Sunday, April 17, 2016

The Opacity of the Algorithm

Sorry this was late! I ended up having to work late today unexpectedly.

Big data is so hard to conceptualize, it is difficult to have a coherent opinion on it.  I installed Ghostery after our discussion about it in class a few weeks ago, and the sheer amount of hidden third-party activity going on in the background of my day-to-day existence on the internet is astonishing.

Before I started getting automatic reports on the dozens of advertising and analytics companies that involve themselves in my web browsing, I think I would have been a little dismissive of Norton’s article on Medium.  Sure my information is being collected, but it’s not a big deal, right? After all, it doesn’t really influence my life.  Norton describing how he would play with people like toys, shaping their tastes and buying patterns, would seem like an exaggeration. Maybe some people are influenced by internet advertising, but not me.

Of course, this line of thought is the textbook example of third-person bias.  Psychologically, we are predisposed to think that things like advertising don’t really affect us – it only affects other people. But that’s wrong.  And if 30 different companies are monitoring my clicks when I visit The Atlantic, surely my data must be more valuable that I would have thought.

But as our readings make clear, big data is about more than just stalking us as we waste time on the internet.  It can be used for everything from optimizing HIV/AIDS research to misidentifying photos of concentration camps to shoring up the glass ceiling by targeting ads for high-paying jobs at men rather than women.


So many elements of life are becoming data-driven and algorithmic, often without us even realizing it. And in a way, the problem with this is the same as the problem with the algorithms that track my clicks to determine my ad preferences. It is so unclear what is happening, what data goes where and how that data triggers events in our lives, it is impossible to know what needs to be changed.  And as these algorithms get more and more complex, transparency seems less and less likely.

No comments:

Post a Comment