Archives For Bing

[TOTM: The following is part of a digital symposium by TOTM guests and authors on the law, economics, and policy of the antitrust lawsuits against Google. The entire series of posts is available here.]

The U.S. Department of Justice’s (DOJ) antitrust case against Google, which was filed in October 2020, will be a tough slog.[1] It is an alleged monopolization (Sherman Act, Sec. 2) case; and monopolization cases are always a tough slog.

In this brief essay I will lay out some of the issues in the case and raise an intriguing possibility.

What is the case about?

The case is about exclusivity and exclusion in the distribution of search engine services; that Google paid substantial sums to Apple and to the manufacturers of Android-based mobile phones and tablets and also to wireless carriers and web-browser proprietors—in essence, to distributors—to install the Google search engine as the exclusive pre-set (installed), default search program. The suit alleges that Google thereby made it more difficult for other search-engine providers (e.g., Bing; DuckDuckGo) to obtain distribution for their search-engine services and thus to attract search-engine users and to sell the online advertising that is associated with search-engine use and that provides the revenue to support the search “platform” in this “two-sided market” context.[2]

Exclusion can be seen as a form of “raising rivals’ costs.”[3]  Equivalently, exclusion can be seen as a form of non-price predation. Under either interpretation, the exclusionary action impedes competition.

It’s important to note that these allegations are different from those that motivated an investigation by the Federal Trade Commission (which the FTC dropped in 2013) and the cases by the European Union against Google.[4]  Those cases focused on alleged self-preferencing; that Google was unduly favoring its own products and services (e.g., travel services) in its delivery of search results to users of its search engine. In those cases, the impairment of competition (arguably) happens with respect to those competing products and services, not with respect to search itself.

What is the relevant market?

For a monopolization allegation to have any meaning, there needs to be the exercise of market power (which would have adverse consequences for the buyers of the product). And in turn, that exercise of market power needs to occur in a relevant market: one in which market power can be exercised.

Here is one of the important places where the DOJ’s case is likely to turn into a slog: the delineation of a relevant market for alleged monopolization cases remains as a largely unsolved problem for antitrust economics.[5]  This is in sharp contrast to the issue of delineating relevant markets for the antitrust analysis of proposed mergers.  For this latter category, the paradigm of the “hypothetical monopolist” and the possibility that this hypothetical monopolist could prospectively impose a “small but significant non-transitory increase in price” (SSNIP) has carried the day for the purposes of market delineation.

But no such paradigm exists for monopolization cases, in which the usual allegation is that the defendant already possesses market power and has used the exclusionary actions to buttress that market power. To see the difficulties, it is useful to recall the basic monopoly diagram from Microeconomics 101. A monopolist faces a negatively sloped demand curve for its product (at higher prices, less is bought; at lower prices, more is bought) and sets a profit-maximizing price at the level of output where its marginal revenue (MR) equals its marginal costs (MC). Its price is thereby higher than an otherwise similar competitive industry’s price for that product (to the detriment of buyers) and the monopolist earns higher profits than would the competitive industry.

But unless there are reliable benchmarks as to what the competitive price and profits would otherwise be, any information as to the defendant’s price and profits has little value with respect to whether the defendant already has market power. Also, a claim that a firm does not have market power because it faces rivals and thus isn’t able profitably to raise its price from its current level (because it would lose too many sales to those rivals) similarly has no value. Recall the monopolist from Micro 101. It doesn’t set a higher price than the one where MR=MC, because it would thereby lose too many sales to other sellers of other things.

Thus, any firm—regardless of whether it truly has market power (like the Micro 101 monopolist) or is just another competitor in a sea of competitors—should have already set its price at its profit-maximizing level and should find it unprofitable to raise its price from that level.[6]  And thus the claim, “Look at all of the firms that I compete with!  I don’t have market power!” similarly has no informational value.

Let us now bring this problem back to the Google monopolization allegation:  What is the relevant market?  In the first instance, it has to be “the provision of answers to user search queries.” After all, this is the “space” in which the exclusion occurred. But there are categories of search: e.g., search for products/services, versus more general information searches (“What is the current time in Delaware?” “Who was the 21st President of the United States?”). Do those separate categories themselves constitute relevant markets?

Further, what would the exercise of market power in a (delineated relevant) market look like?  Higher-than-competitive prices for advertising that targets search-results recipients is one obvious answer (but see below). In addition, because this is a two-sided market, the competitive “price” (or prices) might involve payments by the search engine to the search users (in return for their exposure to the lucrative attached advertising).[7]  And product quality might exhibit less variety than a competitive market would provide; and/or the monopolistic average level of quality would be lower than in a competitive market: e.g., more abuse of user data, and/or deterioration of the delivered information itself, via more self-preferencing by the search engine and more advertising-driven preferencing of results.[8]

In addition, a natural focus for a relevant market is the advertising that accompanies the search results. But now we are at the heart of the difficulty of delineating a relevant market in a monopolization context. If the relevant market is “advertising on search engine results pages,” it seems highly likely that Google has market power. If the relevant market instead is all online U.S. advertising (of which Google’s revenue share accounted for 32% in 2019[9]), then the case is weaker; and if the relevant market is all advertising in the United States (which is about twice the size of online advertising[10]), the case is weaker still. Unless there is some competitive benchmark, there is no easy way to delineate the relevant market.[11]

What exactly has Google been paying for, and why?

As many critics of the DOJ’s case have pointed out, it is extremely easy for users to switch their default search engine. If internet search were a normal good or service, this ease of switching would leave little room for the exercise of market power. But in that case, why is Google willing to pay $8-$12 billion annually for the exclusive default setting on Apple devices and large sums to the manufacturers of Android-based devices (and to wireless carriers and browser proprietors)? Why doesn’t Google instead run ads in prominent places that remind users how superior Google’s search results are and how easy it is for users (if they haven’t already done so) to switch to the Google search engine and make Google the user’s default choice?

Suppose that user inertia is important. Further suppose that users generally have difficulty in making comparisons with respect to the quality of delivered search results. If this is true, then being the default search engine on Apple and Android-based devices and on other distribution vehicles would be valuable. In this context, the inertia of their customers is a valuable “asset” of the distributors that the distributors may not be able to take advantage of, but that Google can (by providing search services and selling advertising). The question of whether Google’s taking advantage of this user inertia means that Google exercises market power takes us back to the issue of delineating the relevant market.

There is a further wrinkle to all of this. It is a well-understood concept in antitrust economics that an incumbent monopolist will be willing to pay more for the exclusive use of an essential input than a challenger would pay for access to the input.[12] The basic idea is straightforward. By maintaining exclusive use of the input, the incumbent monopolist preserves its (large) monopoly profits. If the challenger enters, the incumbent will then earn only its share of the (much lower, more competitive) duopoly profits. Similarly, the challenger can expect only the lower duopoly profits. Accordingly, the incumbent should be willing to outbid (and thereby exclude) the challenger and preserve the incumbent’s exclusive use of the input, so as to protect those monopoly profits.

To bring this to the Google monopolization context, if Google does possess market power in some aspect of search—say, because online search-linked advertising is a relevant market—then Google will be willing to outbid Microsoft (which owns Bing) for the “asset” of default access to Apple’s (inertial) device owners. That Microsoft is a large and profitable company and could afford to match (or exceed) Google’s payments to Apple is irrelevant. If the duopoly profits for online search-linked advertising would be substantially lower than Google’s current profits, then Microsoft would not find it worthwhile to try to outbid Google for that default access asset.

Alternatively, this scenario could be wholly consistent with an absence of market power. If search users (who can easily switch) consider Bing to be a lower-quality search service, then large payments by Microsoft to outbid Google for those exclusive default rights would be largely wasted, since the “acquired” default search users would quickly switch to Google (unless Microsoft provided additional incentives for the users not to switch).

But this alternative scenario returns us to the original puzzle:  Why is Google making such large payments to the distributors for those exclusive default rights?

An intriguing possibility

Consider the following possibility. Suppose that Google was paying that $8-$12 billion annually to Apple in return for the understanding that Apple would not develop its own search engine for Apple’s device users.[13] This possibility was not raised in the DOJ’s complaint, nor is it raised in the subsequent suits by the state attorneys general.

But let’s explore the implications by going to an extreme. Suppose that Google and Apple had a formal agreement that—in return for the $8-$12 billion per year—Apple would not develop its own search engine. In this event, this agreement not to compete would likely be seen as a violation of Section 1 of the Sherman Act (which does not require a market delineation exercise) and Apple would join Google as a co-conspirator. The case would take on the flavor of the FTC’s prosecution of “pay-for-delay” agreements between the manufacturers of patented pharmaceuticals and the generic drug manufacturers that challenge those patents and then receive payments from the former in return for dropping the patent challenge and delaying the entry of the generic substitute.[14]

As of this writing, there is no evidence of such an agreement and it seems quite unlikely that there would have been a formal agreement. But the DOJ will be able to engage in discovery and take depositions. It will be interesting to find out what the relevant executives at Google—and at Apple—thought was being achieved by those payments.

What would be a suitable remedy/relief?

The DOJ’s complaint is vague with respect to the remedy that it seeks. This is unsurprising. The DOJ may well want to wait to see how the case develops and then amend its complaint.

However, even if Google’s actions have constituted monopolization, it is difficult to conceive of a suitable and effective remedy. One apparently straightforward remedy would be to require simply that Google not be able to purchase exclusivity with respect to the pre-set default settings. In essence, the device manufacturers and others would always be able to sell parallel default rights to other search engines: on the basis, say, that the default rights for some categories of customers—or even a percentage of general customers (randomly selected)—could be sold to other search-engine providers.

But now the Gilbert-Newbery insight comes back into play. Suppose that a device manufacturer knows (or believes) that Google will pay much more if—even in the absence of any exclusivity agreement—Google ends up being the pre-set search engine for all (or nearly all) of the manufacturer’s device sales, as compared with what the manufacturer would receive if those default rights were sold to multiple search-engine providers (including, but not solely, Google). Can that manufacturer (recall that the distributors are not defendants in the case) be prevented from making this sale to Google and thus (de facto) continuing Google’s exclusivity?[15]

Even a requirement that Google not be allowed to make any payment to the distributors for a default position may not improve the competitive environment. Google may be able to find other ways of making indirect payments to distributors in return for attaining default rights, e.g., by offering them lower rates on their online advertising.

Further, if the ultimate goal is an efficient outcome in search, it is unclear how far restrictions on Google’s bidding behavior should go. If Google were forbidden from purchasing any default installation rights for its search engine, would (inert) consumers be better off? Similarly, if a distributor were to decide independently that its customers were better served by installing the Google search engine as the default, would that not be allowed? But if it is allowed, how could one be sure that Google wasn’t indirectly paying for this “independent” decision (e.g., through favorable advertising rates)?

It’s important to remember that this (alleged) monopolization is different from the Standard Oil case of 1911 or even the (landline) AT&T case of 1984. In those cases, there were physical assets that could be separated and spun off to separate companies. For Google, physical assets aren’t important. Although it is conceivable that some of Google’s intellectual property—such as Gmail, YouTube, or Android—could be spun off to separate companies, doing so would do little to cure the (arguably) fundamental problem of the inert device users.

In addition, if there were an agreement between Google and Apple for the latter not to develop a search engine, then large fines for both parties would surely be warranted. But what next? Apple can’t be forced to develop a search engine.[16] This differentiates such an arrangement from the “pay-for-delay” arrangements for pharmaceuticals, where the generic manufacturers can readily produce a near-identical substitute for the patented drug and are otherwise eager to do so.

At the end of the day, forbidding Google from paying for exclusivity may well be worth trying as a remedy. But as the discussion above indicates, it is unlikely to be a panacea and is likely to require considerable monitoring for effective enforcement.

Conclusion

The DOJ’s case against Google will be a slog. There are unresolved issues—such as how to delineate a relevant market in a monopolization case—that will be central to the case. Even if the DOJ is successful in showing that Google violated Section 2 of the Sherman Act in monopolizing search and/or search-linked advertising, an effective remedy seems problematic. But there also remains the intriguing question of why Google was willing to pay such large sums for those exclusive default installation rights?

The developments in the case will surely be interesting.


[1] The DOJ’s suit was joined by 11 states.  More states subsequently filed two separate antitrust lawsuits against Google in December.

[2] There is also a related argument:  That Google thereby gained greater volume, which allowed it to learn more about its search users and their behavior, and which thereby allowed it to provide better answers to users (and thus a higher-quality offering to its users) and better-targeted (higher-value) advertising to its advertisers.  Conversely, Google’s search-engine rivals were deprived of that volume, with the mirror-image negative consequences for the rivals.  This is just another version of the standard “learning-by-doing” and the related “learning curve” (or “experience curve”) concepts that have been well understood in economics for decades.

[3] See, for example, Steven C. Salop and David T. Scheffman, “Raising Rivals’ Costs: Recent Advances in the Theory of Industrial Structure,” American Economic Review, Vol. 73, No. 2 (May 1983), pp.  267-271; and Thomas G. Krattenmaker and Steven C. Salop, “Anticompetitive Exclusion: Raising Rivals’ Costs To Achieve Power Over Price,” Yale Law Journal, Vol. 96, No. 2 (December 1986), pp. 209-293.

[4] For a discussion, see Richard J. Gilbert, “The U.S. Federal Trade Commission Investigation of Google Search,” in John E. Kwoka, Jr., and Lawrence J. White, eds. The Antitrust Revolution: Economics, Competition, and Policy, 7th edn.  Oxford University Press, 2019, pp. 489-513.

[5] For a more complete version of the argument that follows, see Lawrence J. White, “Market Power and Market Definition in Monopolization Cases: A Paradigm Is Missing,” in Wayne D. Collins, ed., Issues in Competition Law and Policy. American Bar Association, 2008, pp. 913-924.

[6] The forgetting of this important point is often termed “the cellophane fallacy”, since this is what the U.S. Supreme Court did in a 1956 antitrust case in which the DOJ alleged that du Pont had monopolized the cellophane market (and du Pont, in its defense claimed that the relevant market was much wider: all flexible wrapping materials); see U.S. v. du Pont, 351 U.S. 377 (1956).  For an argument that profit data and other indicia argued for cellophane as the relevant market, see George W. Stocking and Willard F. Mueller, “The Cellophane Case and the New Competition,” American Economic Review, Vol. 45, No. 1 (March 1955), pp. 29-63.

[7] In the context of differentiated services, one would expect prices (positive or negative) to vary according to the quality of the service that is offered.  It is worth noting that Bing offers “rewards” to frequent searchers; see https://www.microsoft.com/en-us/bing/defaults-rewards.  It is unclear whether this pricing structure of payment to Bing’s customers represents what a more competitive framework in search might yield, or whether the payment just indicates that search users consider Bing to be a lower-quality service.

[8] As an additional consequence of the impairment of competition in this type of search market, there might be less technological improvement in the search process itself – to the detriment of users.

[9] As estimated by eMarketer: https://www.emarketer.com/newsroom/index.php/google-ad-revenues-to-drop-for-the-first-time/.

[10] See https://www.visualcapitalist.com/us-advertisers-spend-20-years/.

[11] And, again, if we return to the du Pont cellophane case:  Was the relevant market cellophane?  Or all flexible wrapping materials?

[12] This insight is formalized in Richard J. Gilbert and David M.G. Newbery, “Preemptive Patenting and the Persistence of Monopoly,” American Economic Review, Vol. 72, No. 3 (June 1982), pp. 514-526.

[13] To my knowledge, Randal C. Picker was the first to suggest this possibility; see https://www.competitionpolicyinternational.com/a-first-look-at-u-s-v-google/.  Whether Apple would be interested in trying to develop its own search engine – given the fiasco a decade ago when Apple tried to develop its own maps app to replace the Google maps app – is an open question.  In addition, the Gilbert-Newbery insight applies here as well:  Apple would be less inclined to invest the substantial resources that would be needed to develop a search engine when it is thereby in a duopoly market.  But Google might be willing to pay “insurance” to reinforce any doubts that Apple might have.

[14] The U.S. Supreme Court, in FTC v. Actavis, 570 U.S. 136 (2013), decided that such agreements could be anti-competitive and should be judged under the “rule of reason”.  For a discussion of the case and its implications, see, for example, Joseph Farrell and Mark Chicu, “Pharmaceutical Patents and Pay-for-Delay: Actavis (2013),” in John E. Kwoka, Jr., and Lawrence J. White, eds. The Antitrust Revolution: Economics, Competition, and Policy, 7th edn.  Oxford University Press, 2019, pp. 331-353.

[15] This is an example of the insight that vertical arrangements – in this case combined with the Gilbert-Newbery effect – can be a way for dominant firms to raise rivals’ costs.  See, for example, John Asker and Heski Bar-Isaac. 2014. “Raising Retailers’ Profits: On Vertical Practices and the Exclusion of Rivals.” American Economic Review, Vol. 104, No. 2 (February 2014), pp. 672-686.

[16] And, again, for the reasons discussed above, Apple might not be eager to make the effort.

The CPI Antitrust Chronicle published Geoffrey Manne’s and my recent paperThe Problems and Perils of Bootstrapping Privacy and Data into an Antitrust Framework as part of a symposium on Big Data in the May 2015 issue. All of the papers are worth reading and pondering, but of course ours is the best ;).

In it, we analyze two of the most prominent theories of antitrust harm arising from data collection: privacy as a factor of non-price competition, and price discrimination facilitated by data collection. We also analyze whether data is serving as a barrier to entry and effectively preventing competition. We argue that, in the current marketplace, there are no plausible harms to competition arising from either non-price effects or price discrimination due to data collection online and that there is no data barrier to entry preventing effective competition.

The issues of how to regulate privacy issues and what role competition authorities should in that, are only likely to increase in importance as the Internet marketplace continues to grow and evolve. The European Commission and the FTC have been called on by scholars and advocates to take greater consideration of privacy concerns during merger review and encouraged to even bring monopolization claims based upon data dominance. These calls should be rejected unless these theories can satisfy the rigorous economic review of antitrust law. In our humble opinion, they cannot do so at this time.

Excerpts:

PRIVACY AS AN ELEMENT OF NON-PRICE COMPETITION

The Horizontal Merger Guidelines have long recognized that anticompetitive effects may “be manifested in non-price terms and conditions that adversely affect customers.” But this notion, while largely unobjectionable in the abstract, still presents significant problems in actual application.

First, product quality effects can be extremely difficult to distinguish from price effects. Quality-adjusted price is usually the touchstone by which antitrust regulators assess prices for competitive effects analysis. Disentangling (allegedly) anticompetitive quality effects from simultaneous (neutral or pro-competitive) price effects is an imprecise exercise, at best. For this reason, proving a product-quality case alone is very difficult and requires connecting the degradation of a particular element of product quality to a net gain in advantage for the monopolist.

Second, invariably product quality can be measured on more than one dimension. For instance, product quality could include both function and aesthetics: A watch’s quality lies in both its ability to tell time as well as how nice it looks on your wrist. A non-price effects analysis involving product quality across multiple dimensions becomes exceedingly difficult if there is a tradeoff in consumer welfare between the dimensions. Thus, for example, a smaller watch battery may improve its aesthetics, but also reduce its reliability. Any such analysis would necessarily involve a complex and imprecise comparison of the relative magnitudes of harm/benefit to consumers who prefer one type of quality to another.

PRICE DISCRIMINATION AS A PRIVACY HARM

If non-price effects cannot be relied upon to establish competitive injury (as explained above), then what can be the basis for incorporating privacy concerns into antitrust? One argument is that major data collectors (e.g., Google and Facebook) facilitate price discrimination.

The argument can be summed up as follows: Price discrimination could be a harm to consumers that antitrust law takes into consideration. Because companies like Google and Facebook are able to collect a great deal of data about their users for analysis, businesses could segment groups based on certain characteristics and offer them different deals. The resulting price discrimination could lead to many consumers paying more than they would in the absence of the data collection. Therefore, the data collection by these major online companies facilitates price discrimination that harms consumer welfare.

This argument misses a large part of the story, however. The flip side is that price discrimination could have benefits to those who receive lower prices from the scheme than they would have in the absence of the data collection, a possibility explored by the recent White House Report on Big Data and Differential Pricing.

While privacy advocates have focused on the possible negative effects of price discrimination to one subset of consumers, they generally ignore the positive effects of businesses being able to expand output by serving previously underserved consumers. It is inconsistent with basic economic logic to suggest that a business relying on metrics would want to serve only those who can pay more by charging them a lower price, while charging those who cannot afford it a larger one. If anything, price discrimination would likely promote more egalitarian outcomes by allowing companies to offer lower prices to poorer segments of the population—segments that can be identified by data collection and analysis.

If this group favored by “personalized pricing” is as big as—or bigger than—the group that pays higher prices, then it is difficult to state that the practice leads to a reduction in consumer welfare, even if this can be divorced from total welfare. Again, the question becomes one of magnitudes that has yet to be considered in detail by privacy advocates.

DATA BARRIER TO ENTRY

Either of these theories of harm is predicated on the inability or difficulty of competitors to develop alternative products in the marketplace—the so-called “data barrier to entry.” The argument is that upstarts do not have sufficient data to compete with established players like Google and Facebook, which in turn employ their data to both attract online advertisers as well as foreclose their competitors from this crucial source of revenue. There are at least four reasons to be dubious of such arguments:

  1. Data is useful to all industries, not just online companies;
  2. It’s not the amount of data, but how you use it;
  3. Competition online is one click or swipe away; and
  4. Access to data is not exclusive

CONCLUSION

Privacy advocates have thus far failed to make their case. Even in their most plausible forms, the arguments for incorporating privacy and data concerns into antitrust analysis do not survive legal and economic scrutiny. In the absence of strong arguments suggesting likely anticompetitive effects, and in the face of enormous analytical problems (and thus a high risk of error cost), privacy should remain a matter of consumer protection, not of antitrust.

Recent years have seen an increasing interest in incorporating privacy into antitrust analysis. The FTC and regulators in Europe have rejected these calls so far, but certain scholars and activists continue their attempts to breathe life into this novel concept. Elsewhere we have written at length on the scholarship addressing the issue and found the case for incorporation wanting. Among the errors proponents make is a persistent (and woefully unsubstantiated) assertion that online data can amount to a barrier to entry, insulating incumbent services from competition and ensuring that only the largest providers thrive. This data barrier to entry, it is alleged, can then allow firms with monopoly power to harm consumers, either directly through “bad acts” like price discrimination, or indirectly by raising the costs of advertising, which then get passed on to consumers.

A case in point was on display at last week’s George Mason Law & Economics Center Briefing on Big Data, Privacy, and Antitrust. Building on their growing body of advocacy work, Nathan Newman and Allen Grunes argued that this hypothesized data barrier to entry actually exists, and that it prevents effective competition from search engines and social networks that are interested in offering services with heightened privacy protections.

According to Newman and Grunes, network effects and economies of scale ensure that dominant companies in search and social networking (they specifically named Google and Facebook — implying that they are in separate markets) operate without effective competition. This results in antitrust harm, they assert, because it precludes competition on the non-price factor of privacy protection.

In other words, according to Newman and Grunes, even though Google and Facebook offer their services for a price of $0 and constantly innovate and upgrade their products, consumers are nevertheless harmed because the business models of less-privacy-invasive alternatives are foreclosed by insufficient access to data (an almost self-contradicting and silly narrative for many reasons, including the big question of whether consumers prefer greater privacy protection to free stuff). Without access to, and use of, copious amounts of data, Newman and Grunes argue, the algorithms underlying search and targeted advertising are necessarily less effective and thus the search product without such access is less useful to consumers. And even more importantly to Newman, the value to advertisers of the resulting consumer profiles is diminished.

Newman has put forth a number of other possible antitrust harms that purportedly result from this alleged data barrier to entry, as well. Among these is the increased cost of advertising to those who wish to reach consumers. Presumably this would harm end users who have to pay more for goods and services because the costs of advertising are passed on to them. On top of that, Newman argues that ad networks inherently facilitate price discrimination, an outcome that he asserts amounts to antitrust harm.

FTC Commissioner Maureen Ohlhausen (who also spoke at the George Mason event) recently made the case that antitrust law is not well-suited to handling privacy problems. She argues — convincingly — that competition policy and consumer protection should be kept separate to preserve doctrinal stability. Antitrust law deals with harms to competition through the lens of economic analysis. Consumer protection law is tailored to deal with broader societal harms and aims at protecting the “sanctity” of consumer transactions. Antitrust law can, in theory, deal with privacy as a non-price factor of competition, but this is an uneasy fit because of the difficulties of balancing quality over two dimensions: Privacy may be something some consumers want, but others would prefer a better algorithm for search and social networks, and targeted ads with free content, for instance.

In fact, there is general agreement with Commissioner Ohlhausen on her basic points, even among critics like Newman and Grunes. But, as mentioned above, views diverge over whether there are some privacy harms that should nevertheless factor into competition analysis, and on whether there is in fact  a data barrier to entry that makes these harms possible.

As we explain below, however, the notion of data as an antitrust-relevant barrier to entry is simply a myth. And, because all of the theories of “privacy as an antitrust harm” are essentially predicated on this, they are meritless.

First, data is useful to all industries — this is not some new phenomenon particular to online companies

It bears repeating (because critics seem to forget it in their rush to embrace “online exceptionalism”) that offline retailers also receive substantial benefit from, and greatly benefit consumers by, knowing more about what consumers want and when they want it. Through devices like coupons and loyalty cards (to say nothing of targeted mailing lists and the age-old practice of data mining check-out receipts), brick-and-mortar retailers can track purchase data and better serve consumers. Not only do consumers receive better deals for using them, but retailers know what products to stock and advertise and when and on what products to run sales. For instance:

  • Macy’s analyzes tens of millions of terabytes of data every day to gain insights from social media and store transactions. Over the past three years, the use of big data analytics alone has helped Macy’s boost its revenue growth by 4 percent annually.
  • Following its acquisition of Kosmix in 2011, Walmart established @WalmartLabs, which created its own product search engine for online shoppers. In the first year of its use alone, the number of customers buying a product on Walmart.com after researching a purchase increased by 20 percent. According to Ron Bensen, the vice president of engineering at @WalmartLabs, the combination of in-store and online data could give brick-and-mortar retailers like Walmart an advantage over strictly online stores.
  • Panera and a whole host of restaurants, grocery stores, drug stores and retailers use loyalty cards to advertise and learn about consumer preferences.

And of course there is a host of others uses for data, as well, including security, fraud prevention, product optimization, risk reduction to the insured, knowing what content is most interesting to readers, etc. The importance of data stretches far beyond the online world, and far beyond mere retail uses more generally. To describe even online giants like Amazon, Apple, Microsoft, Facebook and Google as having a monopoly on data is silly.

Second, it’s not the amount of data that leads to success but building a better mousetrap

The value of knowing someone’s birthday, for example, is not in that tidbit itself, but in the fact that you know this is a good day to give that person a present. Most of the data that supports the advertising networks underlying the Internet ecosphere is of this sort: Information is important to companies because of the value that can be drawn from it, not for the inherent value of the data itself. Companies don’t collect information about you to stalk you, but to better provide goods and services to you.

Moreover, data itself is not only less important than what can be drawn from it, but data is also less important than the underlying product it informs. For instance, Snapchat created a challenger to  Facebook so successfully (and in such short time) that Facebook attempted to buy it for $3 billion (Google offered $4 billion). But Facebook’s interest in Snapchat wasn’t about its data. Instead, Snapchat was valuable — and a competitive challenge to Facebook — because it cleverly incorporated the (apparently novel) insight that many people wanted to share information in a more private way.

Relatedly, Twitter, Instagram, LinkedIn, Yelp, Pinterest (and Facebook itself) all started with little (or no) data and they have had a lot of success. Meanwhile, despite its supposed data advantages, Google’s attempts at social networking — Google+ — have never caught up to Facebook in terms of popularity to users (and thus not to advertisers either). And scrappy social network Ello is starting to build a significant base without data collection for advertising at all.

At the same time it’s simply not the case that the alleged data giants — the ones supposedly insulating themselves behind data barriers to entry — actually have the type of data most relevant to startups anyway. As Andres Lerner has argued, if you wanted to start a travel business, the data from Kayak or Priceline would be far more relevant. Or if you wanted to start a ride-sharing business, data from cab companies would be more useful than the broad, market-cross-cutting profiles Google and Facebook have. Consider companies like Uber, Lyft and Sidecar that had no customer data when they began to challenge established cab companies that did possess such data. If data were really so significant, they could never have competed successfully. But Uber, Lyft and Sidecar have been able to effectively compete because they built products that users wanted to use — they came up with an idea for a better mousetrap.The data they have accrued came after they innovated, entered the market and mounted their successful challenges — not before.

In reality, those who complain about data facilitating unassailable competitive advantages have it exactly backwards. Companies need to innovate to attract consumer data, otherwise consumers will switch to competitors (including both new entrants and established incumbents). As a result, the desire to make use of more and better data drives competitive innovation, with manifestly impressive results: The continued explosion of new products, services and other apps is evidence that data is not a bottleneck to competition but a spur to drive it.

Third, competition online is one click or thumb swipe away; that is, barriers to entry and switching costs are low

Somehow, in the face of alleged data barriers to entry, competition online continues to soar, with newcomers constantly emerging and triumphing. This suggests that the barriers to entry are not so high as to prevent robust competition.

Again, despite the supposed data-based monopolies of Facebook, Google, Amazon, Apple and others, there exist powerful competitors in the marketplaces they compete in:

  • If consumers want to make a purchase, they are more likely to do their research on Amazon than Google.
  • Google flight search has failed to seriously challenge — let alone displace —  its competitors, as critics feared. Kayak, Expedia and the like remain the most prominent travel search sites — despite Google having literally purchased ITA’s trove of flight data and data-processing acumen.
  • People looking for local reviews go to Yelp and TripAdvisor (and, increasingly, Facebook) as often as Google.
  • Pinterest, one of the most highly valued startups today, is now a serious challenger to traditional search engines when people want to discover new products.
  • With its recent acquisition of the shopping search engine, TheFind, and test-run of a “buy” button, Facebook is also gearing up to become a major competitor in the realm of e-commerce, challenging Amazon.
  • Likewise, Amazon recently launched its own ad network, “Amazon Sponsored Links,” to challenge other advertising players.

Even assuming for the sake of argument that data creates a barrier to entry, there is little evidence that consumers cannot easily switch to a competitor. While there are sometimes network effects online, like with social networking, history still shows that people will switch. MySpace was considered a dominant network until it made a series of bad business decisions and everyone ended up on Facebook instead. Similarly, Internet users can and do use Bing, DuckDuckGo, Yahoo, and a plethora of more specialized search engines on top of and instead of Google. And don’t forget that Google itself was once an upstart new entrant that replaced once-household names like Yahoo and AltaVista.

Fourth, access to data is not exclusive

Critics like Newman have compared Google to Standard Oil and argued that government authorities need to step in to limit Google’s control over data. But to say data is like oil is a complete misnomer. If Exxon drills and extracts oil from the ground, that oil is no longer available to BP. Data is not finite in the same way. To use an earlier example, Google knowing my birthday doesn’t limit the ability of Facebook to know my birthday, as well. While databases may be proprietary, the underlying data is not. And what matters more than the data itself is how well it is analyzed.

This is especially important when discussing data online, where multi-homing is ubiquitous, meaning many competitors end up voluntarily sharing access to data. For instance, I can use the friend-finder feature on WordPress to find Facebook friends, Google connections, and people I’m following on Twitter who also use the site for blogging. Using this feature allows WordPress to access your contact list on these major online players.

Friend-Finder

Further, it is not apparent that Google’s competitors have less data available to them. Microsoft, for instance, has admitted that it may actually have more data. And, importantly for this discussion, Microsoft may have actually garnered some of its data for Bing from Google.

If Google has a high cost per click, then perhaps it’s because it is worth it to advertisers: There are more eyes on Google because of its superior search product. Contra Newman and Grunes, Google may just be more popular for consumers and advertisers alike because the algorithm makes it more useful, not because it has more data than everyone else.

Fifth, the data barrier to entry argument does not have workable antitrust remedies

The misguided logic of data barrier to entry arguments leaves a lot of questions unanswered. Perhaps most important among these is the question of remedies. What remedy would apply to a company found guilty of leveraging its market power with data?

It’s actually quite difficult to conceive of a practical means for a competition authority to craft remedies that would address the stated concerns without imposing enormous social costs. In the unilateral conduct context, the most obvious remedy would involve the forced sharing of data.

On the one hand, as we’ve noted, it’s not clear this would actually accomplish much. If competitors can’t actually make good use of data, simply having more of it isn’t going to change things. At the same time, such a result would reduce the incentive to build data networks to begin with. In their startup stage, companies like Uber and Facebook required several months and hundreds of thousands, if not millions, of dollars to design and develop just the first iteration of the products consumers love. Would any of them have done it if they had to share their insights? In fact, it may well be that access to these free insights is what competitors actually want; it’s not the data they’re lacking, but the vision or engineering acumen to use it.

Other remedies limiting collection and use of data are not only outside of the normal scope of antitrust remedies, they would also involve extremely costly court supervision and may entail problematic “collisions between new technologies and privacy rights,” as the last year’s White House Report on Big Data and Privacy put it.

It is equally unclear what an antitrust enforcer could do in the merger context. As Commissioner Ohlhausen has argued, blocking specific transactions does not necessarily stop data transfer or promote privacy interests. Parties could simply house data in a standalone entity and enter into licensing arrangements. And conditioning transactions with forced data sharing requirements would lead to the same problems described above.

If antitrust doesn’t provide a remedy, then it is not clear why it should apply at all. The absence of workable remedies is in fact a strong indication that data and privacy issues are not suitable for antitrust. Instead, such concerns would be better dealt with under consumer protection law or by targeted legislation.

After more than a year of complaining about Google and being met with responses from me (see also here, here, here, here, and here, among others) and many others that these complaints have yet to offer up a rigorous theory of antitrust injury — let alone any evidence — FairSearch yesterday offered up its preferred remedies aimed at addressing, in its own words, “the fundamental conflict of interest driving Google’s incentive and ability to engage in anti-competitive conduct. . . . [by putting an] end [to] Google’s preferencing of its own products ahead of natural search results.”  Nothing in the post addresses the weakness of the organization’s underlying claims, and its proposed remedies would be damaging to consumers.

FairSearch’s first and core “abuse” is “[d]iscriminatory treatment favoring Google’s own vertical products in a manner that may harm competing vertical products.”  To address this it proposes prohibiting Google from preferencing its own content in search results and suggests as additional, “structural remedies” “[r]equiring Google to license data” and “[r]equiring Google to divest its vertical products that have benefited from Google’s abuses.”

Tom Barnett, former AAG for antitrust, counsel to FairSearch member Expedia, and FairSearch’s de facto spokesman should be ashamed to be associated with claims and proposals like these.  He better than many others knows that harm to competitors is not the issue under US antitrust laws.  Rather, US antitrust law requires a demonstration that consumers — not just rivals — will be harmed by a challenged practice.  He also knows (as economists have known for a long time) that favoring one’s own content — i.e., “vertically integrating” to produce both inputs as well as finished products — is generally procompetitive.

In fact, Barnett has said as much before:

Because a Section 2 violation hurts competitors, they are often the focus of section 2 remedial efforts.  But competitor well-being, in itself, is not the purpose of our antitrust laws.

Access remedies also raise efficiency and innovation concerns.  By forcing a firm to share the benefits of its investments and relieving its rivals of the incentive to develop comparable assets of their own, access remedies can reduce the competitive vitality of an industry.

Not only has FairSearch not actually demonstrated that Google has preferenced its own products, the organization has also not demonstrated either harm to consumers arising from such conduct nor even antitrust-cognizable harm to competitors arising from it.

As an empirical study supported by the International Center for Law and Economics (itself, in turn, supported in part by Google, and of which I am the Executive Director) makes clear, search bias simply almost never occurs.  And when it does, it is the non-dominant Bing that more often practices it, not Google.  Moreover, and most important, the evidence marshaled in favor of the search bias claim (largely adduced by Harvard Business School professor, Ben Edelman (whose work is supported by Microsoft)) demonstrates that consumers do, indeed, have the ability to detect and counter allegedly biased results.

Recall what search bias means in this context.  According to Edelman, looking at the top three search results, Google links to its own content (think Gmail, Google Maps, etc.) in the first search result about twice as often as Yahoo! and Bing link to Google content in this position.  While the ICLE paper refutes even this finding, notice what it means:  “Biased” search results lead to a reshuffling of results among the top few results offered up; there is no evidence that Google simply drops users’ preferred results.  While it is true that the difference in click-through rates between the top and second results can be significant, Edelman’s own findings actually demonstrate that consumers are capable of finding what they want when their preferred (more relevant) results appears in the second or third slot.

Edelman notes that Google ranks Gmail first and Yahoo! Mail second in his study, even though users seem to think Yahoo! Mail is the more relevant result:  Gmail receives only 29% of clicks while Yahoo! Mail receives 54%.  According to Edelman, this is proof that Google’s conduct forecloses access by competitors and harms consumers under the antitrust laws.

But is it?  Note that users click on the second, apparently more-relevant result nearly twice as often as they click on the first.  This demonstrates that Yahoo! is not competitively foreclosed from access to users, and that users are perfectly capable of identifying their preferred results, even when they appear lower in the results page.  This is simply not foreclosure — in fact, if anything, it demonstrates the opposite.

Among other things, foreclosure — limiting access by a competitor to a necessary input — under the antitrust laws must be substantial enough to prevent a rival from reaching sufficient scale that it can effectively compete.  It is no more “foreclosure” for Google to “impair” traffic to Kayak’s site by offering its own Flight Search than it is for Safeway to refuse to allow Kroger to sell Safeway’s house brand.  Rather, actionable foreclosure requires that a firm “impair[s] the ability of rivals to grow into effective competitors that erode the firm’s position.”  Such quantifiable claims are noticeably absent from critic’s complaints against Google.

And what about those allegedly harmed competitors?  How are they faring?  As of September 2012, Google ranks 7th in visits among metasearch travel sites, with a paltry 1.4% of such visits.  Residing at number one?  FairSearch founding member, Kayak, with a whopping 61% (up from 52% six months after Google entered the travel search business).  Nextag.com, another vocal Google critic, has complained that Google’s conduct has forced it to shift its strategy from attracting traffic through Google’s organic search results to other sources, including paid ads on Google.com.  And how has it fared?  It has parlayed its experience with new data sources into a successful new business model, Wize Commerce, showing exactly the sort of “incentive to develop comparable assets of their own” Barnett worries will be destroyed by aggressive antitrust enforcement.  And Barnett’s own Expedia.com?  Currently, it’s the largest travel company in the world, and it has only grown in recent years.

Meanwhile consumers’ interests have been absent from critics’ complaints since the beginning.  And not only do they fail to demonstrate any connection between harm to consumers and the claimed harms to competitors arising from Google’s conduct, but they also ignore the harm to consumers that may result from restricting potentially efficient business conduct — like the integration of Google Maps and other products into its search results.  That Google not only produces search results but also owns some of the content that generates those results is not a problem cognizable by modern antitrust.

FairSearch and other Google critics have utterly failed to make a compelling case, and their proposed remedies would serve only to harm, not help, consumers.

In my series of three posts (here, here and here) drawn from my empirical study on search bias I have examined whether search bias exists, and, if so, how frequently it occurs.  This, the final post in the series, assesses the results of the study (as well as the Edelman & Lockwood (E&L) study to which it responds) to determine whether the own-content bias I’ve identified is in fact consistent with anticompetitive foreclosure or is otherwise sufficient to warrant antitrust intervention.

As I’ve repeatedly emphasized, while I refer to differences among search engines’ rankings of their own or affiliated content as “bias,” without more these differences do not imply anticompetitive conduct.  It is wholly unsurprising and indeed consistent with vigorous competition among engines that differentiation emerges with respect to algorithms.  However, it is especially important to note that the theories of anticompetitive foreclosure raised by Google’s rivals involve very specific claims about these differences.  Properly articulated vertical foreclosure theories proffer both that bias is (1) sufficient in magnitude to exclude Google’s rivals from achieving efficient scale, and (2) actually directed at Google’s rivals.  Unfortunately for search engine critics, their theories fail on both counts.  The observed own-content bias appears neither to be extensive enough to prevent rivals from gaining access to distribution nor does it appear to target Google’s rivals; rather, it seems to be a natural result of intense competition between search engines and of significant benefit to consumers.

Vertical foreclosure arguments are premised upon the notion that rivals are excluded with sufficient frequency and intensity as to render their efforts to compete for distribution uneconomical.  Yet the empirical results simply do not indicate that market conditions are in fact conducive to the types of harmful exclusion contemplated by application of the antitrust laws.  Rather, the evidence indicates that (1) the absolute level of search engine “bias” is extremely low, and (2) “bias” is not a function of market power, but an effective strategy that has arisen as a result of serious competition and innovation between and by search engines.  The first finding undermines competitive foreclosure arguments on their own terms, that is, even if there were no pro-consumer justifications for the integration of Google content with Google search results.  The second finding, even more importantly, reveals that the evolution of consumer preferences for more sophisticated and useful search results has driven rival search engines to satisfy that demand.  Both Bing and Google have shifted toward these results, rendering the complained-of conduct equivalent to satisfying the standard of care in the industry–not restraining competition.

A significant lack of search bias emerges in the representative sample of queries.  This result is entirely unsurprising, given that bias is relatively infrequent even in E&L’s sample of queries specifically designed to identify maximum bias.  In the representative sample, the total percentage of queries for which Google references its own content when rivals do not is even lower—only about 8%—meaning that Google favors its own content far less often than critics have suggested.  This fact is crucial and highly problematic for search engine critics, as their burden in articulating a cognizable antitrust harm includes not only demonstrating that bias exists, but further that it is actually competitively harmful.  As I’ve discussed, bias alone is simply not sufficient to demonstrate any prima facie anticompetitive harm as it is far more often procompetitive or competitively neutral than actively harmful.  Moreover, given that bias occurs in less than 10% of queries run on Google, anticompetitive exclusion arguments appear unsustainable.

Indeed, theories of vertical foreclosure find virtually zero empirical support in the data.  Moreover, it appears that, rather than being a function of monopolistic abuse of power, search bias has emerged as an efficient competitive strategy, allowing search engines to differentiate their products in ways that benefit consumers.  I find that when search engines do reference their own content on their search results pages, it is generally unlikely that another engine will reference this same content.  However, the fact that both this percentage and the absolute level of own content inclusion is similar across engines indicates that this practice is not a function of market power (or its abuse), but is rather an industry standard.  In fact, despite conducting a much smaller percentage of total consumer searches, Bing is consistently more biased than Google, illustrating that the benefits search engines enjoy from integrating their own content into results is not necessarily a function of search engine size or volume of queries.  These results are consistent with a business practice that is efficient and at significant tension with arguments that such integration is designed to facilitate competitive foreclosure. Continue Reading…

My last two posts on search bias (here and here) have analyzed and critiqued Edelman & Lockwood’s small study on search bias.  This post extends this same methodology and analysis to a random sample of 1,000 Google queries (released by AOL in 2006), to develop a more comprehensive understanding of own-content bias.  As I’ve stressed, these analyses provide useful—but importantly limited—glimpses into the nature of the search engine environment.  While these studies are descriptively helpful, actual harm to consumer welfare must always be demonstrated before cognizable antitrust injuries arise.  And naked identifications of own-content bias simply do not inherently translate to negative effects on consumers (see, e.g., here and here for more comprehensive discussion).

Now that’s settled, let’s jump into the results of the 1,000 random search query study.

How Do Search Engines Rank Their Own Content?

Consistent with our earlier analysis, a starting off point for thinking about measuring differentiation among search engines with respect to placing their own content is to compare how a search engine ranks its own content relative to how other engines place that same content (e.g. to compare how Google ranks “Google Maps” relative to how Bing or Blekko rank it).   Restricting attention exclusively to the first or “top” position, I find that Google simply does not refer to its own content in over 90% of queries.  Similarly, Bing does not reference Microsoft content in 85.4% of queries.  Google refers to its own content in the first position when other search engines do not in only 6.7% of queries; while Bing does so over twice as often, referencing Microsoft content that no other engine references in the first position in 14.3% of queries.  The following two charts illustrate the percentage of Google or Bing first position results, respectively, dedicated to own content across search engines.

The most striking aspect of these results is the small fraction of queries for which placement of own-content is relevant.  The results are similar when I expand consideration to the entire first page of results; interestingly, however, while the levels of own-content bias are similar considering the entire first page of results, Bing is far more likely than Google to reference its own content in its very first results position.

Examining Search Engine “Bias” on Google

Two distinct differences between the results of this larger study and my replication of Edelman & Lockwood emerge: (1) Google and Bing refer to their own content in a significantly smaller percentage of cases here than in the non-random sample; and (2) in general, when Google or Bing does rank its own content highly, rival engines are unlikely to similarly rank that same content.

The following table reports the percentages of queries for which Google’s ranking of its own content and its rivals’ rankings of that same content differ significantly. When Google refers to its own content within its Top 5 results, at least one other engine similarly ranks this content for only about 5% of queries.

The following table presents the likelihood that Google content will appear in a Google search, relative to searches conducted on rival engines (reported in odds ratios).

The first and third columns report results indicating that Google affiliated content is more likely to appear in a search executed on Google rather than rival engines.  Google is approximately 16 times more likely to refer to its own content on its first page as is any other engine.  Bing and Blekko are both significantly less likely to refer to Google content in their first result or on their first page than Google is to refer to Google content within these same parameters.  In each iteration, Bing is more likely to refer to Google content than is Blekko, and in the case of the first result, Bing is much more likely to do so.  Again, to be clear, the fact that Bing is more likely to rank its own content is not suggestive that the practice is problematic.  Quite the contrary, the demonstration that firms both with and without market power in search (to the extent that is a relevant antitrust market) engage in similar conduct the correct inference is that there must be efficiency explanations for the practice.  The standard response, of course, is that the competitive implications of a practice are different when a firm with market power does it.  That’s not exactly right.  It is true that firms with market power can engage in conduct that gives rise to potential antitrust problems when the same conduct from a firm without market power would not; however, when firms without market power engage in the same business practice it demands that antitrust analysts seriously consider the efficiency implications of the practice.  In other words, there is nothing in the mantra that things are “different” when larger firms do them that undercut potential efficiency explanations.

Examining Search Engine “Bias” on Bing

For queries within the larger sample, Bing refers to Microsoft content within its Top 1 and 3 results when no other engine similarly references this content for a slightly smaller percentage of queries than in my Edelman & Lockwood replication.  Yet Bing continues to exhibit a strong tendency to rank Microsoft content more prominently than rival engines.  For example, when Bing refers to Microsoft content within its Top 5 results, other engines agree with this ranking for less than 2% of queries; and Bing refers to Microsoft content that no other engine does within its Top 3 results for 99.2% of queries:

Regression analysis further illustrates Bing’s propensity to reference Microsoft content that rivals do not.  The following table reports the likelihood that Microsoft content is referred to in a Bing search as compared to searches on rival engines (again reported in odds ratios).

Bing refers to Microsoft content in its first results position about 56 times more often than rival engines refer to Microsoft content in this same position.  Across the entire first page, Microsoft content appears on a Bing search about 25 times more often than it does on any other engine.  Both Google and Blekko are accordingly significantly less likely to reference Microsoft content.  Notice further that, contrary to the findings in the smaller study, Google is slightly less likely to return Microsoft content than is Blekko, both in its first results position and across its entire first page.

A Closer Look at Google v. Bing

 Consistent with the smaller sample, I find again that Bing is more biased than Google using these metrics.  In other words, Bing ranks its own content significantly more highly than its rivals do more frequently then Google does, although the discrepancy between the two engines is smaller here than in the study of Edelman & Lockwood’s queries.  As noted above, Bing is over twice as likely to refer to own content in first results position than is Google.

Figures 7 and 8 present the same data reported above, but with Blekko removed, to allow for a direct visual comparison of own-content bias between Google and Bing.

Consistent with my earlier results, Bing appears to consistently rank Microsoft content higher than Google ranks the same (Microsoft) content more frequently than Google ranks Google content more prominently than Bing ranks the same (Google) content.

This result is particularly interesting given the strength of the accusations condemning Google for behaving in precisely this way.  That Bing references Microsoft content just as often as—and frequently even more often than!—Google references its own content strongly suggests that this behavior is a function of procompetitive product differentiation, and not abuse of market power.  But I’ll save an in-depth analysis of this issue for my next post, where I’ll also discuss whether any of the results reported in this series of posts support anticompetitive foreclosure theories or otherwise suggest antitrust intervention is warranted.

In my last post, I discussed Edelman & Lockwood’s (E&L’s) attempt to catch search engines in the act of biasing their results—as well as their failure to actually do so.  In this post, I present my own results from replicating their study.  Unlike E&L, I find that Bing is consistently more biased than Google, for reasons discussed further below, although neither engine references its own content as frequently as E&L suggest.

I ran searches for E&L’s original 32 non-random queries using three different search engines—Google, Bing, and Blekko—between June 23 and July 5 of this year.  This replication is useful, as search technology has changed dramatically since E&L recorded their results in August 2010.  Bing now powers Yahoo, and Blekko has had more time to mature and enhance its results.  Blekko serves as a helpful “control” engine in my study, as it is totally independent of Google and Microsoft, and so has no incentive to refer to Google or Microsoft content unless it is actually relevant to users.  In addition, because Blekko’s model is significantly different than Google and Microsoft’s, if results on all three engines agree that specific content is highly relevant to the user query, it lends significant credibility to the notion that the content places well on the merits rather than being attributable to bias or other factors.

How Do Search Engines Rank Their Own Content?

Focusing solely upon the first position, Google refers to its own products or services when no other search engine does in 21.9% of queries; in another 21.9% of queries, both Google and at least one other search engine rival (i.e. Bing or Blekko) refer to the same Google content with their first links.

But restricting focus upon the first position is too narrow.  Assuming that all instances in which Google or Bing rank their own content first and rivals do not amounts to bias would be a mistake; such a restrictive definition would include cases in which all three search engines rank the same content prominently—agreeing that it is highly relevant—although not all in the first position. 

The entire first page of results provides a more informative comparison.  I find that Google and at least one other engine return Google content on the first page of results in 7% of the queries.  Google refers to its own content on the first page of results without agreement from either rival search engine in only 7.9% of the queries.  Meanwhile, Bing and at least one other engine refer to Microsoft content in 3.2% of the queries.  Bing references Microsoft content without agreement from either Google or Blekko in 13.2% of the queries:

This evidence indicates that Google’s ranking of its own content differs significantly from its rivals in only 7.9% of queries, and that when Google ranks its own content prominently it is generally perceived as relevant.  Further, these results suggest that Bing’s organic search results are significantly more biased in favor of Microsoft content than Google’s search results are in favor of Google’s content.

Examining Search Engine “Bias” on Google

The following table presents the percentages of queries for which Google’s ranking of its own content differs significantly from its rivals’ ranking of that same content.

Note that percentages below 50 in this table indicate that rival search engines generally see the referenced Google content as relevant and independently believe that it should be ranked similarly.

So when Google ranks its own content highly, at least one rival engine typically agrees with this ranking; for example, when Google places its own content in its Top 3 results, at least one rival agrees with this ranking in over 70% of queries.  Bing especially agrees with Google’s rankings of Google content within its Top 3 and 5 results, failing to include Google content that Google ranks similarly in only a little more than a third of queries.

Examining Search Engine “Bias” on Bing

Bing refers to Microsoft content in its search results far more frequently than its rivals reference the same Microsoft content.  For example, Bing’s top result references Microsoft content for 5 queries, while neither Google nor Blekko ever rank Microsoft content in the first position:

This table illustrates the significant discrepancies between Bing’s treatment of its own Microsoft content relative to Google and Blekko.  Neither rival engine refers to Microsoft content Bing ranks within its Top 3 results; Google and Blekko do not include any Microsoft content Bing refers to on the first page of results in nearly 80% of queries.

Moreover, Bing frequently ranks Microsoft content highly even when rival engines do not refer to the same content at all in the first page of results.  For example, of the 5 queries for which Bing ranks Microsoft content in its top result, Google refers to only one of these 5 within its first page of results, while Blekko refers to none.  Even when comparing results across each engine’s full page of results, Google and Blekko only agree with Bing’s referral of Microsoft content in 20.4% of queries.

Although there are not enough Bing data to test results in the first position in E&L’s sample, Microsoft content appears as results on the first page of a Bing search about 7 times more often than Microsoft content appears on the first page of rival engines.  Also, Google is much more likely to refer to Microsoft content than Blekko, though both refer to significantly less Microsoft content than Bing.

A Closer Look at Google v. Bing

On E&L’s own terms, Bing results are more biased than Google results; rivals are more likely to agree with Google’s algorithmic assessment (than with Bing’s) that its own content is relevant to user queries.  Bing refers to Microsoft content other engines do not rank at all more often than Google refers its own content without any agreement from rivals.  Figures 1 and 2 display the same data presented above in order to facilitate direct comparisons between Google and Bing.

As Figures 1 and 2 illustrate, Bing search results for these 32 queries are more frequently “biased” in favor of its own content than are Google’s.  The bias is greatest for the Top 1 and Top 3 search results.

My study finds that Bing exhibits far more “bias” than E&L identify in their earlier analysis.  For example, in E&L’s study, Bing does not refer to Microsoft content at all in its Top 1 or Top 3 results; moreover, Bing refers to Microsoft content within its entire first page 11 times, while Google and Yahoo refer to Microsoft content 8 and 9 times, respectively.  Most likely, the significant increase in Bing’s “bias” differential is largely a function of Bing’s introduction of localized and personalized search results and represents serious competitive efforts on Bing’s behalf.

Again, it’s important to stress E&L’s limited and non-random sample, and to emphasize the danger of making strong inferences about the general nature or magnitude of search bias based upon these data alone.  However, the data indicate that Google’s own-content bias is relatively small even in a sample collected precisely to focus upon the queries most likely to generate it.  In fact—as I’ll discuss in my next post—own-content bias occurs even less often in a more representative sample of queries, strongly suggesting that such bias does not raise the competitive concerns attributed to it.

Last week I linked to my new study on “search bias.”  At the time I noted I would have a few blog posts in the coming days discussing the study.  This is the first of those posts.

A lot of the frenzy around Google turns on “search bias,” that is, instances when Google references its own links or its own content (such as Google Maps or YouTube) in its search results pages.  Some search engine critics condemn such references as inherently suspect and almost by their very nature harmful to consumers.  Yet these allegations suffer from several crucial shortcomings.  As I’ve noted (see, e.g., here and here), these naked assertions of discrimination are insufficient to state a cognizable antitrust claim, divorced as they are from consumer welfare analysis.  Indeed, such “discrimination” (some would call it “vertical integration”) has a well-recognized propensity to yield either pro-competitive or competitively neutral outcomes, rather than concrete consumer welfare losses.  Moreover, because search engines exist in an incredibly dynamic environment, marked by constant innovation and fierce competition, we would expect different engines, utilizing different algorithms and appealing to different consumer preferences, to emerge.  So when search engines engage in product differentiation of this sort, there is no reason to be immediately suspicious of these business decisions.

No reason to be immediately suspicious – but there could, conceivably, be a problem.  If there is, we would want to see empirical evidence of it—of both the existence of bias, as well as the consumer harm emanating from it.  But one of the most notable features of this debate is the striking lack of empirical data.  Surprisingly little research has been done in this area, despite frequent assertions that own-content bias is commonly practiced and poses a significant threat to consumers (see, e.g., here).

My paper is an attempt to rectify this.  In the paper, I investigate the available data to determine whether and to what extent own-content bias actually occurs, by analyzing and replicating a study by Ben Edelman and Ben Lockwood (E&L) and conducting my own study of a larger, randomized set of search queries.

In this post I discuss my analysis and critique of E&L; in future posts I’ll present my own replication of their study, as well as the results of my larger study of 1,000 random search queries.  Finally, I’ll analyze whether any of these findings support anticompetitive foreclosure theories or are otherwise sufficient to warrant antitrust intervention.

E&L “investigate . . . [w]hether search engines’ algorithmic results favor their own services, and if so, which search engines do most, to what extent, and in what substantive areas.”  Their approach is to measure the difference in how frequently search engines refer to their own content relative to how often their rivals do so.

One note at the outset:  While this approach provides useful descriptive facts about the differences between how search engines link to their own content, it does little to inform antitrust analysis because Edelman and Lockwood begin with the rather odd claim that competition among differentiated search engines for consumers is a puzzle that creates an air of suspicion around the practice—in fact, they claim that “it is hard to see why results would vary . . . across search engines.”  This assertion, of course, is simply absurd.  Indeed, Danny Sullivan provides a nice critique of this claim:

It’s not hard to see why search engine result differ at all.  Search engines each use their own “algorithm” to cull through the pages they’ve collected from across the web, to decide which pages to rank first . . . . Google has a different algorithm than Bing.  In short, Google will have a different opinion than Bing.  Opinions in the search world, as with the real world, don’t always agree.

Moreover, this assertion completely discounts both the vigorous competitive product differentiation that occurs in nearly all modern product markets as well as the obvious selection effects at work in own-content bias (Google users likely prefer Google content).  This combination detaches E&L’s analysis from the consumer welfare perspective, and thus antitrust policy relevance, despite their claims to the contrary (and the fact that their results actually exhibit very little bias).

Several methodological issues undermine the policy relevance of E&L’s analysis.  First, they hand select 32 search queries and execute searches on Google, Bing, Yahoo, AOL and Ask.  This hand-selected non-random sample of 32 search queries cannot generate reliable inferences regarding the frequency of bias—a critical ingredient to understanding its potential competitive effects.  Indeed, E&L acknowledge their queries are chosen precisely because they are likely to return results including Google content (e.g., email, images, maps, video, etc.).

E&L analyze the top three organic search results for each query on each engine.  They find that 19% of all results across all five search engines refer to content affiliated with one of them.  They focus upon the first three organic results and report that Google refers to its own content in the first (“top”) position about twice as often as Yahoo and Bing refer to Google content in this position.  Additionally, they note that Yahoo is more biased than Google when evaluating the first page rather than only the first organic search result.

E&L also offer a strained attempt to deal with the possibility of competitive product differentiation among search engines.  They examine differences among search engines’ references to their own content by “compar[ing] the frequency with which a search engine links to its own pages, relative to the frequency with which other search engines link to that search engine’s pages.”  However, their evidence undermines claims that Google’s own-content bias is significant and systematic relative to its rivals’.  In fact, almost zero evidence of statistically significant own-content bias by Google emerges.

E&L find, in general, Google is no more likely to refer to its own content than other search engines are to refer to that same content, and across the vast majority of their results, E&L find Google search results are not statistically more likely to refer to Google content than rivals’ search results.

The same data can be examined to test the likelihood that a search engine will refer to content affiliated with a rival search engine.  Rather than exhibiting bias in favor of an engine’s own content, a “biased” search engine might conceivably be less likely to refer to content affiliated with its rivals.  The table below reports the likelihood (in odds ratios) that a search engine’s content appears in a rival engine’s results.

The first two columns of the table demonstrate that both Google and Yahoo content are referred to in the first search result less frequently in rivals’ search results than in their own.  Although Bing does not have enough data for robust analysis of results in the first position in E&L’s original analysis, the next three columns in Table 1 illustrate that all three engines’ (Google, Yahoo, and Bing) content appears less often on the first page of rivals’ search results than on their own search engine.  However, only Yahoo’s results differ significantly from 1.  As between Google and Bing, the results are notably similar.

E&L also make a limited attempt to consider the possibility that favorable placement of a search engine’s own content is a response to user preferences rather than anticompetitive motives.  Using click-through data, they find, unsurprisingly, that the first search result tends to receive the most clicks (72%, on average).  They then identify one search term for which they believe bias plays an important role in driving user traffic.  For the search query “email,” Google ranks its own Gmail first and Yahoo Mail second; however, E&L also find that Gmail receives only 29% of clicks while Yahoo Mail receives 54%.  E&L claim that this finding strongly indicates that Google is engaging in conduct that harms users and undermines their search experience.

However, from a competition analysis perspective, that inference is not sound.  Indeed, the fact that the second-listed Yahoo Mail link received the majority of clicks demonstrates precisely that Yahoo was not competitively foreclosed from access to users.  Taken collectively, E&L are not able to muster evidence of potential competitive foreclosure.

While it’s important to have an evidence-based discussion surrounding search engine results and their competitive implications, it’s also critical to recognize that bias alone is not evidence of competitive harm.  Indeed, any identified bias must be evaluated in the appropriate antitrust economic context of competition and consumers, rather than individual competitors and websites.  E&L’s analysis provides a useful starting point for describing how search engines differ in their referrals to their own content.  But, taken at face value, their results actually demonstrate little or no evidence of bias—let alone that the little bias they do find is causing any consumer harm.

As I’ll discuss in coming posts, evidence gathered since E&L conducted their study further suggests their claims that bias is prevalent, inherently harmful, and sufficient to warrant antitrust intervention are overstated and misguided.

Former TOTM blog symposium participant Joshua Gans (visiting Microsoft Research) has a post at TAP on l’affair hiybbprqag, about which I blogged previously here.

Gans notes, as I did, that Microsoft is not engaged in wholesale copying of Google’s search results, even though doing so would be technologically feasible.  But Gans goes on to draw a normative conclusion:

Let’s start with “imitation,” “copying” and its stronger variants of “plagiarism” and “cheating.” Had Bing wanted to do this and directly map Google’s search results onto its own, it could have done it. It could have set up programs to enter terms in Google and skimmed off the results and then used them directly. And I think we can all agree that that is wrong. Why? Two reasons. First, if Google has invested to produce those results, if others can just hang off them and copy it, Google’s may not earn the return on its efforts it should do. Second, if Bing were doing this and representing itself as a different kind of search, then that misrepresentation would be misleading. Thus, imitation reduces Google’s reward for innovation while adding no value in terms of diversity.

His first reason why this would be wrong is . . . silly.  I mean, I don’t want to get into a moral debate, but since when is it wrong to engage in activity that “may” hamper another firm’s ability to earn the return on its effort that it “should” (whatever “should” means here)?  I always thought that was called “competition” and we encouraged it.  As I noted the other day, competition via imitation is an important part of Schumpeterian capitalism.  To claim that reducing another company’s profits via imitation is wrong, but doing so via innovation is good and noble, is to hang one’s hat on a distinction that does not really exist.

The second argument, that doing so would amount to misrepresentation, is possible, but I’m sure if Microsoft were actually just copying Google’s results their representations would look different than they do now and the problem would probably not exist, so this claim is speculative, at best.

Now, regardless, I doubt it would be profitable for Microsoft to copy Google wholesale, and this is basically just a red herring (as Gans understands–he goes on to discuss the more “innocuous” imitation at issue).  While I think Gans’ claims that it would be “wrong” are just hand waiving, I am confident it would be “wrong” from the point of view of Microsoft’s bottom line–or else they would already be doing it.  In this context, that would seem to be the only standard that matters, unless there were a legal basis for the claim.

On this score, Gans points us to Shane Greenstein (Kellogg).  Greenstein writes:

Let’s start with a weak standard, the law. Legally speaking, imitation is allowed so long as a firm does not violate laws governing patents, copyright, or trade secrets. Patents obviously do not apply to this situation, and neither does copyright  because Google does not get a copyright on a search result. It also does not appear as if Googles trade secrets were violated. So, generally speaking, it does not appear as if any law has been broken.

This is all well and good, but Greenstein goes on to engage in his own casual moralizing, and his comments are worth reproducing (imitating?) at some length:

The norms of rivalry

There is nothing wrong with one retailer walking through a rival’s shop and getting ideas for what to do. There is really nothing wrong with a designer of a piece of electronic equipment buying a rival’s product and studying it in order to get new ideas for a  better design. 

In the modern Internet, however, there is no longer any privacy for users. Providers want to know as much as they can, and generally the rich suppliers can learn quite a lot about user conduct and preferences.

That means that rivals can learn a great deal about how users conduct their business, even when they are at a rival’s site. It is as if one retailer had a camera in a rival’s store, or one designer could learn the names of the buyer’s of their rival’s products, and interview them right away.

In the offline world, such intimate familiarity with a rival’s users and their transactions would be uncomfortable. It would seem like an intrusion on the transaction between user and supplier. Why is it permissible in the online world? Why is there any confusion about this being an intrusion in the online world? Why isn’t Microsoft’s behavior seen — cut and dry — as an intrusion?

In other words, the transaction between supplier and user is between supplier and user, and nobody else should be able to observe it without permission of both supplier and user. The user alone does not have the right or ability to invite another party to observe all aspects of the transaction.

That is what bothers me about Bing’s behavior. There is nothing wrong with them observing users, but they are doing more than just that. They are observing their rival’s transaction with users. And learning from it. In other contexts that would not be allowed without explicit permission of both parties — both user and supplier.

Moreover, one party does not like it in this case, as they claim the transaction with users as something they have a right to govern and keep to themselves. There is some merit in that claim.

In most contexts it seems like the supplier’s wishes should be respected. Why not online? (emphasis mine)

Where on Earth do these moral standards come from?  In what way is it not “allowed” (whatever that means here) for a firm to observe and learn from a rival’s transactions with users?  I can see why the rival would prefer it to be otherwise, of course, but so what?  They would also prefer to eradicate their meddlesome rival entirely, if possible (hence Microsoft’s considerable engagement with antitrust authorities concerning Google’s business), but we hardly elevate such desires to the realm of the moral.

What I find most troublesome is the controlling, regulatory mindset implicit in these analyses.  Here’s Gans again:

Outright imitation of this type should be prohibited but what do we call some more innocuous types? Just look at how the look and feel of the iPhone has been adopted by some mobile software developers just as the consumer success of graphic based interfaces did in an earlier time. This certainly reduces Apple’s reward for its innovations but the hit on diversity is murkier because while some features are common, competitors have tried to differentiate themselves. So this is not imitation but it is something more common, leveraging without compensation and how you feel about it depends on just how much reward you think pioneers should receive.

It is usually politicians and not economists (other than politico-economists like Krugman) who think they have a handle on–and an obligation to do something about–things like “how much reward . . .pioneers should receive.”  I would have thought the obvious answer to the question would be either “the optimal amount, but good luck knowing what that is or expecting to find it in the real world,” or else, for the Second Best, “whatever the market gives them.”  The implication that there is some moral standard appreciable by human mortals, or even human economists, is a recipe for disaster.

One of my favorite stories in the ongoing saga over the regulation (and thus the future) of Internet search emerged earlier this week with claims by Google that Microsoft has been copying its answers–using Google search results to bolster the relevance of its own results for certain search terms.  The full story from Internet search journalist extraordinaire, Danny Sullivan, is here, with a follow up discussing Microsoft’s response here.  The New York Times is also on the case with some interesting comments from a former Googler that feed nicely into the Schumpeterian competition angle (discussed below).  And Microsoft consultant (“though on matters unrelated to issues discussed here”)  and Harvard Business prof Ben Edelman coincidentally echoes precisely Microsoft’s response in a blog post here.

What I find so great about this story is how it seems to resolve one of the most significant strands of the ongoing debate–although it does so, from Microsoft’s point of view, unintentionally, to be sure.

Here’s what I mean.  Back when Microsoft first started being publicly identified as a significant instigator of regulatory and antitrust attention paid to Google, the company, via its chief competition counsel, Dave Heiner, defended its stance in large part on the following ground:

All of this is quite important because search is so central to how people navigate the Internet, and because advertising is the main monetization mechanism for a wide range of Web sites and Web services. Both search and online advertising are increasingly controlled by a single firm, Google. That can be a problem because Google’s business is helped along by significant network effects (just like the PC operating system business). Search engine algorithms “learn” by observing how users interact with search results. Google’s algorithms learn less common search terms better than others because many more people are conducting searches on these terms on Google.

These and other network effects make it hard for competing search engines to catch up. Microsoft’s well-received Bing search engine is addressing this challenge by offering innovations in areas that are less dependent on volume. But Bing needs to gain volume too, in order to increase the relevance of search results for less common search terms. That is why Microsoft and Yahoo! are combining their search volumes. And that is why we are concerned about Google business practices that tend to lock in publishers and advertisers and make it harder for Microsoft to gain search volume. (emphasis added).

Claims of “network effects” “increasing returns to scale” and the absence of “minimum viable scale” for competitors run rampant (and unsupported) in the various cases against Google.  The TradeComet complaint, for example, claims that

[t]he primary barrier to entry facing vertical search websites is the inability to draw enough search traffic to reach the critical mass necessary to become independently sustainable.

But now we discover (what we should have known all along) that “learning by doing” is not the only way to obtain the data necessary to generate relevant search results: “Learning by copying” works, as well.  And there’s nothing wrong with it–in fact, the very process of Schumpeterian creative destruction assumes imitation.

As Armen Alchian notes in describing his evolutionary process of competition,

Neither perfect knowledge of the past nor complete awareness of the current state of the arts gives sufficient foresight to indicate profitable action . . . [and] the pervasive effects of uncertainty prevent the ascertainment of actions which are supposed to be optimal in achieving profits.  Now the consequence of this is that modes of behavior replace optimum equilibrium conditions as guiding rules of action. First, wherever successful enterprises are observed, the elements common to these observable successes will be associated with success and copied by others in their pursuit of profits or success. “Nothing succeeds like success.”

So on the one hand, I find the hand wringing about Microsoft’s “copying” Google’s results to be completely misplaced–just as the pejorative connotations of “embrace and extend” deployed against Microsoft itself when it was the target of this sort of scrutiny were bogus.  But, at the same time, I see this dynamic essentially decimating Microsoft’s (and others’) claims that Google has an unassailable position because no competitor can ever hope to match its size, and thus its access to information essential to the quality of search results, particularly when it comes to so-called “long-tail” search terms.

Long-tail search terms are queries that are extremely rare and, thus, for which there is little user history (information about which results searchers found relevant and clicked on) to guide future search results.  As Ben Edelman writes in his blog post (linked above) on this issue (trotting out, even while implicitly undercutting, the “minimum viable scale” canard):

Of course the reality is that Google’s high market share means Google gets far more searches than any other search engine. And Google’s popularity gives it a real advantage: For an obscure search term that gets 100 searches per month at Google, Bing might get just five or 10. Also, for more popular terms, Google can slice its data into smaller groups — which results are most useful to people from Boston versus New York, which results are best during the day versus at night, and so forth. So Google is far better equipped to figure out what results users favor and to tailor its listings accordingly. Meanwhile, Microsoft needs additional data, such as Toolbar and Related Sites data, to attempt to improve its results in a similar way.

But of course the “additional data” that Microsoft has access to here is, to a large extent, the same data that Google has.  Although Danny Sullivan’s follow up story (also linked above) suggests that Bing doesn’t do all it could to make use of Google’s data (for example, Bing does not, it seems, copy Google search results wholesale, nor does it use user behavior as extensively as it could (by, for example, seeing searches in Google and then logging the next page visited, which would give Bing a pretty good idea which sites in Google’s results users found most relevant)), it doesn’t change the fundamental fact that Microsoft and other search engines can overcome a significant amount of the so-called barrier to entry afforded by Google’s impressive scale by simply imitating much of what Google does (and, one hopes, also innovating enough to offer something better).

Perhaps Google is “better equipped to figure out what users favor.”  But it seems to me that only a trivial amount of this advantage is plausibly attributable to Google’s scale instead of its engineering and innovation.  The fact that Microsoft can (because of its own impressive scale in various markets) and does take advantage of accessible data to benefit indirectly from Google’s own prowess in search is a testament to the irrelevance of these unfortunately-pervasive scale and network effect arguments.