Automation Limitations with Judy Johnson

**SPEAKER_1** (0:01)
Automation is key when it comes to scaling your engineering efforts. But can you automate yourself out of a job?
On this week's episode of Dev Interrupted, I brought in tenured senior software engineer and automation expert, Judy Johnson, to answer my questions about what we shouldn't be automating and how to convince your business to take on the upfront costs of an automation initiative.

**SPEAKER_2** (0:23)
This episode is sponsored by Linear B. Give your dev team the power to improve with team-based metrics, high-risk code alerts, and the world's first project board based on real-time Git activity.
Sign up free at linearb.io.

**SPEAKER_1** (0:38)
What we're here to talk about today is automation, right? And you did some great presentations, and I've read some of your stuff about automation.
And I would love to get your perspective. You can see things in the world like everything is kind of being automated right now, and maybe it's normal, maybe it's not normal.
How did that timeline look for you? Like, were we always automating stuff back in the day? Is this a new thing that's happening now? Like, how do you see it?

**Judy Johnson** (1:15)
Okay, so yeah, so I mean, obviously with a computer science degree and being a program for all those years, like that's what programming is. It's automating things, right? So whether you start with automating a calculation or whatever, then that's really automation. We're not on advocacy anymore.
So yes, but as we progress, obviously, like for example, even going from assembly language to Fortran, to Java, with C I guess is in between, you're getting farther and farther from the machine. And as you get farther from the machine, is what makes it more... Yeah, I mean, all the way it's automating, but it makes it easier for us. So now that we kind of have it down how to write code, now we're automating the processes more, right?

**SPEAKER_1** (2:07)
One thing that I read from your presentation, you had a presentation called Automation Justification.

**Judy Johnson** (2:15)
That's right.

**SPEAKER_1** (2:16)
Which is a great title. Love that title.

**Judy Johnson** (2:18)
Thank you.

**SPEAKER_1** (2:19)
There's one item on there that I saw. You had something like Preserve Enterprise Knowledge or Knowledge Preservation. That caught me because you have, okay, why do I automate? Well, it gets to save everybody time. Customers could be happier because there's not as many bugs in production, MTTR, meantime to restore all of that.
But that bullet point, Preserve Enterprise Knowledge, caught me off guard. What does that mean?

**Judy Johnson** (2:52)
Yeah, so like I was saying earlier with the DevOps, where you write it and then you script it, I think at some point, I am saying I'm the tester. Let's always go with automated tests because that's what people like.
I know that I have to check, make sure that numbers and dashes and question marks and underscores and carrots can all fit into that field and be treated appropriately. If they're illegal, get the right message and stuff like that. So I know that and I test all those things. They need me to say, what were these subtle little things you were testing? What were some of the edge cases that you wouldn't think of if you were just an engineer writing tests for yourself? So if you would extrapolate that to anything else anybody does, then you know that you need to automate. So the other thing I was going to bring up, and that's because it's kind of what I do, is security-related stuff.
So one of the things, two of the big problems that we have with our engineers that I'm sure you've seen them both is works on my machine, and I lost my changes.

**SPEAKER_1** (4:01)
Well, works on my machine is the worst.

**Judy Johnson** (4:05)
Works on my machine.
So what we do is you can automate your infrastructure, right? You can use Puppet, you can use Chef, you know, Ansible Salt, Terraform, whatever. So you're actually setting up the machine in an automated way so that your development, your test, and your production platform are exactly the same, and then nobody has that excuse anymore, right? So that's a way that automation is really useful. And in terms of losing data, that was just a configuration management using GitHub, GitLab, whatever. And I know you guys are pretty familiar with GitHub, right?

**SPEAKER_1** (4:43)
Yeah, absolutely. I have to tell you a works-on-my-machine story. We all have to. The most epic one that I can remember, I was probably, yeah, like 23 years old, 24 years old at this time. It was my second job out of school. I won't name where this is or any of the names.