X-Message-Number: 27571 Date: Fri, 3 Feb 2006 22:35:55 -0500 Subject: Uploading Myths, Continued... From: <> >> Myth #1: Identity will undergo radical revision. This could be a valid prediction, but it needs to be clarified as being very very far in the future. << The definition of identity derives from philosophical and logical requirements---not from the state of technological progress. Therefore, it is not subject to change with increasing technological advance. If the definition should ever be revised, it will be done so by the hands of a philosopher or logician, not by a technologist. Consider the simple statement, 'Look, I placed that coffee cup on the countertop an hour ago.' This statement, and the vast majority of other ones regarding existent things, rely on the definition of identity I put forth earlier. Without it, such statements would be meaningless. There is no reason to expect that the development of new technologies would invalidate such a definition (and every reason to trust it would not). What may change is what people value, and what they profess to value. A person might come to believe that his 'identity' will continue existence in a spirit world after his death; or that his 'identity' will continue to exist even if his brain is destroyed and all that survives is a Turing tape ticking away for the ages. People can and do change their beliefs, and their values. But what won't change, are those things not subject to values, such as mathematics, logic, and certain truths gleaned from philosophy. The necessary nature of identity for real-world objects is one of these things. >> Myth #2: Merging of selves. Again, this is a very distant prediction. And again, I still think that taken in that context, it is a valid prediction. If I am constantly adding neurons one at a time to my future brain in a very ordered and organized way, then I would still be a general purpose machine, but a constantly growing one. << This sentiment of yours depends on another myth, which I hereby christen, The Myth of the Intelligent Blob. This myth states that the brain is a blob of neurons, and that you can increase its intelligence and/or functionality by gluing on more neurons. You can't imagine how this sounds to me. You work in IT, so you have some experience with digital electronics. Consider how ridiculous it would sound to you if someone on the street suggested he could increase the processing power of his Pentium computer by adding one transistor at a time. The Pentium has a specific architecture; it's designed to do a certain job. You can't make it any better by adding another transistor. You can't add another transistor that does any useful purpose. If you want to improve the Pentium by a significant margin, you need to completely re-engineer it. Lengthen the pipeline, replace the branch prediction algorithm with a better one, etc. In a similar fashion, *radical* improvements to the functionality of the brain will involve a redesign. And let me be clear about this: a redesigned brain won't be like your brain; it won't have the same circuits, the same neural architecture, the same division of tasks, or possibly even the same neurons. In particular, there will not exist a step-by-step migratory path from your existing brain to a redesigned brain---no more than you can, step by step, transform an 8086 into a Pentium by incremental addition or removal of transistors (with each step being fully functional). What may be possible is infusing a redesigned brain with your memories (maybe), but what is certainly not possible, for the vast majority of possible redesigns, and for all radical improvements, is a neuron-by-neuron upgrade of your existing brain. Infuse a different kind of brain with your memories, and call it 'you', but don't believe for a second that it is. And don't think it will help if you mash your atoms around, bringing your brain through non-functional states to get to a desired configuration. You die when your brain becomes non- functional. >>How can you possibly predict that I will not be more competitive than a special purpose machine?<< This is a simple matter of resources: in a given volume of space, an optimal machine that is able to use all of its resources to accomplish a task will be more successful than an optimal machine able to use only part of its resources to accomplish that task. Humans suck at math. That's why we use calculators. They have more recently started sucking at chess. Eventually they will suck at every specific task---exceeding only at the one for which they were designed: self-reproduction. Logic circuits designed explicitly to factor numbers perform several orders of magnitude better than general purpose CPUs containing a hundred times more transistors. I go further and point out that specific modes of computing, that we have not invented but are beginning to imagine, will perform exceptionally well in extremely narrow ranges. For example, an optical computer whose very principles of operation lend itself amazingly suited to performing packet routing on an all-optical Internet, but which can do absolutely nothing else well. The future, in every area of knowledge and service production, lies with special purpose machines. General purpose AI, as it's called, can't hope to compete. I'm surprised you haven't seen this happening all around you. Humans lose ground to in every area; perhaps it's so common you are blinded to it. >> Myth #4: The human brain can be simulated. Of course it can be simulated. << That depends on what you mean by 'simulated'. If you mean, will computers be able to produce human-level AI by direct simulation of biological brains, then no---they will never be simulated. In my previous message, I pointed out the absurdity of simulating a SINGLE neuron. Do you know the Schrodinger equation is unsolvable for most atoms, and that even with the fastest approximations, we can't predict the motion of a few hundred atoms for more than femtoseconds? Now imagine 10,000,000,000,000 atoms over the span of a human lifetime. This feat may be forever impossible (or at least, forever impractical), even given the most advanced nanotechnology. If you somehow succeed, you have succeeded only in simulating A SINGLE NEURON. The scope of the problem simply cannot be comprehended. The hope of the uploaders is that each neuron can be modeled by a simple mathematical formula that can be computed cheaply. Not a chance. If that were the case, evolution would have found a way to model it with fewer than 10,000,000,000,000 atoms. Not many computer scientists have come to appreciate the fact that each neuron is an extremely complicated molecular and quantum computer. Which is, incidentally, why all the attempts to simplify them to one-line summation equations have produced nothing but toys used principally to recognize shapes and features in images. In any case, an extraordinary burden of proof lies on someone making an extraordinary claim such as, 'The brain can be simulated as a collection of neurons, where each neuron is modeled precisely by a terminating mathematical equation.' >> Myth #5: We could live life in a virtual world Come on now. Stretch your mind and think about the very distant future. Like you said, not everyone will choose the same path. This is certainly possible. Why do you think it could never happen? << Let me ask you this. Do you think you are living right now, inside a toilet? If you think that question is absurd, and that the answer is obviously 'No,' then you will precisely understand my reasons for rejecting the possibility of virtual life. A virtual thing is by definition one that does not exist except in the mind of the beholder. Such things don't possess factually existences. What exists are CPUs, that is, silicon substrates; and brains; but such a thing as a virtual brain does not, and cannot exist. That's why it's called 'virtual'. >> Or did you see something on my website that I overlooked?<< I want each and every cryonics provider to provide a statement stating they will only attempt to repair damage to cryonics patients in a non-destructive fashion; and that, should they transfer their patients or their obligation to care for their patients to another entity, they will force that entity to contractually abide by this agreement. Richard B. R. Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=27571