minstrel
12-09-2003, 12:53 AM
Too much Net data?
Saturday, December 6
STEPHEN STRAUSS
Globe and Mail Update
One of the defining features of modern life is the Net's ability to link disparate things in ways that they never have been linked before.
Part of what ensues is the frustration of a Google search for something such as ''black pearls,'' which in the first 20 hits ends up referring you variously to Elizabeth Taylor's line of perfume, a kind of Polynesian pearl and black female blues singers of the 1920s.
Despite the search engine's drawbacks, Simon Fraser University student Geoff Peters has come up with an interesting way to make use of comparative Googlology.
At his site, called GoogleDuel, you can type in a word and compare its prevalence on the Internet with that of another word. Thus, you can see the result of the citation fight of Good versus Evil or Cat versus Dog.
The technique theoretically could tell businesses such as Coke how they are doing in terms of Internet market consciousness when compared with Pepsi, or give political strategists an idea of how widely known Paul Martin is in relation to Sheila Copps.
Unfortunately, Mr. Peters has learned how grossly inaccurate comparisons on the Internet can be. When he typed in a "dog" versus "cat" duel over a two-month period, he found vast variations. Early one day, there might be 15.9 million sites mentioning "dog" and later the same day 17 million. A month later, "dog" had slumped to 13.5 million sites and been overtaken by more than 15 million cat references.
Why the swings? Given the vast and increasing number of sites Google has to search in a few seconds, "it doesn't actually count all the pages that contain the word," Mr. Peter says. "So it must have a very quick way of estimating how many pages contain that result. Unfortunately, Google doesn't reveal the method that it uses."
Given the increasing reliance of researchers, not to mention ordinary human beings, on Internet search engines, the existence of this error-prone shorthand makes one nervous about all of Google's results and raises the possibility that the huge amount of information on the Web is overpowering the master information gatherer.
Source: http://sympatico.globeandmail.com
Saturday, December 6
STEPHEN STRAUSS
Globe and Mail Update
One of the defining features of modern life is the Net's ability to link disparate things in ways that they never have been linked before.
Part of what ensues is the frustration of a Google search for something such as ''black pearls,'' which in the first 20 hits ends up referring you variously to Elizabeth Taylor's line of perfume, a kind of Polynesian pearl and black female blues singers of the 1920s.
Despite the search engine's drawbacks, Simon Fraser University student Geoff Peters has come up with an interesting way to make use of comparative Googlology.
At his site, called GoogleDuel, you can type in a word and compare its prevalence on the Internet with that of another word. Thus, you can see the result of the citation fight of Good versus Evil or Cat versus Dog.
The technique theoretically could tell businesses such as Coke how they are doing in terms of Internet market consciousness when compared with Pepsi, or give political strategists an idea of how widely known Paul Martin is in relation to Sheila Copps.
Unfortunately, Mr. Peters has learned how grossly inaccurate comparisons on the Internet can be. When he typed in a "dog" versus "cat" duel over a two-month period, he found vast variations. Early one day, there might be 15.9 million sites mentioning "dog" and later the same day 17 million. A month later, "dog" had slumped to 13.5 million sites and been overtaken by more than 15 million cat references.
Why the swings? Given the vast and increasing number of sites Google has to search in a few seconds, "it doesn't actually count all the pages that contain the word," Mr. Peter says. "So it must have a very quick way of estimating how many pages contain that result. Unfortunately, Google doesn't reveal the method that it uses."
Given the increasing reliance of researchers, not to mention ordinary human beings, on Internet search engines, the existence of this error-prone shorthand makes one nervous about all of Google's results and raises the possibility that the huge amount of information on the Web is overpowering the master information gatherer.
Source: http://sympatico.globeandmail.com