Tuesday, December 04, 2007

Handsome Carmody Dashing Genius

David Pogue on using number of search results to prove an argument: "On Google, ‘chicken armadillo’ gets 595,000 hits. ‘Banana carburetor’ gets 132,000 hits. ‘Liquefy purple warthogs’ is just about the most ridiculous improbable phrase I could come up with, and even that one gets 303 results, for crying out loud.

Once you start down this road, you’ll discover that you can prove any point at all. If you’re arguing that Microsoft tortures puppies, you could write, ‘A Google search for ‘Microsoft puppy torture’ returned 264,000 results,’ and you’d be correct (about the results, I mean—not the conclusion)."



2 comments:

LPS said...

You know, I actually used precisely this method in a paper to show that "che guevara t-shirt" is by far more popular than either "mickey mouse t-shirt" or "jesus t-shirt." Unlike Pogue and Consumer Reports, however, I did put my phrases in quotes. That drastically cuts down on the number of results returned, as I have also found when googling statistically improbable-sounding phrases from my students' papers to check for plagiarism. Of course, I wasn't attempting to prove that there are more che guevara t-shirts in the world than mickey mouse t-shirts, or that more money is spent on che swag, just that the che shirt is a significant phenomenon in popular culture.

Matt Penniman said...

Another potentially valid use of search results is to argue for the prevalence of one semantic construction over another -- for example, "unalienable rights" gets about 400,000 results, while "inalienable rights" gets 750,000.