a metal framework tower holding up power lines is silhouetted against a dark sky that appears to be photoshopped because it is purple
Unrelated image from pexels.com to make this post look nicer in social media shares.

A little bit of self-doubt is good for you! That might sound weird but hear me out. Doubting yourself just enough to accept the fact that you don’t know everything and will be wrong sometimes is incredibly useful in software development. So many projects have died when the thing that everyone assumed would be easy (or at least doable) turned out not to be. If you can accept that you could be wrong and do a little double checking, you can save yourself so much pain.

While we’re on the subject, if you ever hear the words “it’s done, we just need to integrate it” you should be very very afraid. Not only is the thing not done, but it’s not going to be done any time soon. I admit I’m generalizing from a small sample here, but I’ve never seen the assumption that integrating a new feature will be easy turn out to be accurate.

In general, if you’ve never used that API before, don’t assume it will be easy. Read the docs really thoroughly too – just because it sounds like that API does what you need doesn’t mean it actually does. I’ve been burned by that before and you probably will be too. Same thing for libraries you haven’t used before. Just because you think a library should work a certain way doesn’t mean it actually does. In case you weren’t already worried enough, don’t forget that sometimes the docs are outdated, missing, or just plain bad, too. There’s more than one reason building a very simple proof of concept first is such a good idea.

Like I said earlier, don’t assume you understand the problem either. Talk to people until you’re sure you understand why that feature should exist, not just what it should do. Talk to your team lead, talk to other departments, ideally talk to someone who is actually going to use the thing (I know this isn’t feasible for everyone but it’s really great if you can manage it). The assumption that you’re right about what people need kills projects too. You would think that would be an obvious thing to check, but judging by the horror stories in the news about massive government projects that turn out not to do what the front line staff actually need, it’s not obvious at all.

Seriously, I can’t stress enough how dangerous assumptions are to projects. That’s how projects end up months if not years late and massively over budget if they get delivered at all. Refusing to admit you could ever be wrong kills projects too. The real problem, underneath the assumption that integrating that API will be easy to integrate, is the assumption that you’re right and don’t need to check.

In addition to killing projects, failing to question your assumptions can also lead you to make a complete fool of yourself, a la google bro. If he had questioned any of his assumptions, if he could have faced the possibility he could ever be wrong about anything at any time ever, he would be a far better engineer. As it is he clearly can’t be trusted not to make wild assumptions that will doom his projects, and if you can’t trust someone to check their own assumptions you simply can’t trust them to work as a senior engineer. One simple definition of engineering levels is that junior developers need adult supervision, intermediate developers can work on their own, and senior developers are adult supervision. If you can’t bring yourself to accept that you could ever be wrong about anything, then you can’t competently supervise anyone.

Too much self doubt is obviously completely unproductive, and you can never know everything there is to know about your project before you build it, but a little self doubt is enormously useful and tremendously underrated. In tech we seem to place so much emphasis on always being perfectly sure of yourself but not only is that an unreasonable goal, it’s not even useful. What’s useful is understanding that the most brilliant writers still need editors, and the most brilliant programmers still need input from their teams. Nobody is right absolutely all of the time, the word for people who think they are is “delusional.”

Admit you could be wrong and you’ll end up with the right answer a lot more often.