Where Y.A.G.N.I. falls short

I am a big believer in the YAGNI (You aren’t gonna need it) principle. But it applies to features, not coding practices. It is more of a way to say: Code the MVP (minimal viable product) first. However, when applied outside of that realm, it can be harmful. For example, when talking about ‘how’ to write code, it often doesn’t apply because there are so many instances where you are gonna need it.

YAGNI is for what you code, now how you code.

With how you code, you are gonna need it:

  1. Quality – You are gonna need quality.
  2. Maintainability – You are gonna maintain it.
  3. Replaceability – You are gonna replace it.
  4. Testability – You are gonna test it
  5. Security – You are gonna need security.

Quality – You are gonna need quality

Code should always be written with quality. There are a lot of principles that are guidelines for quality, starting with SOLID. If you struggle to understand the SOLID principles or think they are too general, then I would suggest you follow my SOLID Training Wheels until you understand them better.

You may have heard of the Quality-Speed-Cost Triangle. The triangle is used in manufacturing with the following rule: You can only choose two of the three. Because code is like manufacturing, some believe this triangle applies to code. It doesn’t. Software is not physical, it is virtual. Once you create a piece of code, you can run that code a million times in what a human would consider almost instant.

You can use the Quality-Speed-Cost Triangle with code, but with code, the triangle does not have the same rules. For example, the rule that you only get two doesn’t apply.  Why? Because Quality is the only way to get speed and low cost.

In code, the rule is: Quality gives you the other two.

Unlike manufacturing physical products, software actually gets faster and cheaper when you increase the quality. You can’t have code speed without code quality. You can’t have low cost with code quality.

So focus on SOLID. The S (Single Responsibility Principle) and I (Interface Segregation Principle) both actually mean smaller objects (mostly classes and methods) in code. Smaller building blocks lead to faster code. When you write with smaller building blocks, there is less duplication. Every line of code has a cost to both create and maintain. Duplicate code destroys speed and raises costs. So smaller or ‘single’ will always be cheaper and faster.

Maintainability – You are gonna maintain it

If your company goes out of business (or your open source project dies), maybe your code isn’t maintained. But is going out of business your goal? If not, then your code is going to last. Every line of code has a maintenance cost.

The smaller or more ‘single (S in SOLID)’ the code is, the easier it is to unit tests, the less likely it is to have bugs, the less likely it is to change (part of O in SOLID), and the more like it is to be finished and never touched again. If most of your code is SOLID, small, unit-tested, replaceable building blocks, your maintenance costs will stay extremely low. Finished code leads to low maintenance costs.

Replaceability – You are gonna replace it

Systems die and get old. Code that uses systems will die with those systems. If the code is like the arteries in a human, entwined in everything everywhere, there is no replacing it. If the code is more like car parts, you can replace it with some work. If the code is more like computer peripherals, where you can add them, remove them, disable them in place, and replace them, then you are going to be faster. Quality is going to be easier because replacing pieces is going to be cheaper.

In SOLID, the S and I make things smaller, and usually (though not always) smaller is easier to replace. The O hints that you should replace code instead of changing code. The L and D are all about replaceability. Replaceability directly affects quality and future cost. If you are using MS SQL or Oracle and suddenly need to scale to hundreds of database servers on the cloud, you want your repository to be replaceable so you can migrate easily to a cloud database by replacing your repository.

Many companies who wrote monoliths without replaceable parts are now suffering this reality, as they either fail to replace their code with modern cloud architecture or spend exorbitant amounts to rewrite.

Every letter of SOLID in some way hints at replaceability. S – single, and it is easier to replace a single thing. O means the code is closed for changes, and so any new functionality goes in new code that likely replaces or extends the old code, the L is about replacing parent classes with child classes, and the I is about having small interfaces that are easily replaced with other implementations, and the D is literally about being able to inject any implementation from one to many interchangeable (or replaceable) implementations by inverting the dependency (The D is dependency inversion not depending injection, but a lot of people are now saying it is the latter).

Testability – You are gonna test it

This is almost the same as Maintainability, but while similar, it is different. Unit Tests help finish and close code without bugs. The more tests you have, the less likely you are to break anything as you add features to your code base.

SOLID doesn’t cover testing. But other principles, such as TDD, and industry standards, such as having greater than 70% code coverage (I say you should have 100% and close your code), all indicate that testability is key to speed and keeping costs down.

If everytime a dev makes a change, they introduce a bug due to lack of unit tests or automated tests, the costs will grow and the speed will slow as work stops to fix bugs.

However, if the tests warn you of the bugs as your are coding, the cost will stay low and there won’t be bumps (aka bugs) in the road slowing you down.

Security – You are gonna replace it

SOLID also doesn’t discuss security, but you are gonna need it.

So many people said they were never replace their database and yet are trying now to replace their database with cloud options.

Law suites from breached data are not cheap. Trying to bolt on security is not cheap. Unless you made everything easily replaceable, then it can be a low-cost task to replace implementations with secure implementations later in the process. If code is replaceable, adding security becomes much easier.


YAGNI is a good rule of thumb to live by for what you code, i.e. features and the MVP. However, YAGNI, will ruin your code base if you apply it to how you code.


One Comment

Leave a Reply

How to post code in comments?