Days ago I came across an article from the series “The long read” of The Guardian. The article, titled “Crash: how computers are setting us up for disaster” analyzes the downsides of overly reliance on computer-controlled systems and automation in general. It’s an incredibly well written essay and I wanted to contribute my own two cents to the topic.
For those who don’t know me, I have been a Programmer, a System and Network Administrator, an IT Engineer and a Team Leader, and currently Head of IT for Telenor Digital. I have been pursuing automation for my whole professional life: every time I found myself doing the same thing over and over, I always tried to automate the process or at least streamline it. This attitude was one of the drivers for me to learn the Perl 4 and 5 programming languages in the late 90s; in 2004, when Configuration Management was not much of a thing yet, I started fiddling with CFEngine 2 to manage a set of servers in an integration environment at work; later, around 2011, it was Puppet and CFEngine 3; today I am ploughing my way through Terraform to manage infrastructures in Amazon Web Services with code. How much automation is something I live with, you got the idea.
During my career, I have enjoyed the upsides of automation and I have been bitten by the downsides of it, and considering both of them in my approach became second nature: I learned naturally that “automation is like a big hammer: it helps you shove a big nail into the wall, or it makes a huge disaster if you use it wrong”. I know that there are limits, areas where you still want human beings to have the last word, actions that you don’t want to be triggered without a second check by a skilled person with a deep understanding of the system.
This was true in the past, and more so today.
There is a word that we hear and use a lot: digitalisation. Different people give the word different meanings but let’s be frank, there is only one thing that companies mean when they talk about digitalisation: zeroing human interactions between customers and the company (whether customers like it or not) and automating everything that can possibly be automated; this allows companies to reduce their workforce, save more cost than they lose by reducing their prices (if ever), and thus increase their margins.
Some companies are pursuing these objectives head down. Headcounts are reduced in the name of efficiency, costs are cut here and there to try to focus on the one the thing that brings even more revenue: get those damn humans out of the picture. Oh well.
There is one thing these companies are not seeing and it’s well explained in the article. The more you rely on the computers, the more people overseeing the system lose their knowledge of the system, the more they lose their ability to manage it in case of a failure. The simplest of the examples used in the article is about phones: most of us have completely lost their ability to remember a phone number: our phones have the capacity to save thousands of them, we rarely type one on a dialpad so even our “muscle memory” or our “visual memory” to remember how to call a family member is gone. Should we find ourselves in an emergency where our phone is unusable, it would be pointless to borrow one to call, say, our partner, simply because we would not know which number to call. If this example seems too naive, you can also read about what was the contribution of over-automation to the disaster of Air France flight 447.
Automating a process and delegating it to a computer creates a paradox: where you needed, say, 10 skilled engineers to oversee the process you now need only two less skilled personnel to oversee that the computer is doing the right thing, but when your automation fails and things start going sideways you’ll need as many engineers, and even more skilled than the ones you replaced, if you want to get back on track before you start losing millions. And those skilled engineers will be very difficult to find because, hey! the system has been doing fine most of the time, they have not been interacting directly with the system for months and who the hell remembers how it actually works!
I’m not going to annoy you further: you have the article from The Guardian and I will never be able to match such a great piece of writing. One thing I’ll say though: automation requires wisdom, wisdom requires experience, experience requires people, smart people. If your digitalisation strategy (if you have one) is to replace many skilled humans with a few cheaper ones, then you may just being digging your company’s grave.