Recognizing Bottlenecks on the Net
posted by Frank Pasquale
Adam Raff’s editorial on “search neutrality” in the NYT today has already provoked critical commentary and schadenfreude. But I found the editorial both informative and compelling. Principles of search neutrality are bound to be more complex than the network nondiscrimination rules made notable in the net neutrality debate. But we must realize (as forward-thinking cyberlaw activists like Sherwin Siy do) that “bottlenecks” at any layer of the internet—physical, social, applications, or content—can be problematic.
Some commentators have suggested that it would be impossible to make search engines as “fair” to the sites they index as cable and telephone companies should be to the customers whose traffic they carry. However, some basic principles of transparency should guide both fields. Raff articulates serious worries about Google’s influence over spheres of internet activity adjacent to search:
With 71 percent of the United States search market (and 90 percent in Britain), Google’s dominance of both search and search advertising gives it overwhelming control. . . One way that Google exploits this control is by imposing covert “penalties” that can strike legitimate and useful Web sites, removing them entirely from its search results or placing them so far down the rankings that they will in all likelihood never be found. For three years, my company’s vertical search and price-comparison site, Foundem, was effectively “disappeared” from the Internet in this way.
Admittedly, Raff likely can’t prove conclusively that Google’s prioritization practices purposefully hurt any particular company, because they are so secretive. I would not be surprised if Google PR responded to this editorial by saying that entirely neutral, albeit private, ranking practices led to the obscurity of Foundem. (For an interesting discussion of that style of justification, see James Grimmelmann’s discussion of Google’s SearchKing litigation.) But that opacity is itself concerning.
Public interest groups have made some inroads in holding carriers accountable, but even they appear reluctant to take the next step to recognize the parallel power of a dominant search engine like Google. They will soon have no choice but to confront this dominance, given that the obstacles to holding Google accountable—trade secret protection for its ordering algorithms—will also interfere with network neutrality regulation. Like search engines, carriers face an information overload problem, as spam, viruses, and high-demand applications threaten to overwhelm their networks. They are likely to make key network management practices as confidential as search engine rankings, and trade secret protection has already been deployed in other technological settings to block critical review of questionable corporate behavior.
Dominant search engines and carriers are the critical infrastructure for contemporary culture and politics. As these dominant intermediaries have gained more information about their users, they have shrouded their own business practices in secrecy. Internet policy needs to address the resulting asymmetry of knowledge and power. I’m glad to see people like Raff bringing these concerns to a public forum. Consider, for instance, the problems he identifies here:
Another way that Google exploits its control is through preferential placement. With the introduction in 2007 of what it calls “universal search,” Google began promoting its own services at or near the top of its search results, bypassing the algorithms it uses to rank the services of others. Google now favors its own price-comparison results for product queries, its own map results for geographic queries, its own news results for topical queries, and its own YouTube results for video queries. . . . Without search neutrality rules to constrain Google’s competitive advantage, we may be heading toward a bleakly uniform world of Google Everything — Google Travel, Google Finance, Google Insurance, Google Real Estate, Google Telecoms and, of course, Google Books.
In my 2007 article Copyright in an Era of Information Overload, I presented Google as a company that could break the dominance of concentrated cultural industries, joining a chorus of cheerleaders for “disintermediation.” However, as Google becomes more of an online conglomerate, it may create problems in new areas similar to the ones it once helped solve. Consider the complexities caused by Google‘s ownership of YouTube. Does the fact that a company does business with Google lead Google to make it more salient in search results than a company that (ceteris paribus) does not? How well are YouTube‘s rivals doing in searches on Google for videos?
Again, Google may have perfectly legitimate reasons for ranking YouTube items on top. (For example, if it faces more lawsuits like this, it may want to prioritize its subsidiary’s results because it suspects that YouTube’s “objectionable content police” are more active than upstarts’ bowdlerizers.) But just as Google wants the carriers to be open about how they manage traffic, it should be transparent about exactly how its commercial relationships affect the ranking of its business partners and customers. Without such transparency, regulators will not be able to assess whether the company is engaged in stealth marketing, which can be a deceptive trade practice.
Some commentators may say that principles of search neutrality are impossible to specify. I agree with Greg Lastowka’s concerns about Raff’s own ideas about search neutrality. But if you want to see a more precise specification of what search neutrality might look like, just look at page 27 of my article comparing dominant search engines and carriers. Or the series of comparisons in the chart on this page. Or take a look at Dawn Nunziato’s discussion of Google News in her book, Virtual Freedom. Siva Vaidhyanathan also has interesting insights. Thankfully, I’m no longer a voice crying in the wilderness on this issue. As Viva Moffat shows, there is a wide range of opinion on regulating search.
PS: More discussion at TPM, Marketing Pilgrim, and Business Edge.
X-Posted: Balkinization.
December 28, 2009 at 5:46 pm
Posted in: Cyberlaw, Google & Search Engines, Technology
Print This Post
Responses (11)
Jon Garfunkel - December 28, 2009 at 6:45 pm
Frank– thanks for posting and adding the links above. Certainly Raff & Foundem have a PR battle to fight as well as a legal-theoretical one. And I’m a bit amazed that there’s no reference of either of them (or the term “search neutrality”) in the archives of Search Engine Land. So right now they’ve got a big uphill battle to convince the techies that search neutrality is as important as network neutrality. (One doesn’t have to call it astro-turfing, but Google has certainly helped propel the NN debate).
The Madisonian blog linked to an August 2009 assessment by an SEO consultancy on the Google vs. Foundem debate, and Foundem didn’t fare well then (their conclusion was that Foundem resembled a link directory in Google’s eyes).
The bottom line here is that Foundem op-ed could end up be a very weak case for search engine oversight. Anyone interested in search engine oversight might have to wait for a better case to come along.
Jon
James Grimmelmann - December 28, 2009 at 7:22 pm
I’ll come out and say it: Foundem is a near-worthless site, and the quality of its search results is poor. I can’t imagine any search, other than on “foundem” itself, on which I would want these jokers coming up at all prominently in my Google results. I want Google to be able to rank them poorly, and any regulation that would interfere with that process would make the Web that much less usable.
I’m intellectually sympathetic to the idea that concentrated online power is dangerous and that transparency is crucial. But every time someone actually accuses Google of unfair discrimination, it turns out to be yet another shady “competitor” that no well-informed searcher would want anything to do with. Some of them are grey-to-black SEOs who’ve been caught in the act; others are not-ready-for-prime-time search engines looking for a PR hit. Each such incident counts as evidence against search neutrality regulations, not for it.
Chris C - December 29, 2009 at 7:11 am
Foundem’s primary function is to spam search engines, clogging results with worthless links. Relegation to the depths of search results is what they deserve.
On another point raised by the op-ed, MapQuest was great a decade ago, but lack of innovation is the reason almost no one uses it anymore, not search placement.
Seth Finkelstein - December 29, 2009 at 9:24 am
James – there’s something of a paradox, in that the more legit sites which have had problems with Google, either eventually “settle” the dispute, or don’t scream loudly. But on a *relative* scale, if the FCC held hearing on Comcast, I’d say there’s more than enough similar evidence for it to hold hearings on Google (putting aside the jurisdictional issue). There’s plenty of people who had exactly the same reaction about “I want [Comcast] to be able to [traffic-manage P2P], and any regulation that would interfere with that process would make the Web that much less usable.”.
James Grimmelmann - December 30, 2009 at 11:08 pm
Which “legit” sites, Seth? What evidence? Let’s talk specifics, because Foundem and its ilk certainly don’t provide a principled basis for regulatory scrutiny.
Seth Finkelstein - December 31, 2009 at 4:42 am
Remember what I said (emphasis) – “But on a *RELATIVE* scale, if the FCC held hearing on Comcast, I’d say there’s more than enough SIMILAR evidence for it to hold hearings on Google (putting aside the jurisdictional issue).”
Before we talk specifics, I’d want to nail down what’s the standard. As I see it, all that seems to need to be shown to hold FCC hearings is some minimal conduct which has anti-competitive *implications*. It’s not necessary that the target has clean hands – for heaven’s sake, filesharing (glossing) certainly doesn’t! You don’t need any evidence that anti-competitive motives were the reason either, per the “traffic management” argument, only that it’s a plausible reason (because of course the company is always going to make some claim about quality).
Ver briefly, *IF* trying to handle practical conditions of saturating a network for ~99% copyright infringement is a principled basis for regulatory scrutiny (P2P/Comcast), *THEN* Foundem and SearchKing should be more than sufficient.
A top-heavy political blogopshere? « Democracy Distorted - January 1, 2010 at 5:23 am
[...] US scholars, such as Frank Pasquale, have looked into the possibility of regulating search engines to address the power held by Google [...]
James Grimmelmann - January 1, 2010 at 10:17 am
I’d like to add an additional criterion: it’s a good use of the FCC’s time and attention, given all the other things on its plate. And that brings into the balance: (a) how likely we are to think we’ll find a bad motive, (b) the scale of anti-competitive effects, (c) the importance of a neutrality principle, given where in the Internet stack we are, and (d) the potential of governmental action itself to suppress competition.
And with that in mind, I say: SearchKing? SearchKing? Having spent a fair amount of time looking through the SearchKing site and those of its clients at the time of the lawsuit, I’m glad that Google kept their SEO tactics from working. The general principle Google asserted — absolute First Amendment immunity — is dangerous and probably wrong, but on the facts of the case, Google was overwhelmingly in the right.
Seth Finkelstein - January 1, 2010 at 11:40 am
But that’s very much the sort of thing which said about throttling filesharing – that many people were glad ISPs were keeping copyright-infringing bandwidth-hogs from ruining network performance for everyone else, and there was nothing government action could do because it’s an intrinsic problem of resource allocation and a classic tragedy of the commons (worse, more like appropriation). And that given this very real and obvious technical issue, it was absurd to think there was an anticompetitive motive involved.
But there were hearing held nevertheless. This is what I mean by *RELATIVE* standard. If you add in a qualification of likelihood of bad motive, then the Comcast hearings wouldn’t reach the standard.
James Grimmelmann - January 2, 2010 at 11:31 am
Google has no plausible anti-competitive motive from impeding access to SearchKing or Foundem. The only people who think that Foundem poses any kind of competitive threat to Google are its own employees and the commentators misled by their overheated claims. Comcast, on the other hand, had at least a plausible potential motive — reduced demand for its own video offerings.
Moreover, the “bandwidth hog” argument depends on unproven empirical facts about usage and capacity — and could be fixed either through different pricing schemes or through increased capacity. There’s inherently only one #1 ranking for each search; that problem won’t go away.
And let’s add to this the fact that with Comcast, there is a baseline of “neutral” treatment and it was clear Comcast had deviated from it. We can debate whether this is the right normative or practical baseline, but it was well-established and reasonably easy to define. With search, there is no such baseline, and it’s far from clear that Google had even “acted” in a way that could start us asking questions. Your test would lead to FCC hearings for any search engine any time it returns a ranked list of results with more than one entry.
Seth Finkelstein - January 2, 2010 at 1:23 pm
“[Comcast] has no *plausible* anti-competitive motive from [throttling ~99% *coopyright infringing* filesharing]. The only people who think that [P2P] poses any kind of [legal!] competitive threat to [Comcast] are [pirate business] and the commentators misled by their overheated claims. [Google], on the other hand, had at least a plausible potential motive – reduced demand for its own [shopping] offerings.”
Regarding bandwidth, it’s a law of human nature that demand expands to consume supply. And it shouldn’t be an argument that filesharing programs try to saturate a network. That problem won’t go away.
“And let’s add to this the fact that with [Google], there is a baseline of “neutral” treatment and it was clear [Google] had deviated from it. (that’s what SearchKing was all about, remember?) Your test would lead to FCC hearings for any ISP any time it makes a network management decision (and in fact, such scrutiny is exactly what some people want!)”
Quite seriously, ISP’s have to make traffic management decisions, otherwise various bandwidth-exploiting schemes will degrade performance, and search engines have to make ranking managements decision, otherwise spammers will degrade results. While I wouldn’t claim they’re 100% identical, structurally, they seem similar enough to provide some uncomfortable comparisons.
Leave a Reply