We are thinking of using either www.servint.net and/or www.rackspace.com to host a database backed web-site. Anyone have experience with either or both of these? I'll be happy to summarize whatever comes out of my investigations to anyone interested. --A.S. Wait
I think the advantage of "Open-source" vs "Closed-source" is larger market-share or more users (paying or not). That's why investors love dot-comms that don't make any money (slashdot included).
The next step is turning those users into customers and sharing the profits with the developers. Personally I'm not too sure these guys have the right model but so-far only market-share matters. (IMHO)
It would be easy for a programmer to try and fix a bug, and inadvertantly introduce another bug, simply because he doesn't have the mathematical knowledge that the maple programmers have. You can rely on the fact that the maple programmers know what they are doing, and since there is a centralised place (i'll admit, linus is good at doing this for the kernal), you know who is rensponisble for making or breaking the code.
Remember that these packages are being used to prove or disprove theories, you dont want mistakes.
Um, first of all many open-source programmers have PHDs in Mathematics, Engineering or the Sciences-- is that enough math back-ground? Second of all-- regression testing is essential for a symbolic math package whether you are a monkey (no offense to monkeys reading this) or a Field's medalist! For example check out the regression test suite of this number theorist's tool-kit.
I'm a graduate student in Quantum Computing (my advisors discovered Quantum Teleporation and Quantum Cryptography). In 1991/1992 I developed the arbitrary precision math code for a symbolic math-package that had hoped to compete with Maple/Mathematica by including an integrated word-processor and spread-sheet. That project, "Lab-writer", died by 1992 or so but I think it's time to try it again.
There has been some interest on the LyX mailing list regarding extending that software with a symbolic math-package. I am happy to coordinate this project and can personally work on QC simulation code (mostly linear algebra). (The reason we need yet another QC simulator is a topic for another day!) If you are interested in this idea please Email me! Be sure to try out LyX if you are not already familiar with it. Math-package(s) will be selected based on their ease of interface with LyX and selfishly if they have the operations I need. Obviously people who can code or test are very important but at this stage Email from any potential user will be helfpful in convincing the LyX team this is a project they should support.
A flood of thoughtful email would be great-- I think it's pathetic that universities around the world are pouring millions into proprietary closed-source "Scientific Workplace" type solutions when they could be using that money to support an open-source effort instead!
Bob Metcalfe is accusing Linus and anyone who believes in open-source of hypocrisy. He expects people to see that it's ridiculous to open the designs of electronic circuits and furthermore that it's ridiculous for Transmeta to free their patents etc. BUT WHY NOT? I'm not suggesting that we all become whiners like the people who screamed for the Slashdot source release but I am suggesting that we should consider this-- even support it.
Profit and the contribution of IP to the community at Transmeta, Rehdat/Cygnus, VA, Corel, etc. is NOT mutually exclusive so long as the timing is right. Everyone who believes in open-source needs to earn a living and sometimes that means keeping things proprietary in the short-term; wouldn't Bob really be forced to eat his words if Transmeta did "open-source" its technology: chip designs, patents etc. Not only would Metcalfe be proven wrong but the philosophy itself will be proven right-- yet again.
Bob Metcalfe is accusing Linus and anyone who believes in open-source of hypocrisy. He expects people to see that it's ridiculous to open the designs of electronic circuits and furthermore that it's ridiculous for Transmeta to free their patents etc. BUT WHY NOT? I'm not suggesting that we all become whiners like the people who screamed for the Slashdot source release but I am suggesting that we should consider this-- even support it.
Profit and the contribution of IP to the community at Transmeta, Rehdat/Cygnus, VA, Corel, etc. is NOT mutually exclusive so long as the timing is right. Everyone who believes in open-source needs to earn a living and sometimes that means keeping things proprietary in the short-term; wouldn't Bob really be forced to eat his words if Transmeta did "open-source" its technology: chip designs, patents etc. Not only would Metcalfe be proven wrong but the philosophy itself will be proven right-- yet again.
You'd have to purchase enough licenses to cover your users to make it legit. One copy of MS Office or Windows wouldn't cut it.
Even if the users don't use the sofware at the same time? Licenses that restrictive should be illegal [in my oppinion]. The benefit of a server is to eliminate Windows-Linux dual boot on every machine and replace it with just VMWare on one machine. (but with 10mbit that can suck.) Without a solution like this you can never be sure if a machine with that pesky-windows-only-app will be free.
In our office we used to speculate that gigabit ethernet would make it fairly practical to serve multiple simultaneous apps (say MSOffice or Quicken) from a linux-box running am emulator (say VMWare to be specific). Then a small office could use Linux everywhere (and only have a single license of whatever Windows apps are wanted) but is this legal? Any other issues besides bandwidth? (memory requirements on the server machine etc.)
The reason that Nature can "roll" the script of MacBeth (and in fact it did) is because it is much more likely to "roll" a program (Shakespeare's DNA) that does something complicated then it is to "roll" the complicated thing directly. The program, the living and breathing Shakespeare, then interacts with its environment and writes plays. (or late-night posts) There was no guarantee Shakespeare's DNA would result in a man that wrote Macbeth-- as we may find the day we start cloning humans for fun and profit[tm]-- the important thing is that evolution made that man possible.
What is the chance, for example, of randomly producing the digits of Pi? Zero since the sequence is of infinite length. On the other hand the chance of randomly generating a computer program that generates PI is better. (The probability is still: 1/2^[length of the program in bits].) The odds are best if the particular programming language we use has the right primitives. In that case just the characters "Pi" could be enough and assuming a 8bit encoding per character the chance of rolling this program is only 1/65536! Easy.
Evolution is a kind of computer program; at each stage the DNA whose host has lots of children (ie. is succesful at being copied) will give those children even better primitives to work with. At each stage only a small "leap" is required and that's why evolution works.
Previously I wrote: Copeland believes, like Penrose, that the mind is NOT a Turing machine!
And you replied: I think (hope) that "believes" is too strong a word. I haven't read his book but I got the idea from the encyclopedia article that Copeland was simply expressing reservations about what is proven and what isn't.
Later you wrote: Penrose on the other hand is so wrapped up in his own qmind theory that he's completely lost touch with reality:o(
As it turns out Copeland uses Penrose to support his own arguments! Penrose, essentially, says that modern physics must be wrong because, as I said previously, modern quantum-gravity theories are consistent with the idea that a classical (probabilistic) Turing machine can simulate any physical system perfectly and a Quantum Turing machine can simulate any physical system efficiently (and perfectly). Penrose is trying to find a theory that unified Quantum mechanics and gravity in such a way that some permitted physical processes cannot be simulated by any kind of Classical or Quantum computer. This is an extremely ambitious goal! If he does find such a theory, however, we are going to look pretty stupid. His machine based on this new physics will be able to do things that Turing machines and Quantum computer cannot do in principle.
From the paper I mentioned, which everyone can download and read for themselves, Copeland says on QC: "It is perhaps somewhat surprisng that not all classical algorithms are manual methods. That this is in fact the case has emerged from recent work on quantum computation. Algorithms for quantum Turing machines are not in gernal manual methods, since not all of the primitive operations made available by the quantum hardware can be performed by ap person unaided by mahcinery. Nevertheless the algorithms exected by Deutsch-Solovay-Yao quantum Turing machines are all classical in the sense used here."
In that paper he goes on to cite Penrose: "Others who have speculated about the existence of physical proceses that are not Turing-machine-computable include Geroch and Hartle 1986, Komar 1964, Penrose 1989,1994 and Vergis et al. 1986." At least Penrose is actively working on a new theory of physics that will support the (somewhat) outlandish arguments he makes about consciouness; Copeland on the other hand is saying that someone else (like Penrose) will figure out a new physics that will support his outlandish arguments!
Probably I'm spending too much time here but honestly I think these ideas are close to the most interesting research on the planet and that's why I think we should be careful before we burn people at the stake for pseudo-science.
The book by Dr. McFadden is supported by published research. And I think it is worth repeating one abstract here:
Biosystems 1999 Jun;50(3):203-11
A quantum mechanical model of adaptive mutation.
McFadden J, Al-Khalili J
Molecular Microbiology Group, School of Biological Sciences, University of Surrey, Guildford, UK. j.al-khalili@surrey.ac.uk
The principle that mutations occur randomly with respect to the direction of evolutionary change has been challenged by the phenomenon of adaptive mutations. There is currently no entirely satisfactory theory to account for how a cell can selectively mutate certain genes in response to environmental signals. However, spontaneous mutations are initiated by quantum events such as the shift of a single proton (hydrogen atom) from one site to an adjacent one. We consider here the wave function describing the quantum state of the genome as being in a coherent linear superposition of states describing both the shifted and unshifted protons. Quantum coherence will be destroyed by the process of decoherence in which the quantum state of the genome becomes correlated (entangled) with its surroundings. Using a very simple model we estimate the decoherence times for protons within DNA and demonstrate that quantum coherence may be maintained for biological time-scales. Interaction of the coherent genome wave function with environments containing utilisable substrate will induce rapid decoherence and thereby destroy the superposition of mutant and non-mutant states. We show that this accelerated rate of decoherence may significantly increase the rate of production of the mutated state.
Can anyone with a strong bio-molecular background comment on this? Rather than the press-release?!?
Thank you for mentioning Copeland's entry in the Stanford Encyclopedia of Philosophy! In my term paper I make the following quotation from his article: Beyond the Universal Turing Machine:
``...two sets of functions are of special--although certainly not exclusive--interest. These are the functions that are computable by an idealized human being who is unaided by machinery, and the functions that are in principle computable in the real world, which is to say, are computable by machines, or organs, or in general entities, that physically could exist, given the resources on offer in the real world, even if they do not actually exist, nor ever do so. Turing argued, we think persuasively, that the first of these sets is coincident with the set of Turing-machine-computable functions. We believe that the extent of the second set is an open, empirical question. [...]
...it would--or should--be one the greatest astonishments of science if the activity of Mother Nature were never to stray beyond the bounds of Turing-machine-computability.''
Copeland believes, like Penrose, that the mind is NOT a Turing machine! On the other hand the best analysis of modern physics indicates that any finite physical process, understood by modern physics, can be simulated efficiently, i.e. with a number of gates polynomial in the size of the system, by a Quantum network. Copeland believes that Deustch is wrong not because he is too radical but because he is too conservative. (A Turing machine can perfectly simulate a Quantum computer although it is VERY UNLIKELY it can do it efficiently-- if it could we could crack RSA in real-time on your linux box.)
I'm sorry if I "invoke mysterious quantum effects" but let me try to be more specific. If we can build quantum computers then those computers will have well understood new capabilities. These things are being written about all the time in Science, Nature and discussed at the major computing conferences (STOC, FOCS...) For Cryptography it will mean we can factor efficiently and do unconditionally secure key exchange-- surely a spectacular start?
So "What can QC do for AI?" Well, if humans are Turing machines, then in principle you can write down my algorithm and run it faster than my own brain can. On a quantum computer you could run my algorithm (ME essentially) not just a constant amount faster (more MHZ or a constant number of parallel processors) but quadratically faster and maybe even exponentially faster. It's not clear that "faster" would lead to more intelligence but I'd be surprised if you never ran out of time on an exam-- in those cases Faster would be smarter.
So, "What can QC do for Evolution?" We are in the proccess of decoding all 3 billion bits of the human genetic code. The specific arrangement of 3 billion bits came into existence through an evolutionary process over generations and generations of organisms (ultimately humans) on this planet. Now imagine we simulate that process. [We used to do this for fun in high-school] On a quantum computer entirely new ways of searching the available state space emerge-- once again we have a minimum of quadratic improvement on exhaustive search (for a QC) and exponential improvements are possible. That means my evolutionary simulations on a QC will be much richer and more interesting than your simulations on a classical computer.
Clearly, the simulation of evolution is not the same thing as evolution itself. My genetic algorithms will evolve more interesting behaviour than yours if mine run on a quantum computer-- that's the best --I-- can do. In "Quantum Evolution" a book that I haven't read McFadden tried to make a strong connection between QC and Evolution and previously I posted an Amazon link (lots of reviews there) that explores this QC-Evolution connection. I've read that book and I still can't explain the Many-worlds-evolution thing! Deutsch book is, for the most part, sound so maybe this new "Quantum evolution" thing is sound too. You can NEVER judge science by the press-releases. Look at this as an example. The journalists are actually talking about an experimentally verified technique (Quantum teleportation) that might be used to help us build a practical quantum computer. Did you get that from the article?
Slashdot Mirror
We are thinking of using either www.servint.net and/or www.rackspace.com to host a database backed web-site. Anyone have experience with either or both of these? I'll be happy to summarize whatever comes out of my investigations to anyone interested. --A.S. Wait
The next step is turning those users into customers and sharing the profits with the developers. Personally I'm not too sure these guys have the right model but so-far only market-share matters. (IMHO)
--Sasha.
It would be easy for a programmer to try and fix a bug, and inadvertantly introduce another bug, simply because he doesn't have the mathematical knowledge that the maple programmers have. You can rely on the fact that the maple programmers know what they are doing, and since there is a centralised place (i'll admit, linus is good at doing this for the kernal), you know who is rensponisble for making or breaking the code.
Remember that these packages are being used to prove or disprove theories, you dont want mistakes.
Um, first of all many open-source programmers have PHDs in Mathematics, Engineering or the Sciences-- is that enough math back-ground? Second of all-- regression testing is essential for a symbolic math package whether you are a monkey (no offense to monkeys reading this) or a Field's medalist! For example check out the regression test suite of this number theorist's tool-kit.
--Alexander (Sasha) Wait
There has been some interest on the LyX mailing list regarding extending that software with a symbolic math-package. I am happy to coordinate this project and can personally work on QC simulation code (mostly linear algebra). (The reason we need yet another QC simulator is a topic for another day!) If you are interested in this idea please Email me! Be sure to try out LyX if you are not already familiar with it. Math-package(s) will be selected based on their ease of interface with LyX and selfishly if they have the operations I need. Obviously people who can code or test are very important but at this stage Email from any potential user will be helfpful in convincing the LyX team this is a project they should support.
A flood of thoughtful email would be great-- I think it's pathetic that universities around the world are pouring millions into proprietary closed-source "Scientific Workplace" type solutions when they could be using that money to support an open-source effort instead!
--Alexander (Sasha) Wait
I posted a link to Open Design Circuits a while back. Maybe these guys should get together?
Profit and the contribution of IP to the community at Transmeta, Rehdat/Cygnus, VA, Corel, etc. is NOT mutually exclusive so long as the timing is right. Everyone who believes in open-source needs to earn a living and sometimes that means keeping things proprietary in the short-term; wouldn't Bob really be forced to eat his words if Transmeta did "open-source" its technology: chip designs, patents etc. Not only would Metcalfe be proven wrong but the philosophy itself will be proven right-- yet again.
A. Wait
Profit and the contribution of IP to the community at Transmeta, Rehdat/Cygnus, VA, Corel, etc. is NOT mutually exclusive so long as the timing is right. Everyone who believes in open-source needs to earn a living and sometimes that means keeping things proprietary in the short-term; wouldn't Bob really be forced to eat his words if Transmeta did "open-source" its technology: chip designs, patents etc. Not only would Metcalfe be proven wrong but the philosophy itself will be proven right-- yet again.
A. Wait
Even if the users don't use the sofware at the same time? Licenses that restrictive should be illegal [in my oppinion]. The benefit of a server is to eliminate Windows-Linux dual boot on every machine and replace it with just VMWare on one machine. (but with 10mbit that can suck.) Without a solution like this you can never be sure if a machine with that pesky-windows-only-app will be free.
A. Wait
A. wait.
What is the chance, for example, of randomly producing the digits of Pi? Zero since the sequence is of infinite length. On the other hand the chance of randomly generating a computer program that generates PI is better. (The probability is still: 1/2^[length of the program in bits].) The odds are best if the particular programming language we use has the right primitives. In that case just the characters "Pi" could be enough and assuming a 8bit encoding per character the chance of rolling this program is only 1/65536! Easy.
Evolution is a kind of computer program; at each stage the DNA whose host has lots of children (ie. is succesful at being copied) will give those children even better primitives to work with. At each stage only a small "leap" is required and that's why evolution works.
A. Wait
And you replied: I think (hope) that "believes" is too strong a word. I haven't read his book but I got the idea from the encyclopedia article that Copeland was simply expressing reservations about what is proven and what isn't.
Later you wrote: Penrose on the other hand is so wrapped up in his own qmind theory that he's completely lost touch with reality :o(
As it turns out Copeland uses Penrose to support his own arguments! Penrose, essentially, says that modern physics must be wrong because, as I said previously, modern quantum-gravity theories are consistent with the idea that a classical (probabilistic) Turing machine can simulate any physical system perfectly and a Quantum Turing machine can simulate any physical system efficiently (and perfectly). Penrose is trying to find a theory that unified Quantum mechanics and gravity in such a way that some permitted physical processes cannot be simulated by any kind of Classical or Quantum computer. This is an extremely ambitious goal! If he does find such a theory, however, we are going to look pretty stupid. His machine based on this new physics will be able to do things that Turing machines and Quantum computer cannot do in principle.
From the paper I mentioned, which everyone can download and read for themselves, Copeland says on QC: "It is perhaps somewhat surprisng that not all classical algorithms are manual methods. That this is in fact the case has emerged from recent work on quantum computation. Algorithms for quantum Turing machines are not in gernal manual methods, since not all of the primitive operations made available by the quantum hardware can be performed by ap person unaided by mahcinery. Nevertheless the algorithms exected by Deutsch-Solovay-Yao quantum Turing machines are all classical in the sense used here."
In that paper he goes on to cite Penrose: "Others who have speculated about the existence of physical proceses that are not Turing-machine-computable include Geroch and Hartle 1986, Komar 1964, Penrose 1989,1994 and Vergis et al. 1986." At least Penrose is actively working on a new theory of physics that will support the (somewhat) outlandish arguments he makes about consciouness; Copeland on the other hand is saying that someone else (like Penrose) will figure out a new physics that will support his outlandish arguments!
Probably I'm spending too much time here but honestly I think these ideas are close to the most interesting research on the planet and that's why I think we should be careful before we burn people at the stake for pseudo-science.
The book by Dr. McFadden is supported by published research. And I think it is worth repeating one abstract here:
Biosystems 1999 Jun;50(3):203-11
A quantum mechanical model of adaptive mutation.
McFadden J, Al-Khalili J
Molecular Microbiology Group, School of Biological Sciences, University of Surrey, Guildford, UK. j.al-khalili@surrey.ac.uk
The principle that mutations occur randomly with respect to the direction of evolutionary change has been challenged by the phenomenon of adaptive mutations. There is currently no entirely satisfactory theory to account for how a cell can selectively mutate certain genes in response to environmental signals. However, spontaneous mutations are initiated by quantum events such as the shift of a single proton (hydrogen atom) from one site to an adjacent one. We consider here the wave function describing the quantum state of the genome as being in a coherent linear superposition of states describing both the shifted and unshifted protons. Quantum coherence will be destroyed by the process of decoherence in which the quantum state of the genome becomes correlated (entangled) with its surroundings. Using a very simple model we estimate the decoherence times for protons within DNA and demonstrate that quantum coherence may be maintained for biological time-scales. Interaction of the coherent genome wave function with environments containing utilisable substrate will induce rapid decoherence and thereby destroy the superposition of mutant and non-mutant states. We show that this accelerated rate of decoherence may significantly increase the rate of production of the mutated state.
Can anyone with a strong bio-molecular background comment on this? Rather than the press-release?!?
A. Wait
``...two sets of functions are of special--although certainly not exclusive--interest. These are the functions that are computable by an idealized human being who is unaided by machinery, and the functions that are in principle computable in the real world, which is to say, are computable by machines, or organs, or in general entities, that physically could exist, given the resources on offer in the real world, even if they do not actually exist, nor ever do so. Turing argued, we think persuasively, that the first of these sets is coincident with the set of Turing-machine-computable functions. We believe that the extent of the second set is an open, empirical question. [...]
Copeland believes, like Penrose, that the mind is NOT a Turing machine! On the other hand the best analysis of modern physics indicates that any finite physical process, understood by modern physics, can be simulated efficiently, i.e. with a number of gates polynomial in the size of the system, by a Quantum network. Copeland believes that Deustch is wrong not because he is too radical but because he is too conservative. (A Turing machine can perfectly simulate a Quantum computer although it is VERY UNLIKELY it can do it efficiently-- if it could we could crack RSA in real-time on your linux box.)
A. Wait.
So "What can QC do for AI?" Well, if humans are Turing machines, then in principle you can write down my algorithm and run it faster than my own brain can. On a quantum computer you could run my algorithm (ME essentially) not just a constant amount faster (more MHZ or a constant number of parallel processors) but quadratically faster and maybe even exponentially faster. It's not clear that "faster" would lead to more intelligence but I'd be surprised if you never ran out of time on an exam-- in those cases Faster would be smarter.
So, "What can QC do for Evolution?" We are in the proccess of decoding all 3 billion bits of the human genetic code. The specific arrangement of 3 billion bits came into existence through an evolutionary process over generations and generations of organisms (ultimately humans) on this planet. Now imagine we simulate that process. [We used to do this for fun in high-school] On a quantum computer entirely new ways of searching the available state space emerge-- once again we have a minimum of quadratic improvement on exhaustive search (for a QC) and exponential improvements are possible. That means my evolutionary simulations on a QC will be much richer and more interesting than your simulations on a classical computer.
Clearly, the simulation of evolution is not the same thing as evolution itself. My genetic algorithms will evolve more interesting behaviour than yours if mine run on a quantum computer-- that's the best --I-- can do. In "Quantum Evolution" a book that I haven't read McFadden tried to make a strong connection between QC and Evolution and previously I posted an Amazon link (lots of reviews there) that explores this QC-Evolution connection. I've read that book and I still can't explain the Many-worlds-evolution thing! Deutsch book is, for the most part, sound so maybe this new "Quantum evolution" thing is sound too. You can NEVER judge science by the press-releases. Look at this as an example. The journalists are actually talking about an experimentally verified technique (Quantum teleportation) that might be used to help us build a practical quantum computer. Did you get that from the article?
A. Wait.