A big theme to this was "It doesn't matter what your code looks like." The awkward part is that I was speaking about Clean Code and the importance of caring about the readability of code later that day.
Fortunately, a lot of people showed up for my presentation, and we had a really great time together. But I did get an interesting question afterward from someone who worked at a startup:
How do I reconcile "It doesn't matter what the code looks like" with the principles of craftsmanship?The developer believed in craftsmanship and readable code, and he was a bit distressed that his boss was saying that they had to do things fast and dirty. As a bit of a surprise to myself, I actually had a good answer to give (at least I thought it was a good answer):
Treat the version 1 product like a prototype.Let's explore this a bit more.
The Importance of Moving Quickly in Multiple Directions
I understand the startup mentality of "it doesn't matter what the code looks like" -- but with full understanding of the consequences of that approach.
The problem is that we only have an idea to start with, and we're not even sure if anyone will be interested in it yet. So we put together just enough code to get things working. And a lot of times, we don't even get that far; we just create a landing page to see if anyone is interested in the product before we even write a single line of code.
And once we start coding, we cannot invest too much time or effort into this. If it is a complete flop, we discard the code and move on to the next idea. We don't know what we need at this point. We're really just exploring. Ultimately, we're working toward a minimum viable product, which is just enough functionality to be useful to someone.
Even then, we don't know what feature will really resonate with our users, so we have to be ready for unexpected responses and to move quickly in another direction.
Just a Prototype
The only way we can use this approach is if we treat this code as prototype code -- even if we are releasing it to the users. This is our exploratory code -- where we figure out what excites our users and differentiates us from the rest of the pack.
There Will Be Problems
When we write code like this, there will be problems. We will have bugs that are hiding in the code. We will have issues scaling for more users. We will reach a point where it is difficult to add a new key feature.
And this is the point were "it doesn't matter what the code looks like" really starts to cause us pain. The best thing to do is discard what we have and write the code for our real product based on the prototype.
Clean Code Does Matter
It's really easy to fall into the trap of thinking that clean, readable code doesn't matter. After all, if we can be successful without it, who cares?
This works short term, but things fall over in the long term.
Once we have our viable product, and we start coding "for real", we need to pay attention to craftsmanship principles. We have to make sure that we don't introduce bugs in our code. And we do this through automated testing (the proof that our code does what we think it does) and through readable code (if the code is easily readable, then the bugs have nowhere to hide).
In addition, since we now know what features make our application successful, we can leave spots for abstractions and extensibility points so we can easily add new features.
And this is where we start to think about the long-term viability of our product. If we treat our version 1 as the real version rather than the prototype, we're just setting ourselves up for a lot of pain. But if we take the learnings from the prototype to then create the actual product, we can reap the benefits of our quick, maneuverable process and use them to create a sustainable product.
In the Wednesday keynote with Max Lynch, the focus was really on starting many companies (one after the other, and sometimes several at the same time) based on ideas that we don't know will work. We need to figure out what resonates with our users quickly so that we can focus our efforts.
So, my advice to people who are coding "fast and dirty" and don't feel right about the process: think about this as a prototype. It is an exploration of what might be useful. And just because we release that product to our users does not make it any less of a prototype. Expect to discard this code once we have gotten what we need to know from it.
It Does Matter What Code Looks Like
Ultimately, it *does* matter what the code looks like. Having readable code is what makes it easier for us as developers to make changes to meet our users needs. If we can't quickly meet our users needs, then we've failed. If we don't go back to replace that prototype code, then we'll end up falling over at some point.
If we create a mess so that we can get some learnings out of it, that's fine. But we can't leave things that way. For long-term viability, we need to think about the quality of our code which ultimately impacts the quality of the product.