I notice that "Measuring bias in 'organic' Web search" by Edelman and Lockwood, is simply "posted" on a Website (http://www.benedelman.org/searchbias/), rather than being a peer-reviewed paper, so I'm not sure why anyone is paying it any attention. The authors suggest that Google is biased on the basis of what is probably the most trivial bit of data collection one could imagine. It's easy to replicate. One search term they used was "mail" and, sure enough, when I use this in Google, the first item retrieved is a link to Gmail. However, when I use email - presumably equally likely to be used by searchers, the first item is a link to Hotmail and the second is to Gmail; when I use "e-mail", the first link is to Yahoo's mail service, the second is to Wikipedia, the third is to Hotmail, the fourth is to news on e-mail and only the fifth is to Gmail.
In other words, to base a proposition of bias on one possible form of a concept is to write absolute rubbish. Still, these are Harvard Business School people and no doubt self-advertisement is the primary motivation here.
In other words, to base a proposition of bias on one possible form of a concept is to write absolute rubbish. Still, these are Harvard Business School people and no doubt self-advertisement is the primary motivation here.
No comments:
Post a Comment