Re: BOI Chapter 8 vs Schmidhuber
On 09 Oct 2011, at 18:30, John Clark (FOR list) wrote: On Sat, Oct 8, 2011 Bruno Marchal marc...@ulb.ac.be wrote: If you are a machine, you are duplicable (in principle). Yes. And if pure randomness is important to the feeling of self then you are not a machine and can not be duplicated because randomness (which by its very nature is non deterministic and is in fact the very definition of random) It is a definition of one precise form of randomness. Others exist, but it would distract us from the topic. can not be reliably duplicated. But my feeling of self continues from one moment to the next so the only logical conclusion is that randomness is not important in generating a feeling of continuity of self. OK. But that is not in the conclusion. That is the starting assumptions. And below you might grasp that the first person indeterminacy plays a key role in the ability to remain conscious in lawful physical reality. But I am anticipating. you cannot predict in advance where you will feel to be. You will feel to be where your sense organs are, the location of your brain is irrelevant provided it is not so distant that the speed of light becomes important, and if recent developments turn out to be true perhaps not even then. The speed of light is not relevant. If you are told that you will be reconstituted in a far away galaxy, or even in a different universe, or a different multiverse, you have to take that reconstitution into account to predict your next experience, no matter what. If not, you would introduce a magical ability (that is, a non Turing emulable ability) to persons. If your eyes send you an image of the Kremlin then you will feel like you're in Moscow, and if your eyes send you an image of the White House you will feel like you are in Washington. Of course in this example the two identical copies of yourself have received non-identical stimulation and as a result are no longer identical and will have diverged, but both of them have an equal right to call themselves Bruno Marchal because both can remember being Bruno Marchal yesterday. Yes. And those two Bruno Marchal have to recognize they could not have guessed in advance where they will be reconstituted. Ypu are pleading for comp here (alias computationalism, or DM = digital mechanism). You cannot predict the result of the next self-localization. So what? It would only take you a few minutes to write a computer program that will look at all the even numbers greater than 4 until it finds the smallest one that is not the sum of two odd primes and then stop, but if the computer this program is running on has X amount of memory you can not in general predict if it will stop before it reaches X. That is a third person indeterminacy, which concerns some event in the long run, and is quite different from the first person indeterminacy which concerns the result of an immediate experiment, from the first person point of view. You can see a similar difference between the deterministic chaos (like with weather) and the observation of an electron in the base {up, down} when it is in the state up+down. So what? Well, it is the discovery of the notion of first person indeterminacy in the deterministic frame of digital mechanism. The first person experience, in this case, cannot be emulated by any deterministic process I'll bet you feel like your first person experience has continued from yesterday to today, but I have a secret to tell you. Last night when you were asleep I scanned your body to the atomic precision Mr. Heisenberg allows and recorded the position and momentum of all the atoms in your body, and then I destroyed your body, and then I used that information to construct a new body, and you knew nothing about it until I told you just now. And yet you still feel like you because you remember being Bruno yesterday. Yes. That's what I call digital mechanism. Tthe assumption that indeed I would still be Bruno Marchal after an experience like that. More precisely DM assumes that there is a level of description of myself such that I survive for a substitution made at that level. Who you? Which you? How to do that? I honestly don't understand the question, or what the alleged problem is supposed to be, or what that is that I'm supposed to explain. I'm sure all the copies of me will have an opinion on whether they are John K Clark or not, and if you ask them nicely they would be happy to tell you what that opinion is. This is because I have explained only the key ideas. As I said, the first person cannot be aware of the delays of reconstitution, nor of the virtual or physical nature of those reconstitution. So if we assume that there is a (primary) physical universe, and that it is robust enough to run a universal dovetailer, then you can understand that physics is in principle entirely
Re: BOI Chapter 8 vs Schmidhuber
On 03 Oct 2011, at 01:40, Russell Standish wrote: On Sun, Oct 02, 2011 at 01:42:19PM +0200, Bruno Marchal wrote: Hi Russell, On 02 Oct 2011, at 11:37, Russell Standish wrote: In David Deutsch's Beginning of Infinity chapter 8, he criticises Schmidhuber's Great Programmer idea by saying that it is giving up on explanation in science, Actually I did address this point on the FOR list years ago. Somehow, I share David's critics on Schmidhuber's idea of a great programmer when seen as an *explanation* (of everything). My (older, btw) publications makes this point clear. The Universal Dovetailer (which can be seen as an effective and precise version of the great programmer, and which is a tiny part of elementary arithmetical truth) makes it possible to *formulate* (not solve!) the mind body problem mathematically, but Schmidhuber use it as an explanation gap. He missed the fact that if we are machine we cannot know in which computations we are and we have to recover the physical laws, not from one computation but from an internal (self-referential) statistics on infinities of computations, and that statistics has to be recovered entirely from the self-reference ability of machine. Sure - Schmidhuber, with his speed prior, assumed that the specific implementation of the universal reference machine has physical consequences, but we, thanks to your work, know better. David's criticism was quite specific - because the specific implementation of the UTM doesn't have any physical consequences, therefore one is somehow giving up on obtaining the ulimate explanation. My response was that surely the question becomes uninteresting (David's terminology) - or even meaningless (as you state below). It means that from a comp view, to fix a UTM for physical reality like a quantum computer, would be a treachery, would be bound to be wrong, and would miss the opportunity of the necessary Solovay split for distinguishing quanta and qualia. ... snip ... But this was an answer to David's remark that the great programmers explains too much, and so don't explain anything. I'm aware of this criticism, which applies to ensemble theories in general. IMHO, the only way to address that critique is with some sort of observer-relative anthropic selection - but that is a whole other topic! OK. The time has not yet come to dig on the heart of the ASSA/RSSA thread :) as the hardware on which the Great Program runs is unknowable. Of course the contrary is true. If we are machine, we know (up to some recursive equivalence) what runs us, and where the possible hardware come from. Any first order specification of any universal machine or theory will do the job. I use elementary arithmetic because we are all familiar with it. The laws of physics cannot depend on that choice. We're actually saying the same thing here. OK. David, why do you say that? Surely, the question of what hardware is implementing the Great Simulators simply becomes uninteresting, much like the medieval arguments about the number of angels dancing on the head of a pin. It is unknowable, and it doesn't matter, as any universal machine will do. I disagree with this. The notion of primitive hardware is precisely shown to be meaningless. The laws of physics are shown to be machine independent. Eventually the initial universal system plays the role of a coordinate system, and the laws of physics does not depend on it. Isn't this stating the above in a stronger form? Meaningless, rather than unknowable? Yes. But it is important to chose one which facilitates the derivation of the couplings consciousness/matter, or quanta/qualia. We can choose a quantum computer, but this will make the extraction of quanta very confusing, with a risk of treachery at each step. Given that self-reference is born in arithmetic (OK, in Gödel's Principia Mathematica, but soon on much weaker theories, which makes the result more general), and given that arithmetic is taught in high school, I think it is the better choice. Especially that arithmetic distinguish nicely universality (the everything) and Löbianity (the observer person multiplied in the everything), by the passage from Robinson arithmetic to Peano Arithmetic. It helps to use mathematical logics to solve a problem in computer science. The universality notion used in AUDA is the notion of sigma_1 completeness. The restriction of the probability to UD accessible states, in translated in arithmetic by the restriction of p to the sigma_1 sentence. And then I appreciate the numbers and number theory, but well I appreciate also the combinators and more abstract applicative algebra too. To do the work we have to choose a theory (as conceptually simple as possible), then this gives the phi_i and the w_i, which defines the computations and their domains. Bruno --
BOI Chapter 8 vs Schmidhuber
In David Deutsch's Beginning of Infinity chapter 8, he criticises Schmidhuber's Great Programmer idea by saying that it is giving up on explanation in science, as the hardware on which the Great Program runs is unknowable. David, why do you say that? Surely, the question of what hardware is implementing the Great Simulators simply becomes uninteresting, much like the medieval arguments about the number of angels dancing on the head of a pin. It is unknowable, and it doesn't matter, as any universal machine will do. The second question I have to David is why you say The whole point of universality is lost if one conceives of computation as being somehow prior to the physical world.? I do appreciate that mathematically, hypercomputers exist, an example being the infinity hotel example you give in your book. So a consequence of something like Schmidhuber's theory is that hypercomputers can never exist in our physical world. I suppose you would say that if physics were generated by machine, why the class of Turing universal machine, and not some hyper-(hyper-) machine? Whereas in a physics-first scenario, physics can only support Turing computation. Surely though, we can reverse the question in the physics-first case - why can't physics support hypercomputation? Cheers I'm copying this to the everything-list, as people there are interested in this topic too. -- Prof Russell Standish Phone 0425 253119 (mobile) Principal, High Performance Coders Visiting Professor of Mathematics hpco...@hpcoders.com.au University of New South Wales http://www.hpcoders.com.au -- You received this message because you are subscribed to the Google Groups Everything List group. To post to this group, send email to everything-list@googlegroups.com. To unsubscribe from this group, send email to everything-list+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/everything-list?hl=en.
Re: BOI Chapter 8 vs Schmidhuber
Hi Russell, On 02 Oct 2011, at 11:37, Russell Standish wrote: In David Deutsch's Beginning of Infinity chapter 8, he criticises Schmidhuber's Great Programmer idea by saying that it is giving up on explanation in science, Actually I did address this point on the FOR list years ago. Somehow, I share David's critics on Schmidhuber's idea of a great programmer when seen as an *explanation* (of everything). My (older, btw) publications makes this point clear. The Universal Dovetailer (which can be seen as an effective and precise version of the great programmer, and which is a tiny part of elementary arithmetical truth) makes it possible to *formulate* (not solve!) the mind body problem mathematically, but Schmidhuber use it as an explanation gap. He missed the fact that if we are machine we cannot know in which computations we are and we have to recover the physical laws, not from one computation but from an internal (self-referential) statistics on infinities of computations, and that statistics has to be recovered entirely from the self-reference ability of machine. The consequence is that, a priori, the laws of physical cannot be digital, the physical reality cannot be Turing emulable, nor can consciousness. Both matter and mind becomes global feature of the fabric of reality. Mechanism (I am a machine) entails that the everything which is not me, cannot be a machine (like arithmeyical truth cannot be emulated by any machines). In fact mechanism is incompatible with digital physics. Mechanism (I am a machine) entails that the everything which is not me, cannot be a machine (like arithmetical truth cannot be emulated by any machines). In fact mechanism is incompatible with digital physics. But this was an answer to David's remark that the great programmers explains too much, and so don't explain anything. In fact it explains nothing, but its effective version makes it possible to formulate the mind body problem, and to solve it both conceptually, and technically (but this leads to mathematical open problems, some of which have been solved since). as the hardware on which the Great Program runs is unknowable. Of course the contrary is true. If we are machine, we know (up to some recursive equivalence) what runs us, and where the possible hardware come from. Any first order specification of any universal machine or theory will do the job. I use elementary arithmetic because we are all familiar with it. The laws of physics cannot depend on that choice. David, why do you say that? Surely, the question of what hardware is implementing the Great Simulators simply becomes uninteresting, much like the medieval arguments about the number of angels dancing on the head of a pin. It is unknowable, and it doesn't matter, as any universal machine will do. I disagree with this. The notion of primitive hardware is precisely shown to be meaningless. The laws of physics are shown to be machine independent. Eventually the initial universal system plays the role of a coordinate system, and the laws of physics does not depend on it. The second question I have to David is why you say The whole point of universality is lost if one conceives of computation as being somehow prior to the physical world.? Good question. I am interested in what David can say about this. The notion of universality has been discovered by mathematician, and is indeed a provably arithmetical property of numbers, relatively to numbers. I do appreciate that mathematically, hypercomputers exist, an example being the infinity hotel example you give in your book. I will have to read that. In fact I think that hypercomputation is a red herring. Basically our reality must seem hypercomputed (and even worst that that) once we are digital machine. The analytical (which is above the arithmetical, which is itself above the computable (sigma_1 arithmetical), and the physical are internal aspect of the computable, once we assume that we (not the universe) are Turing emulable. So a consequence of something like Schmidhuber's theory is that hypercomputers can never exist in our physical world. The opposite conclusion than mechanism. But digital physics entails mechanism, and mechanism entails the falsity of digital physics. This means that digital physics is a contradictory notion. I suppose you would say that if physics were generated by machine, why the class of Turing universal machine, and not some hyper-(hyper-) machine? Whereas in a physics-first scenario, physics can only support Turing computation. If I (whatever I am) is a machine, then the universe (whatever responsible for me to exist) cannot be a machine, nor explicitly generated by a machine (but it can be, and need to be *apparent* to machines points of view). This follows from the Universal Dovetailer Argument (and I wait some replies on it on the FOR list).
Re: BOI Chapter 8 vs Schmidhuber
On Sun, Oct 02, 2011 at 01:42:19PM +0200, Bruno Marchal wrote: Hi Russell, On 02 Oct 2011, at 11:37, Russell Standish wrote: In David Deutsch's Beginning of Infinity chapter 8, he criticises Schmidhuber's Great Programmer idea by saying that it is giving up on explanation in science, Actually I did address this point on the FOR list years ago. Somehow, I share David's critics on Schmidhuber's idea of a great programmer when seen as an *explanation* (of everything). My (older, btw) publications makes this point clear. The Universal Dovetailer (which can be seen as an effective and precise version of the great programmer, and which is a tiny part of elementary arithmetical truth) makes it possible to *formulate* (not solve!) the mind body problem mathematically, but Schmidhuber use it as an explanation gap. He missed the fact that if we are machine we cannot know in which computations we are and we have to recover the physical laws, not from one computation but from an internal (self-referential) statistics on infinities of computations, and that statistics has to be recovered entirely from the self-reference ability of machine. Sure - Schmidhuber, with his speed prior, assumed that the specific implementation of the universal reference machine has physical consequences, but we, thanks to your work, know better. David's criticism was quite specific - because the specific implementation of the UTM doesn't have any physical consequences, therefore one is somehow giving up on obtaining the ulimate explanation. My response was that surely the question becomes uninteresting (David's terminology) - or even meaningless (as you state below). ... snip ... But this was an answer to David's remark that the great programmers explains too much, and so don't explain anything. I'm aware of this criticism, which applies to ensemble theories in general. IMHO, the only way to address that critique is with some sort of observer-relative anthropic selection - but that is a whole other topic! as the hardware on which the Great Program runs is unknowable. Of course the contrary is true. If we are machine, we know (up to some recursive equivalence) what runs us, and where the possible hardware come from. Any first order specification of any universal machine or theory will do the job. I use elementary arithmetic because we are all familiar with it. The laws of physics cannot depend on that choice. We're actually saying the same thing here. David, why do you say that? Surely, the question of what hardware is implementing the Great Simulators simply becomes uninteresting, much like the medieval arguments about the number of angels dancing on the head of a pin. It is unknowable, and it doesn't matter, as any universal machine will do. I disagree with this. The notion of primitive hardware is precisely shown to be meaningless. The laws of physics are shown to be machine independent. Eventually the initial universal system plays the role of a coordinate system, and the laws of physics does not depend on it. Isn't this stating the above in a stronger form? Meaningless, rather than unknowable? -- Prof Russell Standish Phone 0425 253119 (mobile) Principal, High Performance Coders Visiting Professor of Mathematics hpco...@hpcoders.com.au University of New South Wales http://www.hpcoders.com.au -- You received this message because you are subscribed to the Google Groups Everything List group. To post to this group, send email to everything-list@googlegroups.com. To unsubscribe from this group, send email to everything-list+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/everything-list?hl=en.
Re: [foar] BOI Chapter 8 vs Schmidhuber
You could simply point to Goedel's Incompleteness Theorem which kind of mandates that not everything in a universe is explicable from *within* that universe. This is not to give up on explanation. This *is* the explanation. It is neither good nor bad as an explanation - but it does require an 'act of faith'. As Bruno Marchal says, there are realities we can never prove, merely bet on. These things tend to end up being classified as 'religion' but, bless me, that's the value of religion! It allows us to have a point of view on things we can never prove in this life. Doesn't have to be organised, public religion - PERSONAL religion is the only authentic religion. Religion is simply what you happen to *believe*. David clearly adheres to a religion of Optimism as he calls it. That's fine; I admire this gigantic optimism of his and Popper's, it's very inspiring and will yet give birth to a great many new insights and discoveries. Just occasionally, one's personal religion does something positive for the world like that. Organised religion isn't really religion after all - it's a club with rules and creeds and punishments if you fall foul of the regulations. That's politics. Yoga is the science of the East and Science is the Yoga of the West - Someone or Other. Kim Jones On 02/10/2011, at 8:37 PM, Russell Standish wrote: In David Deutsch's Beginning of Infinity chapter 8, he criticises Schmidhuber's Great Programmer idea by saying that it is giving up on explanation in science, as the hardware on which the Great Program runs is unknowable. -- You received this message because you are subscribed to the Google Groups Fabric of Alternate Reality group. To post to this group, send email to f...@googlegroups.com. To unsubscribe from this group, send email to foar+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/foar?hl=en. -- You received this message because you are subscribed to the Google Groups Everything List group. To post to this group, send email to everything-list@googlegroups.com. To unsubscribe from this group, send email to everything-list+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/everything-list?hl=en.