Law in the Internet Society
MY REVISED ESSAY

It is strongly recommended that you include your outline in the body of your essay by using the outline as section titles. The headings below are there to remind you how section and subsection titles are formatted.

Google Inc has not cooked its search results to favor its own products and listings, Executive Chairman Eric Schmidt told a U.S. Senate hearing looking into whether the search giant abuses its power. Members of the Senate Judiciary Committee's antitrust panel said last September that Google had grown into a dominant and potentially anti-competitive force on the Internet. This hearing should come as no surprise to anyone who has been following Google’s ongoing squabbles with the FTC and the EC. Practically every player in the digital economy is gunning for Google these days with some accusing Google of operating a “black box” algorithm that lacks transparency or accountability. Others say Google stacks the deck against rival services, such as maps or shopping services, when it display its own affiliated sites or content prominently in search results.

“Search neutralists,” as they call themselves, articulate their argument against Google as follows: If search engines have become an undisputed gateway to Internet, and are now arguably as essential a component of its infrastructure as the underlying network itself, does that not create a basis on which to argue for algorithmic transparency? Given that Google, the overwhelmingly dominant search engine, can apparently assert full and undisclosed editorial control of what content you see and what you don’t, does it follow that this endangers the fundamental openness of the internet?

Disregarding transparency’s obvious problems with execution ((1) the more transparent the algorithm, the more vulnerable it is to gamesmanship by spammers or worse, the greater the chance of the algorithm being rendered useless; (2) if the algorithm is transparent to regulators, they are unlikely to adapt fast enough to spur innovations), the concept is only worthwhile to prosumers, not consumers, and it is vital to remember that antitrust law, at least in theory, is supposed to be about protecting consumers. All consumers see is the supposedly objective final results, not the intervention by the gatekeeper. Unless the search manipulation is drastic (i.e. no relevant result appears), corrupted results are an “unknown unknown” and so no one cares. People will continue to see the search as a credence good, whose value is difficult to determine even after consumption.

Transparency’s relation to prosumers: The prosumer campaigns for a system that allows a visitor to conduct any or all three types of a search task: develop information, compare options, and find where to execute transactions. Algorithmic opacity would not be ideal for the prosumer because the prosumer, the active, tech-savvy customer who gains information from digital media or online, and interprets and influences mass consumers in terms of lifestyle and brand choices, desires increased facility with the technology in order to maximize his ability to engage critically with it and collaborate with others. Collaboration or federation is value to the prosumer. Currently, web search engines like Google function as weak federation mechanisms either by bringing up relevant web pages for user queries or via directories of related sites. A federated architecture, however, would offer a single point of entry allowing users to employ specific applications optimized for their searches. To be clear, the emerging paradigm is based on the combination of a multi–domain query approach with the integration of heterogeneous data sources capable of scouring the deep Web.

Conceptualizing a prosumer- ideal search architecture, or as Professor Moglen puts it, “a system of federated search technology, in which we all do searching for one another in some fast and efficient manner” can prove difficult, however, for a number of reasons, not least of which is because there would need to be a revenue mechanism different from the “pay-per-click” method that we are accustomed to. Existing revenue sharing agreements between search engines and publishers, where each receives a fixed share of the profit, are no longer feasible. Consider Google’s model: once a user clicks on a sponsored link, the search engine receives the payment from the corresponding advertiser and gives part of this to the publisher. The payment ratio of the search engine is defined by a commercial contract, existing independently of the specific search. When it comes to federated search, however, the contracts between the publisher and the domain-specific search engines must account for the fact that each search engine plays a role in the generation of the search process. In order for there to be a disciplined way to estimate the search value of each domain-specific search engine, the monetization must be performed after the ranking. This would avoid issues of trust (often a search engine may want to bid strongly or decrease the bid on a query for purely economic reasons) around the domain-specific engines.

Google's Algorithmic Cat and Mouse Game: The Case against Greater Transparency

-- By FarayiMafoti - 21 Oct 2011

Google, Give us a Peak

Google is now the subject of antitrust scrutiny, having been accused of leveraging its dominance in the “"keyword targeted internet advertising" space (consider this attempt number 1 at defining the relevant market here) to favor its own services to the detriment of a competitive market for the purpose of marginalizing rival companies.

Depends on your definition of "scrutiny." This subject is much discussed by theorists in what used to be thought of as "law reviews," and the FTC engages in occasional throat-clearing. But it was apparent to all knowledgeable observers even before Christine Varney departed US DoJ? for Cravath that there will be no serious dust-up between this Administration and Google in during this or any second term. Similarly, the EC has no political stomach for, and precious little jurisdictional armament with which to maintain, such a prolonged confrontation.

I was initially motivated to write this paper after the senate antitrust committee threatened to subpoena Eric Schmidt as a means of forcing his appearance before the committee. The Senate, likely aware of (1) the reluctance on the part of companies to open up their executives to interrogation when there are federal investigations under way and (2) the serious legal consequences confronting the government if there is any indication that the probes are not the result of good faith independent judgment, still went ahead with such an aggressive tactic, which sparked my interest. By “scrutiny,” I am merely referring to the existence of the various probes that you have referred to without making any judgment about the likelihood of an actual case. Perhaps I’ll just get the ball rolling with a neutral reference to the September hearing as to avoid any confusion.

Many have argued that Google tweaks its algorithm in ways that push down certain sites in search results that compete for eyeballs, allegedly because these sites have unoriginal content. Disgruntled site owners have cited Google giving Yelp the “TripAdvisor treatment” in support of their accusations. Yelp CEO Jeremy Stoppelman, when questioned about Google Places, remarked: "Google’s position is that we can take ourselves out of its search index if we don’t want them to use our reviews on Places…. But that is not an option for us, and other sites like us – such as TripAdvisor? – as we get a large volume of our traffic via Google search…We just don’t get any value out of our reviews appearing on Google places and haven’t been given an option other than to remove ourselves from search, how to improve this situation."

These are two completely different questions being conflated. At the beginning of the paragraph you are discussing the supposed unfairness of Google's algorithmic rankings. Later you are reflecting on a completely different issue; whether a site operator has some reason to object when material that is searchable to Google spiders--and to the entire rest of the universe on the same terms--is re-aggregated by Google in contexts that bring no immediate monetary advantage to the site operator whose page was searched. Whatever analysis is appropriate to each of these questions (about which I suspect in both cases we might differ), there is no appropriate mode of analysis that begins by confusing them.

I’ll delete everything from “Disgruntled…” onwards as I am only concerned with the former issue.

The "Everyone Wins" Artifice

Google’s competitors have bolstered their campaign against Google’s black box algorithms using the pretext of “consumer welfare” – which is defined almost exclusively in terms of consumers reaping the benefits of a fair marketplace that produces real innovation presumably through merit-based competition.

This sentence doesn't make any actual sense. I'm not aware, at the moment, of any search engine operator proposing to establish service on the basis of transparently published ranking algorithms. If there were such an operator, it's argument in favor of transparency would not be to consumers, except in the general sense that all propaganda in a society where consumption represents the predominant share of GDP is conducted in the meaningless bullshit vocabulary of "consumer welfare." Slight analysis reveals that it's not the consumer of search results who cares whether the algorithm is transparent. The very people who think that free software is unimportant because the bulk of ordinary PC consumers don't use it should see at once why this argument is nonsense. Transparent search algorithms are valuable to prosumers of search: that is, to the creation of a system of federated search technology, in which we all do searching for one another in some fast and efficient manner we haven't devised yet, and no party in the world knows what all the rest of us are thinking about.

But such a system, in which we spread search out throughout the Net, deconcentrating it, would remove the power over advertising that is held by a centralized search engine that knows what lots of people are looking for right now. In other words, the real competition in search is not between parties centralizing search to control advertising, for each of whom ranking algorithms that enable auction markets are quite legitimately and necessarily hypervaluable trade secrets, but between the architecture of centralized search and the architecture of federated search. As I've already explained in class, centralized search architecture—which makes Google, Bing, and almost all the other known variants—has what at present seems an insurmountable lead. The one-way link design of the Web is apparently overwhelmingly in its favor. And no one has yet built a model of successful, general, instantaneous, Web-wide federated search, so there is not even an unviable competitor to the reigning architecture.

On a twenty-year time scale, however, that's not true. The Web is less than 7,000 days old now. When it is 15,000 days old, federated search will be the dominant architecture, and ranking algorithms will be transparent, just as operating system kernel implementations are becoming transparent, because they too will be free software.

That has important implications for the future of advertising, and the distribution of economic power, that are probably the big story here. But you can't find them if you don't look for them.

If search engines have become an undisputed gateway to the Internet, and are now arguably as essential a component of its infrastructure as the underlying network itself, does that not create a basis on which to argue for algorithmic transparency? In other words, if we accept that Google, the overwhelmingly dominant search engine, can assert full and undisclosed editorial control of what content you see and what you don’t, could you not deduce, as Foundem and other search neutralists have in the past, that this endangers the fundamental openness of the internet? I am simply trying to give Google’s critics a leg to stand on here even though I believe that public utility style regulation of a digital economy is pointless.

I agree with your comment about consumers. All users see is the supposedly objective final results, not the intervention by the gatekeeper. Unless the search manipulation is drastic (i.e. no relevant result appears), corrupted results are an “unknown unknown” and so no one cares. People will continue to see the search as a credence good, whose value is difficult to determine even after consumption.

Prosumers: I suppose one could argue that unless the search technology is bereft of individualized advertising, a federated search architecture, with its implications of democratizing the media towards participatory systems, would only spur the commodification of human creativity. The more users make use of advertisement-based free online platforms and the more time they spend online producing, and exchanging content, the higher the advertisement prices will rise and the higher the prosumer’s value will become as a commodity. The people who “google” data constitute a user commodity that is sold to advertisers; unlike with the radio and television, search users would be much more active, continually creating content, which is likewise commodified (exploited?). I do not see any other way to avert this problem without doing away with the advertisement space altogether a la Linux or Wikipedia, wherein the platform has use value and no exchange value. But then what would make federated search any different from a P2P? network?

An Overview of Online Search

Google's sponsored links are produced for businesses interested in advertising and willing to pay Google when users click on their ads. Advertisements are generated by the keywords a user enters into Google's search engine. The amount that Google charges for sponsored links is calculated according to a keyword auction conducted through Google's AdWords? platform. These auctions are automated based on a set of parameters specified by each advertiser, and they occur instantaneously each time a keyword is entered into Google's search engine. An advertiser who places a higher bid for a keyword will receive better placement of its advertisements when a user enters that keyword as part of his search. Additionally, Google employs an innovative quality metric that adjusts the placement and cost to the advertiser of sponsored links based on the links' relevance to the search query and the quality of the underlying webpage. The heart of the dominant theory of Sherman Act, Section 2 liability against Google relates to Google's use of quality scoring in influencing the outcome of its AdWords? auctions. The quality score employs advanced algorithmic technology to maximize the relevance of search results and thus the value of the search engine to users, the likelihood of revenue-producing impressions to advertisers, and revenue to Google. Allegations of anticompetitive conduct surrounding the quality score turn less on its existence--all major search engines use quality scores to improve the relevance of search results—and more on its arcane nature. The specific determinants of quality scores are kept hidden by design and Google has repeatedly justified this opacity with the obvious axioms: (a) No company wants to share its secret formulas with its competitors; (b) by making the ranking formulas too accessible, it would be easier for people to game the system.

There's no "dominant theory" of Google's antitrust liability because there's no viable theory. Not one serious party, public or private, has decided to make the vast bet required to test any theory, because there is no theory that demonstrates both harm and causation sufficient to invoke legal process by any party with a claim to propose.

The other problem here is that you aren't discussing ranking algorithms anymore, you're discussing advertising placement algorithms for sponsored links. These are not the same things, not the same algorithms, and not the same analytic issues with regard to transparency. You need to be clear from the beginning that you're not asking anything about search, the purpose of which is to find things in the web that the reader wants to see. You're asking about advertising placement, which is the algorithm matching stuff that wants to be seen to people who could be induced to click on it whether they actually want it or not. There is no analytic case whatever for transparency of the latter algorithms, except the general belief that people should be able to stufy the software they interact with, which is too weak a claim to overset the obvious operator interest in trade secrecy.

What a Transparent Google Might Look Like

Some argue that now is the time for Google to admit that its power in the ecosystem is so great that the company owes it to the rest of the ecosystem to become more transparent in how it ranks sites. One outcome of this arrangement would be the creation of an independent, transparent, quality scoring system. This system would simply rate pages and search results and publish a Klout-like score. The details of the report would be open for everyone to see and scrutinize. Popular dot-com era entrepreneur and blogger Jason Calacanis has also called for Google to provide a calendar of algorithm updates and the informing of the ecosystem of its plans.

But that's about search ranking. The case is weak, but you're not discussing it. Are you? The confusion on technical points is harmful even to basic coherence at this point.

The Case for Pandora’s Box

From an antitrust perspective, no business, even a monopolist (assuming that Google is one and Google has refused to deal in lieu of short-term profits), has an antitrust duty to reveal to competitors formulas that it uses to set prices (see http://newscenter.berkeley.edu/2011/06/07/digital-democracy/ - the quest for transparency on the internet in general is a red herring). Moreover, courts are skeptical to intervene on the basis of complaints about product design by rivals because of the presumption that such intervention will chill innovation.

From the User’s Perspective

The problem for Google is not transparency but rather making us feel that it is transparent. Google undoubtedly has some perverse incentives in the ranking process with their own content properties and that alone leaves us wanting for some transparency in the process, although Google owes us nothing. As end users, we are attuned to the belief that an open internet is a desirable internet. It is this openness however, that creates a spam economy and eventually, an unusable web. If anything, an opaque, ever changing algorithm allows Google to be one step ahead of the spammers, even if Google profits from some of the spam (i.e. sponsored links). I acknowledge that the logical outcome of this argument is dangerous, however: we essentially would have to rely on Google to make things more transparent, more open, and more independent (note: the user-controlled, transparent quality scoring system mentioned earlier would admittedly be appealing at this stage of the argument even though it would nibble away at Google’s ability to do as it sees fit with its essential intellectual property).

Here’s what I think the roadmap of my paper should be as to address both search manipulation and the advertisement space:

1. Critics argue that Google determines search relevancy or where certain things show up in search results, accusing Google of operating a “black box” that lacks transparency or accountability. If this accusation is true, what’s the remedy?

2. One possible solution is to erect an independent search commission that would monitor search results in the interest of search neutrality.

3. Antitrust law, however, is about protecting consumers. This proposed solution fails to realize that transparent search would only be valuable to prosumers, not consumers. at this point I will have decided to no longer entertain the critics’ ideas and moved on to a few implications of federated search with respect to advertising

Given that this template would change my entire discussion, I wanted to submit my comments first.


You are entitled to restrict access to your paper if you want to. But we all derive immense benefit from reading one another's work, and I hope you won't feel the need unless the subject matter is personal and its disclosure would be harmful or undesirable. To restrict access to your paper simply delete the "#" on the next line:

# * Set ALLOWTOPICVIEW = TWikiAdminGroup, FarayiMafoti

Note: TWiki has strict formatting rules. Make sure you preserve the three spaces, asterisk, and extra space at the beginning of that line. If you wish to give access to any other users simply add them to the comma separated list

Navigation

Webs Webs

r8 - 10 Dec 2011 - 18:44:16 - FarayiMafoti
This site is powered by the TWiki collaboration platform.
All material on this collaboration platform is the property of the contributing authors.
All material marked as authored by Eben Moglen is available under the license terms CC-BY-SA version 4.
Syndicate this site RSSATOM