Monday, December 6, 2010

RE: Cablegate from Wikileaks: a case study

Posting to NGC4LIB

Michele Newberry wrote:
<snip>
Oh my goodness -- isn't this exactly what we do when we "tweak" our relevance ranking algorithms in our own systems? We call it the "On the road" tweak -- doing what we need to do to make this obvious titles appear on the first page of the results preferably near the very top.
You could also call it the "Gone with the wind" tweak or even the "Nature" tweak.
</snip>
On 12/6/2010 2:10 PM, Jonathan Rochkind wrote:
<snip>
Of COURSE Google's algorithms are the result of subjective human judgements in how to best apply available technology to meet user needs. This should surprise nobody that knows that software isn't magic, it just does exactly what programmers tell it to.
</snip>
Interesting reactions. Google very clearly tweaked its results based on a story from the NY Times, and the purpose was to downgrade certain results based on what they considered to be the "greater good" or something like that. The articles very clearly pointed out that being able to do this is *incredibly powerful* in terms of societal impact, and I agree. After all, people trust Google.

I confess, I have never understood relevance ranking in library catalogs, although I do understand the concept rather clearly in general search engines such as Google, which is based on various types of citation analysis. In this article, Google pretty much admitted that they tweak results based on political considerations (i.e. articles in the NY Times). How would Google have tweaked things during the US Civil War? Or during WWI? What else is Google doing today that we don't know? Do libraries tweak results based on political considerations? I hope not.

I brought up these articles as examples of some very difficult matters that the entire information world needs to deal with today, since these matters often have very tangible consequences for society.

No comments:

Post a Comment