In a recent study, Indiana University researchers conclude that search engines actually produce an egalitarian effect - challenging the notion of 'Googlearchy'
The notion that search engines such as Google and Yahoo monopolise world wide web traffic has taken a sharp detour at the Indiana University School of Informatics.
In a recent study, IU researchers conclude that search engines actually produce an egalitarian effect.
This challenges the so-called 'Googlearchy' hypothesis, the commonly held belief that search engines make popular pages more popular at the expense of new and lesser-known ones.
The researchers' paper, The Egalitarian Effect of Search Engines, was cited in the 19 November 2005 issue of The Economist, a publication known for its analysis of world business and current affairs; and in Physicsweb, the online version of the Physics World magazine.
Since then, the paper has generated much debate and discussion in more than 100 blogs around the world.
"The web is a place where the rich-get-richer dynamic is well established," says Santo Fortunato, a post-doctoral fellow at the School of Informatics.
"We believe the empirical data does not support the belief that search engines amplify this phenomenon.
"In fact, our evidence strongly suggests new pages have a greater chance to be discovered through the use of search engines".
Joining Fortunato in the study are Filippo Menczer, associate professor of informatics and computer science; Alessandro Flammini, assistant professor of informatics; and Alessandro Vespignani, professor of informatics.
A search engine is a complex system designed to help find information stored on the web.
It allows users to look for content meeting specific criteria - typically those containing a given word or phrase - and retrieves a list of references that match those criteria.
The resulting 'hits' are returned and ranked by relevance.
The researchers, who pooled their expertise in areas such as web mining, complex systems and networks, constructed models detailing two different cases.
In the first scenario, users browsed the web only by choosing random links; in the second, users visited only pages returned by search engines.
From there, the researchers tracked and compared traffic to many web sites against the number of incoming links.
They concluded the existing bias toward popular sites caused by the link structure of the web is actually mitigated by the queries submitted to search engines and by the way in which users view the results.
The methodology employed by the IU researchers has raised some objections, owing to its reliance on so-called 'noisy data' provided by search engines.
"We used the best data available," says Fortunato.
"We conducted extensive statistical analysis to obtain a reliable traffic trend.
"Our results were confirmed using two independent sources to count the incoming links (Yahoo and Google), and collecting the data twice in a three-month period".
The IU study might help understand the social impact of search engines, as well as aid in the design of future search technology, says Menczer.
"Search engine companies always are trying to improve the ways they rank their search results, which really are the secret recipes to their success," he notes.
"Our work points to a key ingredient that search engines can manipulate to become more egalitarian or more elitist".
