Re: Lack of evolution (computers and living things)

In message <e4c3b544-7d2f-40f5-b7c4-ec70cfef1ed2@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>, Seanpit <seanpit@xxxxxxxxx> writes
I hope you didn't injure yourself whilst lifting those goalposts.

I've always said that increasing or decreasing the level or degree of
a particular type of function, once it is present in a gene pool, is
easy for the evolutionary algorithm to achieve. This is not a problem
at all. It happens all the time in real life - - rapidly.

So you have no objection to a unique function forming in a small sequence then that sequence growing over time in both size and complexity to optimise that function? That satisfies your request as to how 1000aa systems come about then doesn't it?

I'm very interested to see how you move the goalposts to rule out that route since you have both admitted that new functions can evolve in small sequences *and* that small sequences can grow over time.

What evolutionary pressure do you think be on a program to generate a
game such as checkers?  You do understand what I mean by "evolutionary
pressure" don't you?  There has be a genetic advantage to develop
something, if there is no advantage then it will not evolve.

Even when there is an advantage such a program will not evolve because
of the non-beneficial gap problem that exists between what is present
in the softward "genome" and what might be present to a beneficial
degree if it could ever be found via random walk in sequence space.

Assertion with no evidence nor reasoning. You haven't coded up any evolutionary algorithms have you?

In the case of our computer program evolving checkers we know what the end result is and use that to apply our selection pressure, so it's most definitely *not* a random walk.

Any system which has environmental pressure which affects which genes are passed on to the next generation will optimise over time. No random walking over any distance is done unless the results are selectively neutral.

If you had a program which could produce random boards, random game
pieces, random rules and random strategy for games and the selection

BTW: Bad phrasing there on my part using the word "random". I meant that all possible game boards etc. could be produced by the system i.e. all game space was reachable. You would probably seed it with random boards etc. to speed up the process but it wouldn't be necessary.

pressure was such that the closer it was to checkers then the more
likely that genes would be passed on then you will find that it would
evolve quite quickly.

Not true. Why? Because, not all of the intermediate lines of code
would be sequentially beneficial. In fact, very few of the
intermediate lines of code would be benefical at all. These non-
beneficial gaps are what kill evolutionary progress dead in the water
this side of trillions upon trillions of years of time.

Nope, pure assertion on your part. The intermediate stages would all be better than that which had gone before.

For example, sticking to your example of the checkers game, lets look at the board itself:

If you had a population of game boards which was seeded with random shapes and colours and bred them, giving the boards which were most like checkers boards an advantage (more offspring) and stirred in occasional mutations to randomly change the colour, shape and layout of the board then you would reasonably quickly end up with a checkers board. I could even write the program myself!

A board with blue squares would have more offspring than one with green circles for example (because squares would score more highly) etc.

So there you have a nice, beneficial intermediate board with blue squares scattered in a non-grid pattern which would fair better than a board with green circles spread in a similar pattern.
--
sapient_usenet02@xxxxxxxxxxxxxx ICQ #17887309 * Save the net *
Grok: http://spam.abuse.net http://www.cauce.org * nuke a spammer *
Find: http://www.samspade.org http://www.netdemon.net * today *
Kill: http://spamsights.org http://spews.org http://spamhaus.org

.

Relevant Pages

  • Re: .99999... still=/= 1
    ... sequence of n ups is +n and the sequence of n downs is -n. ... surreals obtained by sequences of ups and downs based on ... out of his game theory. ...
    (sci.math)
  • Re: solve for chess
    ... >total game sequence of moves. ... CHESS IS THE *EXACT SAME TASK*!!! ... is ambiguous in this fashion" TO A GAME OF TIC-TAC-TOE. ...
    (rec.games.chess.computer)
  • Re: Best game intro ever
    ... just a flashy title sequence) ... of any game intro movie that I found all that interesting. ... Ultima V actually presented the opening sequence using the game engine, ...
    (comp.sys.ibm.pc.games.rpg)
  • Re: Calling All Pirates...
    ... game starts, hit the button to capture the image of the software. ... > What I was thinking was that there might be an identifiable code ... > to this sequence on boot-up and simply activate it before the intro ... > cracked program that finds the space bar start sequence, ...
    (comp.sys.cbm)