"COMPUTER POWER TO THE PEOPLE! DOWN WITH CYBERCRUD!" - Theodor Nelson
Monday, August 16, 2004
I am addicted to achieving great design and I love reading about how to do it better. To me, design has always been about getting the most out of the least. I think this is what we admire most in art and music. Artists and musicians making the most out of the least. If you look at the greatness of Bach and Mozart, they were masters at making the most out of a small number of ideas. It's the ingenuity of stretching those few (great) ideas to their maximum while delighting the listener. I think of program design the same way. I want to write the least amount of code to get the most functionality. So, I tend to favor techniques that allow me to type less and don't make me repeat myself. This is why I favor dynamic languages like Smalltalk, Lisp, and Ruby. I also favor those languages because they give more tools to get the most out of my ideas.
Now, there is a huge push in the academic circles of computer science for mathematical purity. I even see this in the industry. I think the static typing advocates like the mathematical purity of typing. Haskell is one of those languages loved by academia. I know I was impressed with the typing system of Haskell initially. It's typing system is a lot better than what you see in Java/C++. It's cool until you get past some of the simple examples and things start to get a little messy. The point I'm trying to get to here is that computers are built on simple mathemtical principles outlined by Turing and Von Neuman. What a lot of folks want to do is extend that model to everything a computer does. But, just because what you are building is based on pure mathematics, doesn't mean the whole system needs to be. I know it sounds absurd. But, buildings have a basis in mathematics too. But, there's that squishy area where user requirements and construction and design meet. It's this squishy area that can not be expressed in math. It's the area where years of contracts have made the art of building business software complicated and hard. The languages we use and ideas we have should be amendable to this fact. I think the dynamic languages are all built on simple mathematics as well, but they relish the fast that the real world is squishy.
So, my point is that I see a lot of designers trying to make great design by making them mathematically perfect or at least adhering to mathematical beauty. I think while this is a lofty goal, they tend to break down in the real world. The best designs make the most out of the least number of moving parts. These designs do not target mathematical purity as the goal. But, the funny thing is that these designs tend to be beautiful in a mathematical sense. Thus, achieving the mathematical purity goal anyway. When I look at good designs (ala Seaside). It seems the goal was to get the most with the least. They certainly achieved it since I believe Seaside is still the easiest and takes the least amount of coding of any of the web frameworks. Also, if you look at the code, there is a sense of wonderment and mathematical beauty. I would be curious as to what design goals they had for it.
The real world is squishy and has lots of rules that don't seem to make sense. Let's embrace this randomness and chaos instead of shunning it because it's not pretty. Let's make our designs of the real world beautiful by writing the least amount of code.