Why Models Don't Work
To model something requires defining the most important parts of the thing to be modeled, finding or creating something that has those same important parts, then declaring that these objects will behave in the same way. This can not work. Say you want to model something as complex as a human being - what kind of processing power would it take? Why, something at least as complex as a human of course. Could I make a human model in clay? If I handed you a clay model of a human and a real human, do you think you could tell the difference? Probably. What about a computer model of human thought? Could you talk with a computer program and know that it wasn't a human? Probably. And to model an entire human, can you image what it would take to do that from the ground up?
How much more complex is the Earth? What does it take to model the Earth? How about the Earth's Climate? Do you think a computer model can accurately model the Earth's future weather? A computer that fits under your desk or in a very large room is still way to small and way to simple to model something as large as this planet.
"Sure," you say, "you could model those things that are important and forget about things that are not. As long as you model the important stuff, you can model anything! Anything!" No, sorry, the words I am putting in your mouth are wrong. That is the whole point of chaotic systems - variances too small to predict lead to changes too large to compensate for. I believe Wolfram addresses all of these concerns in A New Kind of Science.
So, in my own rambling way, what am I trying to say? All models are flawed. Knowing the flaws doesn't help you. You can get some sort of rough estimates of how something might work (ie about how much pressure to snap a human rib bone) but the only way to have an exact number is to measure it after the fact. The more dynamic the system, the more exact your estimate has to be, or else you will be chaotically off. The Butterfly Effect.
-Edward
PS And most models are too thin - they real should eat something and look more like real people do. Do they really look like most people do? And that answers the larger question on models...
2 comments:
Wolfram is half genius, half quack. I've learned to take a few things he says with a grain of salt (especially stuff in his latest book).
However what you say about modelling is quite true and this is in fact a major area of study. One of the really cool things about chaos theory is the way it flip flops between order and disorder as you continously restrict your region of interest. Fun stuff.
He may be half and half, but his book is all good. Well, the back jacket is, because I haven't slugged my way through all 10,000 pages of it yet. Or anything beyond the jacket actually. But, I think I figured out the complete contents from reading that and can therefore speak with authority when I say, "Yes."
The Limits of Knowledge has always been a topic of keen interest for me. Even as little as that is studied, there is a topic even more Arcane, Extraction of Seemingly Missing Information. Just like a healing brush, communication and reality have much information that seems to be missing, but is actually filtered out. So, by cleaning one's filters (or in-painting), one can see a whole New Arcane World, different from what others see...
Post a Comment