X-Message-Number: 3091 From: (Nick Szabo) Subject: CRYONICS: Determinism & reversibility Date: Fri, 9 Sep 1994 01:35:55 -0700 (PDT) There may be no practical difference between quantum mechanical processes that are truly random (as most physicists now believe) and ones that are driven by very chaotic deterministic processes. Even the most commonplace process of radioactive decay is at least so chaotic, that it is beyond the precision of our current instruments, and the statistical analysis capacity of our fastest computers, to tell it apart from a truly random process. A fundamental result in computation theory is that it is easy to generate practically random sequences with a fully deterministic process. The Blum-Blum-Shub generator is one example of such an algorithm. The gap between the effort needed to generate randomness, and the effort needed to reverse that process to find the underlying determinism, grows quickly as amount of generated information grows. The amount of information in our brain is a large indeed; if a significant nonredundant fraction of it undergoes computationally one way quantum change we're dead beyond hope. It doesn't effect the problem significantly to assume faster computers; the difference quickly becomes so large that if every atom in the universe were a Cray, they still wouldn't be able to reverse the process. The only way I could then be reanimated is if we reached the hypothetical "Omega Point", and even then I would only witness an infinitesimal fraction of the future. I'm only infinitesimally satisfied with that. Ralph Merkle has put forth a perceptive model called information theoretic death. Along those lines, we might model death as a process of increasing randomization that at some point becomes computationally infeasible to reverse. In the equilibrium or living state of the brain, the only computationally irreversible processes are channeled into waste heat; structural processes are more or less at an equilibrium of randomizing and ordering processes. During ischemia the randomization quickly grows, until (at some point around currently defined medical death) the amount of damage grows beyond the capacity of any in vivo ordering process. By the time the corpse has rotted, restoring the brain back to its original state has become computationally infeasible for any process. The main question then becomes how reversible are the ischemia, perfusion, freezing, and vitrification processes undergone by neural tissue during suspension. Ralph has started to tackle this with his excellent articles on the repair of the brain and "cryptanalysis" of freezing damage. Much more knowledge of memory and its redundancy, as well as a much more detailed analysis of damage incurred in freezing and vitrification, may be needed before we can make confident predictions about the reversibility of the suspension process. Some parts of the dying process are known to be reversible with a well observed probability. For example, minutes of ischemia at low temperatures. I find Mike Darwin's research to be quite valuable, as its goal is to physically demonstrate (a much stronger result than theoretical proof) a perfusion/cooldown process that is reversible with high probability under a reasonable range of conditions. Research on freezing and vitrification is also valuable, but the results are much more ambiguous. For example, detailed STM pix of vitrified synaptic structure, with atom by atom accounting of the damage, would be quite valuable, but would only allow improvement in our indirect guesses of probability. Atomic accounting is a kind of worst-case scenario that doesn't allow us to model memory redundancy. It would be easy to have atomic displacements that in themselves are computationally irreversible, but there exists redundancy elsewhere in the brain, or even in other brains, by which the original location can be inferred in a computationally feasible manner. On the other hand, there may well be varieties of damage for which there is no computationally feasible redundancy. The area where brain research and information death theory may have the most impact, is to identify areas of low redundancy, difficult to reverse damage, and to redesign suspension protocols to eliminate these forms of damage. Assuming away irreversible damage is very dangerous; that is precisely the kind of damage we should be looking for in order to eliminate it. Nick Szabo Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=3091