Introduction1
Google’s well-known mission is “to organize the world’s information”. It is, however, impossible to organize the world’s information without an operating model of the world. Melvil(le) Dewey (1851-1931), working at the height of Western colonial power, could simply take the Victorian world view as the basis for a universal classification system, which, for example, put all “religions other than Christianity” into a single category (no. 290). Such a biased classification scheme, for all its ongoing usefulness in libraries, cannot work in the irreducibly multi-cultural world of global communication. In fact, no uniform classification scheme can work, given the impossibility of agreeing on a single cultural framework through which to define the categories.2 This, in addition to the scaling issues, is the reason why internet directories, as pioneered by Yahoo! and the Open Directory Project (demoz)3, broke down after a short-lived period of success.
Search engines side-step this problem by flexibly reorganizing the index in relation to each query and using the self-referential method of link analysis to construct the ranking of the query list (see Katja Mayer in this volume). This ranking is said to be objective, reflecting the actual topology of the network that emerges unplanned through collective action. Knowing this topology, search engines favor link-rich nodes over link-poor outliers. This objectivity is one of the core elements of search engines, since it both scales well and increases the users’ trust in the system.