|
Author
|
Topic: Reverse Engineering Evolution; evolvability and the primitives responsible for life
|
Jack Foster
Member
Member # 79
|
posted 29. August 2003 18:07
No more meetings! I've a few minutes for a few thoughts:
quote:
RBH: Interestingly, that's precisely how Lenski, et al.'s Avida simulation defined things: The programs - sequences of assembly language instructions - are thought of as the genotype and the operation of the programs as the phenotype. I'm glad Jack included that "traditionally written" phrase because I don't think the notion that evolved programs generally easily "break" is the case - while the various programs evolved in the Lenski, et al, study lost some functions in their knockout tests, they also occasionally gained functions when some instructions were knocked out. So the observation that traditionally-written programs are brittle, occupying narrow fitness space islands, does not necessarily generalize to evolved programs.
I think we're in general agreement here. Traditionally written programs are brittle because they aren't designed to evolve. A programmer didn't need to consider making a program robust due to the dependability of the CPU. But because they aren't designed to evolve, when they ARE subjected to the basic Darwinian algorithm of replication, variation, and selection . . . they don't evolve.
I'll have to go back and re-read Lenski, but I might quibble with Lenski, et al's definition of genotype (even though it's the same as mine)! I believe in the Lenski case there is a portion of the program which is subjected to mutation, which then through program infrastucture maps to operation. In this case, I would not consider the infrastructure part of the program (including the g-p map) a part of the genotype; it is more properly viewed as a part of the phenotype or operation of the program. There's no such distinction in traditionally-written programs. (Perhaps this isn't a quibble at all; Lenski et al might agree.)
I think both situations have good analogues in biology. Translation/transcription machinery is not a part of the genotype; the machinery is part of the phenotype. With hypothetical non-mapped replicators, the genotype IS the phenotype; or rather a certain conceptual quality of the replicator is the genotype, and a certain physical quality (i.e., the operation of the conceptual quality) of the replicator is the phenotype.
I agree that the analogy between computer program evolution and biological evolution is not perfect (which is why the former requires a new definition for Irreducible Complexity) but I think it is still a fantastic aid to understanding. I've got some more thoughts which have been unleashed by your posts, but they will have to wait.
Regards,
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 29. August 2003 18:39
Jack F wrote quote:
I agree that the analogy between computer program evolution and biological evolution is not perfect (which is why the former requires a new definition for Irreducible Complexity) but I think it is still a fantastic aid to understanding.
Um, well actually I was trying to argue that the analogy between human-written programs and biological phenomena is problematic; I think the analogy with the evolution of computer programs is much closer (though not, as you say, perfect) and is potentially a very fruitful route to some increased understanding.
I look forward to your further thoughts. It being Friday evening, my brain has officially gone on break. Besides, I have to zip out and buy "The Two Towers"!
RBH
IP: Logged
|
|
nobody
Member
Member # 145
|
posted 30. October 2003 12:41
I've been gone for a while, but this morning I did a google for these three words: "Reverse engineer life" and this thread came up in the fourth position.
The first link is this one over at ARN:
http://www.arn.org/boards/ubb-get_topic-f-1-t-001540.html
Jack has started a very interesting thread on a topic that I believe is going to get a lot more attention over the next decade, so I think it's worth a bump.
IP: Logged
|
|
|
Printer-friendly view of this topic