I For One Welcome Our New Robot Overlords

Programmers and system administrators love control. We like to micro-manage eensy-bitsy details so they’re just how we want them, even if the defaults would be perfectly within a normal person’s parameters.

As a Linux sysadmin, I find myself particularly this way when it comes to my servers. I hand-configure all the individual pieces of software and lovingly place them together. This gives me an incredibly deep knowledge of the system that’s useful for debugging, but it has its issues.

As your fleet of servers grows, it becomes increasingly more difficult to run shell commands and edit files across all machines. Tools can make this better, but it’s still imperfect. And in times of high stress (e.g. the website is down), humans make even more mistakes.

But that’s why we have robots. Robots make no mistakes.

Which isn’t really true. Robots make all sorts of mistakes, because you gave them imperfect programming. But we can test their behavior during a low-stress time, when we’re at our best. Then, when the shit’s hit the fan, we can trust the robot to stay cool.

Of course, I’m not talking about actual robots - I’m talking about automating system administration tasks. Yes, those very tasks that we so love to do ourselves. But in addition to us being mistake-prone, we’re not easily replicable, while our programs can be copied and run simultaneously on hundreds on machines.

I know you think you’re awesome (you are!), but look at the facts. Look, and weep for your humanity.