Dancing to Silent Algorithms

More and more, we live our lives according to the unknown auspices of machine codes

I’ve been on the phone to my bank again. This time I spoke with Rob, who sounded Canadian. I guess the advantage of doing business with the world’s local bank is that it’s always business hours somewhere. I was talking to Rob because my cash card was refused again. I don’t know why. I can never know why, and that’s the problem.

“Are you travelling at the moment?” asks Rob. Well, yes and no. Right now I’m living in Barcelona, but last week I was in Groningen. Before that I was in London, by way of Manchester and Brighton. Before that, Bangkok via Penang, Istanbul, Amsterdam. I admit this is not normal.

Unfortunately, every time I do take money out of the bank, the bank has to decide if things are normal. It needs to decide if it’s really me at the other end. I have identity tokens and a PIN key that I can show to automated teller, a squat block of plastic, glass and motors, but really these are just talismans for an intricate machine brain that rests somewhere in a climate-controlled mountain fortress or a London basement mainframe or who knows where. This electronic brain is the automated teller’s manager, and my transaction must be approved by this manager before the teller can authorise it.

The manager doesn’t care too much for tokens or PIN keys — if either of these are wrong my request for cash won’t even make it that far. What the electronic brain consults is a vast database of behaviours, including my presumed location, my recent transactions, the kind of transaction I’m requesting, the time elapsed since my last transaction, the amount of money I’m asking for, the time and date of this transaction, and most likely dozens of other metrics I’m not even aware of. It weighs up all of these factors, and decides if I’m really who I say I am. It weighs up all of these and decides whether to give me money or not.

I’m constantly engaged in a kind of dance with the algorithm, one where I can’t hear the music

If the machine decides not to give me my money, it will refuse the transaction. And — this is rather crucial — it won’t tell me why. This means I’m constantly engaged in a kind of dance with the algorithm, one where I can’t hear the music, and the only feedback I get is when I step on my partner’s toes. I don’t know the rules of the algorithm, so I have to learn as I go. I also don’t know when the rules of the algorithm change.

This is, in effect, exactly what artists such as Timo Arnall and James Bridle have been taking pains to point out to us: that our lives are influenced by technologies not simply as objects but invisible systems that surround us, and whose architecture shapes the patterns of our lives. To live inside these invisible systems, whether they be insurance forms, loan requests, job applications or dating websites, we consciously adjust ourselves, providing the information we think is most suitable — or the least open to misinterpretation by the system.

The machine doesn’t really understand normal, not beyond a hump-shaped statistical curve that my behaviour should fall into. My job is to second guess the shape of that curve, and try to draw myself toward its centre. The bank’s algorithms are tuned to detect fraudulent transactions, and they’re not about to share the secrets of what sets alarms buzzing in the electronic brain.

The end result is that like a lot of security infrastructure, criminals know more about these algorithms than the public do. They can better navigate the machine brain’s various quirks to empty my bank account. As Bridle pointed out in MATTER’s Ring Of Steel, British criminals have already learned the evasive driving techniques needed to slip through the UK’s dragnet road surveillance, while most law-abiding citizens are unaware that the system even exists.

After I chatted with Rob, I discovered where the problem lay: I’d cancelled a transaction on one of the cash machines here in Barcelona, after it had queried my UK account. As far as my British bank was concerned, the money had been paid out, and all subsequent requests for cash were refused as pushing me over my daily limit. This grants me a tiny peek into the architecture of the banking system, namely that a cancelled transaction isn’t reported back to the central bank, at least on Sundays, and the ledgers of Spanish banks aren’t tallied up until the following day (whereupon my cash, still effectively trapped inside the ATM, was credited back to my account). It’s not much, but I’ve learned one more step in the never-ending dance with the silent algorithms.

Enjoy this essay? Click the Recommend button below.