Intelligent Evolution?

[jimbob]

Some designs did have "irriducible complexity" comparetd to their predecessors.

There is no such thing as irreducible complexity - merely steeper gradients to move from one point in a design space to another.

Fair enough, I was exaggerating for effect, but the probability of a large change being successfull is smaller. That is probably why many organisms demonstrate neotonous traits, as this allows a simple mutation to have a large effect and still have a viable organism.

Then suddenly they all start sprouting jet engines,

And yet we have sea creatures that use the principle of turbulent flow as a propulsion mechanism.

Of course analogous propulsion systems evolved in sea creatures, however jet propulsion did not evolve in aircraft, it was designed. Octopodes and clams seperately evolved these propulsion methods, they did not occur by "intelligent copying".

it is analogous to a crocodile suddenly sprouting gills, or a temperate lizard having a full set of mamalian fur (possible with genetic modification, but not evolution).

I'm afraid you're simply wrong jimbob - a steeper gradient can be overcome in a shorter time with an increase in the severity of mutation.

That it doesn't happen in nature is NOT a fundamental restraint on the abstract concept. It is an observation about the mutation rates in nature.

Notice that I said "mamalian" fur. The chances of structures with identical genetic sequences seperately occuring from mutation in two different organisms after they diverged from their common ancestor is miniscule*. Indeed it has been often mentioned as a way that evolution is a falsifiable theory.

Structures that perform the same task can and do evolve many times, but not using identical genes.

*Won't occur in the lifetime of the planet Earth.


I am not arguing that it is useless to discuss how man-made systems develope over time. But I am saying that if you are using it as an analogy for evolution it could play right into the hands of IDers.

How does pretending that engineered systems evolve, not bolster Behe?

"Further, I find the idea of common descent (that all organisms share a common ancestor) fairly convincing, and have no particular reason to doubt it."

An ID proponent could say

Devil's Advocate: "

If that is what you meant by evolution, with intelligent agents altering the design, then I do agree with evolution

"ETA

If I define the fitness criteria, then I am acting almost exactly as "The Intelligent Designer Who Works Through Evolution, His Wonders To Perform".

Yes, you are. But then words are just words and how you choose to describe something has no impact on what it is. People who want to describe things in terms of gods are going to do so anyway.

By define I mean "set", as in telling the algorithm what the criteria for acceptability are. There are no specifications in evolution, but there are when using evolutionary algorithms.

A deity that sets the fitness criteria would control the direction of evolution. Evolution does not need this, but ID does.

If I want to evolve a better transistor, I will define the selection criteria. Otherwise I will not get a better transistor. Fiorst I need to decide what defines "better". (Cost, speed, size, blocking preformance, on-performance, yield, lifetime, reliability, manufacturability etc.)

 

[jimbob]

I think I've already offered a view as to how design with forethought is simply natural evolution short-circuited. Those 'revolutionary' and 'ambitious' designs might well have occurred purely through incremental change, had the designer been unambitious and extremely patient, not to mention completely uncommercial.

But an evolutionary route from an early engineered system to a more advanced system would follow a different "trajectory" of iterations compared to one that was intelligently designed.

Aircraft:

The Supermarine Spitful was a piston-engined replacement for the Spitfire, and the Wing of the Spiteful was used in the jet-engined Supermarine Attacker (Ah the joys of wiki)

Is it useful to say that the Attacker was a jet-engined evolution of the spitfire? I would say no, because you could relpace "evolution" with "development".

The Spitfire/Spiteful design did not "evolve" a jet engine, it was designed in because the performance advantages of jets were clear.

In the DeHaviland Mosquito, a prototype apparently broke its back on landing. If the design evolved, this would be a failure, so would not "breed". Instead, extra plywood strengthening was applied at just the point of the break. This is the sort of learning from failures that evolution can't do.

It is fundamental to how evolution works.

In the soap nozzle, someone wrote the specification that was then implimented in the selection algorithm. The person would not have known beforehand what the design would be, but they did know what it would do.

 

[jimbob]

Self-replication is IRRELEVANT. If you want to insist that copying done by a machine that is not entangled with the machine being copied is not equivalent then, ultimately, you will have to come to the ludicrous conclusion that because atoms don't split cells don't really 'self' reproduce either. What happens in a cell is lots of tiny machines work on the inside to produce a new copy. If you want to insist that doing it from the outside makes the process fundamentally different then go ahead. It's silly but whatever.

The cell, is, after all just an abstraction for a collection of smaller machines working together.

Abstractions lie.

How is self-replication irrelevent? Self-replication is the only process that removes a predefined and thus arbitary selection criteria on the iterations, and replaces it with the criterion used in evolution. "Does this variant make a self-replicating copy of itself?" Not one that can replicate, but one that actually does.

This does not need to be physical, it just needs to be self-replication with errors; computer worms and viruses could be subject to evolution if they self-replicated with errors.

Perfect copying won't work, as the design is then frozen.

The extra information in evolution comes form "useful mistakes".

 

[jimbob]

No. I've always been consistent ID. And yes, I am being general - that's the whole damn point of abstraction!

Now, let's go slow.

If I have no knowledge about the mechanics of the universe how do I, as a supposedly intelligent actor, design a flying machine?

You could use an evolutionary algorithm, but the selection criteria would include "the closest to getting off the groun, or flying".

You would still know that you wanted a flying machine.

I am assuming that you know enough to spot if the iterations are closer to getting off the ground.

 

[cyborg]

air enough, I was exaggerating for effect, but the probability of a large change being successfull is smaller.

Yes jimbob, which is why it all comes down to the actual numbers, not the particulars of the process.

In nature it doesn't happen because mutation rates don't allow it. (And if they did things would not be stable enough for evolution to occur anyway: none of which matters if you place the process in the right context of course.)

*Won't occur in the lifetime of the planet Earth.

That is, it comes down to the numbers again. Not anything fundamental about the description.

But I am saying that if you are using it as an analogy for evolution it could play right into the hands of IDers.

And, as I said earlier, they don't understand what they are talking about. I do not, therefore, give too much of a fig if they view such reasoning as a way to insert their god as a special axiom because they find 4 + 6 = 10 very appealing to them.

How does pretending that engineered systems evolve, not bolster Behe?

Well I'm sure it would bolster Behe but then he's already come to a conclusion.

Other people might better spend their time pondering the fundamental mechanisms of developing machinery (both those biological and non-biological) rather than getting all excited about whom it was developed by: as if that was the most important question.

How the **** do I persuade Behe that question doesn't really matter? Beats me but saying I'm playing into his hands doesn't make it so.

There are no specifications in evolution, but there are when using evolutionary algorithms.

Again I ask you to consider jimbob what is the difference, in your mind, between this situation:

Change design X using evolutionary algorithm to satisfy parameters Y.

And this one:

Change design X using evolutionary algorithm to survive simulation Y.

A deity that sets the fitness criteria would control the direction of evolution. Evolution does not need this, but ID does.

Yet again jimbob if someone is determined to insist a god was involved then they will insert it as a special and totally useless axiom in the process anywhere they deem fit: a useless lie.

And, if the history of this behaviour is anything to go by, they will keep on pushing it back in the process as far as their acceptance of science will allow.

If I want to evolve a better transistor, I will define the selection criteria. Otherwise I will not get a better transistor.

Au contrairé - you could define the selection criteria for a better computer and maybe get a better transistor in the process - or maybe something totally un-transistor like. Or maybe you just define the selection criteria as whatever the market will buy - and if you can extract a definite selection criteria for that then I suggest you enter the investment market.

Really all you have to do to achieve this is back off the strictness of the goal. That's really what we see in nature: the strictness of what constitutes a successful goal is so loose that we don't even identify them as such any more. That's where 'natural' is handed off from 'artificial' - but you've always got to bear in mind there will be grey areas in the middle where making a choice as how to describe it looks very arbitrary indeed.

 

[jimbob]

I wasn't asking how to persuade Behe, but people who might be convinced by his arguments.

It is not a useful lie, it is an unhelpful lie.

 

[cyborg]

How is self-replication irrelevent?

Because, as I've tried to point out earlier, it doesn't really occur!.

Cells do not 'self-replicate' if you are going to be strict about it. New cells are built by tiny cell machines that pump out copies of other tiny machines that are not copies of the machines building those machines itself that then go on a build more stuff etc...

The important part here is REPLICATION - i.e. a reproduction of something recognisable as belonging to the same or similar class as something else.

Perfect copying won't work, as the design is then frozen.

The concept of a 'perfect' copy has nothing to do with the replication process. The argument being made is that if we cannot identify a reproduction as being capable of producing itself then it cannot, essentially, be copied. This is clearly a nonsense.

I am assuming that you know enough to spot if the iterations are closer to getting off the ground.

A posteriori, not a priori. That's dumb, not smart.

 

[cyborg]

I wasn't asking how to persuade Behe, but people who might be convinced by his arguments.

Anthropocentric thinkers are likely to think that. You appear to be in this category from what you are saying.

 

[Unalienable]

Technological development, to my mind, is closely analogous to natural selection. Small but significant changes are made here and there over time such that each new variant becomes more and more sophisticated and capable. The ‘obsolete’ variety becomes less efficient or capable and is ‘discontinued’.

I think most ID proponents will jump down your throat on that one. This is the big difference between evolution and technology, and you dismiss it as "analogous".

With technology, we can make very tiny improvements, but we can also make completely sweeping changes based on entirely new theories. E.g., you can look at the evolution of the car, and it's easy to see how 1969 Ford Mustang was the result of making a lot of tiny changes to a model-T Ford. But where would the Mazda RX-7, with it's rotary engine, fit into your phylogeny? That wasn't a tiny change of anything, it was a sweeping change. The Mustang has parts directly analogous to parts found on the Model-T, but the rotary engine has parts which are not to be found in any car before it, and not just one part, but a number of them, arranged in a completely new way.

I guess to keep your analogy fit, if cars are like insects, then the Mazda is like an arachnid--to the uneducated an arachnid looks like just another kind of insect but when you 'get under the hood' you realize it's quite different.

It’s widely acknowledged that many major advancements in technology and medicine have come about through chance or fluke. That’s analogous to mutations in my mind.

That's a stretch. Serendipity (like penicillin) happens now and again, but it's not the driving force. As a software designer, I can tell you that if I type "+" when I really meant "-" it's very unlikely that I am going to make an improvement and certainly not a revolution.

Probably a tighter fit would be that mutations are analogous to new ideas, but of course that would undermine the point you're trying to make.

 

[cyborg]

You guys who seem to think 'revolutionary' design changes 'just happen' and don't involve many years of detailed physical analysis (effectively creating limited models of the world where designs can be 'cheaply' tested) and prototyping (the expensive type of testing that only occurs once you're relatively sure it will work).

I have no idea where you get these bizarre ideas that this is how human engineering works. It doesn't. This is one reason why the whole ID debate is stupid - it's propagated mainly by people who have no idea what designing a successful system actually entails. And I know from experience that successful systems rarely spring up from spec to implementation unless they're really basic. There's a whole history of design failures to back me up on this too.

 

[Unalienable]

There is no such thing as irreducible complexity - merely steeper gradients to move from one point in a design space to another.

I question that. Perhaps you mean to say "we have never witnessed irreducible complexity in nature", in that case I might agree (although I'm hardly qualified to judge).

Behe's definition of "Irreducible complexity" is this:

A system composed of several well-matched, interacting parts that contribute to the basic function, wherein the removal of any one of the parts causes the system to effectively cease functioning.

Take any digital computer, for example the original ENIAC--that system, in my estimation, is irreducibly complex. The ENIAC had millions of parts and if any of them were removed, the system would either malfunction severely or not function at all.

You might argue "there are some parts that could be removed and it would still work", you would be technically correct. You could take out screws from the backpanel, for instance. So go ahead and remove those parts, one by one, until you can't remove any more without it being nonfunctional. Once you get to that point, what you've got left, is the irreducibly complex core of the ENIAC. And you'll still be left with millions of parts.

In what sense does the ENIAC not meet the definition of "irreducibly complex" or am I simply misinterpreting your assertion?

 

[cyborg]

The ENIAC had millions of parts and if any of them were removed, the system will either malfunction severely or not function at all.

There's always a physical 'function'. The terms 'malfunction' and 'no function' are really just descriptors about how something falls shorts of the desired function.

Since there's no desired function in evolution the term is quite useless.

Of course it is just as easy for me to state I 'desire' this 'malfunction' as it is for you to state you do not. This is the real problem with the notion of irreducible complexity: it merely asserts that if you change the design enough it's, uh, no longer the design you started off with. Wow. Thanks for that Behe.

And that's the point really: unless you're talking about irreducible complexity in terms of how some change to a design affords one function compared to another then it's relevance to the evolution of some design is moot. And since it's generally framed in terms of the creation of some new, identifiably modular, functionality as arises by the change in design if there is no limitation on how the design can change there is nothing irreducible about it.

 

[jimbob]

I do have a fair idea of how engineering works, and sometimes there are "revelations".

In normal engineering, there are plenty of cases where causes of failures are corrected. The failure is analysed, the cause found, and the remedy applied. This is not an evolutionary approach, as information was obtained from the failure.

 

[cyborg]

sometimes there are "revelations".

Examples - and not of the 'it appeared just by magic at the consumer stage' variety. Or of the, 'random discovery found and then marketed' variety. No, I want an example of something totally hitherto unthinkable that went from idea to product in a flash with no messy development bits in the middle.

This is not an evolutionary approach, as information was obtained from the failure.

A priori from a posteriori. Now you're getting somewhere.

 

[jimbob]

I am not arguing for the lack of development. I am saying the development is not evolution.
EDIT: Earlier post of mine hidden in spoiler for clarity

Nobody is arguing that engineered systems do not develop; I (and I imagine ImaginalDisc) am simply aguing that their incremental development over time via human designers is not evolution.

The mechanism is completely different.

If you are showing that complex systems can (be deveolped/develop) from simpler ones, then your analgy is fine.

My question is whether anyone is actually contesting that? (Some extreme biblical literalists?)

I do know of Intelligent design proponents who say that evolution couldn't work, and the (obvious) incremental changes need (at least*) one designer.

I can not see how your analogy helps in this situation, as the examples given
had many designers contributing to the development of the systems.

*this is implicit, so they can pretend to not be putting forward a solely fundamentalist Christian viewpoint.

Of course there is a lot of development. Some of it theoretical, some of it practical. But unless evolutionary algorithms are used, evolutinary mrethods are not used.

Evolutionary algorithms are good analogies for evolution, just like selective breeding, but the direction is defined for the specifier's benefit. They are still not evolution, though they share many common features.

ETA.

Even "revelations" require a lot of work to impliment them. The idea behind the first TV was a novel idea, but it required a lot of work.

 

[Soapy Sam]

It's a side issue, but I honestly doubt a decent Roman Army engineer (for example) would be stupefied into silence by all modern technology. He would be baffled by electronics (though no more so than maybe 50% of the current population), impressed by internal combustion power and frankly contemptuous of much modern carpentry and building "skills".

 

[Unalienable]

OK, so let me get this straight. The ENIAC is wired with a program to output the first 10,000 digits of e, and I claim it's irreducibly complex, meaning that if you remove any part, it will not be functional.

You then take me up on that challenge. You remove a wire, the machine stops working, but you declare "I removed a part, and now it's still 100% functional, except that its new function is to press against the floor with 30 tons of weight, while generating heat."

It seems to me that you could summarize your position thusly: "Functional is a meaningless word."

To some people this would seem bizarre, but only because they are human beings who are trianed to see the world in terms of function: my heart is a pump, my liver is a chemical processing factory, etc. But to nature, your statement is entirely true, for nature is blind to human concepts like "function", and nature "desires" nothing.

Am I following your train of thought or have I derailed somewhere?

 

[cyborg]

Of course there is a lot of development. Some of it theoretical, some of it practical. But unless evolutionary algorithms are used, evolutinary mrethods are not used.

Well as I said before tautologies aren't helpful for anything other than defining the landscape of discussion.

The point here is that the assertion is: "Intelligent design is so different because you can build things wholesale without having to bother with anything messy like the trial and error of evolution."

The point is that intelligent design, at least as done by humans, is messy. The real qualifying point is that a posteriori knowledge about design gets fed into a priori models so we can avoid some of the messiness next time around. How much we can avoid is entirely dependent on how good those models and development techniques are.

As such the ID crowd is completely wrong from the get-go about how they think intelligences design things in the first place. Of course, they say, a god, being infinitely intelligent, would not have these problems (which would be perfectly true but, again, a completely useless lie because it can't tell us anything interesting). But that's not the point. The point is that their whole argument is built up on a false impression of what it means for humans to create things. The whole, "painting needs a painter" argument.

Evolutionary algorithms are good analogies for evolution, just like selective breeding, but the direction is defined for the specifier's benefit.

Yet again jimbob I ask you to consider what it means to start with a simple, obvious goal in some evolutionary algorithm to expanding that notion such that goals are only implicit in some simulation that selects.

It's still entirely determined in the computer. The difference is how hard it is to construct a nice, easy and simple to formulate explanation of what the 'goal' is.

Am I following your train of thought or have I derailed somewhere?

Pretty much right. The really big distinction in human vs natural design is that humans design things so that the parameters of design are easily distinct and hence easy to vary so you can modify the design: e.g. I want my car to have a different colour - simple variable to change in the design.

Natural design, having no such desire to have easily separable design parameters, tends to produce things with highly entangled parameters. Want to change skin colour? Whoops, now you've affected a whole host of other things you didn't want to. Try again. Oh dear, now something else has been affected. And so on.

This is why, as you should understand, software design avoids things like 'spaghetti code' as a bad design process. It's not that spaghetti code can't work: it's just that trying to modify it is so much more difficult.

 

[jimbob]

The point here is that the assertion is: "Intelligent design is so different because you can build things wholesale without having to bother with anything messy like the trial and error of evolution."

The point is that intelligent design, at least as done by humans, is messy. The real qualifying point is that a posteriori knowledge about design gets fed into a priori models so we can avoid some of the messiness next time around. How much we can avoid is entirely dependent on how good those models and development techniques are.

As such the ID crowd is completely wrong from the get-go about how they think intelligences design things in the first place. Of course, they say, a god, being infinitely intelligent, would not have these problems (which would be perfectly true but, again, a completely useless lie because it can't tell us anything interesting). But that's not the point. The point is that their whole argument is built up on a false impression of what it means for humans to create things. The whole, "painting needs a painter" argument.

Ah,

I am working form a slightly different premise:

I would say that the situation as described above isn't "ID" but "perfect creation" (I don't know the correct term, but I imagine that you understand what I mean). In such a situation, showing real examples of imperfect iteritive development could be useful.

As far as I see it, ID proponents accept evolution wherever they really can't avoid it, but magically claim that a designer miust have guided it. I imagine that as they acept common descent of man and chimp, that their contention is that "god had a [explicit] goal of sentient worshipers, and organised evolution to reach that goal".

This group of ID proponents (the most vocal group, including Behe) accept messy development, and (implicitly) "guided evolution". What they don't get, is that this guiding is not needed. Indeed any guiding intelligence must be pretty sick given the many wonderous parisites.

I do not think we could change Behe's belief, any more than Kleinmann's, but we should be wary of inadvertantly lending them spurious credability with weak analogies.

 

[cyborg]

I'm not interested in trying to convince small minds of their smallness anymore. Let them live in ignorance.