QR code

How to Be Lazy and Stay Calm

  • Moscow, Russia
  • comments

Managementmanagement Zerocracyzerocracy

What frustrates me most in my profession of software development is the regular necessity to understand large problem scopes before fixing small bugs, especially if the code is legacy and not mine. Actually, it’s even more frustrating when the code is mine. The “deep thinking,” as they call it, which is always required before even a small issue can be resolved, seriously turns me away from programming. Or did turn me away. Until I started to think differently and encourage myself to be lazy. Here is how.

Sin City (2005) by Frank Miller
Sin City (2005) by Frank Miller

I wrote about this a few years ago in this blog post: How to Cut Corners and Stay Cool. However, in our Telegram group, where we talk about Zerocracy, some programmers keep asking me the same question over and over again: What should I do when the project is absolutely new to me, I have just 30 minutes, and the bug is very complex?

One of the core principles of Zerocracy is #NoAltruism. This literally means that you should always and only think about yourself and your personal profit. You should not try to improve the project, to increase its quality, to fix the code, or to refactor anything… unless you are paid for it.

First of all, when the task, which you are going to be paid for, is in front of you and you can’t understand how to solve it, don’t blame yourself. You are not supposed to be an expert in the legacy code you just opened up. Strictly speaking, you are not supposed to be an expert in anything. A project, unlike your mom, doesn’t expect you to be intelligent or tech-savvy. It needs you to close tickets.

Who do you blame, if not yourself, when the bug is serious, the code is messy, and you have no idea how much time it will take just to understand it, let alone fix it? Well, you can blame everybody around you, but first of all you should blame the code base itself. How do you blame it? You report its low quality by creating new tickets, which may sound like this:

  • “The class X is not sufficiently documented, I don’t understand how it works.”
  • “The method X is too complex, I don’t know what it does.”
  • “The algorithm X is messy, I can’t figure out what it does.”
  • “The library X is used here, but I don’t understand why you don’t use library Y.”
  • “The rules of class naming are not clear, document them please.”
  • “The principle of data organization is not obvious, document it.”

However, don’t make the mistake many programmers are making when we tell them that tickets are the only right way to solve problems. They start asking questions and seeking help in the tickets, just like this:

  • “How can I unit test class X, please explain.”
  • “Please help me create class X.”
  • “Where should I put class X, in which package?”
  • “Which library should I use for doing X?”

The project is not a school, it’s not interested in making you smarter or more of an expert in its code. Nobody will explain anything to you, because it’s a waste of money and time. What the project will do instead is fix its code base so that it becomes cleaner and more obvious for you and everybody else. Thus, never ask for explanation or help, ask for documentation and source code fixes.

What do you do next? You sit and wait, until those tickets are resolved. Who will resolve them? You don’t care. That’s a problem for the project manager. Maybe he/she will even assign those tickets back to you and it will be your problem to resolve them. But if that happens, the scope of work will be different for you. You won’t need to fix the bug anymore, you will have to document some functionality or refactor some module.

You will have other problems in this new and smaller scope. You will create new tickets, blaming everybody around you, and they also may come back to you. And so on and so forth. Ultimately, the scope of a ticket will be as small as it’s possible to fix in 30 minutes.

See the algorithm? I’m sure you do, but it’s very difficult to apply it to real life and real software projects, for a few obvious psychological reasons:

  • You are ashamed. You are trained to feel guilty when you are not smart enough. What can I say? Just stop it!

  • You are a perfectionist. You want to complete the entire ticket, solve the entire problem, and understand the entire scope. What can I say? This won’t be solved while the project continues to pay you by the hour/month. Once they start paying for results, this disease will be cured.

  • You have no passion. You just don’t care about the quality of code at all. You don’t want it to look clean, you can’t even tell what clean is or what messy is. You just want them to pay you by the end of the month. In this case you won’t even know what tickets to report. What can I say? I guess you have to try and find another job. Maybe a manager?

  • You are afraid. Blaming the project and reporting tickets may look like you have a negative attitude towards the code base, and people who created it, which is not true. Instead, your attitude is positive, since you care about it and want it to get better. What can I say? Make your tickets sound extremely polite and gentle. But keep reporting them.

  • You have no time. You have to solve the problem now and you have no time to wait for the resolution of those complaints you reported. What can I say? Blame the management and require more time. Much more time. But never blame yourself.

Software development is perfect territory for cutting corners, being lazy and remaining calm, because our work is often discrete and can be very incremental. Very occasionally it might not be possible to blame the project and put the ticket on pause until all your complaints are addressed. I can’t imagine such a situation though. If you can, please let me know.

sixnines availability badge   GitHub stars