School of Media and Communications, UNSW
Google. That the noun has rapidly become a verb speaks volumes for the influence of this search engine. Powered by PageRank, the accuracy of its results has done more than make Google the premiere search application – it has moved web search into the realm of ‘killer app’ alongside e-mail. When über.nu columnist Adam Mathes tested his theory of the ‘Google Bomb’, he may have realised the potential of his actions, but he perhaps underestimated the power of the pure idea itself. The fashion in which Google could be manipulated highlighted many questions about the nature of the web and its network of linkages. Deleuze and Guattari would see such activity as not only exemplifying the web as rhizome, but that it also demonstrates their conception of the refrain. Google Bombs demonstrate how web link ecologies, particularly those of blog linkages, influence PageRank – Google’s ranking algorithm that effectively implements the idea of the refrain to invoke a hierarchical list of all pages on the web.
On April 6, 2001, Mathes suggested that those reading his column should include a link on their own websites, using a specific piece of text as the link, to one Andy ‘talentless hack’ Pressman, in order to make his friend look stupid. Many of Mathes’ readers are ‘bloggers’ – people who maintain online diary-style websites, or ‘blogs’ (short for web logs) – so the call for readers to add links to websites could lead to a very large number of linkages. While his example was fairly innocuous, the theory behind the action was groundbreaking. He wanted to generate a distributed set of linkages to a specific page in order to raise that page’s ranking for a particular search term. Dubbed ‘Google Bombing’, Mathes intended to exploit the following fact:
Google is unique among search engines in that while it almost always shows you pages that have the exact keywords you are looking for, occasionally it will show you pages that don’t have those keywords, but other pages linked to that page with those words. (Mathes, 2001)
In fact, this is exactly what Google does. According to the founders of Google, Sergey Brin and Lawrence Page, Google ’employs a number of techniques to improve search quality, including page rank, anchor text, and proximity information’ (1998: 12). Google’s ability to return high quality search results is primarily thanks to the PageRank algorithm. This algorithm, as its name implies, ‘is a global ranking of all web pages, regardless of their content, based solely on their location in the Web’s graph structure’ (Page, et al., 1). The ranking algorithm is recursive, with ranks being generated not only by the number of links leading into a page but also the PageRank of the pages that link to the page. The higher the rank of a linking page, the more valuable that link becomes to the page receiving the link. This ranking is then complemented by exactly what Mathes described above; an analysis of not only the text on each web page, but the anchor text used to point to each page is also assigned to that page’s text analysis as well. This was far from a guarded secret, but no one had so openly attempted to manipulate it before.
While it is hard to say how long the ‘bomb’ took to ‘go off’, within days a Google search for the words ‘talentless hack’ delivered Pressman’s website ranked #1. This ranking remained for at least eleven months, such was the power of this first ever Google Bombing. In Mathes’ own words:
In a bizarre surreal bow to the power of perception on the web, what you say about a page becomes just as important as the actual content of the page. The page must be what other people say it is. That Google adheres to this rule and is by far the most effective search engine raises many interesting issues… (Mathes, 2001)
Indeed it does … not the least of which is that more than three years later, a Google search for ‘talentless hack’ now gives the #1 ranking to Mathes’ original article! It seems that over time, as the concept itself has gained coverage, the original search term has folded back onto the article that called for it to be pointed elsewhere. Meanwhile, Pressman’s site is now nowhere to be seen in the ten pages of search results. With connection, multiplicity and temporality all seemingly part of the way the Google Bomb works, this stunt has inadvertently, and thus ever more appropriately, generated a perfect example of Deleuze and Guattari’s rhizome and the territorial markings that form a refrain (1987).
We must remember that through Google’s status as the most precise search engine available on the web, any success in manipulating it must somewhat grant insight into the nature of the World Wide Web itself. As the Google Bomb only works through a highly distributed set of web links (‘simply having tons of the same links with the same phrase on a single page will do nothing’ Mathes notes (2001)) we are shown the distributed, non-hierarchical nature of the web. Of course, with ‘no points or positions in a rhizome… only lines’ (Deleuze and Guattari, 1987: 8) we have the same basic nature of the rhizome reflected in the web. This has been well covered elsewhere, notably in Kathleen Burnett’s essay on hypertext as rhizome more than ten years ago (1993).
By moving deeper into the idea of repeated, distributed linkages – Google Bombs or not – we find the web, particularly when analysed by a system such as PageRank, is full of what Deleuze and Guattari have called ‘refrains’ (1987). The refrain is described as ‘any aggregate of matters of expression that draws a territory’ (323). Importantly, the territorialising marks cause ‘a reorganisation of functions and a regrouping of forces’ (320). To apply these ideas to the web, the links that flow from one site to another form territories for the pages that are linked to. The aggregation of the various links that lead into any particular page or site form a set of territorial markings, and it is these markings that Google has harnessed to achieve its superior search results. In turn, these markings generate a grouping of forces, passing on value to the linked page and enhancing the page’s PageRank.
Through a search on Google, a territory is invoked relating to the search term at hand. This territory is delivered as a ranked list of links, ranked according to a mix of PageRank’s raw analysis of page importance as well as the relevance to the search terms. In essence, Google analyses the refrain of each relevant page and delivers a list of sites ranked according to the relevance and magnitude of the page’s refrain. The relevance relates to the text of the page and most notably the anchor text used to lead to the page. The magnitude relates to the total influence of incoming links and, as mentioned earlier, the influence is adjusted according to the PageRank of the pages from which the links are received.
The fact that the result of such activity changes over time further proves the rhizomatic nature of the web and how PageRank is an efficient, temporal indicator of each page’s refrain on the web. Links move from front pages to archives, in turn reducing the value of such links as they move from pages of higher rank to pages of lower rank. The link may not change but its value to the refrain is reduced as it is no longer on a page with as many incoming links of its own, therefore it possesses less value to pass onto pages it links to. Further, new articles and pages link to different targets with regard to the same phrases, adding further competition to any refrain related to relevant search terms. As the linkages change, so too do the refrains. For these invoked lists of pages to become unchanging would be a sign of stagnation, a sign that the web was no longer dynamic. The chaos of web linkages is what the web – and Google – thrives on.
Since the Google Bomb has become a ‘known exploit’ of PageRank, there are a few examples of its use to perform protest action. By building a page with protest information about a certain organisation or activity, and then coordinating a Google Bomb to attach that page to the name of the organisation itself, the page of anti- information will often appear very close to the organisation’s own page in a Google search. Notable search terms targeted include ‘Verisign’ and ‘Bill Gates’. Of course, this will certainly become more difficult the larger an organisation is. To rank a page alongside ‘Microsoft’, you would need a massively widespread set of linkages, or a smaller number of linkages from some very highly ranked pages. This is further evidence of the nature of the refrain. The larger an organisation, the more important it would appear to PageRank. It would be very likely to have a lot of attention and therefore many incoming links from many other pages, some of which would themselves be important pages – and nearly all of which would be likely to use anchor text related to the company name. Thus their territorial markings, or refrain, would be well established. To challenge their refrain would require a similarly influential set of linkages to grant the page the territory it needs to gain a notable PageRank. The humble site owner or blogger, are less likely to gain attention from important pages, but they can use the blogging community to try to garner support and generate a widely distributed set of incoming linkages to build them a refrain. This new territory on the doorstep of the large company’s own territory hopes to spread an alternative message to anyone searching for the relevant term.
Many commercial enterprises have attempted to influence Google in a similar fashion to the Google Bomb (Sullivan, 2002). Organisations have set up ‘link farms’, in which many servers running many different domains all contain links to try and raise the ranking of target websites relating to certain search terms. But this attempt artificially to extend a site’s territory has been highly ineffective at influencing PageRank. Indeed the PageRank algorithm is known now to account for certain kinds of manipulative link instances, rendering them ineffective.
Is it the case that Google Bombs – and blog-centred link swapping activities in general – are also artificial and should be treated like link farms? There is an important difference. In the former instance, one single entity is attempting to improve/expand a particular refrain across the web. In the latter, we have a distributed group of individuals who, each on their own terms, decide to support a particular refrain. If a Google Bomb attempt does not convince a distributed group of individuals to participate then it simply will not work. Google ‘knows’ the difference between the organised yet genuinely distributed creation of a real territory and an attempt artificially to construct one.
There has been some misuse of the term ‘Google Bomb’ – in one case, relating to the ‘Cannot find Weapons of Mass Destruction’ satire web page (Cox, 2003a). An amusing alteration of the Microsoft Internet Explorer error page, the page recently spread quickly around the Internet through email and blog links. When people realised it had become the number one search term in Google for ‘Weapons of Mass Destruction’ the joke was extended further. Rather than directly linking the page, people instead suggested searching Google for that term and use the ‘I’m Feeling Lucky’ button (which takes you directly to the first page found). This Google relationship got some writers referring to the page as a Google Bomb (Cox, 2003b). But in this case there was no organised, intentional distribution of links – using a specific piece of link text – to give it such a ranking. This was simply a case of a popular page being linked by many people independently – an entirely organic refrain was generated without any intent to capture a particular set of territorial markings.
As touched on above there are some who believe that what the Google Bomb represents is a problem that Google needs to address. Andrew Orlowski, a writer for technology news website The Register, is one such source of concern over the ability of bloggers to influence Google search results. In April 2003 Orlowski wrote about the manipulation (or ‘Googlewashing’, as he called it) of the term ‘Second Superpower’ (2003). Originating in a New York Times article by Patrick Tyler (2003) as a concept referring to ‘world public opinion’, Orlowski claimed that in just 42 days the term had been Googlewashed so that a Google search for ‘Second Superpower’ referred almost exclusively to an article that used the term with reference to Internet users. To Orlowski this was a clear example of a disturbing ability of bloggers to shift meaning on the web – that links from a few ‘”A-list” tech bloggers’ (2003) can quickly lead to such shifts in Google search results. If we return to a Deleuze and Guattari centred reading of such linkages, however, this ability is not so disturbing at all.
Just as a rhizome ‘ceaselessly establishes connections between semiotic chains’ (Deleuze and Guattari, 1987: 7), so too does the web. Blog linkages are often quite volatile as the automated nature of such publishing means links move from positions of importance on the front page of a blog into the sites archives. As touched on earlier, a page in an archive will have fewer links leading into it resulting in a lower PageRank which in turn results in links leading from that page offering a reduced weight of territorial influence. This volatility changes these influences in Google’s system very quickly (Hiler, 2002a). While Orlowski complains about a small group of bloggers causing this shift, Tom Coates soon noted that ‘all it takes is one Register article (picked up on by other bloggers) for this problem to correct’ (2003). Within days of Orlowski’s The Register article, competing information (most notably his article, which quickly became the second highest ranked ‘Second Superpower’ item) was now available in the first page of search results. As it turns out, the term ‘Second Superpower’ was never used in the original New York Times article. So even though Orlowski’s claim was based on an inaccurate reading of the original story, it was still able to assist in correcting what he saw as an online injustice.
Part of this negativity toward seemingly inaccurate or manipulated results may stem from a desire for Google to be something different to what it is. While Google is an index of the web – and the territories and linkages that exist in that space only – many seem to treat Google as a tool that will search for results not only based on those web territories but within the context of the real world too. But Google is not an index of prevailing conditions in the real world – even if it wanted to be. It can only index what exists on the Internet! That the content of the web reflects what is happening in the real world means it is still only a reflection – with all the subtle distortions and differences that go along with that.
In the ‘Second Superpower’ case above, the term may have been used in high-profile offline spaces a number of times with its original meaning. But it had few online instances to generate a refrain that would create a sufficiently large territory. When the term was used in a new ‘online’ context it quickly gained attention and a territory was shaped around it – a greater online territory than the original.
It cannot be ignored here that the Patrick Tyler’s original New York Times article has no presence in Google. It had, in fact, just disappeared around the time Orlowski wrote his article for The Register. Google cannot include pay content in its archives and therefore cannot include such content in its indexing for search terms – and pay-content systems are what many traditional news provider’s archives are moving to. This exacerbates the difficulty, mentioned above, of reflecting the real world – without authoritative archives of news information from traditionally offline sources Google can only refer to information that is available only in purely online sources. In the words of one pundit ‘the “googlewashing” Orlowski talks about was done by the New York Times, not by Google, and not by bloggers’ (Searl, 2003). If the information is not in Google, then it cannot be indexed.
Conversely, the open nature of Google’s indexing efforts grant the ability to participate in the ‘free market’ of ideas and linkages and this also means search terms are always open to territorial ‘dispute’. If enough people want to participate in the territory that is invoked by a particular search term then they can attempt to do so. No single concept will ever own a territory on the web – as long as multiple search results are returned to the user this will mean many pages can compete for attention within particular invoked spaces. Even on the first page of a set of search results you will often see different concepts, particularly where new or generic terms are the territory in dispute. While they are ranked in an order of relevance, or the strength of their refrain, they exist within the same emergent territory.
Google has come under fire not only for their reluctance to redress the influence of blog linkages on search results; they have also been attacked for their outright refusal to remove highly undesirable pages from their search results. One recent case related to the search term ‘Jew’ and the high ranking achieved by an anti-Semitic site. As undesirable as such sites may be, Google stand firmly by their policy and do not want to become a de facto censor for the web. Recall their definition for PageRank included the statement ‘a global ranking of all web pages, regardless of their content’ (Page, et al., 15) as well as another statement that PageRank rates pages ‘objectively and mechanically, effectively measuring the human interest and attention devoted to them’ (1). Perhaps in some instances this later definition is debatable, but as Google argued in the above situation, the search terms used are possibly the major source of creating such results.
If you use Google to search for ‘Judaism’, ‘Jewish’ or ‘Jewish people’, the results are informative and relevant. So why is a search for ‘Jew’ different? One reason is that the word ‘Jew’ is often used in an anti-Semitic context. Jewish organizations are more likely to use the word ‘Jewish’ when talking about members of their faith. The word has become somewhat charged linguistically… In fact, prior to this incident, the word ‘Jew’ only appeared about once in every 10 million search queries. Now it’s likely that the great majority of searches on Google for ‘Jew’ are by people who have heard about this issue and want to see the results for themselves. (Google, 2004)
Which is the path to true objectivity? A measure of all pages on the web, regardless of content? Or a measure of human interest and attention devoted to them? Are they entirely compatible? Clearly the more influence Google’s search holds on people’s use of the web, the more likely it will face increasing pressure to fulfil a role online that should not be the domain of a corporation. As it stands, Google delivers results from the open market of ideas and competing territories that are found on the web and ranked according to PageRank. Perhaps Google’s search engine could be considered the objective arbiter in a no holds barred battle of refrains, ranking the raw power of each page’s territorial markings on the web – and all are welcome to add linkages wherever they choose.
Returning to the idea that ‘a rhizome ceaselessly establishes connections between semiotic chains, organisations of power, and circumstances relative to the arts, sciences, and social struggles’ (Deleuze and Guattari, 1987: 7) Google Bombs – and blog-centric linkages in general – ensure the potential for dynamism and tension across the World Wide Web. The ability to add new linkages where there were previously none is essential to both the rhizome and the refrain. Google Bombed results can rupture territories or mark out new ones, and they have the potential to raise debate and promote discussion across the web that can only further promote the dynamic, temporal and decentred nature of the web. The Google Bomb supports and promotes the powerful potential of the rhizome and the manner in which ideas compete for territory online through their refrains.
It’s time to stop worrying about bloggers and their influence over Google. By learning to love what they, and their Google Bombs, can do opens us all to the widely accessible power that still exists on the web for those who care to engage with it.
Séamus Byrne is Deputy Editor of the Internet culture and lifestyle magazine internet.au, as well as a monthly contributor to Australian MacWorld. When not busy being a tech journalist, he fits in working on a Ph.D. at the UNSW School of Media and Communications. Research interests include online communities, sound design, advertising, activism, and the construction of cool. [firstname.lastname@example.org]
Brin, Sergey and Page, Lawrence. ‘The Anatomy of a Large-Scale Hypertextual Web Search Engine’, (1998), http://www7.scu.edu.au/programme/fullpapers/1921/com1921.htm
Burnett, Kathleen. ‘Toward a Theory of Hypertextual Design’, Postmodern Culture 3.2 (1993): http://www.infomotions.com/serials/pmc/pmc-v3n2-burnett-toward.txt
Coates, Tom. ‘plasticbag.org | weblog | Oh self-correcting Blogosphere…’, (2003), http://www.plasticbag.org/archives/2003/04/oh_selfcorrecting_blogosphere.shtml
Cox, Anthony. ‘Cannot find Weapons of Mass Destruction’, (2003a), http://www.coxar.pwp.blueyonder.co.uk/
_____. ‘The war on the web’, (2003b), http://www.guardian.co.uk/online/story/0,3605,994676,00.html
Deleuze, Gilles and Guattari, Felix. A Thousand Plateaus: Capitalism and Schizophrenia, trans. Brian Massumi (Minneapolis: University of Minnesota Press, 1987).
Google, ‘Google Technology’, (2004), http://www.google.com/technology/index.html
Google, ‘Google: An explanation of our search results’, (2004), http://www.google.com/explanation.html
‘Google Answers: Google Inc., and the Google Bomb’, (2003), http://answers.google.com/answers/main?cmd=threadview&id=179922
Hiler, John. ‘Google Loves Blogs’, (2002a), http://www.microcontentnews.com/articles/googleblogs.htm
_____. ‘Google Time Bomb’, (2002b), http://www.microcontentnews.com/articles/googlebombs.htm
Marks, Kevin. ‘Googlewash? Hogwash’, (2003), http://epeus.blogspot.com/2003_04_01_epeus_archive.html#200094183
Mathes, Adam. ‘Filler Friday: Google Bombing’, (2001), http://uber.nu/2001/04/06/
Mottram, Jack. ‘Submit Response: Googlewash The Second Superpower’, (2003), http://www.submitresponse.co.uk/archives/googlewash_the_second_superpower.php
Orlowski, Andrew. ‘Anti-war slogan coined, repurposed and Googlewashed… in 42 days’ in The Register (2003): http://www.theregister.co.uk/content/6/30087.html
Page, Lawrence, et al. ‘The PageRank Citation Ranking: Bringing Order to the Web’, (1998), http://dbpubs.stanford.edu:8090/pub/1999-66
Searl, Doc. ‘The Doc Searl Weblog: Friday, May 16, 2003’, (2003), http://doc.weblogs.com/2003/05/18
Sullivan, Danny. ‘Google Bombs Aren’t So Scary’, (2002), http://searchenginewatch.com/sereport/02/03-bomb.html
Tyler, Patrick. ‘A new power in the streets’, New York Times (2003), http://www.submitresponse.co.uk/archives/superpower.html (Mirror of original)
Walker, Jill. ‘Links and Power: The Political Economy of Linking on the Web’, (2002), http://huminf.uib.no/~jill/linksandpower.html