticompetitive conduct of which Google
is accused. Thus, the same evidence shows both power and conduct.
This is theoretically reasonable. If a firm has the ability to engage in
anticompetitive conduct, that ability alone is evidence that the firm
possesses market power; otherwise, its lack of power would constrain
its ability to act anticompetitively. Indeed, there is a movement in
competition law toward looking directly to conduct to determine the
existence vel non of market power. This essay applies that approach to
the search-engine market.

22. See infra part VI.

23. See id.
 Google and Search Engine Market Power 9

C. The Google-Bing Comparison

The insignificance of market share has important implications for

Joshua Wright’s very interesting comparison of search “bias” by
Google and Bing.24 Wright provides data that indicates that the fre-
quency of bias by Bing is comparable to, and perhaps greater than,
that by Google. From this data, he concludes that “‘bias’ is not a func-
tion of market power, but an effective strategy that has arisen as a
result of serious competition and innovation between and by search
engines.”25 The basis for drawing this conclusion, presumably, is that
if Bing, which does not have a large market share, engages in bias,
then the practice must be procompetitive. Or, as Wright says, “that
both this percentage [of own-content bias] and the absolute level of
own content inclusion is similar across engines indicates that this
practice is not derivative of one’s misuse of its market power, but an
industry standard.”26
However, this conclusion does not follow unless one assumes that
Bing’s share is in fact evidence of a lack of power. If instead the
source of power for search engines is not market share but the inabil-
ity of searchers to evaluate search results, then all search engines, re-
gardless of market share, could have power. The limits on information
created by restrictions on advertising, for example, have been found to
be associated with higher prices, and those higher prices are charged
by small sellers as well as large ones.27 In the same way, the unavaila-
bility to searchers of information that would allow them to evaluate
the quality of the results they are receiving could allow low quality
results to be provided by small search engines as well as large ones.28
Some of Wright’s commitment to the view that bias is not a prob-
lem for search engines appears to derive from his claim that there is “a
well‐understood economic analysis of the competitive effects of a
vertically integrated firm’s ‘discrimination’ in favor of its own prod-
ucts or services, including widespread recognition that such arrange-
ments generally produce significant benefits for consumers.”29 To be
sure, some discrimination by sellers in favor of their own products is
procompetitive, but other such discrimination is not, as reflected by

24. See Wright, supra note 9.

25. Wright, supra note 9, at 46.
26. Id. at 47.
27. See, e.g., John E. Kwoka, Jr., Advertising and the Price and Quality of Optometric
Services, 74 AM. ECON. REV. 211 (1984).
28. There is also no reason to think that small search engines would have the ability and
incentive to constrain the power of larger ones. First, competing search engines are likely to
have the same problems as do consumers in evaluating the results provided by other search
engines. Second, as the Supreme Court observed in Kodak, competitors with similar sources
of market power may choose to “live and let live,” rather than engage in competition that
could make all of them worse off.
29. Wright, supra note 9, at 5.
10 Harvard Journal of Law & Technology

antitrust rules condemning some tying arrangements and exclusive

dealing. And the search-engine market is new enough, and unstudied
enough, to counsel caution regarding broad claims about what com-
petitive effects discrimination “generally” has, particularly when the
support for such claims is not identified.
Also without support is Wright’s claim that “Google users likely
prefer Google content.”30 Wright quite reasonably objects to those
who do not consider the possibility of competitive product differentia-
tion, pointing to work by Benjamin Edelman and Benjamin Lock-
wood comparing search results.31 But despite his advocacy of
“evidence-based” assessment, Wright offers no evidence that would
suggest that the differences among search engines are the product of
consumer preferences.32 The closest that Wright comes is to quote
Danny Sullivan, a frequent defender of Google: “If someone’s search-
ing for ‘maps’ on Google, they may be more likely to want Google
Maps than Yahoo Maps — and vice versa.”33 This might be true, but
it is also possible — one might even say likely — that searchers want
whatever search engine they are using to deliver the best map availa-
ble, not the one that happens to be affiliated with the search engine.34
This is not to say that delivering affiliated content is necessarily anti-
competitive, as Wright claims some argue. It is merely to say that the
fact that search engines deliver different content cannot be assumed to
show that they are responding to consumer demand, as Wright sug-
gests. The competitive costs or benefits of own-content bias must be
determined independent of assumptions based on market share or on
views regarding the effects of vertical integration in general.

30. Id. at 13.

31. Id.
32. Moreover, if there were different search preferences of this kind, that itself would
suggest that search engines have market power. If search engines were differentiated,
Google could not necessarily exploit those that preferred a Bing-type search engine, but it
could exploit the subset of consumers that preferred a Google-type search engine. In other
words, the Google-type search engine could be a submarket, and Google could have power
in that submarket.
33. Id. at 6 n.10 (quoting Danny Sullivan, Study: Google “Favors” Itself Only 19% of the
Time, SEARCHENGINELAND, (Jan. 19, 2011),
http://searchengineland.com/survey‐google‐favors‐itself‐only‐19‐of‐the‐time‐61675.
34. Wright offers some exceedingly strained explanations of evidence offered by Edel-
man and Lockwood that suggests that consumers may not be satisfied by some own-content
favoritism by Google. Id. at 18. Edelman and Lockwood show that although Google pro-
vides a link to its own Gmail first in its results and a link to Yahoo Mail second, 29% of
clicks go to Gmail while 54% go to Yahoo. Wright says that “any number of other benign
reasons could explain this anomalous ranking; for example, users might realize after running
this search that they know of a more efficient way of accessing Gmail, or they may simply
have clicked on Yahoo Mail first, immediately returned to the search page, and subsequent-
ly clicked on Gmail. Id. (citing Danny Sullivan, Study: Google “Favors” Itself Only 19% of
the Time, SEARCH ENGINE LAND, (Jan. 19, 2011),
http://searchengineland.com/survey‐google‐favors‐itself‐only‐19‐of‐the‐time‐61675).
 Google and Search Engine Market Power 11

III. SEARCH RESULT QUALITY AND INFORMATION COSTS

As stated above,35 although there are several potential sources of

market power for Google,36 this paper will focus on only one: infor-
mation costs. The key question is whether, if Google were to act anti-
competitively, users could determine whether there is a sufficient
benefit to justify switching to another search engine. Or, to put the
question another way, could Google provide low-quality search re-
sults without causing users to switch to other search engines? If so,
then there would be reason to say that Google has market power.

A. Search, Experience, and Credence Goods

Economists divide products into three types of goods with respect

to the means by which consumers can evaluate the quality of the
goods. Search goods are those whose quality can be evaluated before
the good is purchased, often by searching for evaluations of the prod-
uct. Experience goods are those that are difficult to evaluate before
purchase, but that can be evaluated as they are used. Credence goods
are those that are difficult for consumers to evaluate even after they
are used. Thus, a painting would likely be a search good, a restaurant
meal will typically be an experience good, and a vitamin supplement
may be a credence good.
In which of these categories are search results? It depends on the
nature of the search. Rarely will a search be a classic search good, in
that generally one will not often be able to find, before searching, in-
formation about the quality of particular searches. There are many
searches, however, which one can be confident will produce good
results, even before the search is conducted. A search for a word’s
definition, for example, or for a recent sports score, is almost certain
to produce the result sought.
Other search results are experience goods. For example, if one
searches not for a fact, but for more complex information, like an ex-
planation of a natural phenomenon, one will probably not know be-
fore conducting the search what will be the quality of the results, but
it is likely that they can be evaluated once received. Or if one seeks
information about a person, such as a long-lost friend, one cannot be
sure of finding that information, but it will probably be possible to tell
whether the information provided is about the person sought.
Finally, for some searches, and these are the ones of most interest
here, one may not know even after performing the search whether the
quality of the results was high. For example, if one searches with the
keywords “best price iPhone 5” or “nice inexpensive New York ho-

35. See supra text following note 19.

36. See supra text accompanying note 16 and following note 18.
12 Harvard Journal of Law & Technology

tels,” it will be difficult to evaluate the quality of the results. Even

after reviewing the search results, one often will have no way of
knowing whether the search engine performed well. For example,
even if one receives a web site with what one thinks is a good price on
an iPhone 5, one will not always know whether better prices are avail-
able. And a user who searches for “nice inexpensive New York ho-
tels” cannot know whether the results are in fact nice inexpensive
New York hotels, let alone the nicest or most inexpensive New York
hotels, or if better results might have been available. Even if the user
ran the search on another engine and obtained different results, it
would not be clear which results were better. In that sense, search re-
sults can be credence goods.
It is significant that the particular competitors against which
Google is alleged to have discriminated compete against Google in
providing exactly this sort of credence information. Those competitors
have primarily been vertical search engines that claim to provide con-
sumers with product information in particular areas. In seeking such
product information, consumers are unlikely to know, if a search en-
gine (like TradeComet, a vertical search engine at issue in one case
against Google) includes a particular seller in its results, whether that
result is valuable, or more valuable than alternatives. Similarly, if the
vertical search engine (like Foundem, one of the complainants in the
EU case) provides a particular price for a given consumer product, the
consumer may not know whether that is a good price, or if better ones
are available.

B. The Effects of Ignorance

The difficulty of evaluating search results makes it less likely that

Google, or any other search engine, will be constrained by competi-
tion.37 Some searchers, when they receive unsatisfactory search re-
sults, might go to Bing, for example, but others might believe that
Bing will do no better and thus remain with Google. If so, Google will
not be greatly constrained by the presence of Bing or other search
engines. Google is not, of course, completely unconstrained — it can-
not regularly produce poor results — but it likely has some freedom to

37. Although the issue is surely contestable, the history of search engines does not appear
to demonstrate that better algorithms alone determine success or cause consumers to switch
search engines. The web site Search Engine History states that AltaVista lost its position as
market leader “[d]ue to poor mismanagement, a fear of result manipulation, and portal relat-
ed clutter.” SEARCHENGINEHISTORY, http://www.searchenginehistory.com/ (last visited
Mar. 2, 2013). Inktomi “failed to develop a profitable business model.” Id. Overture (now
Yahoo) “had two major downfalls which prevented [it] from taking Google’s market posi-
tion:” it chose not to become a search destination itself, but instead was distributed through
partners; and it did not have as profitable an advertising model as Google. Id. Although the
reference to result manipulation as one cause of AltaVista’s troubles suggests that quality
can be important, this web site, at least, represents business issues as playing a greater role.
 Google and Search Engine Market Power 13

provide less-than-optimal results, particularly if it does so only in cer-

tain areas and not routinely.
The difficulties of evaluating search results can be seen by com-
paring them with the difficulties of evaluating other information
sources. Wikipedia, for example, is a commonly used information
source, despite the fact that it is known to contain inaccuracies. The
accuracy of Wikipedia is fairly easy to assess, though, because in
many cases there are alternative sources, like print encyclopedias, of
the information it provides.38 For search engines, there usually are not
such alternative sources, except other search engines. The fact that
Wikipedia is widely used despite its inaccuracies, combined with the
fact that search engines are more difficult to evaluate than is Wikipe-
dia, suggests that competition is not likely to serve as a strong con-
straint on search-engine quality.
An alternative comparison is that of magazines and web sites that
provide reviews of consumer products. The reviews that these organi-
zations provide are in some respects search goods, because they in-
clude information like product features, and the value of this
information is immediately apparent. Reviews often provide more
qualitative information like “ease of use,” however, which is probably
best thought of as an experience good, in that not every consumer will
view “ease of use” in the same way. Once the reviewed product is
purchased, though, the value of the review is likely apparent, so the
product itself serves as the benchmark against which the review is
compared, just as print encyclopedias can for Wikipedia. As noted
above, though, there may be no such outside benchmark for search
engines, even after search results are obtained.
Google and other search engines may be able to operate with
some degree of freedom from competition, but of course that freedom
is not complete. For example, Google recently changed its algorithm
to make its search results more timely, and it did so, reportedly, in
response to competition.39 More specifically, it was reported that
Google’s motivation was the more timely nature of search on Twit-
ter.40 The very particular nature of the competitive threat of Twitter,
though, is unusual and confined to a particular problem. Although
Twitter provided a competitive benchmark for search engines, it did
so only with respect to a very specific dimension, timeliness. The

38. Moreover, Wikipedia articles often cite external sources, which makes checking the
accuracy of the articles easier. There are no such external sources for comparison in the case
of search engines.
39. Claire Cain Miller, Google Changes Search Algorithm, Trying to Make Results More
Timely, N.Y. TIMES, Nov. 3, 2011, available at http://bits.blogs.nytimes.com/2011/11
/03/google-changes-search-algorithm-trying-to-make-results-more-timely/ (“The new algo-
rithm is recognition that Google, whose dominance depends on providing the most useful
results, is being increasingly challenged by services like Twitter and Facebook, which have
trained people to expect constant updates with seconds-old news.”).
40. Id.
14 Harvard Journal of Law & Technology

specificity of that constraint only emphasizes that there is no such

benchmark for search results more generally.41 The upshot is that it
seems reasonable to suspect that search engines have some degree of
market power over their users, even if these comparisons provide little
to show the magnitude of that power.

C. Search Engine Optimization

Further evidence of the possibility of manipulation of search re-

sults is found in the active market for search engine optimization
(SEO), the “optimization” of web sites to improve their position in
search results.42 To some extent, the very existence of SEO demon-
strates the power of search engines, particularly if, as Google claims,
some SEO lessens the quality of search results.43 Searchers continue
to use Google and other search engines with little concern regarding
the effects of SEO, which suggests that search result quality can be
degraded with little market effect. To be sure, Google seeks to adjust
its algorithm to minimize the negative effects of SEO. Still, given the
amount of money spent on SEO, the effects likely are significant.
It might not be possible to derive any implications regarding
search-engine power if SEO affected all search engines equally. In
that case, the degradation of Google’s results would be accompanied
by similar degradation of Bing’s results and those of other search en-
gines, and competitive positions would be unchanged.44 But the pri-
mary target of SEO is Google.45 Consequently, it is likely that SEO

41. The issue is complicated also because there are no doubt skilled users, or users who
are skilled in certain types of searches, and search engines might have to respond to those
users. But it is likely the larger quantity of unskilled users that provide advertising profits.
Therefore, even if skilled users were to switch to another search engine from Google, it is
possible that Google would be able to lure them back, perhaps with improvements to its
algorithms, before the profitable unskilled users, who probably lag behind the skilled ones,
also leave Google. If this is true, and if other search engines know that it is true, competition
will be weakened. I will explore this issue more fully in an expanded version of this paper,
and I am grateful for discussions on this topic with Fabrizio Dell’Acqua, a student in the
master’s program of the Department of Economics at Bocconi University.
42. The market is active enough to justify a web site that ranks the top providers. See
Best: Top 50 Search Engine Optimization Companies, TOP SEOS (Mar. 2, 2013, 11:52 PM)
http://www.topseos.com/rankings-of-best-seo-companies.
43. There are two kinds of SEO, so-called “black hat” SEO, which distorts search results,
at least in the view of search engines, and “white hat” SEO, which more “accurately” pre-
sents a site to search engines’ algorithms, thus improving search results.
44. In theory, all search engines could find their competitive positions worsened if users
ceased or lessened their use of search engines. It is not clear, however, what could serve as a
substitute for a search engine, at least for a user seeking a web site. A user seeking the an-
swer to a particular factual question could turn to other information sources, though.
45. Sharon Nelson and John Simek, Making Your Web Site Visible: How to Find A Good
Seo Company, L. PRACTICE, May–June 2010, at 24 (“Despite the existence of other search
engines like Yahoo, Bing and Ask, Google remains the indisputable king, with over 65
percent of market share at present. That’s why most search engine optimization (SEO) is
done primarily with Google in mind.”).
 Google and Search Engine Market Power 15

disproportionately affects Google’s search results. If so — that is, if

the quality of Google’s results is lessened more than those of its com-
petitors, yet Google suffers little or no market effect — then Google’s
power must be significant.
Google’s ability to maintain its position despite SEO could be ex-
plained in at least three ways. First, it could be that SEO has little ef-
fect. As suggested above, however, this seems unlikely, given both
the amount of money spent on SEO and the significance of the efforts
of Google and other search engines to respond to it. Second, it could
be that Google’s pre-SEO quality is sufficiently greater than that of
other search engines that the quality can be reduced by SEO and still
remain higher. Third, it could be that Google’s quality is thought to be
high by searchers, regardless of the effect of SEO, so that Google’s
market position is unaffected. The latter two possibilities would both
be evidence of market power.46
Of course, the more significant consequence of the possibility that
minor degradations in search result quality do not cause significant
losses in user share is not the ability of SEO to degrade quality but the
ability of Google to do so. Just as SEO providers can profit from ma-
nipulating search results, so too could Google. Indeed, given that
Google knows its own algorithm, its capacity to manipulate results,
unlike that of the SEO providers, is not limited by trade secrecy but
only by its willingness to risk detection.
The common view is that Google would not engage in this sort of
manipulation because of potential reputational effects, and that view
is plausible. The reputational effects of manipulation are of course
significant for Google in a way that they are not for SEO providers.
But on several occasions Google has been found to have acted in
ways that were inconsistent with its previous statements, and it is not
clear that its reputation has suffered as a result. That is not evidence
that Google has engaged in manipulation on other occasions, but it
does suggest that the reputational threat may not be as great as is
sometimes stated. If so, then the constraint that reputation imposes on
Google may not be great.

IV. SEARCHERS AND WEB SITES

The preceding portions of this paper have focused primarily on

the relationship between Google and searchers who use Google to
find information. But Google has another set, or two other sets, of
consumers: the web sites that appear in its organic results and those
that purchase AdWords advertising. This has several implications.
First, the interaction between Google’s upstream and downstream

46. Note here that the existence of market power does not suggest that either the power or
the method of obtaining it was anticompetitive.
16 Harvard Journal of Law & Technology

customers creates difficult conceptual issues related to the analysis of

two-sided markets. Second, any distortion by Google of its search
results by excluding competing web sites could both demonstrate
power over consumers and injure the web sites and advertisers that are
trying to reach them. It is also possible, however, that the injury to
excluded sites would not be an injury to consumers. Third, as in most
cases of exclusion, the allegedly excluded parties — here, the web
sites and advertisers — are not passive victims of the exclusion, but
have means of responding to any discrimination by Google. Finally,
the paragraphs below suggest that the two-sided nature of the Google
market might provide a means of measuring the competitive effect of
exclusion.

A. Two-Sided Markets

Google operates in a two-sided market: a market with two

different customer groups that provide each other with network bene-
fits. As Rochet and Tirole describe, in two-sided markets, platforms,
like Google, “enable interactions between end-users, and try to get the
two (or multiple) sides ‘on board’ by appropriately charging each
side.”47 In Google’s case, the two classes are searchers and advertis-
ers.48 As in other two-sided markets, the price, quality, and output
relationships for the two sides are interrelated, which may make it
inappropriate to consider the two markets in isolation in assessing
Google’s power.49 For example, in such a market the platform might
be charging high prices to one user group, but if the prices are below
cost to the other, it is not clear that the platform should be viewed as
exercising market power.50 Conversely, if the platform is providing a
service to one user group for free, as Google does, it might be doing
so in order to enable it to charge a higher price to the other, so that the
zero price does not show that the platform lacks power.
It is not clear, however, that we should never consider price-
quality relationships on one side of the market in isolation. After all, if
the price to advertisers is high, it is no less high because the price to
searchers is low. Advertisers receive benefits from lower prices to
searchers, because the resulting greater quantity of searchers effec-

47. Jean-Charles Rochet & Jean Tirole, Two-Sided Markets: An Overview (Mar. 12,
2004) (unpublished manuscript), available at http://faculty.haas.berkeley.edu/
hermalin/rochet_tirole.pdf.
48. In fact, one could count web sites as a third customer class of Google, though the web
sites in the “organic” search results will often overlap with advertisers.
49. See David S. Evans & Richard Schmalensee, The Industrial Organization of Markets
with Two-Sided Platforms, 3 COMPETITION POL’Y INT’L 151, 173–75 (2007); see also David
S. Evans, Two-Sided Markets, in MARKET DEFINITION IN ANTITRUST: THEORY AND CASE
STUDIES 437 (ABA Section of Antitrust Law ed., 2012), available at http://papers.ssrn.
com/sol3/papers.cfm?abstract_id=1396751.
50. Nor is it clear, however, that it should not be.
 Google and Search Engine Market Power 17

tively makes Google a higher-quality provider for advertisers, but it is

not clear that the quality benefits compensate for the higher costs.
This is not an economic question so much as it is a legal one: even if
competition among providers in a two-sided market dissipates su-
pracompetitive profits, the law might take the position that one side of
the market should not be made to subsidize the other.
In fact, of course, Google provides its search results for
free.51 Applying the principle just discussed, the law could focus on
the search market and, despite the two-sided nature of a search engine,
determine that Google is exercising market power in the search mar-
ket.52 For example, suppose, referring to the allegations against
Google, that Google chooses to downgrade in its search results other
information providers that it perceives as posing competitive threats.
In that case, Google would be deliberately providing less valuable
information to its users. Should that be viewed as an exercise of pow-
er?53 Or, considering the two-sided market perspective, should the
answer to that question depend on whether and how Google’s adver-
tisers, the other side of the market, respond? Are we confident that the
advertisers would respond at all, especially given that searchers might
not even be aware of the distortion of search results? For competition-
law purposes, the answers to these questions turn on decisions that
have not yet been made regarding how competition law should treat
two-sided markets.
For Google, though, it may be possible to use the advertiser side
of the two-sided market to infer power on the searcher side. As sug-
gested above, the ability to move a site up or down in search results
without a significant consumer response could be viewed as evidence
of market power. That is, to put it another way, moving a site down
several spots in the results could be viewed as constituting a signifi-
cant lowering of quality that in the absence of market power would
cause a search engine to lose customers. If so, given that price stays
the same at zero, the ability to lower quality significantly could be
viewed as evidence of market power. This method of assessing power

51. Actually, one could take the view that in making searches on a search engine, users
provide that search engine with valuable information for which they are not paid. See supra
text following note 18.
52. Of course, even if Google were not exercising power in the market, it could still pos-
sess power in the market. For both Sherman Act § 2 and Article 102 TFEU, it is possession
of monopoly power or dominance that is the element of the offense. For neither provision is
the relationship of the conduct element of the offense to its power element well defined.
53. See Evans & Schmalensee, supra note 49, at 177 (“As a consequence, the two-sided
platform may impose requirements on side A that do not benefit them directly and which
customers on that side might even reject after comparing private benefits and costs.”). The
authors continue with some added complications: “But such requirements may benefit side
B. And if the demand increases on side B, these requirements may increase the value placed
on the platform on side A — and in fact could increase value so much that the feature pro-
vides a net benefit to side A.” Id. at 177–78.
18 Harvard Journal of Law & Technology

then requires some means of quantifying the loss in quality of moving

sites in search results.
One way to do this, offered tentatively below, is to use the prices
paid for placement in Google’s AdWords results. Because the value of
search positions in the organic results is not quantified, but AdWords
positions are, we can use AdWords as a proxy for the organic re-
sults.54 Indeed, if the prices for positions in the AdWords results differ
significantly, one would expect the difference in value between simi-
lar positions in the organic results to be even greater. That is so at
least if, as seems likely, consumers value more highly the unpaid-for
organic results than the paid-for AdWords results.
A more difficult question is whether we can use prices in the
AdWords market as a proxy for consumer harm. Power over advertis-
ers is relevant, but if searchers are not injured by the manipulation of
the position of web sites, then the significance of such manipulation is
less clear. For example, suppose there are several similar web sites,
and Google disadvantages, or is able to disadvantage, only one. In that
case, searchers might not be injured. But the sections below suggest
that effects in the upstream advertising market could indeed be rea-
sonable proxies for effects in the downstream searcher market.

B. Distortion as Abuse and Power

In the discussion above, Google’s ability to provide less-than-

optimal search results was suggested as evidence of market power.
But the distortion of search results is also the anticompetitive conduct
of which Google is accused. Thus, the same evidence shows both
power and conduct. This is theoretically reasonable. If a firm is able
to engage in anticompetitive conduct, that ability alone is evidence
that the firm possesses market power; otherwise, its lack of power
would constrain its ability to act anticompetitively. Indeed, there is a
movement in competition law toward looking directly to conduct to
determine the existence vel non of market power.
Separating power and conduct is difficult in exclusion cases in
part because the exclusion of a competitor is also the injury to con-
sumer, at least if consumers desire access to that competitor. So, in the
case of Google the exclusion of Yelp from effective access to con-
sumers would cause harm to those consumers if in fact they preferred
Yelp to the alternative they were provided. As the reference to the
arguments of Joshua Wright above describes, though, it is possible
that Google denies, or makes more difficult, access to Yelp because
Google knows that consumers prefer other services. In that respect,
the fact of exclusion tells us little about whether the exclusion is pro-

54. Actually, some of the allegations against Google concern AdWords, not the organic
results, so that the results in the text could be applied directly in that context.
 Google and Search Engine Market Power 19

competitive or anticompetitive, just as it tells us little about whether

the party doing the excluding — Google here — has power.
The argument that Google is merely providing what consumers
want seems more compelling for its organic search results than for
AdWords, though. The organic results are Google’s own product in a
more direct way than are the AdWords links. In a sense, the organic
results are the editorial content, where one would expect Google to
exercise considerable control, but AdWords are advertisements, where
one would expect Google to be more even-handed.55 That is especial-
ly so in that AdWords advertisers pay Google for clicks to their sites,
so that the sites must profit from consumer visits to avoid losing mon-
ey on the advertisements. That would presumably limit the likelihood
that advertisements unwanted by consumers would appear for long.
An analogy can be drawn here between Google and grocery
stores. Grocery stores are often paid by manufacturers for shelf space,
through payments known as slotting allowances or slotting fees, much
as Google is paid for advertising space.56 One of the arguments for
why such payments can be procompetitive is that of signaling: “When
a manufacturer believes its product is highly likely to succeed, it will
be willing to pay a significant slotting fee, knowing that it is likely to
recover this expense through profits earned from future sales.”57 Simi-
larly, one would expect that advertisers who choose to pay for Ad-
Words placement do so because they expect that their products will be
likely to succeed by pleasing consumers. In other words, the prices
that web sites are willing to pay for particular placements on Ad-
Words might in fact reflect the value to consumers of information
about those web sites. If so, then any distortion of that pricing could
be evidence of market power, because it would reflect a deviation
from consumer preferences.

C. Competitive Responses

If Google seeks to exclude competitors or potential competitors,

though, those competitors will respond, and that may limit Google’s
power. That is, even if Google could, in general, provide low-quality
results to searchers by excluding sites that searchers might prefer, and
even though it could be difficult for searchers to recognize the low
quality of the results, the excluded sites themselves might be able to
prevent this. If those sites have alternative means of reaching consum-
ers, any exclusion by Google would be ineffective and therefore prob-

55. Cf. Home Placement Serv., Inc. v. Providence Journal Co., 682 F.2d 274 (1st Cir.
1982).
56. See FEDERAL TRADE COMMISSION, SLOTTING ALLOWANCES IN THE RETAIL
GROCERY INDUSTRY: SELECTED CASE STUDIES IN FIVE PRODUCT CATEGORIES (2003),
available at http://www.ftc.gov/os/2003/11/slottingallowancerpt031114.pdf.
57. Id. at 1.
20 Harvard Journal of Law & Technology

ably would not occur. The question, then, is whether effective alterna-
tive avenues exist.
Certainly web sites like vertical search engines could advertise on
traditional media or on other web sites. Whether those means of ad-
vertising would be effective alternatives is not immediately obvious.
The advantage of advertising on Google is that ads can be targeted to
searchers seeking relevant products, and that same advantage cannot
be offered to the same extent by other sites, let alone by traditional
media. Therefore, although other means of reaching consumers exist,
it is not easy to determine how effective they are.
Again, though, the prices paid for AdWords, and particularly the
relative prices paid for AdWords placements, can provide an answer.
One allegation against Google is that it uses AdWords “Quality
Scores” to raise prices charged to its competitors or potential competi-
tors for placement on AdWords.58 For example, a web site that for-
merly paid $X for placement in the first position might subsequently
be forced, without a legitimate justification,59 to pay the same price
but only receive placement in the second position. If those prices are
signals of value to consumers, this sort of distortion could reflect
harm to consumers.
Moreover, the payments, and differences in payments, are an in-
dicator of the value, or lack thereof, of alternative means of reaching
consumers. For example, suppose that the price for placement as the
first AdWords result were 20% greater than that for placement as the
second result. If the value of the second placement is 20% lower than
the value of the first, then it seems likely that the value of alternative
means of reaching consumers is even less. Thus, not only do the pric-
es paid for AdWords seem likely to reflect value of advertisements to
consumers, differences in value seem likely to be evidence of alterna-
tives. As such, an investigation of these values seems to be a reasona-
ble approach to assessing Google’s power.

D. The Power to Manipulate Search Results

Although actual prices for AdWords are not easy to obtain, one
can use Google’s own “Traffic Estimator”60 to estimate some figures.
For example, using the keyword phrase “kitchen faucet,” the Traffic

58. See infra text accompanying note 62.

59. Of course, Google argues that it has legitimate justifications for such moves, and that
might be true; the purpose of this test is not to prejudge Google’s conduct, but to assess its
power. As suggested above, though, the signaling explanation suggests that AdWords pay-
ments may reflect value to consumers, so that distortion of the AdWords market would be
unjustified.
60. Traffic Estimator, GOOGLE ADWORDS, https://adwords.google.com/select/
TrafficEstimatorSandbox (last visited Mar. 3, 2013) (registration required).
 Google and Search Engine Market Power 21

Estimator provides the following numbers for different specified max-

imum costs per click (“CPC”):61

Table 1: Google AdWords Traffic Estimator for “Kitchen Faucet”

Maximum Estimated Estimated Estimated Estimated

CPC average CPC ad position daily clicks daily cost
(specified)
$2.50 $1.26 1.49 404.08 $510.90
$2.00 $1.11 1.71 379.86 $421.63
$1.50 $0.92 2.14 338.75 $312.35
$1.00 $0.70 3.00 264.30 $184.10
$0.50 $0.44 5.83 119.69 $ 52.16

As can be seen in the table, referring to the second and third col-
umns, the price difference between ad positions 2 and 3 in these esti-
mates is greater than ($0.92 – $0.70) / $0.92 = 23.9%. The difference
between positions 1 and 2 appears to be greater than ($1.26 – $0.92) /
$1.26 = 27.0%. For this keyword phrase, then, if Google could move
a site from position 2 to position 3 or from position 1 to position 2, it
would be decreasing the value of the placement by approximately
25%. The U.S. Merger Guidelines state that a price increase of 5–10%
is evidence of “a significant loss of competition,” so one could infer
that price differences of approximately 25% show the existence of
market power and, because 25% is significantly greater than 10%,
perhaps also monopoly power.
Admittedly, this approach poses some problems. It is not offered
here so much as a precise means of assessing market power as an ex-
ample of the kind of approach that could be used. One of the apparent
problems, though, is not a real one, at least in certain circumstances. It
might be thought that since the price paid goes down as the site’s po-
sition is lowered in the results, there is in fact no exercise of power,
but only a simultaneous reduction in quality and price. This is not
true, at least according to some of the allegations against Google.
Google assigns a Quality Score to sites that use AdWords, and the
effect of the quality score is to adjust the price paid.62 Consequently,
although the prices shown in the table above present an unadjusted
price that goes down with position in the results, Google in fact can
selectively charge higher prices than the unadjusted ones. As a result,
Google could charge a particular web site the same price for position

61. This experiment was conducted on March 3, 2012.

62. See Check and understand Quality Score, GOOGLE ADWORDS HELP (Oct. 14, 2012),
http://support.google.com/adwords/bin/answer.py?hl=en&answer=2454010.
22 Harvard Journal of Law & Technology

3 as it charges others for position 2 simply by adjusting the Quality

Score. (Again, this article takes no position on whether Google actual-
ly does this. The point here is that it could do so, and that if it did, this
conduct could demonstrate market power.)

V. THE OBJECTIVITY BASELINE

Although this paper does not aim to cover all the issues that are
relevant to Google’s market power, one final issue is worth consider-
ing, if only briefly. Google’s statements regarding the objectivity of
its search results have evolved in recent years.63 In 2007, it said “Our
search results are generated completely objectively and are independ-
ent of the beliefs and preferences of those who work at Google.”64
Subsequently it said that it was delivering the results that searchers
wanted. More recently, it has said that “Google’s search results are
ultimately a scientific opinion as to what information users will
find most useful.”65 At times, it makes no reference to the prefer-
ences of searchers at all, saying in its documentation, for example,
that it “tried to clarify where possible that although we employ algo-
rithms in our rankings, ultimately we consider our search results to be
our opinion.”66 And Google, if it chose, could simply say, “We’ll pro-
vide the search results that we want to provide.”
Does objectivity have any implications for the assessment of
Google’s market power? Suppose that Google does in fact manipulate
search results. On the one hand, we can ask if either its power to do so
or the anticompetitiveness of its conduct would be less if it clearly
stated that it manipulated results, or reserved the right to do so. On the
other, we can ask if its power would be greater or its conduct worse if
Google maintained that its results were entirely objective. The analo-
gous issues rarely arise in the context of non-informational products,
perhaps because in that context the competitive effect of conduct is
more apparent, or at least less dependent on consumer perceptions.
Statements and perceptions about objectivity can have implica-
tions for consumer acceptance of information, but whether these ef-
fects should be relevant for competition law is less clear. Information
costs have been recognized as a factor that can create or maintain

63. Cade Metz, Google drops nuke on ‘objective’ search engine utopia, THE REGISTER
(Dec. 12, 2010, 1:25 PM), http://www.theregister.co.uk/2010/12/16/google_
algorithms_are_google_opinions/.
64. Google: An explanation of our search results, GOOGLE (Mar. 21, 2007),
http://web.archive.org/web/20070321092528/http://www.google.com/explanation.html.
65. Schmidt Testimony, supra note 2, at 3 (emphasis in original).
66. Matt Cutts, Comment to In Google’s Opinion . . . . , JOHN BATTELLE’S SEARCH BLOG
(Dec. 1, 2010) http://battellemedia.com/archives/2010/12/in_googles_opinion.php; see also
Schmidt Testimony, supra note 2.
 Google and Search Engine Market Power 23

market power for competition-law purposes,67 and it seems that

statements about information provided could play three possible roles.
First, if no statement is made, then there is presumably no change in
information costs. There is then presumably no effect on market pow-
er, and the failure to make any statement is not likely to be viewed as
anticompetitive. Second, an information provider might publicly state
that the information it provides is manipulated. In that case, the state-
ment probably reduces the provider’s market power, and of course
making the statement would be procompetitive, or at least not anti-
competitive (though the manipulation might remain anticompetitive).
The third possibility is that while providing biased information
the provider states that it is not biasing the information. Although this
would be a misrepresentation, it is not entirely clear that it should be
viewed as raising information costs for competition purposes, though
it certainly would not lower them. If the default understanding of con-
sumers is that information is unbiased, then the misrepresentation
would not change their perceptions, though it might make them less
willing to reconsider them. Moreover, competition law is generally
distinct from the law of false advertising, and it is not clear where the
boundary lies. Perhaps a statement that applies to all of an information
provider’s products, like a statement by a search engine about the ob-
jectivity of its results, is a competition issue, while a statement about a
particular product, like a statement that a food product is “organic,” is
false advertising. Because competition law does not have a well-
developed approach to information products, this issue is unclear.68

VI. CONCLUSION

The purpose of this paper is to take some first steps toward devel-
oping methods of assessing the market power of information provid-
ers like Google. As information products come to constitute a larger
portion of the market, and a larger portion of allegations regarding

67. See Eastman Kodak Co. v. Image Technical Serv., Inc., 504 U.S. 451 (1992).
68. A statement about objectivity could also have significance in the context of two-sided
markets. As noted before, one of the difficulties posed by two-sided markets is the interrela-
tionship between pricing and output on the two sides. That interrelationship means that
exploitation of one side of the market is not necessarily an indicator of market power, be-
cause both sides must be considered. A supracompetitive price on one side of the market, if
balanced by an infracompetitive price on the other, might not be proof of power. In some
sense, though, if Google makes representations about the search side of its market, that
could be viewed as isolating that side of the market for competition purposes. The idea
would be that any statement that defines the price-quality relationship for a product, as a
representation of objectivity in the context of zero price would, sets a baseline against which
competition analysis could be performed. No longer could the provider argue that its con-
duct must be measured by its effect on both sides of the market, because it has chosen to
limit its options on one side. In effect, the provider would have unilaterally assumed the
obligation to act competitively on one side of the market. As suggested in part II.B supra,
competition law might choose to impose that obligation in any case.
24 Harvard Journal of Law & Technology

anticompetitive conduct, competition law must develop techniques for

addressing the special problems posed by information. The value of
information is difficult to determine, in part because it is often used
for the production of other products, rather than for consumption. Re-
latedly, information is often provided for free in order to sell other
products, so that information providers often operate in two-sided
markets characterized by the interaction of two sets of consumers.
These problems are well illustrated by Google. Like many infor-
mation providers in both the old and new media, Google provides its
search results to consumers for free in order to sell advertising. In one
important respect, though, Google differs from other information pro-
viders: there is no clear benchmark against which to evaluate the qual-
ity of the search results it provides. Consequently, the price-quality-
cost relationship that determines market power is difficult to evaluate.
This paper tentatively offers an indirect means of making that evalua-
tion and offers some general observations regarding the treatment of
information products in two-sided markets.