Tuesday, May 22, 2012
Search, ‘Essential Facilities,” and the Duty to Deal
Posted by Marina Lao (Seton Hall)
Do the essential facilities doctrine and the antitrust duty to deal provide an antitrust basis to prohibit search “bias,” as some have suggested? In antitrust discourse, search bias is generally understood to mean a search engine’s favoring its own content over that of a competitor in an adjacent market in the organic search results. It would, for example, involve Google automatically returning a Google map in response to a search for “McDonald’s,” suggesting that the user wants directions, instead of applying some sort of “neutral” standard to determine whether a Mapquest or Bing map should be displayed instead. In an article on “Search, Essential Facilities, and the Duty to Deal” that I hope to be able to post on ssrn shortly, I argue that these principles simply do not “fit” in the search context.
The desire to invoke the duty to deal and essential facility is understandable, since Section 2 of the Sherman Act requires not only possession of monopoly power but also exclusionary conduct. And, seeking the competitive advantages that flow to a firm from integration, alone, is not considered exercise of monopoly power in an antitrust sense. In other words, absent an antitrust duty to deal (whether or not involving an essential facility), Google’s favoring its own content would not constitute exclusionary conduct that could give rise to Section 2 liability. I focus my comments only on Google, though all three major search engines tend to give preference to their own content, because Bing and Yahoo! do not have sufficient share of the general search traffic to have any Section 2 exposure.
Even before Trinko expressed its disfavor of the essential facility doctrine, courts have only sparingly applied it to mandate access. Thus, “essentiality” is strictly construed. Alaskan Airlines and other cases have said that a facility controlled by a single firm will be deemed essential only if control “carries with it the power to eliminate competition in the downstream market.” This strict construction is appropriate since antitrust law generally expects firms to compete with their own resources, not a competitor’s. In my article, I discuss in more detail why Google’s search engine could not reasonably be considered critical to competitive viability in a vertical market. While being visible via a search engine’s organic results is obviously a very attractive way to reach potential customers, it is not essential. There are other ways to reach potential customers, though they may not be as good as being included in the search results, and certainly not free.
Another problematic issue is that there may not be any denial of access in the first place. Denial of access is usually a non-issue in most essential facility cases--the monopolist has either denied access or it hasn’t. But, in the situations where a Google rival is said to be “denied access” and foreclosed from competition, the competitor websites are, in fact, still readily accessible to anyone using various keywords on Google search. For example, enter “map sites” into the Google search box, and Mapquest will top the organic results returned by Google. It is just that, for certain queries, Google may list its own content prominently whereas a rival site might not be shown or is lower in the rankings. For example, search for “McDonald’s,” and a Google map will appear first in the search results, followed by a link to Mapquest. Thus, the implicit premise of the complaints about denial of access seems to be that the organic results list is the alleged essential facility, and that access requires nothing less than having access to high ranking for any search term that might reasonably direct traffic to the competitor site. Given how strictly “essentiality” is defined, there is no basis for construing denial of access this broadly.
This takes us to a related issue that is often overlooked, and that is the feasibility of sharing of the facility (or “nonrivalrousness”). The essential facility doctrine does not require a monopolist of even a clearly essential facility to “share” it with rivals if sharing would be impractical or would detract from the monopolist’s use. In the case of search, is sharing possible? It depends on which is the alleged essential facility. If it is the search engine, then of course it can be shared but, in that case, there is no denial of access. If the ranked results list is alleged essential facility, and being denied a coveted top ranking is the alleged denial of access, then the “facility” is not nonrivalrous—there is only one top rank and so forth. In that case, it would seem that Google would have no antitrust obligation to give up any top slot in the results to a competitor if it needs those slots for its own promotion. The law is quite clear that the monopolist does not have to step aside to let a competitor use an essential facility if the facility cannot accommodate both. And, if there is no obligation to do that, there would logically be no need to adopt some “neutral” standard to allocate the scarce resource.
These are but a few intractable conceptual problems with trying to fit the essential facility doctrine to search that I raise in my article. Finally, I do not mean to suggest that how Google runs its search engine can never give rise to antitrust liability. My point is merely that search bias, or a search engine’s favoring its own content, alone cannot provide the basis for section 2 liability. If there were evidence, for example, that Google specifically demoted a website for doing business with Google’s rival, such as for advertising on Bing or Yahoo!, that would go beyond a pure refusal to deal and would be more akin to Lorain Journal.