Assembled Best Practices for Software Development

I've been curating a list of my favored practices for some time now. These ideas have come from a variety of sources (books, blogs, coworkers, myself, etc.). These make up some of the best ideas and practices that I have seen in software services. There are a whole lot more that I could talk about, but these are the most important that I think everyone that works in software development should be familiar with. Without further ado:

Understand the foundation. This can be one of the most glaringly obvious, yet oft overlooked, points of an application. Every application has a goal; Something that it aims to do better than anyone else. Generally, that goal leads towards two or three core competencies that the whole platform will revolve around. Make sure you know what those competencies are, and don't ever sell them out, or outsource them. Your evenings and weekends depend on it.

Features are like pets. They're fun! But, they have to be groomed, cared for, and carefully maintained. As such, like pets, you should be very careful about which ones you adopt. Whenever someone says, "We should add.." always start with "no". Make any non-core feature work to be in your application.

Scratch your own itch. The next big thing is a lie, and doesn't deserve your attention. Make something that saves you time, and makes your life better. There will always be people trying to figure out how to force the next application to blow up in to something big. Instead, just figure out what your itch is, and scratch it.

Always Be Shippin'. Too many projects die before they even get off the ground. Figure out what the minimum viable set of features are, and then ship it as quickly as possible. After that, iterate on the design, and make sure you get it in front of your (preferably paying) customers, right away! They'll love you all the more for seeing your software evolve in front of them. Speaking of which:

Love your audience. Your software is nothing without your customers. Focus on their needs and collaborating with them, more than the amount of money, or the contracts that they bring in. When they give you information, respond to it. The most important lesson of sales is to remember this one simple truth: People don't buy a product, they buy an emotion. Make them feel good over and over, and you'll have a customer for life.
The customer always believes they are right. But, they're probably not. That's right. Conventional wisdom is wrong. That doesn't mean you treat them as if they're addled. On the contrary. You should always treat your customers with dignity, and respect, if not admiration. Yet, even when you do that, you should help them understand why your software works the way it does, and don't let them dictate terms that will lead you in to a feature haze. That way only leads to heartbreak. Listen to your customer, ask pointed questions, and act only when it is in the best interest of your entire audience. Never become myopic, as that always has an effect on the picture as a whole.

Build simple systems. Some will find it frustrating or condescending, but building three ways to do the same thing in your system doesn't make you nicer, it makes your system more obtuse. Find one way to make your system do the best possible work, and make it as intuitive as you possibly can.

Estimates are pure guesswork. Don't fall in to the trap if you can help it.  Keep shipping features as you are able. If someone makes you guess how long it will take you to implement a particular feature, imagine how long it will take, then weight that by somewhere between 1.5 to 5 times that number, depending entirely on your own personal expertise and the complexity of the system at hand. If you think something that you're not very familiar with will take you a day and you're dealing with a complex system, don't feel bad about saying five days. If it's something you're very comfortable with it's not complex at all, and you think it will take two hours, go with three. The key is this: Always under promise and over deliver. People may think you're daft when you guess five days, but they'll think you're a miracle worker when you deliver it in two, instead of grumbling about it taking twice as long as one.

Remember these three things:
  1. There are things that you know that you know.
  2. There are things that you know that you don't know.
  3. There are things that you don't know that you don't know.
That last category is the largest of the three, guaranteed.

Finally, the best way to learn is to practice, scrutinize, adjust, and repeat. Don't just keep doing same thing that you've always done, else you'll always get what you've always got.

Code on. 

No comments:

Post a Comment