Why you should reinvent the wheel

Below, I will talk about game development specifically, but I think this can be applied to any type of development.

The main theme I want to concentrate on is: if you personally experienced problems which some "wheel" solves, then you will use it better and with more satisfaction.

Consider arrays for example. Do you remember very first time you really understand concept of arrays? Usually this is very close to the beginning of language study. But despite that, usefulness of arrays is immediately obvious. You dont need tens of variables to store multiple values, and of course cycle processing of this values is much easier, to say the least.

Now remember user defined functions. They also come early, but not so early as arrays. Functions benefits is not so obvious until your programs go beyond some 100 lines.

And then, pointers. Double linked-lists. Queue. Stack. All of that strange useless magic. Then classes arrived and OOP with that strange inheritance thing. Why do you need to block access to a structure field? Some of us were forced to all this stuff at university or college. But as time goes you begin to understand why all this exist.

There is, however, a different side of this. Many starting game developers, as I can see, for example, at gamedev.stackexchange.com tend to use heavily some rather complex frameworks for games. The problem is that unless you will create game without usage of some frameworks, unless you didn't try to solve problems which arise with you own hands, you can't quite understand even basic logic that stands behind framework, and can't use it effectively.

You don't use classes from the start. You don't use pointers and complex data structures from the start. Not even functions. You use variables and arrays. Then your code grows, you understand concept of a function. And so on, you use more advanced features, then you face problem and understand how that feature helps to solve it. Before that time, your usage of it would be wrong and unefficient. Similarly enough you don't start game development by creating WoW clone, you are making Tetris clone.

Of course there are exceptions. You don't write XNA youself. You don't write standard C library. But this are well built, long time tested, basic functionality tools from a major brand. They can be considered a part of a language. You don't write image manipulation library, it's not the core kwnoledge that drives your game. But you write your game editor, your game core and physics engine. Even if you fail, your experience will be invaluable.