Yeah, I know, it sounds kinda obvious. Read on. That seemingly obvious statement is not actually what I mean.

I'm not talking about a specific programming language. I'm talking about coding as a practice in general. We're used to understanding common languages, from informal ones (like human languages such as English or Spanish), to more formal or restrictive ones (like math).

But all in all, a language is anything you can use to express ideas with. In that sense art is also a language: for a musician music would be his/her language, painting, and so on. For expert chess players, the rules of chess would constitute a language.

And even when something like this seems obvious (as stated at the beginning), some people in the software industry sometimes fail to recognize the coding activity as a language on itself.

To give an example: let's say you have to communicate an idea about something you want to implement. You can write a perfectly crafted design document (with architecture diagrams, and such), and then send it for review. But is that really the best way to communicate a change? In order words, is it the best language to use when you want to get an idea across?

How about publishing a code review instead? Sometimes instead of just saying 'it would be good for your tool to do X', it's way more efficient, to just go ahead and implement it.

In other cases, a design document might not even be the best suitable language choice to communicate your idea. In the case of a prototype, I would be more confident on seeing actual working code. The same applies for a proof-of-concept. A text document, or presentation slides would always compile and make sense from a high-level perspective. But there are certain things that you won't find out until you dive deep into the task, and that's when coding becomes a more effective communication tool.

After all, the purpose of a pull request is not to merge code into the main branch. It's to communicate an idea using code. When you submit a new pull request you're communicating something. That something, doesn't necessarily need to get merged.

This changes the definition of success for pull requests. Most would argue that a code review is successful if it gets merged without causing issues in production. This makes sense, of course, but it's also limiting. If instead we'd say that a pull request is successful if it effectively communicates an idea, this would bring more value. You can have a first code review submitted only to present a proof of concept. The code could be scrappy and by no means anywhere near what the code base would consider acceptable. But it will be something more concrete to communicate what (and how!) is you're trying to achieve. This can be referenced in mails, or other docs, and then the team can go ahead with the real (and high-quality) implementation. The original code review didn't get merged, not even seriously scrutinized (like a real pull request would), but it still got the job done.