Julia Evans

Fear makes you a worse programmer

Yesterday morning, I asked on Twitter:

Does anyone have good writing about fear + programming (and how being afraid to make important changes makes you a worse programmer?)

and

I feel like there’s this really important line between caution (a++ excellent) and fear (which holds you back from doing necessary work)

A lot of super interesting discussion ensued, and I’d like to talk about some of it.

Before I start, Ryan Kennedy linked me to this slide deck of a presentation he gave called Fear Driven Development which I absolutely loved, and I think you should look at it. I think my favorite sentence from that presentation is “Fear creates local maximums.”

I find that when I’m afraid, I become super conservative. WE CANNOT POSSIBLY MAKE THIS CHANGE WHAT IF IT BREAKS?! And this means worse software! It’s actually kind of disastrous. If you’re scared of making changes, you can’t make something dramatically better, or do that big code cleanup. Maybe you can’t even deploy the code that you already wrote and tested, because it feels too scary. You just want to stick what’s sort-of-working, even if it’s not great.

Better tools & process => less fear

A lot of people brought up tools and processes. (in particular the fantastic Kelsey Gilmore-Innis, who has new-to-me things to say about better processes for testing code every time I talk to her)

Kelsey:

I know you’re talking more psychologically, but this is one of my main reasons I believe investing in tests early is important

[…] obsessive monitoring, CI, canary deploys, chatops, dogfooding, selfserve infra

So! Here are a few ways tools and processes can make us less afraid:

Version control means that you can make changes to your code without being scared of losing the old version. This one is so basic to me now – I can’t even imagine how afraid I would be if I were programming without version control.

So many people mentioned testing as a way to build confidence. My favorite thing I’ve ever read about testing is this book Working Effectively with Legacy Code, where every chapter title is something that’s scary or difficult about working with legacy code. (He defines “legacy code” as “code without tests”, to give you a flavor). For instance: “I Don’t Understand the Code Well Enough To Change It.“, “How Do I Know That I’m Not Breaking Anything?”, “I Don’t Have Much Time and I Have To Change It.”

This tweet from Uncle Bob Martin, explains this idea pretty well:

Even with good design, w/o tests you will fear change and so the code will rot. With good tests, there’s no fear, so you’ll clean the code.

But testing and version control are not the only tools we have! We can also build

  • QA environments where breaking things is totally allowed and encouraged
  • deploys that go out a little bit at a time
  • the ability to roll back a deploy easily
  • QA teams, whose job it is to exhaustively test software
  • tools that will email you if your program throws exceptions

and lots more.

Fear of retribution (and blameless postmortems)

But tools and processes are absolutely not the only thing. Even if I have amazing tools and QA systems and the best deploy tools and well-testing code, I’m still going to make mistakes sometimes. And what happens when I make a mistake is really critical.

Etsy and Google and Stripe (where I work) all have blameless postmortems. This means that if you make a change and that change breaks something, people talk about what happened by focusing on the change and the facts, not on blaming you. (“what about that change caused a problem?” instead of “how did Julia break it?”)

I also realized that this goes much further than programming, and Marc linked me to this amazing site about restorative justice, which you should also go read.

So if you blame people for breaking things, they’ll be more scared to make changes in the future, and you’ll end up with worse programs. Huh.

Irrational fears

One last thing that that Fear Driven Development talk talks about that really resonated with me was – some fears are irrational, and that they can infect other people. If you do a deploy, something goes wrong, and you figure out the cause and fix the problem and nobody yells at you, hopefully future deploys should not be scary!

But because we are only human and not Logical Robots, sometimes they still are, and maybe you’ll feel nervous about doing deploys for a while until you see that things are really usually fine.

I think there’s a lot more to be said about irrational fears, and I would be interested to hear more.

This year was the first year that I worked on large software systems that affect lots of people, and it’s been scary sometimes! Next year will be the second year, and my plan is for it to be easier =)

(thanks to Maggie Zhou and Kelsey Gilmore-Innis and Melissa Santos and many others for all saying excellent things and making me have new thoughts!)

A couple of more talks / posts about fear that I enjoyed:

Reproducing awesomeness Programming doesn't belong to men (it belongs to me)