March 1, 2004
MORE THAN HUMAN:
It's official: even to Steven Hawking, there will always be a little mystery.
(Paul Davies, 1/3/2004, Online Opinion)
The world about us looks so bewilderingly complex, it seems impossible that human beings could ever understand it completely. But dig deeper, and the richness and variety of nature are found to stem from just a handful of underlying mathematical principles. So rapid has been the advance of science in elucidating this hidden subtext of nature that many scientists, especially theoretical physicists, believe we are on the verge of formulating a “theory of everything”.When Stephen Hawking accepted the Lucasian Chair of Mathematics at Cambridge University in 1980 he chose as the title of his inaugural lecture “Is the end in sight for theoretical physics?”. What he meant was that physicists could glimpse the outlines of a final theory, in which all the laws of nature would be melded into a single, elegant mathematical scheme, perhaps so simple and compact it could be emblazoned on your T shirt. Now Hawking has done something of a U-turn by claiming in a lecture that we will never be able to grasp in totality how the universe is put together. [...]
Then there is a deeper question of whether a finite mind can ever fully grasp all of reality. By common consent, the most secure branch of human knowledge is mathematics. It rests on rational foundations, and its results flow seamlessly from sequences of precise definitions and logical deductions. Who could doubt that 1 + 1 = 2, for example? But in the 1930s the Austrian philosopher Kurt Gödel stunned mathematicians by proving beyond doubt that the grand and elaborate edifice of mathematics was built on sand. It turns out that mathematical systems rich enough to contain arithmetic are shot through with logical contradictions. Any given mathematical statement (e.g. 11 is a prime number) must either be true or false, right? Wrong! Gödel showed that however elaborate mathematics becomes, there will always exist some statements (not the above ones though) that can never be proved true or false. They are fundamentally undecidable. Hence mathematics will always be incomplete and in a sense uncertain.
Because physical theories are cast in the language of mathematics, they are also subject to the limitations of Gödel’s theorem. Many physicists have remarked over the years that this will preclude a truly complete theory of everything. Now it seems that Stephen Hawking has joined their ranks.
When we do figure it all out it will turn out to be surprisingly simple, but we won't be merely human anymore. Posted by Orrin Judd at March 1, 2004 7:01 PM
"Iz no eazy vay in Matematics, iz only der rrite vay." -- Prof. Izaak Wirszup, U of Chicago, 1975.
One thing left out of the discussion of undecideablilty is that undecideable statements in one system may be decideable with a different set of axioms. (Sometimes you can use the undecidable statement as an axiom, giving different systems-- the parallel postulate, for example, gives rise to three different systems of geometry.) What Godel showed is that every system, no matter how constructed, will have some statement, somewhere, which is beyond it. A "Unified Theory of Math" is what's impossible. What I've never seen is a proof that you can't multiple math systems, each which has undecidable statement, all of which can intersect or be disjoint, but every statement is provable in at least one system. (In other words, there exist statements that are never provable in any system.) Then again, maybe I just dropped out before they got to that part.
Another fact is that some questions can be show to be false, but never shown to be true. For a long time, the four-color map theorem fell into this category, as you could decide against it by simply drawing a five color map, even though a proof that five colors are necessary may have been impossible to create. (That can be done because such a map would have a finite number of regions, and so in the worst case, you could five colors necessary by just enumerating all the possibilities. (And any real mathematicians out there will know that I've just engaged in a huge amount of handwaving.)) This is why falsifiability is important in science, because it's usually a lot easier and possible.
*shakes head*, the writer isn't exactly correct about what Godel proved in his theorem. Simple mathematics is NOT shot through with logical contradictions. Rather, Godel was able to use simple mathematics to encode mathematical axioms and, more importantly, their proofs. The Computer Science spin on the proof is that, at the time, the theory was that Mathematics was mechanical: It would be possible to feed axioms into some sort of algorithm, turn the crank, and out would pop All True statements. Godel showed that, for any algorithm for cranking out true statements, there was one statement that was demonstrably true, but was also demonstrably unprovable. I.e., no matter how long the crank was turned, that demonstrably true statement would never appear. It would have to be an AXIOM, a statement that had to be true, AND included in the initial state of the algorithm. Godel then showed that, for the NEW and IMPROVED algorithm, there existed yet another statement that was demonstrably true, but was also demonstrably unprovable. He showed that you could keep "improving" the algorithm, but never arrive at the "final" algorithm that would prove all true statements.
(Keep in mind that the crucial distinction is between "true" (I.e. something that is not contradictory or demonstrably false), and "provable" (I.e. being able to go through a reasoning process from a set of axioms to produce that "true" statement.)
The word for this situation is not "undecidable", but "incomplete": A system is incomplete if it cannot prove all truths given the set of axioms and a process of derivation that does not produce contradictions.
"Undecidability" came from a different attack on the problem: Why not build on Godel's system for mapping numbers to symbols that represented axioms and proof processes and develop a meta-process that would generate all possible sets of axioms and proof processes, toss out the "invalidly formed" ones, and analyze the remainder? Alan Turing, IIRC, put the kabosh on that by showing, using the famous Turing Machine, that any metaprocess would be fooled by certain sets. The only way to REALLY determine if the member was valid was to "execute" the proof algorithm on the set (I.e. run the program). He was able to show that some members of the set would cause any proof algorithm to enter an infinite loop: I.e. you could never tell if the program was bad (infintely looping), or was just taking a long time to prove a particularly difficult truth. He boiled this last obstacle down to the question of whether there was a metaprocess for determining if a program halted or not. He was able to prove that any process of proof that was able to determine whether a program could halt did not exist, because you could be able to construct a set of axioms and a proof process that would be false, but which the program would say was true. I.e. it was contradictory, and thus false. Thus, the halting problem is known to be not solvable because any system that claims to do so can be made to produce false statements. It is that finding that causes the whole enterprise of mathematical proof to be "undecidable", since you will never know if you've hit a truth statement that is tough to prove, or which causes your proof system to run endlessly, because it is one of those statements that either makes it cycle in an infinite loop, or happens to be the truth that the system is missing to make it complete.
Whew! Well, THAT'S the computer science version. I'm sure a pure mathematician would cringe at that explanation, because computer science has a very wierd way of proving things, which I never got the hang of...
Oh, the tie in to Hawkins: The holy grail of Mathematics is to develop a branch of mathematics that is totally useless as a means to explain nature or a natural law. They've never found it: Hilbert once developed a branch of mathematics that everyone thought met the criterion, but upon discovering that the math described the behavior of csubatomic particles in a certain state, reportedly muttered, "DAMN!".
It appears that Hawkins is thinking that the reverse is true: If every truth in mathematics has an analog of some sort in physical nature, then there will always be something new to explain in nature, since Godel's theorem says there will always be new truths that cannot be proved.
Posted by: Ptah at March 1, 2004 8:00 PM"But dig deeper, and the richness and variety of nature are found to stem from just a handful of underlying mathematical principles."
Typical confused Platonic Idealism. Nature may be described by a handful of mathematical principles, but they do not "underlie" it, nor does Nature stem from them.
"It turns out that mathematical systems rich enough to contain arithmetic are shot through with logical contradictions."
No, no, no - this is not even remotely what Gdel's Theorem states; it is not a question of contradiction (if it were, arithmetic would be worthless), only completeness. Gdel's (First) Theorem states that any sufficiently rich formal system, if consistent, is incomplete - i.e., there are true statements which can not be derived from the axioms. His second Theorem states that this deficiency can not be solved by the addition of axioms - if you add an axiom making a formally undecidable statement true, then there always exist other formally undecidable statements, ad infinitum.
Alternatively (my view, by the way), Gdel's Theorems can be interpreted as a formal criticism of Tertium non Datur. If you admit undecidable statements in the formal logic, then consistency and completeness both become decidable.
Finally, this application of Gdel's Theorems to physics is bogus. The real problem in physics is the inability to unify Relativity and Quantum Mechanics - a problem which I believe is intractable with the current formulation of Quantum Mechanics.
Posted by: jd watson at March 1, 2004 8:08 PMI believe Ptah is incorrect in stating that all mathematical systems have natural analogues.
I'm not a professional, but my physics adviser says there are many that do not. A friend of his wrote a book, "Mathematics: Queen of the Sciences," to make the point that because physical systems can be described mathematically, and often by apparently abstract math that was developed before the physical system was discovered, people get confused and think that the universe is mathematical.
The universe is the universe. It makes the rules, it does not follow some rules external to it. I read Godel's Theorem last year and could not understand what all the fuss is about. His idea applies to systems created by humans. It says nothing that I can see about natural systems.
Natural systems are describable mathematically, but nature does not need to "decide" the truth of those statements.
Anyhow, the meaning of final theory is beyond the understanding of journalists. The final theory, if found, will be final because the speed of light is constant and therefore no information can be recovered from a space much smaller than the Planck radius.
Therefore, science stops at the Planck mass and Planck radius.
It does not follow, though, that having the final theory on paper allows us to derive from it the character of the Universe. The calculations are too complicated, even if we could get all the preliminary data, which we cannot.
As Steven Weisberg puts it, knowing the genome of a fruit fly does not tell you whether it is alive or not.
Posted by: Harry Eagar at March 1, 2004 9:25 PMRaoul, Ptah, JD, Harry:
Wow--thanks for the outstanding discussion. Speaking as someone who has a (11 yr old) MS in Computer Science, you all refreshed my memory, and taught me a few things.
Much thanks.
Posted by: Jeff Guinn at March 2, 2004 7:19 AMWhat Jeff said. I'm blown away.
Posted by: Peter B at March 2, 2004 8:13 AMRaoul:
I don't think having multiple systems helps; the aggregate of those systems is itself a system, and you won't be able to work out a consistant way of resolving all the contradictions between them.
Posted by: Mike Earl at March 2, 2004 12:34 PMI believe Ptah is incorrect in stating that all mathematical systems have natural analogues.
I did not say that. I said that It looked to me as if HAWKINGS was saying that.
Posted by: Ptah at March 2, 2004 12:56 PMSorry.
Since posting (and misspelling Weinberg's name), I also found this statement of his: mathematics does not explain anything.
Posted by: Harry Eagar at March 2, 2004 2:43 PMBut mathematics does provide astonisingly beautiful descriptions.
Posted by: Jeff Guinn at March 3, 2004 12:28 PM