In his review of Hazel Bell's 'Indexes and Indexers in Fact and Fiction', Andro Linklater says:
What must rank as today�s most useful index � covering 2,073,418,204 pages on the worldwide web and accessible through the Google search engine � is compiled entirely by computers. Google�s superiority over rivals like Hotbot is due to the capacity of its software to catalogue entries according to a crude hierarchy of relevance. But as this charming anthology illustrates, a good index amounts to far more than a catalogue.
Andro misses Google's subtlety here. While it is compiled by computers, it is based on the links between web pages made by humans, who are doing exactly the kind of indexing he praises.
Google's relevance hierarchy is far from crude; it works on the assumption that pages that are linked to by many other pages are more useful, and it takes into account not only the words on the linked-to page, but the words within the links pointing to it. It then takes this idea one stage further, but giving greater weight to links from pages it has already found to be important.
This interesting mixture of democracy and elitism is what gives Google its remarkable ability to list the most useful pages first - it is harnessing the collective indexing decisions of all the people individually publishing web pages that link to each other.
I just searched for 'Andro Linklater indexers' and got no results, because no-one has linked to the article yet. I am going to link to it from my own webpage so that by the time the Spectator publishes this letter, readers will find that search succeeding.
Thursday, 21 March 2002
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment