home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!gatech!ukma!rutgers!igor.rutgers.edu!planchet.rutgers.edu!nanotech
- From: louis@hound.cs.indiana.edu (Sushil)
- Newsgroups: sci.nanotech
- Subject: Re: Evolution and nanotech
- Message-ID: <Sep.15.17.06.45.1992.22238@planchet.rutgers.edu>
- Date: 15 Sep 92 21:06:45 GMT
- Sender: nanotech@planchet.rutgers.edu
- Organization: Indiana University, Bloomington
- Lines: 90
- Approved: nanotech@aramis.rutgers.edu
-
-
- The survey posted by JoSH started me thinking about evolution,
- nanotech and scary scenarios. Glad to see a thread already here.
-
- ------------------------
- In article <Aug.31.01.48.26.1992.24252@planchet.rutgers.edu> radford@cs.toronto.edu (Radford Neal) writes:
- >
- >If we are randomly searching for rare, highly-beneficial mutations,
- >then it shouldn't matter whether we have sex or not. Once a single
- >organism is produced with the good mutation, it's descendents should
- >quickly out-compete the others.
-
- Sexual recombination matters. Think about TWO (2) beneficial
- mutations A and B. For an individual to have both in asexual
- populations, A must occur in an individual followed by the occurence
- of B in a descendant of that individual (or vice-versa).
- In sexual populations, A and B can occur in unrelated
- individuals and come together thru sexual recombination in their descendants.
-
- [ See the "evolution of sex" by John Maynard Smith, Cambridge
- University Press, 1978. for details.]
-
- The other point about mutations is that however small you make
- the probability of mutation, as the population grows (assuming self
- replicating assemblers) the probability that the POPULATION contains a
- mutant individual, increases. Sexual recombination just makes the
- spread of benificial mutations faster. Self replication can be very
- dangerous. Small populations and mutation rates very very close to zero
- are neccessary.
-
- From: szabo@techbook.com (Nick Szabo)
- >I conclude that evolution can be programmed, but also can be easily
- >designed out of a machine. It is probable people will purposefully
- >design evolving machines at some point, just for a neat hack if for
- >no other reason.
-
- >Ref: _Arfificial Life 2_ (2nd Conference on Artificial Life, Santa Fe
- >Institute
-
-
- If fidelity is not guaranteed at 100%, copying errors and
- mutations will exist in large populations. In addition, for natural
- selection to start working you need replicators/assemblers. IF you
- stop copying/replicating you stop evolution. Even with replication,
- there are various ways of controlling runaway evolution, including
- restricting the parts needed for copying. This has already been
- discussed on sci.nanotech. If someone wants to develop assemblers
- that can replicate with EASILY available material, I'd suggest Space
- Station Freedom, or its nanotech made descendants, or the moon with
- all the safegaurds that Drexler proposes and any others that come to
- mind.....(At least one use for SSF).
-
- This IMHO is not as bad as the much more scary scenario of the
- creation/engineering/evolution of an Artificial Intellect in an
- electronic medium. Electronic (and mechanical) nano-supercomputers are
- going to make AI much easier to achieve ;-) Avoiding philosophical
- issues, imagine an AI proliferating and learning on the net. It's rate
- of learning and evolution (and it WILL evolve because copying is so
- easy) will be phenomenally fast. How can you stop replication?
- Electronic books will make things easier to learn. Pretty soon,
- perhaps a few days/months maybe years, such an AI may know more than
- the whole human race. This may not happen as quickly without nanotech
- computers, but IMHO that it will happen. Computer viruses already
- exist and examplify early electronic fauna.
-
- From: tomk@netcom.com (Thomas H. Kunich)
- >Somewhere I read that things evolve as they must. That
- >is an important distinction in evolution. There are many non-competitive
- >mutations that occur in species that are not weeded out because there
- >is no survival pressure. Color distinctions are often involved in
- >this sort of thing. Might not nanotech devices have similar problems?
-
- Are you talking about the neutral theory of molecular evolution?
- Neutral mutations that do not affect the fitness of the organism.
-
- >So 'beneficial mutation' doesn't have anything to do with 'competitive
- >advantage' except in certain circumstances.
-
- Beneficial mutations, by definition, confer a competitive advantage a
- higher fitness. Neutral mutations do not.
-
-
- |=============================================================================|
- Sushil Louis GAs in Design/Adaptive systems
- louis@cs.indiana.edu ------------------------------
- Department of Computer Science Design is an art
- 215 Lindley Hall Natural Selection, an artist
- Indiana University ------------------------------
- Bloomington, IN
- |=============================================================================|
-