AALS Section on Antitrust and Economic Regulation

Google and Antitrust

Moderator:  Michael A. Carrier, Rutgers School of Law - Camden

Mark R. Patterson, Fordham University School of Law

Geoff Manne, Executive Director of the International Center for Law & Economics (ICLE), Lecturer, Lewis & Clark

Frank A. Pasquale, Seton Hall University School of Law

Marina L. Lao, Seton Hall University School of Law

Pamela Samuelson, University of California, Berkeley, School of Law

Samuelson: Google Books—antitrust issues remain interesting, though now on hold because of the rejection of the settlement.

Lao: Disclosed that most recent paper was funded by Google, though earlier work on Google Books wasn’t.  Believed that the rejected settlement was procompetitive.  Had it been approved, Google would have had a nonexclusive license to continue digitizing, and public would have had free previews of many books; Google could have sold blanket licenses, and colleges/public libraries would have gotten at least one free terminal.  Central objection: de facto monopoly over unclaimed works.  But that assumes unclaimed books is the relevant market; if not, then having de facto monopoly wouldn’t give Google monopoly power. 

Even assuming monopoly power, is that anticompetitive?  Foreclosure is the usual theory—DOJ’s theory is that close competitors wouldn’t be able to compete for the unclaimed books.  But monopoly supplier is better than no supplier, and without the settlement, there would by definition be no supplier.  What about effects on search engine?  Improvements to search algorithm would make Google a better competitor in search—ambitious but dubious claim.

Samuelson: disagrees.  Orphan books were to be priced in particular ways, 12 bins: ave. price was $8.  When you think that Amazon is selling commercially available books at $9.99—there could be many out-of-print books that do have commercial value that Google wanted to be able to exploit.  Economics of digital publishing are so different from print that there was value to be extracted.  When she talked to the lawyer for the Authors’ Guild, he said that they added books in the cloud because otherwise it would look like an antitrust violation.  Focusing just on out-of-commerce, unclaimed works makes it seem like they’re low-value.  The database was going to be full of books from major research libraries, dense with knowledge, valuable to have access to.  Also, the license was going to be to scan all books, make nondisplay uses, display up to 20%, classwide releases/safe harbors/negotiations/compulsory arbitration—none of which would be available to anyone else.  Would also resolve ebook rights controversy that right now makes it difficult for everyone else: publishers and authors both claim rights and you essentially need to deal with both.  Google would get enormous leverage to transform the market in troubling ways, especially given that no discounts could be given without Book Rights Registry’s permission, and some agreements not to compete.  Not necessarily a violation, but DOJ did us a favor in raising some of those issues.

Pasquale: Google recognizes that bottlenecks can abuse their power in net neutrality debates; same may apply to search engines.  Google’s competitors complain that Google should be evenhanded and not favor its own services in providing ten blue links—Daniel Crane debunks this reification/naturalization of the way Google was for a certain period.  Debunks hardcore neutrality applied to search engines as in fairsearch.org.  Stealth marketing and stealth deindexation still are important.  A lot of people don’t adequately disclose compensation—we should see more enforcement.  Google’s letter to the FTC explicitly assures they won’t demote entities in google.com just because they opt out of being indexed in Google Shopping.  Enforcement will be an interesting issue there.

SearchKing case: Google’s rankings were protected by the First Amendment.  He doesn’t believe the First Amendment precludes antitrust regulation.  First Amendment ideas of right to access can be extended to Google.

One of Bork’s last papers gave an imprimatur to Google.  Bork’s legacy is very powerful in producing this no-action letter from the FTC.

Lao: would have been wrong for any search engine to deliberately demote a site for refusing to buy ads or because a competitor was buying ads.  Search rankings aren’t objectively neutral, but reflect judgment about relative value of a page.  But she doesn’t think objectively neutral search is possible and thus can’t wrap her head around neutrality as a regulatory principle.  No theory of antitrust liability would work.  “Neutral” presumably means that objective principles must be used in ranking, without intentional demotion for anticompetitive purposes—other than that, who gets to pick the criteria and why are they correct?

Google values links from educational sources more heavily than from others—all else being equal, page with link from .edu will be ranked higher than identical page with link from .com.  As academics, we may agree, but will Fox News?  Except in extreme cases, it’s hard to say that a judgment is wrong.

Tackle specific problems as they arise: if there’s a deliberate change in algorithm to penalize buying ads from someone else/entering into a joint venture with a Google competitor, that would be exclusionary conduct without getting to neutrality in general.

Manne: Market power: it’s easy and commonsensical to talk about search market/search ad market, but doesn’t make a lot of economic sense from POV of advertisers as well as searchers.  Advertisers don’t care what the tech mechanism is—search, social media, billboard on bus—as long as they can reach people.  Defining market around tech has inherent flaw. 

Google’s actually at a disadvantage to potential competitors in serving up most relevant ads—Amazon now offers demand-side advertising, using its data to serve up ads on other sites, as Google does with NYT.  Facebook knows who your friends are; Amazon knows what you’ve bought and what you might well want to buy next; Google knows what you search for.  Whose data would someone who wants you to buy stuff want?  Risk of a separating equilibrium: people searching to buy stuff don’t start on Google; 60% start on Amazon.  Google may be the place people go for information, not products.  Advertisers have very little interest in advertising to people looking for info on the RBI of a baseball player.  FB has done a terrible job of monetizing, but when it develops FB search, it will be an enormously powerful source of info and a place advertisers want to be.  Arguable that Google isn’t a monopolist and lacks market power.  We have to assume FTC recognized the importance of competing technologies and didn’t blindly accept search market as market.

Patterson: dispute over whether Google is a true 2-sided market.  Credit card users want merchants and merchants want credit card users. Some people say that advertisers want searchers, but searchers don’t necessarily want advertisers. 

Suppose someone creates a great new market.  Google might or might not put it on the first page.  Google says it does panel tests with users on different search results.  But Google never has to test whether people like competitors’ sites better. 

Pasquale: Lao made the point that SearchKing says it’s good and Google disagrees; he’s willing to accept that it’s junky.  But he interviewed the founders of a site that had gotten awards in England, had been ranked very highly; they were cut off at the knees when their ad rates went from 5 pence to 5 pounds.  All of a sudden Google Product Search got a huge market share and everybody else got less.  In terms of market definition, sure there are lots of people for the advertising market overall, but search advertising has sufficient differential utility that it can be separated.  It could well be that Google Product Search is demonstrably better than anything else (since he searched for Big & Tall clothes, Google highlights Economy Plus seats).  But that belies Google’s mantra that competition is one click away.  If your competition model is clash of the titans battles to the death, that’s one thing—Microsoft is willing to lose $2 billion/year on Bing.  But if you’re thinking about the zillion little guys, that’s different: as if the electric company decided that it wanted a share of every refrigerator sold or it wouldn’t supply electricity to the manufacturing plant.

Credence goods: black box nature of search means we have to trust Google because we can’t evaluate it.

Manne: you can buy lots of consumer data from other sources.  Big companies have data; smaller companies can buy it. And if they can’t, that’s not a unique barrier to entry.

Pasquale: but the big issue is having the largest set of data.

Manne: if you don’t have the money to have as much data, how do you remedy that?  Antitrust wasn’t meant to correct that financial problem.

Carrier: antitrust isn’t well cut out to solve many of these problems, including privacy.

Lao: Essential facilities?  If Google lacks monopoly power, it can’t have an essential facility. Extremely difficult to define a market, but it can’t be as narrow as general search on user side or search advertising on advertising side.  Switching costs are low, anyway.  Google being better than the alternatives doesn’t make it essential.

Denial of access: usually this is clear-cut, but not so in the case of search. We don’t know what the alleged facility is.  Competitive websites are readily accessible to anyone using Google: if I type in the name of the website, I’ll go there.  It must be the case that the claim is the list is the essential facility, or even placing high on the list.  The implications of that, though, are breathtaking, and courts are unlikely to construe denial of access that broadly.

Feasibility of sharing the facility: essential facilities doctrine doesn’t require sharing if it’s impractical or would detract from monopolist’s use. How would that work in a search engine?  Intrusion on property right would require compensation; could a court order a search engine to change its business model?

If all Google has done is favor its own content over competitors’, no antitrust theory would prevent that.  Different story if it acted affirmatively to block content—targeted a website for entering into joint venture with a competitor, buying ads from competitor/refusing to buy ads from Google.  Very likely that those things are exclusionary; wouldn’t need essential facilities paradigm.  FTC doesn’t seem to have found evidence of this.

Samuelson: Another dimension of GBS is still salient.  One concern about corpus is that Google, especially if it settled, would entrench search dominance because Google would have 20 million books to improve its search/translation tools.  Yahoo and Bing wouldn’t have the same access, and  no one else can settle with the same class.  If Google’s fair use defense succeeds, then we don’t have that problem, though Google could have exclusive agreements with universities that could still be a problem.  What she saw in Judge Chin’s statement is the idea that if one search engine satisfies tail queries and one doesn’t you’ll always choose the former, and that the corpus helps with the tail queries; Chin may have wanted to nudge parties to settle with a license (compensated) to others to use that information as a tool to improve their own algorithms.  Though the essential facilities claim is problematic here, the DOJ probably would have been more happy with data licensing.  There’s more than one layer at work here.

Manne: to the extent the problem is the biasing of Google’s own content, procompetitive justifications are obvious. If indeed there’s a benefit to the content offered through vertical sources (Maps, Travel Search, even Books), then there’s a value to Google of being able to offer the effective answer.  Ironic that a lot of complaints are that Google does what it does too well—it has better data so it can offer more relevant results, faster, with better integration.  This is all to the benefit of consumers.  Google’s maps are not becoming crummier as MapQuest recedes; it’s trying to integrate its maps better into the rest of the services.  We’re too caught in the model of ten blue links, where you search and you get a link to someone else’s site. Given the tech we have now, that’s a terrible way to respond to informational queries, where you have to go and search that page.  To the extent Google can answer your question, that’s a huge benefit.

Pasquale: First, switching costs are enormous: you’ve been training Google for years and it knows enormous amounts about you. Start searching on Bing = they don’t have a vast database of relevant information that can personalize results.  Also, DuckDuckGo as alternative just made complaint to FTC that Google disadvantaged them on Android: consolidation matters.  You could have gotten the same story from Microsoft about switching.

Patterson: Google isn’t going to provide a crappy map, true, but MapQuest is a causal factor in that: without MapQuest, no good Google Maps.  Now hears that people are being told not to compete in areas in which Google has an interest, because it’s not worth it—that’s a problem.  Google as similar to credit rating agency: transparency as remedy?  Algorithms are infinitely malleable; can always justify one.  Real problem: Consumers think Google is something it no longer is.  Eric Goldman was surprised by a search whose results were all paid links.  Antitrust doesn’t have a way of dealing with the information content problem, truthfulness etc., as opposed to access.

Manne: didn’t think Microsoft was right either, but here there aren’t network effects.  Also no strong argument for exclusionary conduct/preclusion of access to important inputs.  Closest argument is that Google has legacy deals with websites that have some exclusionary aspects, though other search engines can operate on those sites; that’s a tiny fraction of their business—only exclusive deal is with AOL.

Samuelson: some of the same issues Google is dealing with in Authors’ Guild—indexing, providing snippets—which she thinks have a strong fair use argument are also at issue here.  Scraping related claims have never struck her as strong; but Google made a representation that, for objecting sites, it will not scrape in ways it’s done before.  (I think a better description is that it won’t display scraped data in ways it’s done before: in its specialized services; you no longer have to choose all or nothing with respect to no-index.)  Google could have pushed harder on that issue given the copyright status, but decided it could live with that concession.

Carrier: new patent obligations for licensing.  Successors will be bound by Google’s promises.