« The Gospels Aren’t Ancient Fiction | Main | Challenge: A Real God Wouldn’t Have Let His Chosen People Suffer Defeats »

November 30, 2015

Comments

In the world of advertisement, whenever the company speaks of the evolutionary advancements of the product ...

these are merely improvements in the design.

Correlations in these two concepts exist.

Intelligent design IS the marks of intelligence in living things

Tim, Tim, Tim. You need to think about this.

My teeth are a mark of intelligence because they chew well, yes?

And so a mark of intelligence on an antenna would be its functioning well as an antenna.

Do you not say so?

Alright then. Assuming that...

The (excellent) antenna pictured here was not designed; it evolved.

By that I mean that no person thought about electromagnetics for a while, considered the particular performance required for this antenna, and then intelligently combined these ideas to decide where to bend the wire, in what direction, and by how much.

No. The antenna is the product of an (unguided) 'genetic algorithm' or 'evolutionary computation'.

Evolutionary computation mimics biological evolution.
Evolution requires: 1) reproduction 2) variation 3) selective Pressure.
Evolutionary programming simulates these 3 components.

The article above and this one describe the components of the analogy so well that I won't.

To be sure, the program was designed (to mimic biological evolution).

But the antenna evolved.

So maybe ID and evolution can both be true. That is...

The designed thing was biological evolution. And the many forms of life and their various features are the products of evolution. The 'marks of intelligence' you see are actually marks of evolution.

The designer of evolution was intelligent, but did not know what evolution would produce.

My question to you: Why not?

This is the question you set out to answer.

But you haven't.

________________

To whoever might reply, please remember that the topic is design/evolution.
Abiogenesis is a different topic.
Tim, if I understand him correctly, has asserted that arms, legs, and eyes are marks of design because they work. The antenna, I say, works yet is not designed.

Reason affirms (actual/ontic) *evil* in the pains of "The Adamic's" privation vis-a-vis the evolutionary schema. Metaphysically speaking, therefore, Naturalism is disqualified from the ground up. Whereas, the peculiar metaphysical paradigm of Christianity is, on all fronts, affirmed.

Even if "goo to you via the zoo" were an accurate representation of how God directed and brought about the diversity of life on earth, the strict definition of evolution being an unguided process puts it in contradiction with an intelligent design theory.

Could an intelligent designer design evolution to run unguided?

Could an intelligent designer design evolution without knowing what would evolve?

Could an intelligent designer design evolution to run unguided?

What do you mean by unguided? If it is designed, then it isn't unguided by its nature.

Could an intelligent designer design evolution without knowing what would evolve?

Computer scientists do this all the time when they write software. It's almost always the case that the intentions for a program do not align with the actual programming, at least without further debugging. Perhaps the question is whether a supremely intelligent being could design a system and not know the outcome of that system?

By that I mean that no person thought about electromagnetics for a while, considered the particular performance required for this antenna, and then intelligently combined these ideas to decide where to bend the wire, in what direction, and by how much.

But, this is exactly what happened. People thought about electromagnetics and wrote down equations to measure the performance of an antenna. Then, they wrote software to simulate the process of building different antennae, to measure the performance of an antenna, as well as algorithms to select better designs. In the end they got their antenna, but it was 100% by human design.

Saying that a few steps in the process vaguely mimic a biological process doesn't make it unguided in any way.

This article explains some of the issues that have arisen with using metaphors for computational procedures.

The product of computer programs isn't a guided, programed, product?

By definition its calculations cannot be non-calculations, hence we're stuck with our own Mind's messy ties. "Predicting" how X got to Z merely affirms that Mind can thusly calculate natural forces.

But the Christian ALREADY knows that about Mind and Nature. In fact, he PREDICTED it.

To claim calculations are free from calculating is simply an unnecessary streatch given that we ALREADY KNOW that Mind Plus Matter can 1) calculate life retrospectively as well as 2) calculate life prospectively, and 3) build life. From the ground up. Both with and without antennas. At least, the Christian has always known that about Mind and Matter. The Naturalist not so much.

The Naturalist in fact seems AMAZED at this finding, as if it's not ALREADY predicted by the Christian's truth claims.

Every time the Mind of Man messes with that pesky benchtop and pushes dirt around, the Christian sits back, relaxes, and awaits yet more affirmation of his paradigm.

Affirmation -- such as (actual/ontic) evil in possible worlds outside of Eden, outside of The Adamic standing between such possible worlds.

The Atheist shouts, "It's all so Evil!".

Affirmation.

The Atheist shouts, "Predictions & Calculations!"

Affirmation.

Nathan,


Do you understand the difference between simulating an antenna and designing one?

You are right that they wrote code to simulate the performance of the antennas. This code is analogous to the lives of biological individuals.

You are right that they wrote code to select the better performing antennas for reproduction with random variation. This code simulates mutation and natural selection.

There is no reasoning about antennas in the code.

There is no mathematical derivation of the form of an antenna from the requirements.

There are no design rules in it.

No design equations.

The closest thing to design is the form of the first generation.

Even that could be randomly generated.


Does your article show or even say otherwise?


Predicting how an X got to, or became, a Z merely affirms that Mind can thusly calculate natural forces.

Which affirms the predictive power of the Christian's truth claims.

What do you mean by unguided? If it is designed, then it isn't unguided by its nature.

The algorithm is designed.

The results are not.

In fact, the more of these predictions and calculations we can put together, the more we discover just how well Mind can "see" natural forces. In directions proximal and distal to the Now.

Sight is an odd capacity.

Especially in a paradigm which has no such metaphysical accounting.

Ron,

Apologies in advance for a long response.

I would use these definitions:

Simulation is the process of evaluating the performance (fitness) of an antenna.

Design is the process of changing the state (of the object in question) until the design objectives have been achieved (or maximized).

If you take away the ability to simulate and evaluate performance, the evolutionary algorithms will be entirely useless. They won't be able to produce a thing. Every evolutionary algorithm needs a fitness function and a representation of the world that they are trying to optimize. Fitness functions are human-designed functions that tell the algorithm what it is looking for.

Representation is also very important. If you have the right fitness function but the wrong representation, you won't get results either. Representations are human-designed and are often optimized for the algorithm that is being used to solve the problem.

I claim that the high-level algorithm which randomly makes changes to a design, and then tests to see if those changes were beneficial, is performing a design process. A human might do the same thing when incrementally testing an antenna. The computer is, in a computational sense, reasoning about what properties are useful in an antenna by trial and error - just much faster than a human could. The computer will also consider many things a human wouldn't because of its computational power.

Some algorithms will be better than others at preserving the best parts of the current antenna, and they will be more effective at finding good solutions (designs). This is why there are many algorithms (besides evolutionary algorithms) for such problems. Algorithms have implicit assumptions about the nature of the problems they are solving and often have to be finely tuned to produce results for a particular problem.

Saying there is "no reasoning about antennas" about the code is a difficult claim, because we have to precisely define what it means to reason. Is Deep Blue thinking when it plays chess? Would you claim that a program cannot play chess well because it can't reason about the game? In many ways Deep Blue is not much different than the antenna design problem. Both programs analyze a large space of possibilities and select that which they think is best according to a (human-supplied) fitness function.

There is nothing special about the fact that computers go through this process differently than humans -- computers have different strengths than humans, and we regularly exploit that fact.

When you deeply understand the computer running the software and the software itself, it is clear that the intelligence needed to create the antenna was entirely human designed. The computers did the process faster than humans could, but that's not because the computers were intelligent. It's because intelligent humans built the computers, built the software, and figured out how to specify the problem and run it on the computer.

With artificial intelligence these days we are getting better at letting computers do more of the heavy lifting, but there is not a single example where the intelligence resulting from a computer was not a product of the people who built and designed the software and hardware. (This doesn't mean that we don't get unexpected results - this happens because of humans in the process. But a computer only does what it is told to do.)

Thus, the antenna, as an example, actually illustrates the need for designer.

What would be amazing to me is if you turned on a brand new computer with random programming and random memory initialization and it suddenly showed you this antenna on the screen.

Could an intelligent designer design evolution to run unguided?

So this “intelligent designer,” under this scenario, would be smart enough and powerful enough to actually design “evolution” but somehow recuse himself of guiding it. (Let’s set aside, for now, all the power that such a being would most likely have).

Let’s just say he’s not that interested in guiding life. He could if he wanted, but he doesn’t want to guide. It bores him.

So he designed the process (“the algorithm”), but he’s just as surprised as we are that humans came about at all (“the outcome”).

He’s also surprised that human beings now have one nose and two ears, ten toes and two nipples, etc.

One could ask what the definition of “guide” is under this scenario and if it matters at all.

What if a creator is so good at guiding that it doesn’t even look like guiding? Wow. Crazy.

There are lots of silly questions one could ask.


Simulation is the process of evaluating the performance (fitness) of an antenna.

This doesn't really matter much to our discussion, but no.

Simulation (I agree with Wikipedia) is the imitation of the operation of a real-world process or system over time.

You can (and often do) evaluate the results of a simulation.

But imitation is a defining element of simulation while evaluation is not.

There are two simulators involved in our discussion: the overall evolution simulator and the antenna simulator (which is part of the evolution simulator).

Evaluation (natural selection) is part of evolution.

So evaluation (comparison to the requirements) is part of the evolution simulator.

Evaluation is not part of an antenna.

So the antenna simulator doesn't evaluate anything; it just imitates an antennae.

Design is the process of changing the state (of the object in question) until the design objectives have been achieved (or maximized).

No, Wikipedia again: Design is the creation of a plan or convention for the construction of an object or a system (as in architectural blueprints, engineering drawings, business processes, circuit diagrams and sewing patterns).

During design, the 'state of the object in question' might 'change until...' as you say.

But changing drafts are not a defining element of design - sometimes the first design is final and optimal.

One defining element of design is plan.

Nobody planned the NASA antenna. Think about that. Admit it.

So lets get this straight.....

A Person can sit at a computer and ultimately predict, demonstrate, map, the path of [Biological System A] getting to, becoming, [Biological System Z].

Both backwards in time and forwards in time.

But wait -- don't we already know that about "Minds amid Nature's Forces".

This is sort of old news, really. For the Christian at least.

RonH,

All of this was predicted, ages ago, by the Christian. Known even.

Think about that.

Admit it.

KWM,

I think the reason for using evolutionary programming in the case of the antenna was precisely because the people who needed the antenna knew how to simulate antennas but not how to design the one they wanted.

They HOPED to get what they wanted in terms of performance by letting it evolve.

They had no idea what it was going to look like.

Even after the winning antenna evolved, the programmers would probably think it looked like a random bent paper clip.

And if you tweak the requirements a bit, the programmers will probably not know how to tweak this antenna to meet the new requirements - they will probably evolve a new one.

If I design a widget factory, are the widgets it produces products of design?

WL,

The widgets are a product of the factory.

You said you designed the factory.

Are the widgets designed or just the factory?

I asked whether the widgets are a product of design.

Let's leave to one side whether they are, therefore, themselves designed.

It seems to me that the widgets are a product of design.

I should have also noted that the factory is sophisticated enough to crank out widgets, and even improve their functionality, without intervention (except that if you destroy the factory, it will obviously not produce widgets anymore).

RonH,

Let’s assume a baker designed cake batter (a new recipe for the first time) and put it in the oven to bake it. Let’s further assume he also designed the oven (a very skilled baker).

The baker doesn’t know what the cake will look like or how it will taste.

Did the baker design the cake or just the batter and oven?

Instead of, “Compliments to the chef,” should it be, “Compliments to the oven”?

WL and KWM,

I said above that plan is a defining element of design.

This plan implies some level of understanding of the connection between, say, the widgets dimensions and its performance. Or between the cake's ingredients and its texture when baked.

The software used in the making of antenna has none of this.

The software makes random changes on the best performers and selects survivors for the next generation based, not on any plan, but purely on their simulated performance.

It matters, WL, what kind of sophistication allows your factory to improve the widgets over time. If it's like the antenna software, then it's not planning and not designing.

But if the factory chooses modifications in some way based on experience (beyond mimicking natural selection like the antenna software) then maybe it's planning and designing.

Being a 'product of design' is nothing if I understand what you mean by the queer-sounding 'un-designed product of design'.

The poor-performing rejected antennas are products of the (designed) software as much as the ones that perform well. What of it?

If the baker, chose the ingredients as part of a plan to make the cake come out a certain way, then he designed the cake.

My wife has baked almost all the bread at our house for a long time. She knows the effects of rise time, kneading, water, sugar, etc. She can, therefore, design/plan a loaf. Complements to her.

If I adapt the antenna program and eventually get a fabulous loaf of bread, then complements to me.

Not for designing the loaf, because I know little of baking, but for designing the program.

Ron,

Be careful about claiming that random changes are being made to the antenna. The changes are not truly random - they are constrained by the representation of the antenna, which is highly crafted by the human designers based on their past experiences. (As well as their understanding of the connections between antenna design and performance.)

From the original research paper:

To encode a single arm of the antenna, the representation that we used consists of an open-ended, generative representation for “constructing” an arm. This generative representation for encoding antennas is an extension of our previous work in using a linear-representation for encoding rod-based robots. Each node in the tree-structured representation is an antenna-construction operator and an antenna is created by executing the operators at each node in the tree, starting with the root node. In constructing an antenna the current state (location and orientation) is maintained and operators add wires or change the current state. The operators are as follows: forward(length, radius), add a wire with the given length and radius extending from the current location and then change the current state location to the end of the new wire; rotate-x(angle), change the orientation by rotating it by the specified amount (in radians) about the x-axis; rotate-y(angle), change the orientation by rotating it by the specified amount (in radians) about the y-axis; and rotate-z(angle), change the orientation by rotating it by the specified amount (in radians) about the z-axis.

An antenna design is created by starting with an initial feedwire and adding wires. The initial feed wire was set to start at the origin with a length of 0.4 cm along the Z-axis. In addition the radius of the wire segments was fixed at the start of a run, with all wire segments in all antenna designs having the same radius. To produce antennas that are four-way symmetric about the Z-axis, the construction process is restricted to producing antenna wires that are fully contained in the positive XY quadrant and then after construction is complete, this arm is copied three times and these copies are placed in each of the other quadrants through rotations of 90◦/180◦/270◦.

The fitness function used to evaluate antennas is a function of the VSWR and gain values on the transmit and receive frequencies. The gain component of the fitness function uses the gain (in dBic) in 5◦ increments about the angles of interest – from 40◦ ≤ θ ≤ 90◦ and 0◦ ≤ φ ≤ 360◦ – and consists of a gainerror component and an gainoutlier component. The gainerror component of the fitness function is a modified version of the Least Squares Error function, and was later modified to evolve the antenna for the revised mission specifications. The gainoutlier component is a scaled count of the number of sample points in which the gain value is below the minimum acceptable. The VSWR component of the fitness function is constructed to put strong pressure toward evolving antennas with receive and transmit VSWR values below the required amounts of 1.2 and 1.5, reduced pressure at a value below these requirements (1.15 and 1.25) and then no pressure to go below 1.1.

The three components are multiplied together to produce the overall fitness score of an antenna design:

F = vswr × gainerror × gain outlier

The objective of the EA is to produce antenna designs that minimize F.

If you read on, you find that their initial design didn't work. They had to revise the specifications and re-design the antenna as a result. But, overall the "construction process is restricted" to produce viable designs. Furthermore, they write here that "the fitness function is constructed to put strong pressure toward evolving antennas with... (the desired design properties)." This is intentional design by the humans making the system.

You seem to be claiming that design by trial and error without some level of understanding is not design. Yet, this is what Edison used in experiments to design batteries, and to a lesser extent, the light bulb. If he had to do 10,000 experiments, it isn't clear that he had a good understanding of what he was doing.

Nathan,

What constraints do you think the representation imposes?

Some of the proponents of intelligent design, such as Michael Behe accept the common descent of species, thus humans came from earlier forms of homoids. So when you guys talk about intelligent design are you agreeing with common descent but it was all planned out? Does everyone in this blog accept the earth is billions of years old? If yes, how do you reconcile that to the first couple chapters of Genesis?

I'm not getting on anybody's case mind you, I am just interested in hearing what you believe.

If you care to reply thank you in advance.

Flying Donkey,

I'm an old Earth type, and don't see any conflict there with Genesis. The Reasons to Believe web site has quite a bit of material on this.

I think common descent, or "goo to you by way of the zoo", is wildly implausible. I'm very skeptical about that. All of these very different body plans springing from a common ancestor? Please. Seriously?

Organic molecules self organizing into living organisms is even more unlikely, kind of like an Atheist "just so" story.

A belated answer to RonH,

"Could an intelligent designer design evolution to run unguided?"

Yes. So?

"Could an intelligent designer design evolution without knowing what would evolve?"

Sure. Again, so what?

Goat Head 5,

So What?

So...

Can Intelligent Design and Evolution Both Be True?

The original question....

"Can Intelligent Design and Evolution Both Be True?"

I suppose so, depending on how you define both terms.

However,

Evolution in the current sense of the word, speciation through random processes coupled with natural selection, begun by Organic molecules self organizing into living organisms....

Intelligent design meaning special creation by a God defined like Brad B. would, with some "micro" evolution making sub species...

No. I don't think they can coexist. Not defined like that.

If you started with a super intelligence who designed a group of original organisms and processes, and then let it all go with maybe occasional intervention... evolution could coexist with that. In that sense, evolution itself would be a designed process.

But even this would be very offensive to those who use evolution as the Atheist creation myth, and those who use Science as their primary life organizing principle.

If you started with a super intelligence who designed a group of original organisms and processes, and then let it all go with maybe occasional intervention... evolution could coexist with that. In that sense, evolution itself would be a designed process.
This is evolution as the sophisticated widget factory I was imagining above.

And you're right GH5, of course atheists cannot countenance that. They'd be winning the battle, but losing the war.

Thank you for the reply Goat Head, I appreciate it.

When time allows I'll check out the articles on this site.

May I ask (this question is addressed to anyone who cares to take the time to answer) how evolution or ID reconcile to Adam being made out of dust and then Eve out of Adam's rib? This is a metaphor or something like that?

GH5,

If you started with a super intelligence who designed a group of original organisms and processes, and then let it all go with maybe occasional intervention... evolution could coexist with that. In that sense, evolution itself would be a designed process.

That is something like what I have been saying.

But even this would be very offensive to those who use evolution as the Atheist creation myth, and those who use Science as their primary life organizing principle.

I guess I'm not in this category, though.

WL,

factory is sophisticated enough to crank out widgets, and even improve their functionality, without intervention (except that if you destroy the factory, it will obviously not produce widgets anymore).

The NASA antenna team did this, no?

And you're right GH5, of course atheists cannot countenance that. They'd be winning the battle, but losing the war.

Here I am, defending the notion that loses me the war. I guess.

Actually, I don't defend it or believe it.

All I'm saying is there's no contradiction in it.

Not if the designer designed evolution rather than its products.

Nathan,

You said

The changes are not truly random - they are constrained by the representation of the antenna, which is highly crafted by the human designers based on their past experiences. (As well as their understanding of the connections between antenna design and performance.)

No.

The representation simply represents the shape of the antenna. How long is each segment? What diameter (radius) wire is it made of? At what angles does it attach to the previously added wire?

Next...

You said...

If you read on, you find that their initial design didn't work.
In what way did it not work?

The representation simply represents the shape of the antenna. How long is each segment? What diameter (radius) wire is it made of? At what angles does it attach to the previously added wire?

I should say:

These numbers are, with a couple of exceptions, randomly generated and mutated by the evolutionary algorithm - not chosen by the programmer.

Exceptions: The programmers chose to use the same diameter wire for every segment and they chose the length and orientation of the first segment.

The NASA antenna team did this, no?
Sure.

I brought up the widget factory to bracket a lot of the details that I don't think have a lot of philosophical import.

In response to the video, the "it looks designed, therefore it was designed" flaw is writ large in this.

Because this is where 'intelligent design' fails. It provides no rigorous testable basis for detecting design in subject items of unknown provenance. Therefore claims of 'design' are just plain false.

So there's no point in comparison. Evolution has broad and deep evidential support - ID has none.

How to Test for Intelligent Design

I'm sorry Amy, is that meant to be a response to my point? How does your link address the significant issue facing ID i.e. detect the 'design' it purports to explain?

Ron,

In what way did it not work?

Their design parameters changed (they needed something that worked while spinning), and the original design didn't work under this environment. The humans in the process had to significantly re-purpose the design parameters and fitness function to get the system to produce what they wanted.

The point is that the software is just an optimization tool that takes design parameters and makes local changes until the parameters are optimized. Without the humans setting the design parameters, it cannot produce anything useful. Despite the software, it still took them three months to design the antenna.

The representation simply represents the shape of the antenna.

Somewhat - more on that below. But note that there are thousands of possible representations.

For instance, they could represent an antenna as a set of start/end coordinates for each segment. But, this would likely fail, because most antennas would end up not being connected, and mutations would not produce usable results.

Note also that in the initial design they restricted the antenna to stay in the positive quadrant so that they could place four antenna next to each other. Removing this design constraint would likely produce too many incorrect designs.

But, if you read their previous work, you see that their full representation is much more complicated than just arm lengths. These are just the operators that are used as part of a high-level language describing the antenna. This is clear because they were able to create splits in the antenna (look at the picture of their initial design), which you can't do by just specifying angles. Their high-level language is designed to work well with evolutionary algorithms.

It is easy to look at the last step in the process and say "wow, look what evolution can do". But, this ignores all the things that didn't work, and we only see a small fraction of the engineering and design that went into the system to make sure that the evolutionary algorithm would run and produce a reasonable result.

Ultimately, the humans are responsible for the antenna - the computer just does the final step of sifting through millions or billions of combinations to find those that are best. It is human intelligence that figures out how things must be set up to make that final step possible.

Getting back to evaluating your original statement (the antenna was evolved not designed), we have to deal with the semantics of both design and evolution.

The term 'evolution' (for what went on in the antenna design) is a metaphor that makes it easier for scientists to explain what went on. But, in the end it is only a metaphor. In the words of the antenna designers, what they did is a "stochastic search and optimization technique". It is "inspired by natural evolution", but it is not the same thing. There are dozens of stochastic search techniques (hill climbing, simulated annealing) each of which uses a different metaphor to describe how it works. We shouldn't be confused by the metaphor. (Hence my first link in a previous post: Metaheuristics—the metaphor exposed.) If someone had invented harmony search before evolutionary algorithms, we'd say that harmonies were used to refine the antenna, not evolution, and we might not be having this discussion.

As for 'design', continuing from the initial definition you quoted from wikipedia: It may involve considerable research, thought, modeling, interactive adjustment, and re-design. (...) In engineering, design is a component of the engineering process. By this broader definition, the antenna was either designed or engineered.

This is to echo a bit of what has been said here elsewhere: ID and evolution could both be true - but evolution would have to be defined as a designed process and not a product of random chance, as is illustrated by the antenna.

Nathan,

Their design parameters changed ...

No.

The 'mission requirements' changed.

Mission requirements are analogous to the environment an organism has to live in.

The mission requirements say nothing about design.

The fitness function expresses the mission requirements in mathematical terms.

So the fitness function is also silent on design.

You'll have to get these things straight to understand what's going on and stop making such big mistakes.


In response to Mike's comment, the "it looks like it passed the test, therefore it passed the test" flaw is writ large.

their full representation is much more complicated than just arm lengths. These are just the operators that are used as part of a high-level language describing the antenna.

No, no, no.

A genetic representation is analogous to a chromosome or the genome or the genotype. It just tells you how to cut and bend the wire to make the antenna. It IS just segment lengths and orientations.

No. The operators are not part of the representation, they are analogous those things that make offspring vary from parents: like mutations and biological crossing over.

Bonus:

In biology, the phenotype is the set of observable traits of an individual).

In this case, the phenotype was both the shape of the antenna and its performance in the simulator - the value of the fitness function.

Wow RonH,

You are open to a designer?

What would it take to convince you?

@Wisdom Lover

I didnt understand your response.

But I assume you disagree with me and I further assume that you disagree with me on the importance of having a rigorous basis on which to detect 'design' as ID posits it

I would note that IDers know this which is why you get ill defined concepts like 'Complex Specified Information' as an attempt to answer that thorny issue

GH5,

Wow RonH, You are open to a designer?
Yes.
What would it take to convince you?
Not the topic here. When it is the topic, maybe I'll weigh in.

Nathan,

I said

The representation simply represents the shape of the antenna. How long is each segment? What diameter (radius) wire is it made of? At what angles does it attach to the previously added wire?

I want to clarify this for you.

The NASA guys used a 'generative representation' as opposed to a 'direct representation'.

The data in a direct representation would be the actual dimensions (the lengths and angles) of the antenna.

The data in a generative representation are code in a specialized programming language.

This code is what gets mutated.

Then, running this code generates the dimensions of the antenna.

The dimensions of the antenna are then fed to the simulator.

The results of the simulation are then used to calculate the value of the fitness function.

Highly fit antennas get selected to reproduce; their representation (the code) is mutated.

Lather. Rinse. Repeat.

Direct representation => Mutate. ........... Simulate. Select. Repeat.

Generative representation -> Mutate. Generate. Simulate. Select. Repeat.

So the generative representation is an indirect representation of the dimensions of the antenna.

Using a generative representation is software design, not antenna design.

If you have programmed, you know the value of loops: they save you from repeating the same code over and over again.

A generative representation allows the same thing.

A generative representation of, say, an 800-legged robot could be the same size (in bytes) as a generative representation of a 4-legged one (instead of being 200 times as big).

There are other benefits too.

Could the NASA guys put antenna engineering or robot engineering into a generative representation (or into a direct representation)?

Yes, they could.

But you can't assume they did put antenna engineering into the representation just because they used a generative representation.

Not seeing this refuted:


".....The term 'evolution' (for what went on in the antenna design) is a metaphor that makes it easier for scientists to explain what went on. But, in the end it is only a metaphor. In the words of the antenna designers, what they did is a "stochastic search and optimization technique". It is "inspired by natural evolution", but it is not the same thing. There are dozens of stochastic search techniques (hill climbing, simulated annealing) each of which uses a different metaphor to describe how it works. We shouldn't be confused by the metaphor. (Hence my first link in a previous post: Metaheuristics—the metaphor exposed.) If someone had invented harmony search before evolutionary algorithms, we'd say that harmonies were used to refine the antenna, not evolution, and we might not be having this discussion...."

Mike-

I was pointing out that you just sawed the branch off that you were sitting on.

How will you ever test anything, rigorously or otherwise, if you don't, at some point, follow this method?

It looks X, so unless and until proven otherwise, I'm going with the assumption that it is X
And if you're stuck with that method, I don't see why the statement "It looks designed, so unless and until proven otherwise, I'm going with the assumption that it is designed" is bad.

Of course, if you've got a nice proof that, in this case, looks are deceiving, then the case is different.

But it doesn't look like you've got such a proof. Instead, it looks like you reject that things are as they seem just because this method is at play. A method that, as already noted, you are going to use yourself, a lot, in other contexts.

===============================================

And if that were not enough, you take things a step farther, you actually say that because this method is used for design, it is therefore false that the universe is designed.

Even if you were right about the invalidity of the method in question (and you are not), this latter argument would be a logical train wreck.

You are claiming that because an invalid method was used to defend the assumption that X is true, X is therefore false.

Given that 'principle', here is my disproof of atheism:

  1. I asked person in a persistent vegetative state whether atheism is true.
  2. Through facilitated communication he said "yes".
  3. That method of discovery described in 1-2 is invalid.
  4. THREFEORE, Atheism is false.

QED?

Well, obviously not. But the point is made: Even if an invalid method was used to defend the assumption that X is true, that is no reason at all to think that X is false.

The comments to this entry are closed.