mem (10/20/82)
c Regarding the concept of a race that is evolutionarily intellectually superior to us: Do you think that human intelligence will advance phylogenically? There is a concept called gerontomorphosis which says that the evolution of a race stagnates as the race becomes more specialized and especially as selective breeding ceases to play a part in the survival of the race. No... this isn't a case against welfare. Or is it? Mark E. Mallett
bcw (10/21/82)
From: Bruce C. Wright @ Duke University Re: Intelligence and mutations Mark Mallett brings up the question of whether human intelligence (and presumably alien intelligence as well) might not have stagnated. This has probably been true for the past 50,000 years or so (man has been able to significantly modify his environment for something on this order of time), since there probably hasn't been too much selection pressure (in a relative sense) during that time. But if our understanding of things like the workings of the brain and artificial intelligence continue at their present rate, we will before very long be able to directly modify intelligence. Selection and evolution arguments really don't matter very much if the race can directly modify it the characteristic under consideration. As for DNA being "programmed" to mutate, this sounds suspicious. The differences between the DNA for humans and apes only has about 2% or less different. It is unclear that this 2% difference was caused by any type of classical mutation (micromutation or macromutation), it is probable that much of the difference is recombination. Many people have the impression that evolution proceeds by the selection of new mutations; in reality, mutations are relatively rare, and most evolution proceeds by the selection of new *combinations*. Finally, I am far from convinced that it is really possible to design an error correcting code which will be guaranteed to remain intact for 20,000,000,000 years (guess of remaining lifetime of the universe) for all of 1e15 robots (give or take a few million) - unless the code is so costly that the time to compute it is of cosmological scale. Without such a guarantee it is possible to imagine the mutations and selection to take place in a manner not too unlike life on earth, which is exactly what some of the other readers have been worried about. Bruce C. Wright @ Duke University
bcw (10/21/82)
References: sii.178 From: Bruce C. Wright @ Duke University Re: Intelligence and mutations Mark Mallett brings up the question of whether human intelligence (and presumably alien intelligence as well) might not have stagnated. This has probably been true for the past 50,000 years or so (man has been able to significantly modify his environment for something on this order of time), since there probably hasn't been too much selection pressure (in a relative sense) during that time. But if our understanding of things like the workings of the brain and artificial intelligence continue at their present rate, we will before very long be able to directly modify intelligence. Selection and evolution arguments really don't matter very much if the race can directly modify the characteristic under consideration. As for DNA being "programmed" to mutate, this sounds suspicious. The differences between the DNA for humans and apes only has about 2% or less different. It is unclear that this 2% difference was caused by any type of classical mutation (micromutation or macromutation), it is probable that much of the difference is recombination. Many people have the impression that evolution proceeds by the selection of new mutations; in reality, mutations are relatively rare, and most evolution proceeds by the selection of new *combinations*. Finally, I am far from convinced that it is really possible to design an error correcting code which will be guaranteed to remain intact for 20,000,000,000 years (guess of remaining lifetime of the universe) for all of 1e15 robots (give or take a few million) - unless the code is so costly that the time to compute it is of cosmological scale. Without such a guarantee it is possible to imagine the mutations and selection to take place in a manner not too unlike life on earth, which is exactly what some of the other readers have been worried about. Bruce C. Wright @ Duke University
REM@MIT-MC@sri-unix (10/23/82)
From: Robert Elton Maas <REM at MIT-MC> When we can directly modify intelligence, traditional biological evolution by selection of DNA will mostly be replaced by evolution by selection of ideas and programs and data. Good ideas such as structured programming and atomic physics will grow and bifurcate and develop further, while bad or useless ideas such as platonic ideals and buggy whips will mostly die out. To some extent this has already been happening. If and when we set loose self-replicating robots throughout the galaxy, we'll see a new kind of evolution which will be a combination of physical evolution of the robots and idea-evolution of their programs. Of course if the programs are used to totally control the physical design, it'll be analagous to DNA totally controlling the design of biological creatures, complete with sharing/exchanging of program fragments as we now know bacteria and other lifeforms share/exchange DNA. <Speculative prediction by REM>