Patrickarbuthnot
Posts: 21 Joined: Feb. 2010
|
Quote (BWE @ Feb. 23 2010,07:04) | I was going to work methodological naturalism into this but opted against.
Godel's theorems have been used to justify a lot of crank-ish science. Usually if you hear someone explain what quantum systems are doing that affects your life somehow, what amazing things fractals can heal or what shocking thing Godel's theorems prove...Odds are you've encountered a crank.
But,
Douglas Hofstadter's Godel Escher Bach as some/many/most of you know from having read it, is not crank science/math/philosophy. In it, Hofstadter raises questions that go far beyond a neural application. One is that to try to add to the system at a certain point you have to say that thinking is non-algorithmic strikes me as pretty unlikely. To justify that, all I can really say is that I've been doing a little homework. It's not that hard to design a simple neural style network anymore and Pougot's bayesian brain model is having some sucess as are Hofstadter's ongoing evolved operators, Dennett too.. etc the list is long.
Any modeling process is algorithmic. If any of you are familiar enough with it to have a copy at your desk, you could verify this for me. Because I don't pack it for travel very often. :) Otherwise, it isn't critical. But as he's describing the process of adding new Godel numbers and running the ordinals up also infinitely and etc, showing how you transcend the levels one at a time he gets to the linits of the system being the self because there is no more complex place to go to get more godel numbers because our brains have just hit their max.
Anyway, Bills thread got me thinking and this is what I wondered:
SCIENCE: LEVEL INDEPENDENT METHODOLOGY? And then I fixed caps lock. Do we have a sort of heuristic ability to expand our models by just throwing shit at things and studying it when it sticks?
Science, the empirical method, works at whatever level of resolution we choose to apply it. In a very simple and straightforward way, the scientific method is simply a highly reliable technique for precise pattern detection. Our ability to shift recursive levels and 'see' the interaction of atoms and chemicals as an independent series of repeating patterns or see the trees behaving according to their own unique repeating patterns affords us our basic tool of awareness. Recognizing patterns and how they repeat enables the organism to navigate those patterns and respond to them appropriately.
We recognize things like frequency of repetition: every time we mix pigment with oil we get paint; every Autumn the trees lose their leaves; every forest provides building material; the berries and bark of every cascara tree induce diarrhea, and so on. We assess the isolation level of a pattern: rocks on the ground near us do not affect the resulting mixture of pigment and oil; leaving the skins on the fruit from which we extract pigment weakens the durability of the resulting paint; cascara trees grow within a certain range of conditions, and so on. We note the duration of a repeating pattern: well mixed paint remains waterproof for a range of time before it starts to break down; it takes approximately three hours to cut down a cascara tree and collect the bark and berries; the embers glow until the wood completely turns to ash; the sun shines a little longer every day until a certain point at which it shines for a little less time until the pattern repeats, and so on. We classify patterns so that we recognize the source and conclusion of its operation. That way, even if we search in vain for seeds, we know that where a tree grows, once upon a time a seed sprouted there. A dog-shaped mat of dog hair clinging tenaciously to the rug is a part of the dog shedding repeating pattern. We know the dog slept there and we might know if is likely to sleep there again. Testing the accuracy and reliability of our pattern recognition keeps us alive, fed, clothed and sheltered. Accuracy completes the modeling-time loop since without accuracy, modeling is normally just called dreaming. The impetus to develop accurate models and predict correctly isn't hard to fathom. Hunting involves understanding the predictable patterns of the prey. Building involves understanding the patterns of material integrity. Farming and agriculture involve knowing the patterns of plant growth and animal reproduction. All of these require accurately predicting the forces involved, estimating the level of control we expect to exert over the process, and predicting the outcome of complicated patterns- It's what we do. That skill, the magic of modeling more time, more than our organism needed for one lifetime, enough to feel confident that it could hatch a plan that would still succeed even beyond the death of the individual, propelled us out of Africa, across the globe, and beyond; at this point it has taken our species to the moon and extended our senses to the entire solar system. The scope of the new model we are building with the expanded storage and computing power of a networked world can hardly be overstated. It seems we know something about 'out there': 'out there' unflinchingly follows repeating patterns, which, once glimpsed, open up doors to new patterns within the universe unimagined by even the previous generation.
Is there any reason to think there are any limits to the vertical/outward ascent of rule transcending? (as opposed to plank lengths etc.) How complicated would we make our models before we stopped figuring stuff out? |
Can't add much except the obvious science is not just about collecting obscure facts. Science is also about constructing, testing, and applying scientific theories, particularly the predictive ones..
-------------- Thomas Edison said: “The doctor of the future will give no medicine, but will interest her or his patients in the care of the human frame, in a proper diet, and in the cause and prevention of disease.”
|