New Media?

The Political Economy of Internet Search Engines

A paper presented to The Communication Technology Policy section 2004 Conference of the International Association of Media & Communications Researchers (IAMCR) Porto Alegre, Brazil, July 25-30

Elizabeth Van Couvering London School of Economics and Political Science Department of Media & Communications Houghton Street London WC2A 2AE e.j.van-couvering@lse.ac.uk

Key words: Internet, search engines, advertising, spam, mass media, political economy

Abstract
This paper presents a structural analysis of the Internet search engine market. Search engines are a core element of the Internet media system, topping the Nielsen NetRatings Top 10 list of online properties in every market measured and generating advertising revenues of billions of dollars each year. Using a search engine is the most popular online activity after reading email. A political economy approach is used to identify the relevant parties in the system, their ownership, their exchanges, and the constraints to which they are subjected. The market is shown to be similar to other media markets in that it is both highly concentrated and global in scope, and funded primarily by advertising. However, there are several important differences. First, although traditional media conglomerates operate online, they do not control the most powerful actors such as Google and Yahoo!. Second, the advertising market is structured differently. Advertising is normally linked to words entered as search terms by users, and this keyword-based advertising is syndicated to other search engines, large and small, primarily by Google and Yahoo!-owned Overture, increasing their influence. Third, the situation is complicated by new commercial actors called search engine optimisation companies who operate within a grey market often in opposition to search engine companies, to strengthen the rank of their clients in search engine results. Recent studies within the computer science discipline have shown search engine results to be systematically biased in favour of commercial sites, popular sites, and US-based sites. With no public service mandate and little regulation, the study raises the question of whether the current system of search provision online serves the public good.

Introduction
The political economy of communication focuses critically on what structural issues in mass media ownership, labour practices, professional ethics, and so on mean for products of those mass media and thus for society more generally. Within the new media field, analysis from a political economy perspective has been relatively lacking, with the exception of studies of the Internet infrastructure1. So is it the case that these factors no longer matter within the area of Internet content? Can we take at face value the idea that the Internet makes it easy for anyone with access to basic technology to have a voice in the new Information Society? It can be argued that the Internet is not a mass medium in the classical sense; that the thousands, or even millions of sites visible on the Web are not the result of an industrial production process, and that nor do they represent a common substrate of everyday life, in the way that Silverstone, for example, characterises television: Watching television and discussing television and reading about television takes place on an hourly basis: the result of focused or unfocused, conscious or unconscious attention. Television accompanies us as we wake up, as we breakfast, as we have our tea and as we drink in bars. It comforts us when we are alone. It helps us sleepWe take television for granted in a similar way to how we take everyday life for granted (Silverstone, 1994: 3).

For example, the distribution of access and more recently skills and competences, also known as the digital divide has been one

area that has received considerable attention (see e.g., Lazarus & Mora, 2000; Mansell, 2001; Norris, 2001) 02 September 2004 E.J. Van Couvering New Media? A Political Economy of Search Engines Page 2

It can even be argued that the Internet does not conform to a basic definition of mass communication, such as that offered by Thompson, who says that mass communication is: the institutionalized production and generalized diffusion of symbolic goods via the transmission and storage of information/communication. (Thompson, 1990: 219). While information and communication are certainly transmitted and stored, can we really take seriously the argument that my friend Joes blog about his daily life represents either institutionalized production or generalised diffusion? Perhaps for this reason, research that approaches the study of Internet content from a political economy of communications perspective is rare. In this paper I suggest that in accepting the argument that some online content is small-scale craft production, scholars are neglecting the study of an important new medium of mass communication as an industry. It would be similarly easy to argue that desktop printing makes everyone a publisher, and thus that there is no need to focus on the economics of the newspaper industry; or that as video editing and handheld cameras come within the reach of many, Hollywood and television networks need receive no further attention. Such a line of argument would be rightly dismissed in those contexts, and a discussion of the Internet along those terms deserves a similar dismissal. In considering the Internet as a medium of mass communication, there is at least one clear set of large industrial players akin to the television networks or the Hollywood studios, and those are the search engines. Four large players dominate the search engine market: AOL, Yahoo!, Google, and Microsoft/MSN. Not only do these companies dominate the search-engine market, it can be argued that they dominate the Internet advertising market as a whole. In 2003, the Internet Advertising Bureau reported that the top 10 ad-selling companies accounted for 70% of US online advertising revenue of $7.3 billion (PricewaterhouseCoopers, 2004). The big four alone contributed $3.4 billion (see Table 1) although as this is total revenue United States revenue will be somewhat less2. If 75% of their advertising revenue is earned in the United States, a conservative estimate, then between them these four earn 32% of all advertising dollars spent online, not just those spent on search engines or portals3. It can even be suggested that the big four search engines dominate Internet content around the world. They come in the top 10 Nielsen//NetRatings listings in all of the markets that Nielsen surveys, apart from Japan (lacking Google) and Hong Kong (lacking Google and AOL). The figure below, derived from the Nielsen//Netratings data for the month of March 2004, shows the enormous reach of these sites globally:
2

Advertising revenues are not broken out by country in the financial reports. However, Google received 26% of its revenue overall

from international sources and Yahoo! 17%. MSN international revenues are not reported separately from Microsoft overall, nor are AOLs international revenues.
3

Although long-term figures are not available, in the week of May 3 to May 9, 2004, Portals and search engines were responsible for

39% of all pages with advertisement viewed by the online public (Nielsen//NetRatings, 2004). 02 September 2004 E.J. Van Couvering New Media? A Political Economy of Search Engines Page 3

Microsoft 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

Google

Yahoo!

Time Warner

Figure 1: The big four search engines, global reach %, March 2004 Data source: Home/Work Panels from Nielsen//NetRatings (*Home panel only, **MSN only, not all Microsoft)

It is easy to see from the graph above how this industry, similarly to the Hollywood studios and the television networks, is concentrated in the hands of a few large players. As further sections of this paper will show, it has other similarities, for example a heavy reliance on advertising revenue and a preponderance of American actors. This paper presents a preliminary investigation of the search engine market, in terms of its ownership, its revenues, its structure, the products it sells, its geographic spread, and the policies and regulations which govern it. A related theme is a critical evaluation of the ways in which the search engine market distinguishes itself from previous mass media. The paper is structured as follows. First, it considers the general economic structure of the search engine market in terms of business models and sources of revenue; second, it describes and compares the advertising products sold to traditional media products; third, it discusses the phenomena of search engine optimisation and search engine spamming; and fourth, it considers the infrastructure currently in place to regulate search engine content.

The Public Sphere and the Public Good

In a 1997 article, Roland Bettig talks of an enclosure of the electronic commons. He explicitly links historical processes, which deprived the common worker of the land that they cultivated and changed them from peasants to wage labourers, to current processes of development of the Internet. He cites three consistent structural tendencies which shape the development of communication technologies. These are first, marketplace concentration; second, expanding intellectual property rights; and third, the growing commercialisation of
02 September 2004 E.J. Van Couvering New Media? A Political Economy of Search Engines Page 4

G er m an H y on g Ko ng * Ja pa n* N /** et he rla nd s* Sp ai n* Sw ed en Sw itz er U la ni nd te * d Ki ng do U ni m te d St at es *

Au

Fr an ce

st

ra

lia

information and cultural output, including the introduction and increase of online advertising (Bettig, 1997: 139-140). These, he argues, are combining to consistently undermine the liberatory potential that may be inherent in the technology (Bettig, 1997: 54). By taking this line of argument, Bettig is working within the tradition of political economy in rejecting the notion that the market alone should be the arbiter of the structure of the media industry, as might be appropriate for other types of products. Instead, he is operating with an assumption that the mass media are sites of public interest and public discussion, or, in short, of the public sphere which Habermas (1992) details as an essential element of a rational and democratic government.4 Many other less radical treatments have shown how the Internet is affecting the processes that govern normal news media increasing the need for speed, altering the balance of funding, changing the way journalists report news, adding alternative viewpoints (refs). Again, the Internet is seen within the paradigm of the media as public sphere, rather than the media as market. The political economy perspective, however, is that the medias effectiveness in fulfilling its role in the public sphere is intimately connected to how media industries operate. The political economy of communication has focused on the study of the historical, political, economic, and material constituents of the communications industry within capitalist societies, whilst viewing them at the same time as integral to fundamental economic, political, social, and cultural processes in society (Mosco, 1996: 71). The study of the mutual constitution of communication and late capitalist society has become more and more important as we enter an age where speed appears to bring places closer together, drawing us into globalization and a space of flows (Castells, 1996). Processes of communication, particularly electronic communication, are fundamental to the current transformation of capitalism; and this is reflected in names such as The Information Society or The Knowledge Society. A political economy of communication, therefore, challenges the pure economic logic of this transformation and enables us to enquire, critically, about the relation of that transformation to social class, gender, race, and place; and more fundamentally, as Golding and Murdock stress, about justice, equity and the public good (2000: 73). Two common metaphors for search engines are the library catalogue or the Yellow Pages index of phone numbers. In other words, search engines, viewed from a nave perspective, bring us objective results from the whole Web based on our query. The reality is that they are shaped at every stage by commercial processes. The significance of this paper derives from a concern with

This is not to say that the media as market perspective is opposite to the media as public sphere perspective. Rather, it is argued

that by virtue of their function in the public sphere, the media should not be treated solely as a market. 02 September 2004 E.J. Van Couvering New Media? A Political Economy of Search Engines Page 5

what that might mean for those of us concerned with constructing an information society that is equally accessible to all.

Search Engines as Media of Mass Communication

Taking to Thompsons definition of mass communication as the institutionalized production and generalized diffusion of symbolic goods via the transmission and storage of information/communication (Thompson, 1990: 219), we have on the one hand, symbolic goods texts, images, sounds, etc. and on the other hand commercial or institutional production, transmission, and storage. This section focuses specifically on the institutional production and diffusion of search engine results, describing and contrasting them with similar processes in other forms of mass media. Thus, in this section and more generally within this paper, we are comparing search engines to other forms of mass communication. As noted in the introduction, it is possible that not everyone will agree with this characterisation. Because media markets are often characterised by their technology of distribution radio, television, the press, cable television it is difficult to say whether or not a new technology is also a new medium. Some have argued that the Internet in general is a medium by implicitly including it in general studies of mass media (for example, McQuail, 2000). However, many media scholars, while acknowledging this possibility, have specifically excluded it from the body of their recent theorizing (for example, Couldry, 2003). Others have noted that a major feature is the commercial exchange between businesses (Mansell & Steinmuller, 2000) and a major strand of research has argued for its function as a means of self-expression and a new form of community. Therefore, this study takes the view, strongly suggested by Miller and Slater (2000), among others, that the Internet is too wide an analytical category to be of practical use for in-depth study. Rather, the Internet is a technological infrastructure that affords a range of uses. This paper therefore makes no argument for the Internet as a whole being a mass medium. Instead, it concentrates primarily on search engines, and argues for their place alongside other mass media, while noting that core differences do exist.

Business models
Economically speaking, it is widely recognized that mass media in general operate in what is known as a dual product market. They produce, on the one hand, cultural commodities for individuals. These products are normally quite tangible books, newspapers, films, magazines. They also produce a second product for their advertising customers. This product is the audience (see Smythe, 1995). The percentage of revenue from each product class is different for different types of media vehicle (Croteau & Hoynes, 2001). Books, for example, derive almost all their revenue from individual consumers, while advertising makes up an average of 80% of a newspapers total income, a substantial fraction of which comes from classified advertising
02 September 2004 E.J. Van Couvering New Media? A Political Economy of Search Engines Page 6

(Sparks, 2000), although this differs from country to country as newspaper subscription is more common in some areas than others. Commercial television derives all of its revenue indirectly through advertising and sponsorship while public television, on the other hand, has the state as its primary support. How do search engines fare when compared to these other more well-known media enterprises? The answer is complex. Table 1 shows the percentage of revenue from advertising for Google, Yahoo!, AOL, MSN, AskJeeves, Lycos, Lycos Europe, and LookSmart5 in 2003, derived from their annual financial reports6.
2003 Total Revenue (Terra Lycos 2002) $1,625,097,000 $961,874,000 $8,600,000,000 $1,953,000,000 621,971,000 $156,229,000 $107,292,000 2003 Advertising Revenue $1,326,905,0007 $916,603,000 $767,000,000 $563,000,000 267,448,000 $140,886,000 $102,767,000 % from Advertising 82% 95% 9% 29% 43% 90% 96%

Yahoo! Google AOL/Netscape MSN Terra Lycos LookSmart AskJeeves

Table 1: Percentage of 2003 revenues due to advertising

This table shows a wide variation, between 95% for Google at number two in terms of advertising revenue and 9% for AOL/Netscape at number three. The explanation for this variation, I suggest, is as follows. The search engine business, as it has developed, has split into two somewhat overlapping camps. First, we have those who operate search services as their primary business (which includes Google, AskJeeves, and LookSmart). This is what I refer to as the search model. Second are those who also provide people access to the Internet via dial-up or broadband(which includes AOL/Netscape, MSN and Terra Lycos). This I refer to as the access model. Finally, Yahoo!, the largest company in terms of advertising revenues, operates on a middle ground of diversified services which include partnerships with access providers and direct-to-subscriber services such as premium email and web-hosting services8. Whether this points to the wider development of a mixed model is still unclear. This division has important implications for the structure of the market. Companies in the first category operating with the search model are heavily dependent upon the search services to bring

5 6 7

Revenue figures are unfortunately not available for Excite (privately held until March 2003, when it was acquired by AskJeeves). Google figures are derived from their S-1 filing for Initial Public Offering (IPO). Includes $1,199,733 in display ads, sponsorships, text-links, paid-performance, paid-includsion, search syndication and transaction Lycos Europe, a purely European provider which operates lycos.de and lycos.co.uk among other sites, shows a similar pattern,

revenue sharing. Also includes 127,172 in listings fees from jobs, autos, real estate and other small-ads services.
8

making 52%, or 43,360,000 of its83,674,000 revenues in 2003 from advertising. Lycos Europe depends heavily on its email and web-hosting businesses. 02 September 2004 E.J. Van Couvering New Media? A Political Economy of Search Engines Page 7

traffic9 and, with it, advertising revenue to their websites. Control of search technology is thus core to their businesses, and in recent years has meant that these companies have acquired a range of companies providing these technologies. To refresh their search capabilities against the threat of Google, AskJeeves purchased the well-regarded search engine Teoma, in 2002. Yahoo! purchased search provider Inktomi in 2003. It also purchased Overture (formerly GoTo) in 2002, an advertising provider which already owned AltaVista and FAST, two big names in search technology. Google and now Microsoft appear to be betting on their own internal technical capacity, although this may change. Businesses operating in the second category, with the access model, have found search less crucial and have been content, in recent years, to license their technology from providers operating with the search-based model10. This practice is known as white-labelling. Access-based models sell services to customers the connection via dial-up or broadband, email addresses, space for their own website and domain names, among other things w ith advertising as a secondary income stream. The archetypal company in this area is AOL, which developed as an online service provider in the early 1990s with a closed network, before the Internet was widely available. It still remains by far the largest earner. This dual-strategy industry has a parallel in cable television, in which cable channels such as Discovery or TNT book and earn money through advertising, while cable operators such as Comcast or Adelphia earn money through direct subscription (Carroll & Howard, 1998) to the network. Figure 2 shows the white-labelling of search results between the major search providers and distributors. Each circle in the figure represents a search-related website. Larger diameter circles are owners, smaller circles touching them are subsidiaries. AOL/Netscape and MSN are publicly traded on the New York Stock Exchange since they are owned by Time Warner and Microsoft, respectively. Yahoo!, AskJeeves, and LookSmart are traded on the NASDAQ exchange. Terra Lycos is traded on the Madrid exchange. At the time of writing, Google was in the process of its initial public offering, and will also be traded on NASDAQ. The thickness of the line around the circles indicates the volume of search traffic on the site. In 2003, Yahoo! terminated its contract for search results provision from Google, and MSN did the same from LookSmart, and is widely expected to drop Yahoo!-owned Inktomi when its MSN search is prepared.

Traffic is a somewhat technical term that refers to both page views and user clicks and is roughly equivalent to a term like In 2003 however MSN terminated its relationship with LookSmart and Microsoft has begun to develop its own web search

audience in other mass communication industries.

10

technology, possibly with the intention to integrate web-search capability into its next operating system (Hu, 2003). 02 September 2004 E.J. Van Couvering New Media? A Political Economy of Search Engines Page 8

Netscape

AOL

Open Directory

Lycos
HotBot

Look Smart

Google
Teoma

MSN

Ask Jeeves
Ex cite Inktomi Yahoo! All The Web

Alta Vista

Overture

Figure 2: Search site ownership and white-labelling of search results, April 2004

Figure 2 shows us that there are, at present, two major providers of search results (Google and Yahoo!) and possibly another in the wings (with Microsofts MSN Search). The minor providers (AskJeeves and LookSmart being the most prominent examples) are probably targets for acquisition. If these acquisitions were to occur, it would be continuing a trend of consolidation that began in the late 1990s and accelerated after the dot-com crash of 2000. In 1998, for example, in a highlycited article, Lawrence and Giles listed six major search engines upon which they based their calculation of the coverage of search engines on the Web. These were: AltaVista, Excite, HotBot, Northern Light, Infoseek, and Lycos. Historically, each of these engines was a separate company, operating separate technology. Today, none of these search engines is still an independent company (with the exception of Lycos), and none of them operates their own search technology. One factor encouraging this consolidation is certainly the scale that search engine providers need to attain to be competitive. Search provision is a capital-intensive industry. The sheer size and scale of the Internet have meant that companies that want to compete in searching the Web in real-time must invest hugely in hardware, software, and connection capacity. One senior program manager at a major search provider who has had a long career in the industry reflected on this: The scale problem is still there, although a bunch of people say its not a big deal any more. Theyre kind of right, and theyre kind of wrong. Its not a big deal if youve got a bunch of money. People have figured out how to solve the scale problem, which is to throw a bunch of machines at it and partition
02 September 2004 E.J. Van Couvering New Media? A Political Economy of Search Engines Page 9

your indexYou can still scale effectively. But now, it costs you a ton of money. This is why ever since 2000, 2001, most of the search research done at the universities is what I call Metacrawler-esque, which is people not building a search engine but doing something on top of a search engine, because they just cant afford to build their own. Which is a shame, because youre not getting these big engines coming out of academia any more.11 Which set of players is more powerful in the market? The answer is very dependent upon traffic, the commodity that fuels the online advertising industry. Large access-based search distributors, such as AOL or Terra Lycos, generate their own traffic, primarily through the use of default home pages for their access subscribers, and by providing other services such as email or chat which ensures regular visits from subscribers. Other distributors, like traditional and local media outlets such as CNN, gain traffic from huge offline promotions on their other media products. This traffic gives large distributors a strong bargaining chip with the search providers, for whom traffic is the major commodity. Search engine providers give such large traffic sources major reductions on the purchase of such services, in exchange for being able to advertise directly to the distributors customers. In the words of the managing director of one large search site: [W]hats happened is that people are now selling search results tied to a deal on paid-for placement, so youll get the paid-for placement quite clearly linked, indicated its a paid-for placement, its not part of the search results, hidden away, its clearly separated, but they will sell the two as a package, so as a distributor of search one can make quite a lot of money out of it, one wont necessarily be paying for the search or paying a very subsidised rate, plus one will be paid for the paid-for search. This dynamic reimbursement by the large search providers for advertisements on a distributors site is extremely important when in comes to evaluating the relative positions of distributors and providers. Advertising run by search providers can be a major source of income for both small and large distributors, a factor which helps to consolidate the influence of the two major search providers on the market as a whole, as we shall see below. In this section we have seen that the search engine industry is dominated by large companies with one of two complementary business models: either a search model, dependent upon advertising for revenue; or an access model, dependent upon the sale of services, particularly Internet access, to consumers. Companies that operate with the search model are search providers and own their own technology. These companies sell results and advertisements to those operating with the access model, who are search distributors. Other companies are also search distributors, most notably other online media companies such as CNN or the BBC.

11

In addition to documentary analysis, this paper is based on data collected from seven interviews with engineers and others at search

providers and search distributors, including Yahoo!, MSN, AskJeeves, Lycos Europe and Nutch as well as former employees of Excite and Webcrawler. These interviews are part of an ongoing research project. The author would like to thank the interviewees for their kind participation in the research. 02 September 2004 E.J. Van Couvering New Media? A Political Economy of Search Engines Page 10

Advertising in the search engine context

From a discussion of the overall industry structure and business models, we now turn to advertising in the search engine context. Advertising, of course, is characteristic of media business models in general. Does search engine advertising differ from other advertising, and if so, how? Both search providers and search distributors have advertising as a major source of revenue, as we saw in Table 1. This section focuses on how advertising is displayed and delivered in the search engine context. In particular, we note the co-existence of more traditional advertising products, similar to those found in offline newspapers and magazines, and a wholly new kind of search-linked advertising that has no direct analogue in other media. This second type of advertising (search-related advertising) is controlled primarily by search providers such as Google and Yahoo! and gives them further influence on the overall online media market. The information in this section was developed from a review of financial statements of Google, Yahoo!, AskJeeves, Terra Lycos, Microsoft, and AOL/Netscape.

Traditional advertising products

Search engines, in common with other online media properties, offer a range of what we will call traditional advertising opportunities to their clients. The term traditional refers to the fact that they have close analogues in other media, particularly press media. The three traditional types of advertisements are display ads, sponsorships and listings. These are summarised in Table 2 below for easy reference.
Tactic Traditional advertising Display advertising Description Advertisements appearing in banners or buttons on any page of a search engine. This can be displayed either in a particular channel (e.g., finance) or on any page of the site as a run-of-site ad. A long-term advertisement on a page, often in a different location than normal display ads. Payment On a cost-per-thousand impressions (CPM), a cost-perclick (CPC) basis, or much more rarely, a cost-per-lead (CPL) or cost-per-sale (CPS) basis. Either on a time basis or on a revenue-share basis. For example, a search engine company might get a share of revenue for each sale it passes to an e-commerce sponsor. Per item, on a time-basis.

Sponsorship

Listings or classified advertisements

A small-ad in a directory, for example a lonely-hearts ad in a dating channel.

Table 2 : 'Traditional' search engine advertising products

See figures 3 and 4, below, for examples of each type of advertisement.

02 September 2004 E.J. Van Couvering

New Media? A Political Economy of Search Engines Page 11

Figure 3: Sponsorship and display advertising on Yahoo!. Note sponsorship logo on upper right hand and NASCAR display ad on right hand side. From http://sports.yahoo.com, last accessed 31 May 2004

Figure 4: Yahoo's classified advertising service, from http://classifieds.yahoo.com, last accessed 31 May 2004

The differences between these formats and press advertising formats are typically in four areas:

02 September 2004 E.J. Van Couvering

New Media? A Political Economy of Search Engines Page 12

1. Pricing models. Press ads are sold in terms of the circulation of the newspaper or magazine and the size, position, and colour of the advertisement. Online ads are often sold in a similar fashion, depending upon the number of times the advertisement is viewed (analagous to circulation figures). This is referred to as cost-per-thousand or CPM. But they can also be sold based on the results of the of the ad12, on what is called a cost-per-click (CPC) basis, or even cost-per-lead (CPL) or cost-per-sale (CPS) basis. Some deals, particularly sponsorship agreements, can also include as payment a share of revenue of sales originating at the search engine. 2. Format. Colour and size are often the only variables in press advertisements, whereas online advertisements may also vary in terms of the technology they use: they can be static pictures, dynamic animations, or even respond interactively to user actions (rich media ads). Figure 6, below, shows a page including an interactive advertisement from travel provider Orbitz. 3. Placement on a page. Like press ads, online ads are charged differently depending upon their placement on a page. Online advertisements can also open in a separate window either above or below the window the user is currently viewing, or when the mouse passes over a particular image; this type of placement also affects the rates charged. 4. Targeting. A press advertisement may run on a particular page, for example the front page or the sports page. This is also true of advertisements on websites. Website advertisements though may also run randomly, based on automated rotations through various pages (known as run-of-site). Despite these differences, all three of these types of advertising the display ad, the sponsorship, and the classified listing have analogues in other mass media, particularly newspapers and magazine. They also characterise other online media sites such as the online version of newspapers and magazines. In this respect, therefore, there is nothing to distance search engines from traditional mass media vehicles, and much that suggests they are closely related. Interestingly in terms of advertising they seem much more closely related to print media than to electronic media such as television and radio, whose ads are typically limited in format and sold by timeslot.

Search-specific advertising products

While some advertising on search engine sites is similar to advertising in other mass media, particularly print media, a large and growing category of search-specific advertising is not. In this

12

Determined by placing a cookie on the users browser when the ad is viewed, and checking for that cookie when a purchase is

completed. 02 September 2004 E.J. Van Couvering New Media? A Political Economy of Search Engines Page 13

section we focus on search-specific ads, sometimes called keyword search advertising. According to the Internet Advertising Bureau (IAB), this type of advertising is currently driving the market. In the 2003 yearly review conducted by PricewaterhouseCoopers for the IAB, keyword search revenues rose from 15% of the total US online advertising market in 2002 to 35% in 2003, while display ads dropped correspondingly from 29% to 21% (PricewaterhouseCoopers, 2004). This section will discuss the mechanics of search specific advertising and reflect on the implications for the industry. Essentially, there are two types of search-specific advertisements, which are summarised in Table 3:
Search-specific advertising Paid performance An advertisement linked to a search term for example, the sponsored result for Expedia that might appear when you type travel. Paid inclusion A fee paid to the search engine to include the site in the search index. Table 3 : Search-specific advertising formats

Run on an auction basis to determine cost-per-click. Per item.

In contrast to most online media, search engines can offer advertisers more than a generic traffic stream. Because users of a search engine site are actively looking for certain information, search engines are able with confidence to sell specific ads to advertisers. In this respect, searchspecific ads are reminiscent of ads placed in the Yellow Pages. By linking the search query term that the user types in with the advertisements that are displayed, search engines have found a way to help advertisers break through the clutter of the mass market and target precisely who they are looking for. This type of advertising is known as paid performance. Paid-performance ads caused a huge controversy when they were introduced (Rogers, 2000), because they called sharply into question the assumption of search engine objectivity by including paid results in the same list as free results. Paid results are now shown on every major search engine results page, although they have become more visible and less controversial due to labelling by search engine providers they are now usually, though not always, indicated by a term such as sponsored results. See Figures 5 and 6 for an example of paid-performance ads from Google and AskJeeves.

02 September 2004 E.J. Van Couvering

New Media? A Political Economy of Search Engines Page 14

Figure 5: Paid performance ads on Google (indicated by sponsored link ) are included at the top and right hand of the page. From http://www.google.com, last accessed 1 July 2004.

Figure 6 : Paid performance ads from Ask Jeeves, indicated by sponsored content on the Orbitz ad and 'featured sponsor' and 'sponsored results on the two bottom sections. From www.ask.com, last accessed 1 July 2004.

Paid-performance ads are typically sold not on mass-market cost-per-impression basis, but on a cost-per-click basis. This charging method is unique to Internet advertising. The cost for each click on a paid-performance ad is determined based on an automated auction system, which adjusts the price upwards depending upon how many advertisers are bidding on the particular search term. Thus a term like flights might be much more expensive than a more explicit and less common term such as flights to New York. One senior manager of a search distributor reflected:
02 September 2004 E.J. Van Couvering New Media? A Political Economy of Search Engines Page 15

Everybody talks about paid-for placements and Overture and Google having pioneered this sort of paid-for search result, which frankly [is] just not true. [W]e were doing it back in 1997. We called them kitty links. We would have three kitty links for any search result as well as a banner. And the kitty links and the banner were paid-for placements, I mean thats exactly what it was. So, everybody was doing that it wasnt invented by Overture and Google. What they invented was being much more focused at it. And distributing more widely so they got critical mass. Indeed, distribution of paid-performance ads has proven to be a critical part of the search providers business model. Since a paid-performance ad must be linked to a search keyword, the rise of paid-performance advertising as a major online advertising vehicle has served to strengthen the position of the search providers. At present, there are two major paidperformance advertising networks: AdWords, owned and operated by Google, and Overture, owned and operated by Yahoo! since October 2003. MSN currently receives its paidperformance advertising from Overture, as does Lycos. Google, on the other hand, provides paid-performance advertising to AOL and AskJeeves. Each operator also provides paidperformance advertisements to a range of smaller sites. In 2003, Google made approximately 26% of its revenue through paid-performance advertising operating on other sites. While Yahoo! does not disclose Overture revenue directly, in the financial year 2002, prior to acquisition, Overture reported $667.7 million in revenue. Assuming no growth in 2003 (a conservative assumption as second quarter growth was up 74% year-on-year from 2003 to 2002), paidperformance sales from Overture would account for slightly less than half of Yahoo!s entire advertising revenue of 1,327 million in 2003. The second type of search-specific advertising is paid inclusion advertising, which is invisible to the customer. Paid inclusion advertising requires a short technical explanation to understand. When search results are displayed on a page, they are generated by a computer program called a search algorithm. The search algorithm does not search the actual Web itself; instead, it searches a previously collected database of sites called the index. The index is generated by another computer program, called a spider which automatically follows links on the web and records details about the pages it finds. This is called crawling. Today, the web is so huge that no spider can crawl the whole thing. Thus, some sites may never be included in the index. Some search engines therefore offer a service that guarantees that the spider will crawl a site an advertiser submits, and include it in the index. Since this index entry looks exactly like an index entry from a normal spider crawl, and since search results are generated from it in the same way, it is invisible to the customer. It is not clear from the company reports how much revenue is generated from paid-inclusion advertising13.

13

Google does not accept paid-inclusion ads as a matter of policy. New Media? A Political Economy of Search Engines Page 16

02 September 2004 E.J. Van Couvering

In this section we reviewed the specific types of products that search engines offer to advertisers. We have found that in many respects they resemble offline media, particularly print media. However, we note that advertising based on search keywords, a different charging structure (costper-result), and different distribution structure in the shape of large advertising networks like AdWords and Overture represent a strengthening of the search providers position with respect to other online publishers and distributors. This type of syndicated advertising distribution network seems to be a unique feature of the search engine marketplace that has no direct parallel in other mass media, which tend to have localised advertising markets.

Extra-search activities
The last section discussed paid-performance and paid-inclusion advertising. However, some enterprising companies would rather be included in search engine results for free. This section discusses search engine optimisers and search engine spammers, both of which work on behalf of clients to increase their prominence in search engine rankings. In order to understand why search engine optimisation and spamming are viable business strategies, and why search engines in particular have spawned these industries, we turn to two related observations about how the Internet media business works. The first point to consider is that links within web pages have an economic value. We have already seen that advertisers will pay sites for clicks from an ad (whether display or paidperformance). This type of payment can extend to something known as a referral fee, where a share of product purchase price is passed to those who send a customer to an e-commerce site. The closest parallel to this is the coupon distributed by manufacturers in local papers, but online the phenomenon is much larger. We have seen that search engines themselves use the referral fee as an additional source of revenue, in all types of advertising including display, sponsorship, listing, and paid-performance14. Thus, sites on which users are likely to click a lot onto other sites that pay for a click or for a referral have a valuable source of income. The second observation is that search engines are the online media properties with the highest traffic, and therefore the most potential clicks. It is known that most searchers do not go beyond the first or second page of search results (Spink, Jansen, Wolfram, & Saracevic, 2002). Thus, the way to capture traffic from search engines directly is by attaining a high place in the rankings of a relevant search. From this it follows that placement within the results rankings assumes great importance economically. This is true whether the site is making money directly from commerce conducted on its website, or from advertisements and referral fees.

14

Unfortunately the amount of revenue gained in this fashion is not reported separately from other advertising revenue. New Media? A Political Economy of Search Engines Page 17

02 September 2004 E.J. Van Couvering

These two factors together have led to the birth of the search engine optimisation (SEO) and search engine spamming industries. Although they are often linked, we will characterise SEO and spamming separately. Simply put, they are the legitimate and illegitimate side of the same business. In the next section, we first characterise legitimate search-engine optimisation, and then move on to discuss search engine spam.

Search engine optimisation

When the publishers of a web site are seeking to attract more traffic to the site, where do they turn? Very often, they look to increase their ranking on major search engines for relevant queries. Amazon.com, for example, may wish to come top of the list for the search books. As companies who seek good press coverage may turn to a public relations agency, the online company may turn to a search engine optimiser (SEO). The SEOs function is in fact very similar to that of a public relations firm in a traditional media context. The business of the search engine optimizer is to structure and refine the web pages of their client so that they appear high in the results rankings for an appropriate search query. That is to say, rather than purchasing on a pay-per-click basis an obvious advertisement linked to keyword search, the client purchases, via the SEO, a long-term placement in the main search listing, which they believe to be more stable and more valuable in the eyes of users, as recent marketing research has shown that people are increasingly ignoring some online advertising altogether (Drze & Husherr, 2003). The search engine optimization process is more risky and often more expensive than keyword advertising, but if done well it may result in a much greater flow of traffic. Many companies, of course, both optimise and advertise. As with public relations, revenues generated from this process do not flow to the search engines themselves, but to a range of small optimization companies. Also unlike the search provision business, but like public relations firms, search engine optimizers tend to be relatively small and localized, requiring know-how but little investment in technology compared to search provision. The SEO and public relations business do have one large difference, however. In the world of print media, the editor or publisher has the final say about what goes into the newspaper or magazine. A search engine, being an automated process, has no such direct say. Therefore, what comes at the top of the list may not be exactly what the search engine provider would like. This leads the discussion neatly into search engine spamming, the subject of the next section.

Search engine spam

If the legitimate actions of SEOs on the pages of their own clients may cause them to rise in the search engine listings, what about dubious or even fraudulent practices? For the purposes of this paper, search engine spam is defined as follows: any content on the Web that is intended specifically to deceive

02 September 2004 E.J. Van Couvering

New Media? A Political Economy of Search Engines Page 18

a search engine s indexing or ranking programs15. Search engine spam has a long history, and in fact has been one of the driving forces in search engine development. An engineer who was involved in creating one of the first search engines in 1993/1994 told me: Ourconcern was, well, you want to make sure that the pages you serve are high-qualityBecause before too long, you got people who tried to cheat the system in various ways. We called them word spammers. That actually took an extraordinary amount of time from us. For a small team that was damaging, you know, we couldnt work on the next generation search because we were dealing with those issues. In the early days, spammers used relatively simple techniques such as black text on a black background to fool search engines into ranking their pages more highly. But spam developed quickly. Another engineer who worked on a search engine in the mid-nineties told me that coping with spam, mostly detected by user complaints, was the day-to-day work of the engineering team: We definitely would see the spam get smarter. For example, we started seeing things that were obviously not readable but that were generated automatically: nonsense text that was designed to fool things that were trying to look for nonsense text. So they got very clever. So its a tough one, it is an arms race. When Google launched in 1997, it was said to be unspammable, largely because instead of using primarily page content as the basis for its ranking, it used the link relationships between different sites as a way to distinguish authoritative content (Brin & Page, 1998). This innovation, called PageRank, has come to dominate modern search engines. But Google and the rest of the search engines are as vulnerable to spam today as they were in 1994, for example by techniques such as the link farm in which a network of heavily cross-linked sites is set up in order to boost the rankings of all the sites included (Perkins, 2003). See Figure 7 for a screenshot of a sample link farm. These link-farm sites then sell paid-performance advertising related to the original search query.

15 This

definition follows Perkinss (2003) quite closely. He defines search engine spam as Any attempt to deceive a search engine's

relevancy algorithm. He also defines what spam is not: Anything that would still be done if search engines did not exist, or anything that a search engine has given written permission to do.

02 September 2004 E.J. Van Couvering

New Media? A Political Economy of Search Engines Page 19

Figure 7 : A sample link farm page. This page, a result from searching 'travel to Eritrea' on Google, has no content but rather a series of links and ads related to the initial search terms, which would be paid for on a cost-per-click basis. From http://travel.worlddirectory.net/dir/1/27.php, last accessed 1 July 2004.

The paradox that is clearly demonstrated by Figure 7 above is that while search engine relevancy teams battle against spam in the interests of bringing a high-quality result to the user, their own advertising formats paid-placement ads are at the heart of a this new generation of spam. Paid-performance ads provide easy revenue generation for spammers who may operate hundreds of link-farmed sites. As content-free, link-farmed sites increase their presence in the top 10 or 20 search results, user satisfaction with search engines is bound to decrease. Thus, search engine results are being subjected to a concerted assault in the name of free enterprise. This longstanding arms race, as discussed, consumes a great deal of the time and attention of the search engine providers, as they constantly tune their indexing and ranking algorithms to try to exclude the new tricks of the spammers. It may even hamper or prevent the development of alternatives. A case in point is the service Nutch (www.nutch.org). Nutch is an open-source search engine, founded specifically in response to concerns about bias in commercial search engines. A standard criticism of Nutch is that it will simply be swamped with spam, having done the spammers work for them. Here is the Nutch reply: Won't open source just make it easier for sites to manipulate rankings? Search engines work hard to construct ranking algorithms that are immune to manipulation. Search engine optimizers still manage to reverse-engineer the ranking algorithms used by search engines, and improve the ranking of their pages. For example, many sites use link farms to manipulate search engines' link-based ranking algorithms, and search engines retaliate by improving their link-based algorithms to neutralize the effect of link farms. With an open-source search engine, this will still happen, just out in the open. This is analagous to encryption and virus protection software. In the long term, making such algorithms open source makes them stronger, as more
02 September 2004 E.J. Van Couvering New Media? A Political Economy of Search Engines Page 20

people can examine the source code to find flaws and suggest improvements. Thus we believe that an open source search engine has the potential to better resist manipulation of its rankings. (The Nutch Organization, 2004) The issue remains unresolved while Nutch is still in a formative stage. This section has distinguished between legitimate SEO and illegitimate spamming practices. In practice, because the laws are weak and both customers and optimisers may be ignorant of the effects of particular practices, the distinction is less clear. Legitimate business owners can become spammers almost by accident. One engineer working on search engine relevance at a large provider related the following story to me: I grew up in a [western US city], and one of the things my family would do, is we would go down to Southern Utah to stay at a couple of places. One of them was called the Hey-di-hi Inn16Well, I ran across their site about a month or two ago, and just kind of looking at it because I thought it was kind of fun, it caught my eye, so I sat down to play with it. It turns out that theyre spamming. And I actually called the guy, and what I found out is that they, along with you know, a zillion other tourist places in the Four Corners Area Arizona, New Mexico, Utah, Colorado all of them are basically hosted by this companywhich was an outfit out southern Utah. And you know, this companys been around for ten years, which is pretty much as long as the Internets been around. So they sound reputable, theyve established good business relationships with all of these mom-and-pop motels, horse ride places, balloon ride places, everything, everything that needs reservations. What they provided was a generic reservations system. So, use our system, well do your reservations, so for all these places its just great. And in addition what this company has branched out to is various forms of link spamming and keyword spamming. Now, its not really hard-core, theyre not the worst of the bunch, but what theyre doing is pretty clearly spam. Its pretty obvious. And thats not a good thing to do. By the same token, even though our friend at the Hey-di-hi Inn is spamming, its not what Mr Smith whos owned the place for 30 years he doesnt know what hes doing. Its not a lot of spam. And as it turns out, heydihi.com is in fact the right URL for the Hey-di-hi Inn. So even though youre spamming, youre still the authoritative source. If you want to stay there, thats still where you want to go. The search engine is now faced with a quandary: what if a quality source, an authoritative source, is also a fraudulent, deceptive source? What does a good quality result consist of in this case? In this section on search engine optimisation and spam, we have highlighted a dynamic that is unique to search engines: search content and search advertising has, in essence, escaped from the control of search providers and is being put to use against them. It is as if the pages of the daily newspaper had risen up against the publisher. In this respect, at least, there is something new about new media.

16

Not the real name of the inn. New Media? A Political Economy of Search Engines Page 21

02 September 2004 E.J. Van Couvering

Regulation and Public Service

In the preceding sections, we have concentrated on the business models and operations of search engines. We have highlighted the growing concentration of the major search engine providers, and the ever-increasing activity of the SEOs and spammers. What, we may wonder, has been the public response to these questions? In this section, we examine the actions of governments to regulate the provision of search engine results. McQuail (1994: 171-173) draws our attention to the three contrasting models of regulation of different media systems: the free press model, the broadcasting model and the common carrier model, each derived from a different perception of how the idea of the public good can apply to media, which are recognised to have a public function in society in addition to a commercial function. The free press model aims to ensure diversity of content by preventing interference by the state in matters of content and distribution. The broadcast model aims at achieving high quality and representativeness through public regulation of content and distribution. The common carrier model aims to achieve universal participation, normally in a peer-to-peer medium such as the post or telephone. McQuail summarises their features as follows:
Free press None None Open Open Broadcasting High High Closed Open Common carrier High None Open Closed

Regulation of infrastructure Regulation of content Sender access Receiver access

Table 4: Three regulatory models compared (from McQuail, 1994: 173)

The Internet is an interesting mixture of regulatory models, reflecting the different origins of its infrastructure (coming from telecommunications) and its content (from existing commercial laws and existing laws regarding print media and intellectual property). Different elements of the Internet system are regulated differently within these regimes. For example, there has been an increasing tendency in both Europe and the US, for example, for Internet service to be considered in universal service provisions17 - in other words, linked to McQuails common carrier model. But in content, the free press model is prevalent: for example, in the US context, content online is constitutionally protected by freedom of speech, as any newspaper or magazine would be. Commercial law not included in McQuails scheme, is also important, for example in enforcing contracts and ensuring that copyright protects other types of proprietary content , such as music files, videos, or computer programs from duplication (Lessig, 1999). Trademark laws are increasingly used to protect domain names and web site designs (Par, 2002). Interesting,
17 The FCC, for example, while not directly regulating Internet access, supports schools and libraries in getting Internet access with

the Universal Service Fund. Again, while Internet access is not mentioned in the EUs Universal Service Directive (2002/22/EC), the European Commissions eEurope 2005 plan (2002/263/COM) has e-inclusion or providing access for everyone in order to combat social exclusion as a general aim.

02 September 2004 E.J. Van Couvering

New Media? A Political Economy of Search Engines Page 22

McQuails broadcast model, including elements of public-service programming, is noticeable by its absence. How do search engine websites fit into this tangle of regulation? As with much Internet regulation, this is a vexed question, because the issue of jurisdiction often arises. However, as the major search engines are based in the United States, we will concentrate here on United States law. Within the US, search engine results (as with other forms of Web content) are protected by the constitutional guarantee of freedom of speech. Thus search engines are entitled to publish whatever content they like, and the presumption is that the Internet user is free to read or not read the results as they choose, just as they would be with any magazine, newspaper, or book. The issues of paid inclusion and paid-performance advertising, however, have attracted some government notice. In 2002, in response to a complaint, the Federal Trace Commission (FTC) issued a letter to search engines regarding paid inclusion and paid performance from their Division of Advertising Practices. The letter said, in part: if your search engine uses paid placement, you make any changes to the presentation of your paid-ranking search results that would be necessary to clearly delineate them as such, whether they are segregated from, or inserted into, non-paid listings. Moreover, the staff recommends that if your search engine uses paid inclusion programs that may distort rankings or placement criteria, you clearly describe how sites are selected for inclusion in your indices. Also, consumers should be able to easily locate your explanation of the paid inclusion program you use, and discern the impact of paid inclusion in search results lists. (Federal Trade Commission, 2002a) The FTC, however, took no formal action in response to the complaint. Based on our observations, paid performance ads are indeed indicated, but paid inclusion remains obscure to the user. The FTCs letter was issued by its Division of Advertising Practices, which protects consumers from deceptive and unsubstantiated advertising, and in this case is acting to prevent advertising from being misleadingly perceived as editorial content. In this sense, the FTC is treating the search engine results very much as it would treat any other mass medium, in the interests of consumer protection. Search engine results, then, are quite lightly regulated by a similar regime to the one that regulates print media, with similar protections for the consumer, and without the guarantees for the citizen that are embodied in the public service model of broadcasting. The concern is with freedom of expression and freedom of trade, based on the view of many competing voices and a competitive marketplace. The next section brings the paper to a conclusion, bearing in mind our initial concern about whether the current system of search engine results provision serves the public interest.

02 September 2004 E.J. Van Couvering

New Media? A Political Economy of Search Engines Page 23

Conclusions
This paper has shown that search engine companies are the major players in the Internet media space, with international if not global reach. It has compared them to other media of mass communication, without arguing that the Internet as a whole is a new medium of mass communication. This mass communication industry, like others, is very highly concentrated. This paper has shown that there are four major distributors and only two major providers of search engine results. These two, Yahoo! and Google, supply both search engine results and advertisements to search engine distributors, large and small, through white-labelling activities and through the growth of their paid-placement advertising networks. While much search engine advertising is analogous to advertising in the print media, these advertising networks are a new development without parallel in other media forms. They syndicate advertising to a variety of large and small sites and pay only for results in the form of clicks or sales. Another unique feature of the industry is the relationship of the search provider to the owner of the content listed in the results. As a result of search engine optimisation and spamming practices, it is not only the search engine providers who determine the relevancy of the results. Instead, SEOs and spammers are engaged in an arms race with search engine providers, a hard-fought and longstanding struggle for control of the results. At stake on the one hand is the economic value of links, and on the other hand the quality of the results. Where does this leave us? From a variety of computer-science studies we know that search engine results tend to be biased in certain directions. Specifically, large sites are over-represented, as are sites with a higher number of links to them, and American sites, probably by virtue of have been online longer and therefore more likely to have higher numbers of links pointing to them (Kleinberg, 1998; Kleinberg & Lawrence, 2001; Lawrence & Giles, 1999; Vaughan & Thelwall, in press). Computer-science commentators have tended to ascribe this tendency to features of the technology. This paper suggests that economic factors may also be at work. Some of the interviewees have indicated the use of old-fashioned editorial practices, such as blacklisting particular sites. However, we remain as yet unclear as to why certain sites are favoured over others. In September of 2002 the FTC issued a Consumer Alert related to the earlier letter it had served on the search engine providers. This said, in part: its a savvy surfer who knows how search results are sorted and ranked. (Federal Trade Commission, 2002b) Indeed. It seems that no one knows very much about this. One of my interviewees, the head of engineering at a large search provider, said: Thats ultimately every search engines secret sauce how relevance is
02 September 2004 E.J. Van Couvering New Media? A Political Economy of Search Engines Page 24

done. The FTCs alert also said: You have a choice. Before you decide which search engine to use, consider whether the use of payment programs for placement or inclusion is important to you. (Federal Trade Commission, 2002b). In fact, as this paper has revealed, the choices are currently quite limited. All major search engines accept paid-placement ads and some require that distributors accept them as well. Finally, the international scale of provision and the concentration of power away from the hands of small distributors means that there may be little incentive for providers to make special provisions for smaller countries who, in turn, may not be able to provide their own services for lack of capital. There is thus some reason to doubt that the current provision of search services, with light regulation according to the free press model, serves the public interest in the information society.

Bibliography
Bettig, R. V. (1997). The enclosure of Cyberspace. Critical Studies in Mass Communications, 14, 138157. Brin, S., & Page, L. (1998). The anatomy of a large-scale hypertextual Web search engine. Computer Networks and Isdn Systems, 30(1-7), 107-117. Carroll, S. L., & Howard, H. H. (1998). The economics of the cable industry. In A. Alexander & J. Owers & R. Carveth (Eds.), Media economics: Theory and practice (2nd ed., pp. 151-174). Mahwah, New Jersey: Lawrence Earlbaum Associates. Castells, M. (1996). The information age: economy, society, and culture: The rise of the network society (Vol. 1). Oxford: Blackwell. Couldry, N. (2003). Media, symbolic power and the limits of Bourdieu's field theory: Media@LSE electronic working papers No. 2 [PDF]. Media@LSE Electronic Working Papers. Retrieved 22 April, 2003, from the World Wide Web: http://www.lse.ac.uk/collections/media@lse/mediaWorkingPapers/ewpNumber2.htm Croteau, D., & Hoynes, W. (2001). The business of media: Corporate media and the public interest. Thousand Oaks, California: Pine Forge Press. Drze, X., & Husherr, F.-X. (2003). Internet advertising: Is anybody watching? Journal of Interactive Marketing, 17(4), 8-16. Federal Trade Commission. (2002a, 27 June). Commercial Alert Complaint Letter Attachment. Retrieved 8 July, 2004, from the World Wide Web: http://www.ftc.gov/os/closings/staff/commercialalertattatch.htm Federal Trade Commission. (2002b, September). FTC consumer alert: Being frank about search engine rank. Retrieved 8 July, 2004, from the World Wide Web: http://www.ftc.gov/bcp/conline/pubs/alerts/searchalrt.pdf Golding, P., & Murdock, G. (2000). Culture, communications and political economy. In J. Curran & M. Gurevitch (Eds.), Mass Media and Society (Third ed., pp. 70-92). London: Arnold. Habermas, J. (1992). The structural transformation of the public sphere : an inquiry into a category of bourgeois society (T. Burger & F. Lawrence, Trans.). Cambridge: Polity. Hu, J. (2003, 6 October). Microsoft drops LookSmart search tool [HTML]. C|Net News.com. Retrieved 10 May, 2004, from the World Wide Web: http://news.com.com/2100-10325087489.html Kleinberg, J. M. (1998). Authoritative sources in a hyperlinked environment. Retrieved 28 August, 2003, from the World Wide Web: http://www.cs.cornell.edu/kleinber/auth.dpf Kleinberg, J. M., & Lawrence, S. (2001). The structure of the Web. Science, 294(-), 1849-1850.

02 September 2004 E.J. Van Couvering

New Media? A Political Economy of Search Engines Page 25

Lawrence, S., & Giles, C. L. (1999). Accessibility of information on the World Wide Web. Nature, 400(-), 107-109. Lazarus, W., & Mora, F. (2000, 20 March). Online content for low-income and underserved Amercians: the Digital Divide's new frontier [PDF]. The Children's Partnership. Retrieved 6 May, 2003, from the World Wide Web: http://www.childrenspartnership.org/pub/low_income/low_income.pdf Lessig, L. (1999). The law of the horse: What cyberlaw might teach. Harvard Law Review, 113(-), 501-549. Mansell, R. (2001). Digital opportunities and the missing link for developing countries. Oxford Review of Economic Policy, 17(2), 282-295. Mansell, R., & Steinmuller, W. E. (2000). Mobilizing the information society. Oxford: Oxford University Pres. McQuail, D. (1994). Mass communication theory: An introduction (3rd ed.). London: Sage. McQuail, D. (2000). McQuail's mass communication theory (4th ed.). London: Sage. Miller, D., & Slater, D. (2000). The Internet : an ethnographic approach. Oxford: Berg. Mosco, V. (1996). The political economy of communication: Rethinking and renewal. London: Sage. Nielsen//NetRatings. (2004, 3 May - 9 May). AdRelevance data glance: Top site genres ranked by impressions. Retrieved 18 May, 2004, from the World Wide Web: http://adrelevance.com/intelligence/intel_dataglance.jsp?flash=true&sr=36808 Norris, P. (2001). Digital divide? : civic engagement, information poverty, and the Internet worldwide. Cambridge: Cambridge University Press. Par, D. J. (2002). Internet governance in transition : just who is the master of this domain? Lanham, MD: Rowman & Littlefield Publishers. Perkins, A. (2003). The classification of search engine spam. E-brand Management. Retrieved 1 July, 2004, from the World Wide Web: http://www.ebrandmanagement.com/whitepapers/spam-classification/ PricewaterhouseCoopers. (2004, April). IAB Internet Advertising Revenue Report, 2003 Full-Year Results [PDF]. Internet Advertising Bureau. Retrieved 9 May, 2004, from the World Wide Web: http://iab.net/resources/ad_revenue.asp Rogers, R. (2000). Introduction - towards the practice of Web espistemology. In R. Rogers (Ed.), Preferred placement - knowledge politics on the Web (pp. 11-24). Maastricht: Jan van Eyck Akademie. Silverstone, R. (1994). Television and everyday life. London ; New York: Routledge. Smythe, D. (1995). On the audience commodity and its work. In O. Boyd-Barrett & C. Newbold (Eds.), Approaches to media : A reader (pp. 222-229). London: Arnold. Sparks, C. (2000). From dead tress to live wires: The Internet's challenge to the traditional newspaper. In J. Curran & M. Gurevitch (Eds.), Mass media and society (3rd ed.). London: Arnold. Spink, A., Jansen, B. J., Wolfram, D., & Saracevic, T. (2002). From e-sex to e-commerce: Web search changes. IEEE Computer, 35(3), 107-109. The Nutch Organization. (2004). Nutch: faq [HTML]. Retrieved 6 July, 2004, from the World Wide Web: http://www.nutch.org/docs/en/faq.html Thompson, J. B. (1990). Ideology and modern culture. London: Polity Press. Vaughan, L., & Thelwall, M. (in press). Search engine coverage bias: evidence and possible causes. Information Processing & Management.

02 September 2004 E.J. Van Couvering

New Media? A Political Economy of Search Engines Page 26