Algorithms

Software algorithms get a lot of bad press at the moment. They’re accused of forcing us into siloes and echo chambers, of shaping our views and attitudes, and of generally being the earthly manifestation of Lucifer.

But that’s not the real problem. The problem with software recommendation algorithms is that they don’t work very well.

We’ve all had the experience of watching Netflix, watching a couple of episodes of a show, and then being spammed to watch a bunch of similar programs. Or listen to a certain type of music, and then the software decides that it’s all that you want to hear. Or, perhaps most insidiously, interacting with someone on your social media of choice, and then seeing every little comment they ever make appear in your stream.

I don’t know about you, but this actively repels me. I can’t think of the last time I paid the slightest attention to what an algorithm recommended to me. All of my best discoveries have been recommendations from friends or sheer serendipity.

Why? Because (at least for now), algorithms can’t understand why you like what you like. Maybe the reason you like a certain band is the lyrical content, but you can’t stand the synthesisers. An algorithm can’t really determine that, so brace yourself for a lot of New Wave.

As someone with quite varied tastes in music, film, and things generally, this is a source of great frustration. But at least I know that I can’t be pigeonholed by Mark Zuckerberg just yet.