James Bridle: Abracadabra algorithms

James Bridle is a writer, artist and technologist from London currentlyliving in Athens. In this short video James shows us behind themagicians curtain, taking the magic out of algorithms and revealing thatsoftware is not programming itself.

James Bridle is a writer, artist and technologist from London currently living in Athens. In this short video James shows us behind the magicians curtain, taking the magic out of algorithms and revealing that software is not programming itself.

 

'Algorithm' has become this magic word that can open closed doors, make sense of large amounts of data and decode hidden patterns. How would you define algorithms? What are they for you?

It is interesting how recently the term 'algorithm', this term from computer science and mathematics, has entered into much wider social discussions, which is great on its own terms because it means that we are starting to discuss the mechanisms which underlie the technologies. But it's also dangerous if it's treated as some kind of magical special thing that we can't really understand. It's just software, it's just the process of code operating. An algorithm is simply a step-by-step software process that does something. So it's not a mystery. The reason they often accrue this mystical intonation is that we're often not aware that software is making decisions. And that's the crucial thing about what an algorithm does. It essentially makes decisions of some kind. And those can be totally innocuous or they may be very serious. A piece of software running a large system may decide whether you get a mortgage for your house, or potentially at some point it may decide your citizenship. Decisions are being made by the software and therefore we need to question that software and understand its function.

The other critical thing of course about it is that so far, software is not programming itself. Embedded within software are huge numbers of explicit, and also implicit, decisions made by its maker. Some of those may be, ‘I want the software to do this’. But other ones may be far more subtle. They may have been encoded by the kind of underlying biases of the people who designed it so they encode potential gender roles, or racial roles or these kind of things within the software that makers may not have even been aware of. But through a kind of critical analysis of software, by not treating the algorithm as magic but really as something that should be challenged and understood, as we would challenge and understand the formation of new laws in society, we can actually start to challenge some of those underlying biases and perhaps change them.

Metadata is yet another fashionable term. What is metadata, how can you define it and is it becoming a dirty word?

I don't think metadata is becoming a dirty word any more than algorithm really. It's quite well-defined, what it is. It's data about data and that's a really important concept to get across. The question is, how do you choose to get this concept across? Do you do it by essentially repeating the kind of sins of the thing that you're arguing against? Or do you use the word to actually talk about other ways in which we can operate, deal with the world, make work that doesn't rely on that as a kind of basic thing. I don't have a problem with having a discussion around it. I just don't want it to necessarily be the motivating factor of the work itself.

First published on November 27, 2015