Computers can’t be biased?!
Krishna Bharat, chief scientist for Google News, told Wired recently,
“The truth is, Google News doesn’t have a point of view. It’s a computer, and computers do not understand these topics the way humans do and can’t be systematically biased in any direction.”
Well it seems the search engine companies have the public on their side on this one - 68% of US users say that search engines are a fair and unbiased source of information according to a recent Pew report on search engine users but if you sit back and think for a second it is hard to see how Google News could be ‘objective’ - what does that even mean? As soon as Google steps in and dictates what counts as a news source and what doesn’t that could arguably introduce bias - and every algorithm has its own tacit ‘tendencies’ to favour certain results over other results whether to improve “relevance” or just accidentally.
It’s a little worrying that most search engine users don’t really grasp these issues. I tried to interview a few of them a couple of years ago about this and found it hard even to get them to understand what I was trying to ask them about…
Thanks to John Battelle’s Searchblog for the link (there are plenty of comments on this issue there).
February 2nd, 2005 at 9:07 pm
Bah! Just lost my original comment.
I was going to say, one of the ways in which you can understand this is that the issue within the search engines is framed in terms of relevance and irrelevance, not in terms of bias and fairness. When you start thinking from the relevance standpoint, the question is whether something is authoritative content or not authoritative content. So from this perspective, mainstream will be better than fringe, because it is more trustworthy and reliable, as measured by how many others give implicit measures of trust by linking to the content.
I just did an interview about this with a high-up in a major search provider. Let me see if I can find a good quote. Yes, here:
***
EVC: Its almost like going back to the Daily Me, if you remember when the web first start out and personalization first started coming in?
Interviewee: Yes. You?ll see a section that?s called My News?
EVC: With all these kinds of ingredients, this is getting more towards being editorial in an old-fashioned journalistic sense of the word, by choosing information providers and by selecting stories and things like that.
Interviewee: Especially in news there?s a strong editorial ? you have to be aware of editorial needs and philosophy. It is different, though, than a broadsheet ? this is not MSNBC or CNN or ABC ? those approaches are much more saying “We have the best human-managed content and we have a specific selection. We are a boutique shop, we?re not Walmart, and we?re going to pick it for you and you should pick us because you like our editorial selection and tone.” That?s how I would characterize the large news media outlets. The newsbot approach, the Google News approach, they?re more about “We?ll let you pick from anything, we?ll try and give you all the sources we can that are at least reasonably trustworthy, and then we?ll personalize it to you specifically, we won?t hold you subject to our view of what?s important.” So it?s a very different approach.
***
You see the issues from this point of view are trust (authority) and some kind of ideology of freedom, rather than the types of fairness issues that characterise journalist ethics.
February 4th, 2005 at 9:47 am
Easy, easy. They just said that syntactics is always truthful, semantics only carries values. I can feel de Saussure turning in his grave, but alas. “I won’t complain” (which eludes Google’s understanding, I’m sure).