“Search Neutrality” and Network Neutrality: Birds of a Very Different Feather - Comments by Marvin Ammori

Posted by Marvin Ammori, Center for Internet and Society at Stanford Law School

Last week, I spoke on an excellent panel with law professors Eugene Volokh, James Grimmelmann, Dawn Nunziato, and Frank Pasquale. We discussed remedies for “search bias” alleged in the Google antitrust inquiry.

“Search neutrality” came up as a usual proposed remedy, with Grimmelmann devoting his talk to critiquing the concept and Frank devoting part of his talk to defending it. One speaker suggested that search neutrality and network neutrality have a lot in common –and that people who support network neutrality should also support search neutrality.

But very few network neutrality proponents support search neutrality or have advocated for it. While some suggest cynical reasons for supporting the latter, there are actually enormous, principled distinctions between network neutrality and search neutrality. Despite the deliberate linguistic similarity, the concepts have about as much in common as George Washington and George Washington Carver.

The goal of this post is to explain some of those distinctions. The economics and policy considerations for network neutrality and “search neutrality” are very different. I think these distinctions help explain why network neutrality has had (and continues to have) enormous support in the consumer, civil liberties, tech, and user communities, while “search neutrality” has had (and should continue to have) such minimal support.

Definitions.

Network neutrality is a requirement imposed on ISPs, like cable and phone companies, forbidding them from blocking or discriminating against websites or software online.

“Search neutrality” would be a requirement that search engines (like Bing) cannot discriminate against sites that compete with the search engine or with another site owned by its parent company (for Bing, Microsoft). But the usual example is Google search.

Distinctions:

Here are several key distinctions, some of which are “distinctions in kind” and some “distinctions in degree.”

Market Problem:

Network neutrality is aiming to solve a problem that the market cannot solve because of the economics and regulation for broadband delivery. Search neutrality is not.

1. Terminating access monopoly. Search engines do not have an appreciable terminating access monopoly. ISPs do. In the broadband context, Netflix or Amazon cannot reach Luke Pelican as a consumer unless they go through his ISP (Comcast). Luke’s ISP has a monopoly on access terminating to him. Even though Luke has an initial choice among ISPs, he has chosen Comcast, is locked into a multi-month contract, and it would be costly to switch merely because of one site like Netflix or Amazon. By contrast, no search engine has a termination access monopoly over Luke regarding websites. Luke can simply punch in the URL for Bing, Yelp, MapQuest, or any other company. Those companies can advertise online and offline to let people know that they offer specialized search results. And while Luke may usually use Google, he can switch at any time, without breaking a contract or paying any fees. He can try DuckDuckGo, Bing, Yelp, or any other search tool.

2. Economic barriers to entry. The barriers to entry and competition are nearly insurmountable in public networks but not in search. Reflecting the huge costs of entering the market for networks, public networks (such as telephone lines and cable lines) were generally built with government-backed monopolies and guaranteed rates of return during eras of public utilities regulation. Today’s networks were not built in competitive environments with risk capital—their return on capital was guaranteed by government monopoly-regulation. To this day, many of these networks continue to receive subsidies through a billion dollar Universal Service Fund and other subsidy-programs such as accelerated depreciation, suggesting that many regions of the nation cannot sustain one carrier, let alone multiple carriers. As Harvard’s Susan Crawford has argued, it is unlikely that wireless networks—which also have huge barriers to entry in terms of available spectrum licenses—can compete with cable wireline networks, as demonstrated by the Verizon-cable deal. Search engines like Yahoo! and Google did not require such expenditures. Google was started by small groups of students working out of dorm rooms with limited resources who attracted risk-capital from investors. Risk-capital is available from a host of sources for technology investors. Even now, companies like Blekko and DuckDuckGo take on Google and Bing with risk-capital from venture capitalists and angel investors. Name a new cable or phone company building a network (other than Google itself in Kansas City).

3. Government-imposed entry barriers. To start a wireless network, you need licenses from the government. There is a finite number of such licenses. A recent episode with a hedge-fund backed company called LightSquared emphasized the difficulty of getting government permission to provide wireless broadband to compete with the existing giants of AT&T and Verizon. For wired networks, the government-imposed barriers include legal hurdles to access rights-of-way to dig up streets and to attach wires on utility poles and franchise requirements in localities--and the government inaction of not regulating the “special access” terms for backhaul for entrants even though the government regulates the access to utility poles that now benefit incumbents. When it comes to search, there are very few government barriers to entry limiting the potential for competition to enter and exit the market.

4. Switching costs. The market for ISPs has more friction than the market for search. Consumers face large costs if they were to switch from one ISP to another because they want to reach a site that is discriminated against. Consumers usually have to pay early termination fees to get out of long contracts. For search, users can just click to another search engine. The main switching costs in search concern changing the default search engine for the browser on your laptop or smartphone and getting used to the interface of another search engine. Neither of these costs is particularly high.

5. Public investment. Similarly, the market for broadband is already deeply affected by government investment, suggesting the right to condition subsidies. Some have argued that since the public largely paid for the public Internet networks of the cable and phone companies, access to the Internet should be provided on a nondiscriminatory basis for the public. Indeed, the stimulus funds for broadband had such strings attached—non-discrimination rules for those who accepted funds. The public does not subsidize search engines through tax dollars.

Digital Industrial Policy:

Not only is the market problem different for broadband and search, but most Americans (from what I can tell) want different things from broadband and from search, as a matter of “digital industrial policy.”

6. Interconnection and balkanization. One argument for network neutrality is ensuring “seamless” and complete interconnection between communications networks and the applications riding atop them. (Kevin Werbach’s Only Connect focuses on this argument, derived from 47 U.S.C. 251.) The Internet, technically speaking, is in fact a system of interconnected networks, hence the name Internet. If some ISPs block some sites, and others discriminate against other technologies, then the “Internet” would represent only partially interconnected networks at best. This would mean that the Internet in China would be different from the Internet in the US; and the Internet on Comcast would be different from the Internet on AT&T. This would impose far more costs on developers of applications and their users, and make users unable to access all the speech that is available on the Internet without discrimination. Network neutrality addresses that problem by ensuring all networks interconnect seamlessly, without discriminations that balkanize. Search and “search neutrality” do not implicate the same interconnection issues. There is no longstanding government policy of interconnection suggesting that search engines should “interconnect” (whatever that would mean here) with each other or specialized competitors. Nor is there a strong economic or democratic argument for such interconnection.

7. General purpose technology. As a matter of digital policy, many argue that the Internet should remain a “general-purpose technology” as such technologies have a disproportionate impact on creating economic growth. The Internet was not traditionally optimized for any specific purpose but for general purposes. Each web “application” literally applies the Internet’s general-purpose technology in a new way. This is perhaps the main reason that the Internet has supported such economic and democratic innovation. Without network neutrality, broadband providers could make their networks more specialized or optimized for specific purposes rather than for general purposes. As a matter of digital policy, many oppose that. Whatever one might think of search, nobody thinks of it as a general purpose technology that can support the nearly infinite number of applications that the Internet can support. Search, while extremely valuable and used for many purposes, is pretty much a single purpose technology.

8. Historical success. The protocols and business practices underlying the Internet have reflected principles of non-discrimination (now often called the broad end-to-end principle and expounded in the classic book on network neutrality). Meanwhile, the entire point of search from the very beginning has been to discriminate. The very purpose of a search engine is to sift through information and deliver the most relevant information to the user. By their nature search engines must discriminate, otherwise they provide little value to users. Communications networks have had nondiscrimination policies since 1910 for telephone lines and earlier on telegraph and postal networks. Sometimes people discuss this in terms of layers—US policy has been to regulate the physical layer with nondiscrimination rules but not the content layer, and it has generally worked. Indeed, it is because of nondiscrimination rules at the physical layer (e.g. network neutrality) that the upper layers can be unregulated and competitive (e.g., no need for search neutrality).

9. Definitional/conceptual issues. It is not clear what the “industrial policy” for search neutrality would be. Because search is based on discriminating among sites to choose and rank the first, second, and third results, “search neutrality” is something of an oxymoron that has almost a dozen definitions (as analyzed by James Grimmelmann and Eric Goldman) Meanwhile, network neutrality has been defined in principle, beginning with the AT&T-BellSouth merger and continuing through the FCC’s most recent orders.

In short, network neutrality and “search neutrality” address different problems and do so in different ways for different purposes.

Those who defend search neutrality cannot merely analogize to network neutrality—the analogy fails pretty decisively.

Each concept must stand or fall on its own arguments and merits, not on linguistic similarities. As these distinctions suggest, there are many reasons why network neutrality would have so much popular and expert support—and why “search neutrality” has so little.

Different Supporters and Opponents.

The godfather of “search neutrality” is Frank Pasquale, a beloved, polymathic law professor at Seton Hall (and a friend), who often steps into the lion’s den with antitrust economists to argue for broader conceptions of the good than mere, American-centered efficiency. He co-authored perhaps the leading article on search neutrality, a follow-up arguing for a government-funded search engine that might obviate the need for search neutrality, and has increasingly staked out a position focused on transparency arguments and disclosure rather than conduct. Opponents of search neutrality include Law School professor James Grimmelmann and Santa Clara Law School professor Eric Goldman who have argued, among other things, that “search neutrality” is incoherent, undefined, and self-contradicting. I don’t know for sure, but I am guessing all three support some version of network neutrality.

At the same time, no nonprofit advocacy organization that fought for network neutrality has spoken out in favor of “search neutrality,” to my knowledge. I am not sure which consumer groups, if any, advocate for search neutrality today. The “advocates” for search neutrality in Washington, DC, have at times seemed purely strategic and half-hearted. In the FCC’s network-neutrality proceeding, cable and phone companies that oppose network neutrality, such as AT&T and Time Warner Cable, cynically invoked “search neutrality” as a bogeyman to distract from the core debates in those proceedings. (See filings here, here, and here for examples.) Today, the leading corporate advocates for “search neutrality” understandably are a coalition of companies, called “FairSearch,” that see themselves as competing with Google are arguing for a “search neutrality” requirement to be imposed on Google. More specialized search companies in that coalition, such as Yelp, MapQuest, TripAdvisor, and Foundem (a product search site or, depending on your viewpoint, a spammy mirror site), have alleged discrimination either by reducing the ranking of the site or elevating the ranking of Google search products. The coalition also includes Microsoft, which owns rival search engine Bing, and is not necessarily alleging search discrimination against its own Bing products in Google Search. Moreover, studies suggest that Microsoft’s Bing and Google’s search both return their own affiliated sites with similar prominence within search results. So Microsoft’s argument on “search neutrality,” like those of cable and phone companies, comes off as somewhat more strategic than principled—since Bing does not profess to offer “neutrality” vis a vis specialized search providers.

Disclosure as usual: I am one of Google's many policy advisors on these issues but don't speak for them.]