Turned up to eleven: Fair and Balanced

Friday, January 11, 2002

The esoterica are flying back and forth on Libertarian Samizdata and The Fly Bottle over the value of Popper's philosophy of science. I have commented on this several times over there, and will not repeat myself too much. Will makes some good points in his latest post, about the potential meta-falsification issues (i.e., when you falsify the falsifying datum, then falsify the falsifier of the falsifier, etc), but I think that that is a philosopher's problem, not a practical one in science. It is possible to get lost in an introspective circle when you think of this in the abstract, but less so in the practical world, at least till you get into quantum mechanics (which I can barely begin to understand, let alone explain). In the real (non-quantum) world, we can accept our observations as true, and allow them, based on the axioms of the system, to guide our hypothesis formation and testing. A lot of this argument seems to me to stem from a fundamental definition problem. Will mentioned the necessity of a formal system and the logic of science flowing therein, as if this were a problem. It is clear, however, from the work of great logicians from Leibnitz thru Russell, Cantor, and finally Godel, that knowledge must be codified into a formal system containing 1) axioms, 2)mechanical rules for application of these axioms, and 3)theorems. Later, Turing showed that any formal system can be turned into a "Universal Computer" (yeah, I am reading a book about this right now, so it's kinda stuck in my head). The Universal Computer then utilize any theorem in the system as an algorithm. The so-called "halting problem" stems from the Formally Undecidable Propositions that Godel identified, or at least showed exist. This all seems off topic, but what it means, in a nutshell, is that the process for acquiring knowledge, science, is a formal system of inquiry, and as such has some questions that it cannot answer. The work of philosophers of science such as Popper, Kuhn, etc, can best be understood as attempts to refine the scientific method so as to practically improve the enterprise of science, rather than to perfect it.
On another note, I have a bone to pick with the entry on Libertarian Samizdata discussing this by Alan Forrester, in which he supposedly refutes the idea that observations can lead to theories, based on the idea that all of the available information present pre-Einstein would lead to Newtonian mechanics and Relativity both, while only one is "correct". While the statement is true, it does not lead to the conclusion that observations don't lead to theories. Of course they do, we don't derive all scientific theories from first principles, that would create a static, rather than dynamic, scientific enterprise. What we do, however, is refine, or even discard, our theories when the facts do not fit them. This is as opposed to "faith-based" reasoning, which discards the facts in order to protect the theory. I can't believe I let myself get sucked into this. I have work to do! First, some observations, followed by some hypothesis formation, followed by some hypothesis testing! Ah, Science!!!