Depends on whether you are talking about ferro-magnetic permanent magnets, rare-earth permanent magnets, regular electromagnets, superconducting electromagnets, regular superconductors or magnetars.
(A regular superconductor is anti-magnetic - it will repel ALL magnetic fields.)
Also, I've said in several posts that the primitives must be Intelligence-complete. Hello World does not involve or include a Turing-Complete set of functions. By introducing "Hello World", you have replaced what I have said with a non-equivalent statement and then merely proved the non-equivalent to be false. This doesn't prove anything other than your inability to choose equivalent statements.
To be equivalent, you must meet ALL the criteria. In case you have forgotten them, they are as follows:
The primitives MUST be Intelligence-complete and Empathy-complete, the corresponding functions in the brain to the requirement of Turing-completeness
The primitives ARE NOT the whole system. The whole system will do other things (which may include "Hello World"). They are a strict subset.
The primitives MUST be a common denominator between ALL brains possessing directly comparable high-level intelligence and empathy functions, as observed
The primitives MUST have changed in ALL brains that do not exhibit ANY high-level intelligence and empathy
The primitives ARE NOT the high level functions themselves, they are merely the components that are required, with nothing extra and nothing missing
The primitives ARE of high-enough level that ONE AND ONLY ONE implementation of intelligence and empathy can be layered on top of them
The primitives are assumed, in this argument, to be wired by means of other mechanisms to produce actual intelligence and actual empathy - something you ignore in favor of sneering. I would appreciate it if you could put your ego to one side for a moment, look at what I am saying, and comment on what is said rather than what you would like me to say because it's easier to poke holes in.
I do not say I'm "right", but I do say that I deserve better than to be walked over.
Much appreciated. The thanks is fine (and more informative:). It's a damn shame there are so many written languages that have been lost through deliberate acts of destruction. (My understanding is that Etruscan books were destroyed first by the Roman Empire and then again by the Holy Roman Empire, which is why so little is known.) Phrygian is another language that we know only fragmentarily, for absolutely no good reason.
The one thing exobiologists are convinced of is that we would NOT be able to identify alien intelligence. It is on this basis that I am taking it that if we can recognize it, it cannot be truly alien.
More to the point: If empathy and toolmaking are possible in extremely simple brains that we could hypothetically simulate, why do they only appear in species with significantly larger brains? Why do ravens have such large brains if they could get away with simple, tiny ones and perform the same tasks? It seems to me that many of the features you're hoping are simple, like tool making, are actually side effects of a general abstract reasoning ability, and that this ability requires a relatively large and complex brain to achieve.
Forgot to deal with this. Simply put, I'm referring to primitives. I don't need to consider more space than is needed to have ALL the building-blocks needed to form intelligence and empathy, of sufficient complexity that intelligence and empathy can be implemented essentially only one way. I do NOT need any of the space needed to cover the actual implementation of either.
I could put the whole of a Turing-complete microprocessor, C library and even the Linux kernel all on one chip. Slashdot has covered web servers running on smaller systems than a matchbox. You couldn't run a decent Enterprise-scale database on anything that small. But there is absolutely nothing in the larger systems that does not exist in the smaller ones, at the primitives level.
My argument is that there is a segment of the brain, more substantial than an individual neuron, that performs key operations that are so absolutely fundamental to what we mean by intelligence and empathy that ALL intelligence and empathy built from those operations will be virtually the same across all animals. The variations in animal intelligence we are seeing are small, implying constraints in the system that keep those variations small.
No, a working implementation of "Hello World" is NOT a primitive. You could never call Hello World in any arrangement to build another arrangement. The C language plus the standard C99 library would be primitives.
In the case of intelligence or empathy, clearly the primitives being used are complex enough that despite a total lack of any definition of intelligence (or, indeed, empathy), we have analytical tools capable of identifying intelligence and empathy in crows. They should be almost as alien as life on another planet.
Primitives tend to evolve once. There are many, many types of eye, for example, and the "eye" as a concept has evolved independently multiple times, but ALL of them are based on the same photosensitive mechanisms that exist in early unicellular life.
Well, yes, which demonstrates that there's plenty of room for divergent evolution. A shark's dorsal fin is triangular, a cetaceans's dorsal fin is curved. However, in the case of Orcas (killer whales), given that they will kill other creatures for amusement (which puts them on the same part of the evolutionary ladder as hare coursers and dog fighters), I'd rather keep my distance anyway.
I am extremely suspicious of "convergent evolution" in cases where there are multiple ways to perform the same general task. The probability of multiple generations converging rather than diverging should be infinitesimal. Convergent evolution does happen, but even there let's pause for thought. Dolphins and whales are descended from animals that moved back into the oceans. Their methods of controlling depth and pressure are unlike that of any fish. They have flukes, which are analogous to fins but do not operate in the same way and are not used in exactly the same way. It converged to a degree, but since then has run more parallel.
Do we see parallel evolution in birds and humans? Possibly. It bothers me, though, that the manner of representation is very human-like - so much so that I'm having a hard time calling it a parallel method of doing the same thing. It seems much more like it's the same method of doing the same thing.
But even if it is parallel, does that matter? OS/X and Linux are parallel lines of evolution in OS', but they both rely on a CPU to provide primitives. Since my argument is that the primitives, the mid-level instruction set necessary to form intelligence, is common, it is immaterial if the implementations were from the same source or evolved wholly independently. They'd still be using the same mid-level instruction set. (In fact, I'm also going to suggest that this is a requirement for "convergent evolution" - that you can't even have parallel implementations if the underlying engines are fundamentally different.)
Our most advanced imaging tool for the human brain is the 9.4T MRI. Our most advanced imaging tool for animal brains is the 12T MRI. These resolve down to single cells and can be used for both static images and fMRI. There are dozens of ways to perform an MRI to get a static image, too. I counted how many other ways there were to monitor brain activity - I came up with a list of about 30. (I was bored.) It is almost unimaginable that the full range of methods and techniques could not be deployed to produce a complete analysis of just the reptilian portion of the human brain. If I'm correct and intelligence is of common descent, then the most primitive constructs on which all later forms of intelligence rest (convergent or otherwise) MUST be in that part of the brain and nowhere else.
Again, if I an correct, then only a tiny subset of that brain will be (a) in common across all animals exhibiting high-level intelligence and/or empathy, AND (b) most active when such intelligence/empathy is in use, AND (c) necessary for high-level intelligence to function, AND (d) not be dedicated to autonomous functions required by the rest of the body. This is not the same as a "seat of intelligence/empathy" or a "seat of consciousness", any more than a node in a masterless computer cluster is the seat of all operations, or an ALU is the seat of all computation. It's merely a device that provides the key primitives. The actual "program" lies elsewhere. (And, according to recent studies, probably "everywhere" in the brain.)
The information from 57 different brain scans (24 MRI + 33 other types of scan) should be plenty of information to seed a Strong AI system, and because we're talking a very tiny number of brain cells (maybe a few thousand to tens of thousands tops) it should be doable on big iron.
Now we're not going to get HAL 9000 out of this, even if I am right. All we're going to get, at best, is a system that is capable of performing a set of very basic operations that can be called Intelligence-complete (in the same way as a Turing Machine performs a few basic operations that equate to anything any digital computer could ever do, no matter how advanced or how programmed). There should be no mental task performable by humans (or any other animal) that cannot be broken down into an algorithm using solely the Intelligence-complete set of operations.
If no such set of instructions can be derived, then one or more of the assumptions is incorrect.
Well, Old Norse is technically based on Old Germanic rather than the other way round, and Old English not only had Old Germanic input but Old Norse input as well. Along with an uncertain amount of Anglic (amazingly little is known about the Angles), possibly some Jute. English uses Norman French, plus modern French (which itself is derived from Norman French). Norman French survives in the modern world in Guernsey, Jersey and maybe some other Channel Islands but became extinct on Alderney.
To bring this Back On Topic, if English were lost, it would be almost impossible to use this program to recover it. English has input from too many sources, resulting in way too many loan-words of incompatible structure and too much incompatible grammar. However, one very interesting test of the program would be to map each of the derived phonemes in Pre-Indo-European to a character, then compare this derived PIE script with each Indo-European language in turn. If the derivation is correct, the number of correct guesses for translations of PIE words into each known IE language aught to be above what would be expected by chance alone AND the translations should remain compatible with the derivations the PIE engineers used in the first place. By comparing across the translations for all languages, the program may discover other word-parts that had not been noticed before.
It may be possible to determine if a language is truly isolate or not, by analyzing against a language multiple times using slightly different data sets and seeing if the results remain about the same. If this test works, then languages of uncertain/unknown ancestry (such as Basque and Etruscan*) can be tested against all 7,200 known languages to see if any of them produce a moderately stable match. No match means no connection with any other existent linguistic family tree.
*Etruscan is a bugbear. There is one book that is completely intact and undamaged. It's made of gold leaf. The academic who currently owns it has not published so much as a single line of the text, merely two of the illustrations. All other Etruscan texts are fragmentary (so you've very little context to work with and not many words that are definitely complete) or too short to be useful. We don't know what Etruscan is related to, but if the above hypothesis is correct, we could find out and then translate the book. But the damaged texts, such as a linen book used to wrap a mummy, are way too fragmentary. You'd never be sure if such a translation was correct. A complete book, on the other hand, would offer no possibility for mistake. It would work or it wouldn't.
Well, a more obvious implication is that if you fed in some percentage of Linear A texts and Cretan pictographic texts, you'd get virtually the same results as feeding in a different set of texts (ie: symbols should always equate to the same opposite number) if they are truly related.
This would at least let you identify if the texts are indeed of the same language, even if you can't read it, which is further along than we are now.
Neither is my great great grandmother's cookbook. Which really is a shame, as I strongly suspect the recipes make something more edible than what's served at the local coffee shop.
Yes, but it means the underlying mechanisms for toolmaking, empathy, etc, were all present no later than the last common ancestor. If a given animal does not have these traits, then the same sections of the brain are presumably used for some other function(s) as well - function(s) more advantageous to those other animals.
It also means that the underlying mechanisms are truly primitive and cannot involve any part of the brain not common to humans and avians. This means basic skills (such as toolmaking, basic problem solving, empathy, etc) should all be achievable with the Strong AI tools that exist today, which are plenty powerful enough to simulate what are relatively trivial neural circuits - compared to the whole human, or indeed avian, brain, that is.
That is absolutely correct. Serial tasks that are parallelized are known as SIMD, and the bulk of parallelism work has gone into this kind of parallelization.
Donald Knuth, and many of the other top names from IEEE, have name-brand power comparable to the IEEE itself. A split is not entirely impossible. If that happens (and in all seriousness, I expect the announcement to be at least a threat of a split) and the rival has reasonable policies and ethics, it will likely capture a fair chunk of the income and PR of the IEEE. (Hell, I've seen arXiv mentioned more in the popular press than the IEEE.) That could cause a serious disturbance in the Force, not to mention a serious disturbance in boardrooms, where there's a heavy reliance on political leverage to get things done. It's extremely difficult to manipulate crusaders - it can be done, but the RIAA and MPAA don't have that kind of Machiavellian skill.
Inmos had problems, I'm sure, but what really finished them off was that Inmos was owned by Thorn EMI. Domestic electronics and music cheapos. What the - were they going to do with a supercomputer grid?
You've got to be careful when talking about threads. There are four basic models: SISD, SIMD, MISD and MIMD. Of those, only SISD is serial, but if you've two independent SISD tasks, you can run them in parallel. Most modern supercomputers are built on the premise that SIMD is good enough. Not sure where MISD is used, MIMD fell out of favour when vector processors became too expensive but may be revived on more modest CPUs with modern interconnects like Infiniband.
The main problem is that it's horribly hard to pass that many messages around without the overheads of the network exceeding the benefit from the parallelization. If they have found a way to reduce this problem, I'd call that a major novelty.
I don't know about this specific project, but Manchester is strongly Open Source. The Manchester Computer Centre developed one of the first Linux distributions (and - at the time - one of the best). The Advanced Processor Technologies group has open-sourced software for developing asynchronous microelectronics and FPGA design software.
Manchester University is highly regarded for pioneering work (they were working on parallel systems in 1971, and developed the first stored-program computer in 1948) and they have never been ashamed to share what they know and do. (Disclaimer: I studied at and worked at UMIST, which was bought by Manchester, and my late father was a senior lecturer/reader of Chemistry at Manchester. I also maintain Freshmeat pages for the BALSA projects at APT.)
Slashdot Mirror
Depends on whether you are talking about ferro-magnetic permanent magnets, rare-earth permanent magnets, regular electromagnets, superconducting electromagnets, regular superconductors or magnetars.
(A regular superconductor is anti-magnetic - it will repel ALL magnetic fields.)
Then how did Doofenshmirtz attract a giant ball of aluminum foil with a giant magnet, then? Huh?
Also, I've said in several posts that the primitives must be Intelligence-complete. Hello World does not involve or include a Turing-Complete set of functions. By introducing "Hello World", you have replaced what I have said with a non-equivalent statement and then merely proved the non-equivalent to be false. This doesn't prove anything other than your inability to choose equivalent statements.
To be equivalent, you must meet ALL the criteria. In case you have forgotten them, they are as follows:
The primitives are assumed, in this argument, to be wired by means of other mechanisms to produce actual intelligence and actual empathy - something you ignore in favor of sneering. I would appreciate it if you could put your ego to one side for a moment, look at what I am saying, and comment on what is said rather than what you would like me to say because it's easier to poke holes in.
I do not say I'm "right", but I do say that I deserve better than to be walked over.
Much appreciated. The thanks is fine (and more informative :). It's a damn shame there are so many written languages that have been lost through deliberate acts of destruction. (My understanding is that Etruscan books were destroyed first by the Roman Empire and then again by the Holy Roman Empire, which is why so little is known.) Phrygian is another language that we know only fragmentarily, for absolutely no good reason.
The one thing exobiologists are convinced of is that we would NOT be able to identify alien intelligence. It is on this basis that I am taking it that if we can recognize it, it cannot be truly alien.
More to the point: If empathy and toolmaking are possible in extremely simple brains that we could hypothetically simulate, why do they only appear in species with significantly larger brains? Why do ravens have such large brains if they could get away with simple, tiny ones and perform the same tasks? It seems to me that many of the features you're hoping are simple, like tool making, are actually side effects of a general abstract reasoning ability, and that this ability requires a relatively large and complex brain to achieve.
Forgot to deal with this. Simply put, I'm referring to primitives. I don't need to consider more space than is needed to have ALL the building-blocks needed to form intelligence and empathy, of sufficient complexity that intelligence and empathy can be implemented essentially only one way. I do NOT need any of the space needed to cover the actual implementation of either.
I could put the whole of a Turing-complete microprocessor, C library and even the Linux kernel all on one chip. Slashdot has covered web servers running on smaller systems than a matchbox. You couldn't run a decent Enterprise-scale database on anything that small. But there is absolutely nothing in the larger systems that does not exist in the smaller ones, at the primitives level.
My argument is that there is a segment of the brain, more substantial than an individual neuron, that performs key operations that are so absolutely fundamental to what we mean by intelligence and empathy that ALL intelligence and empathy built from those operations will be virtually the same across all animals. The variations in animal intelligence we are seeing are small, implying constraints in the system that keep those variations small.
No, a working implementation of "Hello World" is NOT a primitive. You could never call Hello World in any arrangement to build another arrangement. The C language plus the standard C99 library would be primitives.
In the case of intelligence or empathy, clearly the primitives being used are complex enough that despite a total lack of any definition of intelligence (or, indeed, empathy), we have analytical tools capable of identifying intelligence and empathy in crows. They should be almost as alien as life on another planet.
Primitives tend to evolve once. There are many, many types of eye, for example, and the "eye" as a concept has evolved independently multiple times, but ALL of them are based on the same photosensitive mechanisms that exist in early unicellular life.
Well, yes, which demonstrates that there's plenty of room for divergent evolution. A shark's dorsal fin is triangular, a cetaceans's dorsal fin is curved. However, in the case of Orcas (killer whales), given that they will kill other creatures for amusement (which puts them on the same part of the evolutionary ladder as hare coursers and dog fighters), I'd rather keep my distance anyway.
Links follow:
Finding useful information on this book is... hard. You're right, Wikipedia doesn't even mention it. Anywhere.
I am extremely suspicious of "convergent evolution" in cases where there are multiple ways to perform the same general task. The probability of multiple generations converging rather than diverging should be infinitesimal. Convergent evolution does happen, but even there let's pause for thought. Dolphins and whales are descended from animals that moved back into the oceans. Their methods of controlling depth and pressure are unlike that of any fish. They have flukes, which are analogous to fins but do not operate in the same way and are not used in exactly the same way. It converged to a degree, but since then has run more parallel.
Do we see parallel evolution in birds and humans? Possibly. It bothers me, though, that the manner of representation is very human-like - so much so that I'm having a hard time calling it a parallel method of doing the same thing. It seems much more like it's the same method of doing the same thing.
But even if it is parallel, does that matter? OS/X and Linux are parallel lines of evolution in OS', but they both rely on a CPU to provide primitives. Since my argument is that the primitives, the mid-level instruction set necessary to form intelligence, is common, it is immaterial if the implementations were from the same source or evolved wholly independently. They'd still be using the same mid-level instruction set. (In fact, I'm also going to suggest that this is a requirement for "convergent evolution" - that you can't even have parallel implementations if the underlying engines are fundamentally different.)
Our most advanced imaging tool for the human brain is the 9.4T MRI. Our most advanced imaging tool for animal brains is the 12T MRI. These resolve down to single cells and can be used for both static images and fMRI. There are dozens of ways to perform an MRI to get a static image, too. I counted how many other ways there were to monitor brain activity - I came up with a list of about 30. (I was bored.) It is almost unimaginable that the full range of methods and techniques could not be deployed to produce a complete analysis of just the reptilian portion of the human brain. If I'm correct and intelligence is of common descent, then the most primitive constructs on which all later forms of intelligence rest (convergent or otherwise) MUST be in that part of the brain and nowhere else.
Again, if I an correct, then only a tiny subset of that brain will be (a) in common across all animals exhibiting high-level intelligence and/or empathy, AND (b) most active when such intelligence/empathy is in use, AND (c) necessary for high-level intelligence to function, AND (d) not be dedicated to autonomous functions required by the rest of the body. This is not the same as a "seat of intelligence/empathy" or a "seat of consciousness", any more than a node in a masterless computer cluster is the seat of all operations, or an ALU is the seat of all computation. It's merely a device that provides the key primitives. The actual "program" lies elsewhere. (And, according to recent studies, probably "everywhere" in the brain.)
The information from 57 different brain scans (24 MRI + 33 other types of scan) should be plenty of information to seed a Strong AI system, and because we're talking a very tiny number of brain cells (maybe a few thousand to tens of thousands tops) it should be doable on big iron.
Now we're not going to get HAL 9000 out of this, even if I am right. All we're going to get, at best, is a system that is capable of performing a set of very basic operations that can be called Intelligence-complete (in the same way as a Turing Machine performs a few basic operations that equate to anything any digital computer could ever do, no matter how advanced or how programmed). There should be no mental task performable by humans (or any other animal) that cannot be broken down into an algorithm using solely the Intelligence-complete set of operations.
If no such set of instructions can be derived, then one or more of the assumptions is incorrect.
Well, Old Norse is technically based on Old Germanic rather than the other way round, and Old English not only had Old Germanic input but Old Norse input as well. Along with an uncertain amount of Anglic (amazingly little is known about the Angles), possibly some Jute. English uses Norman French, plus modern French (which itself is derived from Norman French). Norman French survives in the modern world in Guernsey, Jersey and maybe some other Channel Islands but became extinct on Alderney.
To bring this Back On Topic, if English were lost, it would be almost impossible to use this program to recover it. English has input from too many sources, resulting in way too many loan-words of incompatible structure and too much incompatible grammar. However, one very interesting test of the program would be to map each of the derived phonemes in Pre-Indo-European to a character, then compare this derived PIE script with each Indo-European language in turn. If the derivation is correct, the number of correct guesses for translations of PIE words into each known IE language aught to be above what would be expected by chance alone AND the translations should remain compatible with the derivations the PIE engineers used in the first place. By comparing across the translations for all languages, the program may discover other word-parts that had not been noticed before.
It may be possible to determine if a language is truly isolate or not, by analyzing against a language multiple times using slightly different data sets and seeing if the results remain about the same. If this test works, then languages of uncertain/unknown ancestry (such as Basque and Etruscan*) can be tested against all 7,200 known languages to see if any of them produce a moderately stable match. No match means no connection with any other existent linguistic family tree.
*Etruscan is a bugbear. There is one book that is completely intact and undamaged. It's made of gold leaf. The academic who currently owns it has not published so much as a single line of the text, merely two of the illustrations. All other Etruscan texts are fragmentary (so you've very little context to work with and not many words that are definitely complete) or too short to be useful. We don't know what Etruscan is related to, but if the above hypothesis is correct, we could find out and then translate the book. But the damaged texts, such as a linen book used to wrap a mummy, are way too fragmentary. You'd never be sure if such a translation was correct. A complete book, on the other hand, would offer no possibility for mistake. It would work or it wouldn't.
Well, a more obvious implication is that if you fed in some percentage of Linear A texts and Cretan pictographic texts, you'd get virtually the same results as feeding in a different set of texts (ie: symbols should always equate to the same opposite number) if they are truly related.
This would at least let you identify if the texts are indeed of the same language, even if you can't read it, which is further along than we are now.
It's also why an inordinate number of mountains are called "your finger, you fool" and "who is this fool who doesn't know what a mountain is?"
Neither is my great great grandmother's cookbook. Which really is a shame, as I strongly suspect the recipes make something more edible than what's served at the local coffee shop.
RMS is not in command-line mode, so is indeed daemonized.
Yes, but it means the underlying mechanisms for toolmaking, empathy, etc, were all present no later than the last common ancestor. If a given animal does not have these traits, then the same sections of the brain are presumably used for some other function(s) as well - function(s) more advantageous to those other animals.
It also means that the underlying mechanisms are truly primitive and cannot involve any part of the brain not common to humans and avians. This means basic skills (such as toolmaking, basic problem solving, empathy, etc) should all be achievable with the Strong AI tools that exist today, which are plenty powerful enough to simulate what are relatively trivial neural circuits - compared to the whole human, or indeed avian, brain, that is.
That is absolutely correct. Serial tasks that are parallelized are known as SIMD, and the bulk of parallelism work has gone into this kind of parallelization.
1) The announcement will be at the TeX convention, but TeX isn't due for a major release, LaTeX is.
2) This explains everything.
Donald Knuth, and many of the other top names from IEEE, have name-brand power comparable to the IEEE itself. A split is not entirely impossible. If that happens (and in all seriousness, I expect the announcement to be at least a threat of a split) and the rival has reasonable policies and ethics, it will likely capture a fair chunk of the income and PR of the IEEE. (Hell, I've seen arXiv mentioned more in the popular press than the IEEE.) That could cause a serious disturbance in the Force, not to mention a serious disturbance in boardrooms, where there's a heavy reliance on political leverage to get things done. It's extremely difficult to manipulate crusaders - it can be done, but the RIAA and MPAA don't have that kind of Machiavellian skill.
...will be released with support for HTML5?
Inmos had problems, I'm sure, but what really finished them off was that Inmos was owned by Thorn EMI. Domestic electronics and music cheapos. What the - were they going to do with a supercomputer grid?
You've got to be careful when talking about threads. There are four basic models: SISD, SIMD, MISD and MIMD. Of those, only SISD is serial, but if you've two independent SISD tasks, you can run them in parallel. Most modern supercomputers are built on the premise that SIMD is good enough. Not sure where MISD is used, MIMD fell out of favour when vector processors became too expensive but may be revived on more modest CPUs with modern interconnects like Infiniband.
The main problem is that it's horribly hard to pass that many messages around without the overheads of the network exceeding the benefit from the parallelization. If they have found a way to reduce this problem, I'd call that a major novelty.
I don't know about this specific project, but Manchester is strongly Open Source. The Manchester Computer Centre developed one of the first Linux distributions (and - at the time - one of the best). The Advanced Processor Technologies group has open-sourced software for developing asynchronous microelectronics and FPGA design software.
Manchester University is highly regarded for pioneering work (they were working on parallel systems in 1971, and developed the first stored-program computer in 1948) and they have never been ashamed to share what they know and do. (Disclaimer: I studied at and worked at UMIST, which was bought by Manchester, and my late father was a senior lecturer/reader of Chemistry at Manchester. I also maintain Freshmeat pages for the BALSA projects at APT.)