Thursday, May 19, 2011

Podcasts of Note: The Filter Bubble

Eli Pariser: "The Filter Bubble" – The Diane Rehm Show, May 17, 2010

On this episode of the Diane Rehm Show, Eli Pariser discussed the personalization of internet search and news feed results, and the problems which this can cause. The premise is that Google, Yahoo, Facebook, and others are using your user history and other data to give you results that you’re more likely to click on.

On the plus side of this, I enjoy the personalized book recommendations on Amazon. I actually prefer the “people you communicate with the most” setting on Facebook to seeing updates from 450+ people, most of whom I only ‘friended’ to be polite. I also prefer seeing advertisements targeted to me. If I’m going to have to see ads on line, I’d rather they be for things I might actually be interested in.

That being said, I am incredibly uncomfortable with the idea of personalized search results from Google, Yahoo, or any other search provider. In theory, such personalized search results would give a conservative Christian Republican vastly different search results than a liberal Atheist Democrat. Instead of providing an accurate, unbiased flow of information, search results will be biased towards what the user is most likely to click on.

I tested this out today. At my workplace, there is a computer that I share with all of my colleagues – a situation which is bound to much up user-personalization algorithms. I googled Newt Gingrich from this work computer. The results I received were overwhelmingly strongly anti-Newt. Additionally, neither his wikipedia entry nor his official site were on the first page of results. When I googled him from my home computer, wikipedia was the first result, and his official site was the second. Additionally, links to his facebook and twitter pages were on the first page of results (and they weren’t on the work computer).

Wikipedia is often the first result to pop up whenever I search for ANYTHING. I also go straight to wikipedia for a lot of things. I had always just assumed the wikipedia articles were popping up first because they had the best information on the topic of my search. Now I wonder if they’ve been popping up because I like to use wikipedia.

I’m also curious as to what my coworkers have been looking up to make the ‘user profile’ for the work computer so avidly anti-Newt. I mean, I’m definitely a liberal, and I don’t like Newt Gingrich, and yet my home computer had much more neutral search results.

I find all of this incredibly disconcerting. I had always just assumed that I was getting the best, most accurate result for my searches, and now I learn that my results depend on if I’m using my computer, a friend’s, or a public (library, internet café, office) computer.

Yes, I’m a liberal. Yes, I prefer certain things. But feeding me the information I want to hear isn’t going to do me any good in the long run. I want accurate facts, even if those facts fly in the face of my personal beliefs.

**One thing from the podcast which irked me: Pariser seemed unaware that one can change from a filtered newsfeed to an unfiltered one on Facebook. He was aware of how to change from ‘Top News’ to ‘Most Recent.’ When one caller pointed out that she was intelligent enough to know how to switch to an unfiltered news feed, Pariser assumed she meant switching from ‘Top News’ to ‘Most Recent’ and basically told her in an incredibly condescending voice that she didn’t know what she was talking about.

[In case you’re wondering, your default setting on Facebook is the filtered ‘people you communicate with the most’ setting. To see updates from ALL your friends, click on the down arrow next to the ‘Most Recent’ tab, then select ‘edit options.’ Next to where it says ‘show posts from’ select ‘all of your friends.’]

2 comments:

Anonymous said...

I believe that Google would only change results if you were 1) Logged in to your Google account or 2) Based on activity within that same session. I bet if you logged out of Google and then restarted your computer you'd get the typical results. In fact, I suspect that logging out would be sufficient.

Annie Nimity said...

I haven't tested logging out then re-starting my computer (don't have time right now), but logging out and searching for Newt Gingrich (I was listening to a podcast about him when I was writing this post and decided to test Pariser's theory; I don't normally do searches for Newt!) gives me the same results as I get while logged in. I have two different Google accounts - I got the same results logged in as either user as I got when logged out. On my work computer, I was not logged in to Google. According to Pariser, the information is logged based on your IP address as well as user name...