OK, here is ImaginalDisc's list again, as I agree with everything on it:
You might recall that I refuted this list at Post #280 insofar as its legitimacy in disproving the analogy goes; the refutation then being developed at Posts #291 and #297 . I don't believe anybody has further challenged what I last said on this matter, which I'll reiterate/rephrase thus:
Design can overhaul, evolution cannot.
Yes, it can, but it only happens sometimes, not all of the time. This can be considered an exception to the general rule postulated in the OP. There are, of course, always exceptions to the rule, but we're looking for
similarities here, not exceptions. That's the very essence of an '
analogy', as cyborg has repeatedly pointed out.
Design can plan for long-term development, evolution cannot.
Yes, it can, but it only happens sometimes, not all of the time. Ditto.
Design can lift elements from one type of thing and apply them to another, evolution cannot.
Yes, it can, but it only happens sometimes, not all of the time. Ditto.
Design can retain the plans of a form indefinately, evolution cannot.
Yes, it can, but does it? What happens if a particular design fails to develop (evolve)? It becomes uncompetitive, inefficient, etc. and hence obsolete, or 'extinct', if you will.
Living things are produced by autonumous reproduction, machines are not.
Irrelevant to the analogy, which relies on observation of outputs only.
Living things have heritable traits, machines do not.
Are you serious? Is every successive model of Compaq PC completely 'traitless' compared to the previous model?!
Living things mutate, and those mutations are passed on. Machines neither mutate, nor pass on mutations.
Open your mind jimbob - please! Changes to materials, components, specifications, etc. (indeed quantum leaps from, say, piston engines to jet engines) can be considered mutations for the purpose of the analogy. DIVORCE YOUR MIND FROM HOW THESE THINGS COME ABOUT AND JUST LOOK AT THE MANIFESTATIONS. THAT'S ALL THAT'S NEEDED FOR THE ANALOGY TO WORK!!!
In my job I design structures (transistors). I model them using the laws of physics, and then develop simplified models of their electrical behaviour in applications. I then look to see what parts need improving, and how to do this. Sometimes, after a lot of analysis, I even get some test structures made to give me some more information from real life, and corroboration. Any information that I develop, before being implimented in silicon has been very deliberately chosen and developed. I also look at failed devices, perform nondestructive tests, and get other people to perform destructive tests to determine what failed, or even if there was a weakness. Any information about the failure is arrived at by analysis. I would not say that the information to correct the device was potentially there before the device. It was developed. There was no random mutation of memes, there was no natural or artificial selection of memes, there are only adequate models and designs. We do sometimes use evolutionary algorithms, but we still determine the fitness criteria, so whilst we don't know what the design will be, we do know what it will do befirehand. This is different to evolution. We also might miss a good design because we have incorrectly missed out the full fitness criteria, as we don't always know them.
As I argued in an earlier post, you don't have to do all of this in such an organized manner. You could, instead, implement random changes (mutations) and simply see what effect they have, for better or worse, just like biological evolution. You could retain the improved versions, to build on even further, and discard the inferior versions. You would either end up with the same or essentially the same product as you might 'design with forethought' (unlikely), or you would end up with a different, equivalent or possibly even better, product. The only difference is that what I've described is a wholly inefficient and uncommercial approach, but that is excatly what biological evolution can be described as. It simply doesn't matter, though, with biological evolution; nobody is counting the cost nor measuring the time taken to get 'there'. There's really no difference when you really stop to think about it. Try it, please.
Back to the more widely known history of the Spitfire. I will talk about evolutionary algorithms and not evoution, as evolution as discussed in the theory of evolution requires imperfect self-replication.
You see, this highlights your problem exactly. Your preconceptions about even what the word 'evolution' means and how it is used is clearly preventing you from applying it in an analogous context. Unless and until you overcome this mental barrier you will be precluded from developing this debate further.
Early versions lacked a fuel injection system, which meant that it, unlike the BF109, couldn't pull negative g without losing power (which was bad in a dogfight). Work was done with the express aim of fixing that prolem.
True, but in theory it didn't have to be that way. Alternatively, random changes could have been made which, given time, would have 'fixed' this particular problem, or even led to an adaptation of an unrelated aspect of the aircraft which would have overcome the problem in another way, or otherwise rendered the problem fixed. Surely you can see what I'm describing here. I can actually picture the 'design' team in my mind flying each of the individual 'mutations'! Can't you?
If there was evolution, then the each aircraft would be slightly different , and only the blueprints of the most successfull aircraft would be copied.
Each type of aircraft is, at least, slightly different, and this is essentially what happens. Only the successful variants persist. What's causing you to believe that this isn't the case?
Apart from being impractical, there would have been no analysis to say that there was a problem. The problem would get solved without any analysis, and probably with a lot of other changes.
Hey, I think you're getting it now jimbob! I think you just need to reconcile the word 'impractical' in your mind and you'll be there. Don't resist it; it's just a word that humans have invented that biological evolution doesn't recognize, and has no need for. Dispense with it!
How does evolution know there is a problem? The only information available, or evolutionary algorithms is that sone design was the best fit to the design criteria, i.e. "fittest". This is different from looking at windtunnel results and deciding that low-level performance is more important, so the wings should be clipped in some variants.
Yes, it is different, but forget about wind tunnel tests. Just change something at random instead and you'll still get there, sooner or later. Wind tunnel testing is simply short-circuiting the evolutionary process for reasons I've explained previously, and above. Wind tunnel testing doesn't have to happen other than in the 'human' world. What happened before wind tunnels came into use?
When a prototype crashes an evolutionary algorithm coul only get the information that there was a failure, not why.
So? Make a random change, and another, and another until the problem gets fixed by chance. What's wrong with that?
If you look at a single example of any aircraft, you can see that it didn't evolve as there is no ability to reproduce.
But you can jimbob, because it did, essentially, 'reproduce'. The fact that it reproduced 'artificially', i.e. by man's hand as opposed to biologically, is completely irrelevant to the analogy. You're getting close jimbob; just try to grasp this concept too and you might be there!
With evolution, because the copy carries the template and "attempts" to reproduce the ability of the template to produce reproducing copies is tested. This is natural selection.
See above. You can make your own analogy now, I'm sure!
If you looked at the "fossil record" in a museum, you would not see the type features that are common to evolution. There really weren't any intermediate stages between a pulse-jet and a piston engine.
See above. Exception to the rule, or, if you like, macro-mutation. Either option works.
I would also argue that there is no room for a "promising technology" with evolutionary algiorithms. The prototype jets had worse performance than the best piston-engined planes, but there was sufficient potential to keep the development going.
And there lies the mental barrier again - 'sufficient potential'. Remove the designers' forethought and replace it with random 'mutations' and you'll still achieve that 'potential', given time.
The first LCD screens were totally inadequate for TV usage, but the potential was there for them to be better than the CRT. That was not an evolutionary approach in the history of televison technology.
Of course it was, and explains why LCD screens were not used in TVs until recently. They had to evolve to the point where they became useful in TVs.
We couldn't design the mamalian eye, and we wouldn't have designed it as it is.
Why not? Have you not seen 'poor' human designs? I could provide a list if you like, and they would only be examples that made it to commercial production!!! Can you not hear GM's Chief Designer gazing at a Citroen 2CV6 and saying: "Well, I wouldn't have designed it like that!"?!?
Minor errors (inelagancies) are not selected out in evolution, especially if the solution would require starting from scratch again, which would have been the case for the maamalian eye. (Design can overhaul).
And they're not always 'selected out' of human designs either. As artuculett pointed out recently, why do commercial airliners still have ashtrays when smoking's prohibited? Because it's probably more costly commercially to take them out. They can still go along 'for the ride' though, for a while, just like the human appendix seems to.
In evolutionary algorithms some intelligent agency has to define the fitness criteria, as there is no option for "did it reproduce?" without self-replication.
Completely wrong - see above.
Evolution requires no intelligence* indeed I would argue that with intelligence "directing" it, evolution becomes an evolutionary algorithm+, or (in the case of ID, deistic selection, aka selective breeding).
Evolutionary algorithms require intelligence* in setting the specifications.
Classical engineering requires intelligence* to set the specifications and to perform the design and engineering
Selective breeding uses evolutionary algorithms.
See above.
What is the objection to using the word "develop" as opposed to "evolve" when talking about technology?
I have no objection. They mean exactly the same to me. Do you think that by using different words affects the validity of the analogy?
*maybe intent is a better word, as a stupidly determined specification would also work, just not produce something that anyone would want...
OK with me, so long as it's clear that 'intent' is equally unnecessary for human design, and that random changes would achieve the same 'result', given time.
. So long as you guys keep resorting (and I don't believe it's tactical, but illustrative of close-mindedness) to regurgitation of your previously, clearly unconvincing, counter-arguments, then it's clear, at least to me, that we all need to change tack slightly.
). If it's the same as making entirely random changes (mutations) to an existing design with no forethought or preconception as to the likely effect on that design, then determining whether the amended design is better or worse suited to its environment compared to its predecessor, measured by its ability to outclass competing designs using whatever metrics are approriate (gross sales, profitability, speed, dog fights won/lost, etc.), then fine, I understand. But I suspect it isn't, in which case you've misunderstood my point. Unfortunately, I don't think I can make it any clearer.
And when information can't adapt, then the non-adapting information dies out so that the workable information can be refined and evolve it ways that it is most useful in perpetuating itself.