Algorithms and the importance of context

· · · · · · | Articles | No Comments on Algorithms and the importance of context

A friend recently complained that Topshop knew exactly which pair of pants she’d looked at online at work, then advertised them on Facebook when she logged on at home. I mentioned algorithms to her, the ones which are at work every moment she browses the web, recording her every move and determining advertising campaigns for countless brands vying for her attention. She retorted: “the algorithm doesn’t know, however, that I am reluctant to spend $90 on a bit of polyester crap.”

 

Quite right. And how could it? An algorithm is merely a step by step procedure to solve a problem; a formula with a set of instructions to look for specific input. In this instance its instructions were to display advertising for Topshop on Facebook based on my friend’s browsing history. What device she uses to access Facebook is irrelevant; it could be her smart phone or a computer in a public library for all the algorithm cares. What matters is that she’s identified herself somehow: logged into Facebook or a Google account perhaps, anything that is using cookies to track her browsing.

 

Retailers like Topshop have the capacity to tailor their online presence specifically to what they think we want to see, based on data supplied by web infrastructure providers (eg. Google), or social networks (eg. Facebook). The beauty of algorithms for any retailer such as Topshop is that they can personalise their Facebook advertising campaigns specifically to what you seem to be interested in.

 

But this system is flawed, of course: context is lacking to truly understand why someone clicked or viewed or liked something. Without knowing the user’s motivation for certain behaviour, the environment the behaviour occurred in, or the particular emotions the user was feeling at the time, the data resulting from the user’s input can’t possibly accurately reflect what he or she wants.

 

Algorithms are only interested in quantitative input: they look for a “Like” here or a keyword there. But we are not defined by our likes and our keywords, as much as Google, Facebook and Amazon would like us to be. Identity is a much more complex phenomenon; a blend of experiences and memories; culturally determined beliefs, socially learned expectations and personally developed ethical values; opinions, suspicions and conclusions; faith and hope.

 

Identity is tidal, shifting depending on many factors; one being context. For algorithms there is no context. There is only input.


No Comments

Leave a comment