A very dear friend of mine and I had a conversation yesterday. Somewhere in the conversation, he said (and I'm paraphrasing) "I'm going to make my own encryption algorithm, and it's going to rock." This struck me as odd so I responded "Why would you try to do that when so many really good encryption algorithms already exist?" By asking this, I didn't doubt his ability to write an encryption algorithm, per se, but more that the AES algorithm was the life's work of Joan Daemon and Vincent Rijmen, two doctors of cryptography. His response was that he just wanted to tinker, which is a valid response, and one that I'm sure every developer has had at some point in their career (and hopefully still have some variant of it).

So I thought long and hard about that. It's good to tinker. I remember the first time I wrote a linked list in C. Advanced concepts like pointers suddenly just made sense. I understand how you could make an array that changed its size. It was a to me. I optimized it as much as I could, and I still have that very code on my system today. Today, I have no need for it, and haven't looked at it since.

Why? Because I know how it works now. I can use someone else's linked list code, one that I don't have to maintain, and one that has had much more work than I'm willing to put into it to make it fast and efficient. I've moved on from that. I could spend my time focusing on how to make my own data structures and make them efficient, or I could make real software using someone else's library, and devote my ingenuity to that.

While still talking to my friend, I used the analogy of building a birdhouse (and did my best not to let my high school shop class experiences taint the analogy). If you write up plans for a birdhouse, and then write up plans for a hammer, and then write up plans for a nail, and then write up plans for a saw, then you'll have quite a mound of work to do. Chances are, you'll probably end your birdhouse project with a half-made hammer, and lots of grand plans. Everyone does this at least once or twice in their career, and if you're a developer, and you claim you don't have a single half-made project somewhere, I will say that you are a liar to supplement your developer title. --:-)

I no longer recompile my linux kernels from vanilla, or even the distro's patched version. I don't need to. The speed difference isn't all that dramatic, and frankly, there's really no purpose. But I know how to do it, should I need it, and I understand how the kernel works. It's not magic to me anymore.