RSS Feed
News
Jan
24
Recommendation Engines
Posted by Troy Newport on 24 January 2013 08:33 am

Helpful? or Diversity Suppressor?

Search engines and social platforms have been stridently moving toward a “recommendation engine” model over the past several years.  That means they track past behavior and try to figure out what information to serve to you based on that past behavior.  In other words, instead of giving you a ‘true’ set of results based on their algorithm, you’re provided with a set of results based on what they think you want because of what you and others have done in the past.

How do you feel about that?

In order to provide recommendations based on your past behavior, that means Google, Facebook, and other platforms using this model must store data about your past online behavior.  Searches you’ve done, websites you’ve visited, how long you’ve stayed there, etc.  Privacy advocates aren’t too happy about that, but the companies in question promise the data is stored without identifiable information attached. 

But beyond privacy, what do you think about the fact that results are based on your past behavior?  This model assumes that your tastes will never evolve, and assumes they can predict what you would like. 

I don’t know about you, but I have a diverse set of tastes.  When I go my favorite restaurant (we’ll call this restaurant ‘Google’) I may have a favorite menu item, but I may want to try different things on the menu just for the sake of trying something new.  But if the restaurant staff immediately serves me Two Friends Panang every time I walk through the door, I won’t have a chance to be exposed to other things on their menu.  Other things I may like better.

Even worse, what if my favorite waitress (we’ll call her ‘Facebook’) forced me to eat what she wanted me to eat every time I came to the restaurant?  I’m pretty sure I would build up some resentment over time.  I don’t like people making decisions for me. 

So while I understand the need for fancy algorithms to aid us in accessing the volumes of information we seek every day, do we need all these “personalization” algorithms?  And anyway, can you fancy schmancy programmers really guess what I want?  Am I that predictable?  When I log onto iTunes and see the ridiculous suggestions displayed by their Genius, the answer is a resounding “no”.

 


Comments (0)