Commentary on the so-called Creation/Evolution/Intelligent Design Debate and Right-Wing nuttery in general - and please ignore the typos (I make lots!)

Tuesday, December 05, 2006

'Looney' - another creationist engineer with all the answers

I recently was amused to see this comment:

Just a reminder for you: Engineers are paid to do Intelligent Design, and scientists (who are paid considerably less) generally don't do intelligent design and they definitely aren't trained on ID or how it works. It should be no surprise that they look down on the field.


here.

Which was odd, for it had no real bearing at all on the post being replied to.

Note a couple things:

- A 'dig' about the amount of money paid in each profession
- the requisite conflation of 'intelligent design' done by humans with the 'Intelligent Design' of the IDcreationism movement
- A 'dig' about how 'scientists' are not trained in ID and so 'look down on it'

Generally silly stuff, so I decided to check out the author.

It is one "Looney", a fundamentalist Christian engineer (what else!). I checked out his blog, and lo and behold - he has a post about how molecular biology has nothing to do with "Darwinism" .

He 'knows' this because he picked up a 13 year old textbook on genetics and molecular biology that did not mention Darwinism once. Wow! Clever stuff! Surely, only an engineer could make such an insightful conclusion! Wait - there is more...

"It has a considerable Physical Chemistry and Organic Chemistry component which would make it intimidating for the large majority of biologists, but this subject is really foundational to understanding the molecular foundations of genetics"



Yeah, because us biologists is just so stupid. Why, we don't understand no kem-er-sty. I mean, we don't have any chemistry in our curricula. No physics, either. And math? Never heard of it. But engineers, why, they have all that stuff, not to mention the copious amounts of biology* they are required to study...

(* note - this is graduate level, and only for biomedical engineering)

And it goes on...

"As I suspected, the modern invoking of DNA and molecular biology in support of Darwinism was merely BS. To further support my contention, professor Schleif routinely invokes factories and computer information concepts (intelligently designed all) to help in understanding the basic concepts, whereas Darwinism is entirely absent."


Of course. Just BS. Something made-up by propagandists. And of course the use of EASY TO UNDERSTAND 'intelligent design' analogies and metaphors to help explain those simple biology concepts. Seems to me that if biological concepts were so simplistic, we would be using biology analogies to help explain engineering concepts.
Doesn't work that way, though.

Buit wait - the intellectual coup de grace:

"My hypothesis is that the field of molecular biology is simply not understood by the majority of biologists and thus pretty secure from rational debate by laymen. By claiming that this discipline (which they probably don't understand either) proves Darwinism and that Darwinism is vital to understanding molecular biology, the Creationists can be silenced, humiliated and put in their place by simply invoking superior knowledge. More malpractice?"


Wondderful! Yeah - biologists don't understand.... molecular biology.
Creationists can be humiliated, silenced, and put in their place all on their own - their intellectual dishonesty, their ignorance-based pseudocertainty, their willful ignorance, their outright fabrications and distortions - no, they do not need any help from us. That another creationist engineer doesn't understand a technical subdiscipline of the field of biology is no surprise. What is a surprise is that here is yet another creationist with an engineering background that exhibits nicely what I call the Kruger-Dunning effect (also, 'Loony' is yet another data point for the Salem hypothesis).

No wait - that is not a surprise either.

54 comments:

King Aardvark said...

I'm an engineer but also did biology at university for one year. Let me say that the math and physics done at the university level by biology students nowhere approaches real math or physics; it's barely beyond grade 12 level. That said, the only biology you get in engineering is a few things about bacteria when it came to municipal water engineering, so not a whole lot, and certainly less than what biologists got in math and physics.

I wish these creationist morons would stop giving my profession a bad name.

Looney said...

Actually, I have 25 years of experience working in multiple countries. My experience is that fewer than 1 in 10 UC Berkeley engineering Ph.D's really understand engineering, but most are really good at parroting what they learned in school. I don't expect anything better of the rest of science.

I stand by my assertion that currently Darwinian theories could just as easily be used to "prove" that the microprocessor could not possibly have been intelligently designed and must therefore have washed up on a beach in Santa Cruz due to random wave action on silicon crystals. Your Darwinism is that bad.

JoeB said...

Wow, just - wow.

There's no reason for a microprocessor-shaped doped silicon structure to be preferred over sand - it wouldn't last longer, it wouldn't create copies of itself that would be longer-lasting than copies of sand.
Evolution is a NON-RANDOM that takes advantage of random mutations to build the most suitable organism slowly, ever so slowly.

Anonymous said...

I stand by my assertion that currently Darwinian theories could just as easily be used to "prove" that the microprocessor could not possibly have been intelligently designed

Except for the fact that evolution applies to biological systems and the microprocessor isn't, you know, biological. At least, the last time I checked my Dell wasn't organic, although that would explain why so many Cheeto's go missing around here ...

I have a suspicion that math isn't actually necessary to engineering, and most engineers don't understand it. They just use it to intimidate us Humanities majors from getting all uppity and building our own bridges. I know this because I picked up an engineering textbook and I couldn't understand any of it, so clearly I am right.

Anonymous said...

I have a friend Ron who is a young-earth creationist. One day he pointed to car parked next to a tree and said something like, "See that tree and that car? Can't you tell that they were both designed?"

My reaction, as a mechanical design engineer of 33 years experience, was that whoever came up with that idea must never have designed anything.

Designs don't just spring up full-blown from the mind of a creator. They evolve. Cars started as primitive farm equipment; features were changed by trial and error, looking for improvements; some worked, some didn't; successful innovations by some manufacturers were cross-bred into other designs; all driven by the goal of survival in the marketplace. We've seen cars evolve in your lifetime! Museums are full of extinct forms, like the Model T and the Edsel. They even have vestigial organs which have been adapted to new functions, such as cigarette lighters being used to power electronic equipment.

Needless to say, Ron didn't buy my argument, but when I started working for Cooper Energy Services (now Rolls-Royce Energy Services), I found the missing link! They have big oil paintings of it on the walls.

Cooper invented the first farm tractor, circa 1850. One painting shows a team of six or eight draft horses, pulling a wagon with a steam engine on it, with another wagon full of coal hooked behind it. Cooper made the steam engines, and farmers hauled them to their fields to run threshers. Somebody at Cooper got the idea to add a bevel gear so the steam engine could turn a wagon axle.

The next picture shows the same wagon train, with the bevel gear - and a team of two horses out front! They still used the horses for steering! So the first car was figuratively a genetic cross between a horse-drawn wagon and steam engine.

Looney said...

So much meat! Too little time!

"Actually, several hard-nosed engineering types of my acquaintance employed by the Department of Defense are paid to do genetic algorithms and evolutionary programming."

Yeah, I am one of them. All of the intelligent design work for non-genetic methods must be done to make a genetic method work. I made a number of posts on this already. The genetic method will burn the most cpu time and give you the worst answer, but they can't work at all without an intelligent designer. Ditto for genetics in life. The order of convergence on genetic algorithms is similar to a random search.

Anonymous said...

Intelligent Design is the intellectual equivilant of seeing the Virgin Mary in a singed piece of toast or Jesus in a water stain on a piece of dry wall. You see the "design" because you want to see the "design" not because it's actually there.

Instead of trying to understand how the world actually works and modify your personal beliefs based on evidence you instead latch on to whatever crackpot theory justifies your pre-conceived superstitions and prejudices.

Intelligent Design isn't a scientifically valid theory... it's a character flaw.

Looney said...

Ben, I think you have enough intelligence to design a better argument!

Kagehi said...

The genetic method will burn the most cpu time and give you the worst answer

Then A) you are doing it wrong and B) any idiot can tell you that a *fixed* and entirely *inflexible* system, like the inside of a computer memory **isn't** a real ecosystem. And yet... Real genetic algorythm experimentors, that create artifical ecosystems, do get unique and better designs than they could have made themselves. You're falling into the "computer vision" trap. Where hand wiring a lot of "neurons" together and finding that they spot the difference between sunny and overcast days, when they are **sopposed** to be identifying scenes with enemy tank in them doesn't mean that evolution is wrong, just that you are an idiot for not knowing enough about biological systems to realize they *don't* evolve the way you *want* them to.

That is the crux of the situation. Just because they are horrible at what you "wanted", doesn't mean they are not 1,000 times better than you could have ever come up with at something you "didn't intend" them to do. Real biologists don't go, "Damn, well that failed..", they go, "Odd... What did I do wrong that produced that result, instead of what I wanted and what 'environmental' change can I make to evolve them in the 'right' direction?" I suggest you read up on "real" genetic algorythms from experts, like the Avida project, and stop fiddling with the bone headedly stupid "applied algorythm" systems that the developers fail to puta, "Warning: Since this is an evolving system, we do not guarantee it will 'evolve' the way you intended.", label on.

Again, I don't care if you are "designing" or just "using" such systems. Since they mimic biological systems, they are evolve to do what "they" think you intended, not what you "want" them to do. Sometimes the two happilly coincide, sometimes the result is something that stubornly refuses to evolve into anything that comes even "close" to what you intended. The result is invariably due to either you having a hooribly screwed up set of assumptions about what you "thought" it should be processing and how, i.e. you thought you where building a bridge, but ended up with a cruise ship, or your test data has inherent repetitions or characteristics that "actually" cause evolution to "see" those, instead of the patterns you "thought" it would see.

Even the experts fall for this. The Avida project tried to selectively "kill" its critters, to prevent them evolving, and instead the A-Life somehow learned to "see" the patterns in the identical sequence of test data they where given each time, when trying to decide if it had "evolved" and kill it. They figured out, entirely by random mutation, and without your "design", all on their own, that, when in the "test" software, they should "pretend" to be less active than normal. They played dead. No one "designed" them to do so, no one "taught them" how to do so, some of them simply where less active when in the test system, got tossed back into the environment, while others where killed off, and the ones that survived where eventually the ones that all seemed slow, dead or non-evolved, when exposed to the "test". That is evolution. The stuff you do is probably producing the same sort of unexpected defects, but since all you see is its failure to work as expected, you never bother to look and see what it "is" doing or where, more to the point, you are screwing up when creating the environment they evolve in.

Its like some clowns in russia trying to bread foxes to be more like dogs and not being able to figure out why they start developing "dog like" coloration, etc. When you force evolution of a biological, or even artifical, organism, you *might* get what you intended, but not always "only" what you intended, and if your methods are wrong, you won't even get that. Try finding someone that understands genetics to run your genetic algorythm projects. You obviously don't have a clue.

Looney said...

Kagehi, engineering has another unique concept besides Intelligent Design. It is called Hype. Now I have been a developer of engineering design tools for 25 years. If what you say is true, then GA should be able to put a lot of us engineers out of business. In fact, if Darwinism is true, then GA shoud be able to replace intelligent designers (engineers) completely. Why waste time arguing with me when you can get rich by investing in a GA company and replacing me?!

Fox1 said...

If what you say is true, then GA should be able to put a lot of us engineers out of business.

What in the seven hells? If this is your response, then you did no more than lightly skim what he said, and thus, have no idea what the ramifications would be if what he said was true.

Now I have been a developer of engineering design tools for 25 years.

Oh... well, there you have it. I'm no YEC, but it looks like this thread is proof that humans and dinosaurs HAVE coexisted! (with apologies to any other senior citizens in the audience)

Looney said...

Acute_agnostia, I have seen zealous young engineers applying GA to get answers that weren't nearly as good as we got 20 years earlier without GA and only 1% of the computing resources. I have seen nothing else. The world class Optimization/GA experts who I have talked to have also told me the same, although some of them publish the same kind of Hype filled papers to gain attention. This is the reality behind the Hype.

saurabh said...

As an actual hard-nosed biologist, I have to scoff at the notion that biologists have a poor understanding of molecular biology. In fact, the opposite is true - most biologists have a poor understanding of evolutionary biology and an excellent one of molecular biology. Many curricula these days are built around molecular biology - DNA replication, cellular metabolism, etc., - and skimp on or completely avoid discussion of evolution, population genetics, etc.

As to the idea that molecular biology contains NO support for evolutionary theory, I fail to see how this is true - the modern understanding of the structure of DNA, the details of replication, recombination, etc., are used every day to validate our understanding of evolution. Of course, a lot of this can be understood without a detailed knowledge of the molecular basis (I don't need to know how exactly DNA polymerase operates in order to appreciate mutation rates, etc.), but it helps, and it is entirely supportive. That molecular biology textbooks contain little treatment of evolutionary questions is simply a result of the fact that they are not, on the whole, germane. That is, molecular biology confirms and extends our understanding of evolutionary processes (by, for example, elucidating the details of inheritance of genes, mutation, etc.), but the reverse is not true. To draw an analogy you might appreciate, molecular biology might be compared to hardware-level stuff - the nuts and bolts of the system - evolution discusses emergent properties of the system, a consequence of the nuts and bolts, but not necessary for an understanding of the former.

Incidentally, if "the order of convergence on genetic algorithms is similar to a random search", I submit that you are merely a shitty programmer. Others can obviously do, and have done, better.

Looney said...

"That molecular biology textbooks contain little treatment of evolutionary questions is simply a result of the fact that they are not, on the whole, germane."

Thankyou.

saurabh said...

You have a remarkable ability to selectively read text. Look, let me make it simple for you:

Molecular biology texts have, as you note, a lot of background on chemistry, because it is important to understand chemistry in order to appreciate some points of molecular biology. However, few chemistry textbooks contain discussions of molecular biology, because the inverse is not true. From this, according to your logic, one should conclude that chemistry provides no support for molecular biology. Therefore, you are a butthead. QED.

Looney said...

The claim is that NOTHING in biology can be understood without reference to Darwinism (Theodosius Dobzhansky). This is what my children are taught in school. Would you like to revise that law?

saurabh said...

Instead of attacking your strawman, why don't you address the actual argument Dobzhansky made, which was that evolution is the only sensible unifying principle one can find to explain the diversity of life? Therein he includes an argument based on orthology at the molecular level, which, I daresay, might be construed as an argument based on molecular biology.

saurabh said...

Or, as he said: "Seen in the light of evolution, biology is, perhaps, intellectually the most satisfying and inspiring science. Without that light it becomes a pile of sundry facts some of them interesting or curious but making no meaningful picture as a whole."

Looney said...

"It is a general postulate to which all theories, all hypotheses, all systems much henceforward bow and which they must satisfy in order to be thinkable and true." - Dobzhansky

I did not overstate Dobzhansky's argument at all. I will rebut that in every application of biology, there are successful practitioners who reject Darwinism, including medicine, farming, and drug design. Dobzhansky was completely wrong.

NelC said...

Looney said

"Actually, several hard-nosed engineering types of my acquaintance employed by the Department of Defense are paid to do genetic algorithms and evolutionary programming."

Yeah, I am one of them.


Um, why would the DoD employ someone to do something he doesn't believe in? That almost guarantees that it's going to get screwed up.

NelC said...
This comment has been removed by a blog administrator.
Looney said...

"Um, why would the DoD employ someone to do something he doesn't believe in? That almost guarantees that it's going to get screwed up."

The DOD doesn't believe in doing anything right! ;-)

According to a Hindu optimization expert: "Don't waste your time on GA."

Actually, I understand GA better than the Darwinists. It is a useful tool, just not in the way it is advertised.

Looney said...

"Just because you didn't see "Darwinism" in the book, does not mean that it is valid."

Don't forget Dobzhansky! I did see ID used to explain molecular biology concepts.

Looney said...

"Anyway, this "Things were better in my day" is crap."

20 years ago, we were doing 3d shape optimization with thousands of degrees-of-freedom using conjugate-gradient methods. It took about 5 or 6 iterations to get a nice convergence which we proceeded to test with good results. That worked out to 1 cpu month of computing. The new kids do 20 degrees-of-freedom on similar problems using GA and run thousands of iterations without much exciting to report. GA's convergence rates are worse than Order(N*N), so you can see where all this is headed. They also ran numerous variations with different mutation sizes and adaptive mutation rates.

eviledv said...

I don't expect anything better of the rest of science.

Ah, the sweet smell of anti-intellectualism. We really shouldn't feed the troll, but he has such a flair for non-sequiturs and even getting them wrong that I can't resist.


The genetic method will burn the most cpu time and give you the worst answer

Nonsense, this is entirely problem/model dependent. But you are correct that it will usually burn a lot of cpu cycles. What does that tell you? That evolution takes time? You also seem to suffer from the erronous assumption that GA somehow is supposed to locate the global optima. There are no such guarantees neither for GAs nor evolution which any quick look at human anatomy should tell you. The fact that intelligent design in some cases can produce a better solution doesn't invalidate the fact that evolution always produce some solution.


if Darwinism is true, then GA shoud be able to replace intelligent designers

Nonsense. Intelligently produced solution might still do a better job than evolved solution. Evolution just does the best with what it has available at them moment, this isn't an argument against evolution not having happened. Actually wheter evolution or ID can produce the best solution is totally irrelevant. Given the multitude of what we would term design errors in organisms this if anything surely speaks in favour of evolution at least given a somewhat competent designer.


20 years ago, we were doing 3d shape optimization with thousands of degrees-of-freedom using conjugate-gradient methods. It took about 5 or 6 iterations to get a nice convergence which we proceeded to test with good results. That worked out to 1 cpu month of computing. The new kids do 20 degrees-of-freedom on similar problems using GA and run thousands of iterations without much exciting to report. GA's convergence rates are worse than Order(N*N)

Convergence rates for a GA as I'm sure you know is model dependent. It also sounds like you are trying to compare a local and a global optimization technique. Comparing running time would then be a tad bit unfair. This is also beyond irrelevant. You have demonstrated that a poor simulation of evolution in a limited environment uses time and doesn't necessarily give an optimal solution and performs poorer than some other optimization. So what?


I will rebut that in every application of biology, there are successful practitioners who reject Darwinism, including medicine, farming, and drug design.

I would assume that almost all successful practicioners reject Darwinism, evolution on the other hand. But you are correct that one doesn't necessarily need to understand evolution to work with your listed subjects. You would of course be severely limited. Creating flu vaccines for instance would be beyond your understanding or breeding plants for that matter. And we would probably be stretching the definition of successful to be synonymous with "making money" or something similar.

It amazes me that you actually possess the hubris to challenge biologist without actually possessing a single relevant argument.

Doppelganger said...

Loony writes:
Don't forget Dobzhansky! I did see ID used to explain molecular biology concepts.

Note the infantile/intellectually dishonest conflation of disparate concepts.

Thye "ID" loony refers to is human contrivance. NOT the big ID as in the IDcreationism of the ID movement.

Why would so super-educated an expert on all science like Loony the creationist engineer continue to rely on so silly and irrelevant a conflation?

Doppelganger said...

if Darwinism is true, then GA shoud be able to replace intelligent designers

WTF???


If creationism is true, should not God have replaced human intelligent designers by now?

See how incredibly stupid an argument you are making, Loony?

Looney said...

"Note the infantile/intellectually dishonest conflation of disparate concepts."

Let me break things down more for you:
1) Molecular biology is the foundation or biology, but there was no reference to Darwinism, only ID in a key textbook that probably isn't out of date at all.
2) Darwinist insist that NOTHING in biology can be understood without a BELIEF in Darwinism (Dobzhynsky died in 1975).
3) Darwinists have been roughly repeating 2) since the end of the 19th century when they obviously knew next to nothing about the -ology behind biology.

It looks to me like a religion, not a science.

Doppelganger said...

Thanks, oh Loony the engineer expert on all, for condescending to grant us all the TRUTH...

Let me break things down more for you:
1) Molecular biology is the foundation or biology, but there was no reference to Darwinism, only ID in a key textbook that probably isn't out of date at all.


Let me break something down for YOU:

1. Watch your sentence structure. I have no idea what that first part of your sentence is supposed to mean.
2. The book did NOT use "ID" as in the IDCreationism you espouse, it used analogies to human activities, which is not the same thing, Mr.Wizard.

2) Darwinist insist that NOTHING in biology can be understood without a BELIEF in Darwinism (Dobzhynsky died in 1975).

You are either 1. lying or 2. monumentally ignorant.
Again, watch your grammar - it should be either "Darwinists inist" or "[the] Darwinist insists".
Dobzhynsky did not say what you ascribe to him - that is intellectual dishonesty on your part. He said nothing about 'BELIEF' nor did he mention 'DARWINISM.'
In actual science, you cannot just make up quotes and such - I know that is standard creationism (I hope it is not standard engineering practice...), but it will not fly here.


3) Darwinists have been roughly repeating 2) since the end of the 19th century when they obviously knew next to nothing about the -ology behind biology.

Right - of course, we must rely on creationist engineeers to explain it all to us.

Do you know what "ology" means? It seems not.


It looks to me like a religion, not a science.

Well, I guess a fundamentralist creationist would know more about religion than much of anything else.

That much you have amply demonstrated.

Doppelganger said...

I will rebut that in every application of biology, there are successful practitioners who reject Darwinism, including medicine, farming, and drug design.

As has been mentioned, most modern scientists reject a strict Darwinism, your use of it is quite informative. On the other hand, there is definitely a monumentally larger number of people in those fields that reject Dembskiism. By your logic, then, we should reject it. And why shouldn't we reject the nonsense of Bill "The Ted Haggard of Information Theory" Dembski?

Looney said...

Doppelganger, that is quite interesting. It looks like the gap might not be nearly so large as I had assumed. As evolution bundles a lot of things together there is a large amount that is compelling together with things that I find most implausible.

Doppelganger said...

And what is it about evolution that you find implausible?

Looney said...

Well, to begin with the Miller-Urey implications of spontaneous generation of initial life forms. Beyond this, most of the extrapolations that fall outside of recorded history.

Other than that, I accept everything.

Doppelganger said...

Of course you must know that Miller-Urey was about the origin of life, not evolution?

And really, it was not about the origin of life, either - it was a series of experiments designed to wsee what organic molecules - if any - could have formed on the primitive earth.

And so you also are suspect of anything outside of recorded history...

Most interesting. Silly and shallow, but interesting.

Looney said...

"Most interesting. Silly and shallow, but interesting."

One of the fundamental laws of engineering: Extrapolation is dangerous. Shallow, but very practical.

Looney said...

Doppelganger, I am puzzled about something:

If someone asked me if the Brooklyn Bridge and the Golden Gate Bridge had common ancestors or common designers, I would probably answer with "who cares?". From a civil engineering viewpoint, this is a useless question.

If someone asks if apes and men have common ancestors or common designers, why does biology not similarly answer with "who cares?". Why can't the Creationist/Darwinist argument be moved entirely out of the classroom?

Rev. BigDumbChimp said...

Because people such as your self continue to misrepresent the theory of and the scientific evidence for evolution. By using such terms as Darwinism and by calling evolutionary biologists and other scientist as Darwinists you also misrepresent where the theory has
evolved. Yes Darwin started the ball rolling but many thousands of scientists have worked to move the theory along for 150 years, all the time changing and fine tuning which mechanisms and environmental factors act on species to cause them to evolve.

The creationism/ID side misrepresents the facts and injects pseudo-science into the curriculum under the guise of some non-existent controversy. By creating a false sense of equality in the creationists stories and good hard scientific theory they are being dishonest to the subject and are doing a disservice to the children. The constant attempts by the creationists to move bad information into the classroom is why it's not moved out of the classroom.

ID/creationism is theology. There is no hard science with experiments, results or even a smattering of published papers as well as the fact it is rife with mis-information and dishonesty. When ID comes up with even a small fraction of the hard evidence that evolution has on its side, then maybe we can talk. Unfortunately all they have now is a giant PR campaign and people such as your self who make wild claims that "Well that man made in a factory speaker is designed, so of course everything in nature is designed..... Just ignore that growing 150 year pile of evidence over there." Using the God did it excuse because science hasn't discovered how something works automatically takes you out of the scientific realm and into the supernatural. Scientists are always willing to change a theory and are always trying to disprove them. Religious explanations take that away. As soon as God did it, you've found your answer. And the answer is, it's not science its religious faith.

--------

please excuse any typos and probable bad sentence structure, my wife is tapping her foot waiting to go to dinner i didn't have much time to proofread.

TabAtkins said...

Wait, wait, wait. GA is useless and finds bad solutions? I believe I have a nice link to many, many, many examples showing how blatantly false that is.

Here it is:
http://www.talkorigins.org/faqs/genalg/genalg.html

I'm currently paying special attention to a particular paper referenced there, as I'm doing a project that runs along very similar lines. Read the paper linked to here:
http://www.talkorigins.org/faqs/genalg/genalg.html#chellapilla2001

It talks about using GA to evolve a checkers-playing program that can play at the master level. And this was done in 2000 (there is an earlier paper discussing their project as well) on a computer that is laughably out of date today. There, it took them 6 months of computer time to evolve about 800 generations. On today's desktop computers, you could likely cut that time by a factor of 10. A good dedicated rig could do the same in a matter of days at most.

And that's only one example out of many on that one site, which is nowhere near exhaustive. They simply chose a single example in each field.

GAs are not useful for everything, and one should try to find analytical solutions when possible, but they are very well suited to problems involving a large number of nonlinear variables.


I humbly submit that anyone who can make such claims about GAs while simultaneously claiming that they actually work with GAs is either 1) a very, very bad programmer or 2) completely full of shit.

Looney said...

Jackalmage, if you know the answer, then you can engineer (ID) the GA method to get the right answer. Even when we don't know the right answer, we use ID to tune the GA to work better.

If you spend any time working with GA, you will quickly find that GA without ID is impossible. In fact, not a single aspect of ID can be circumvented by employing GA.

I will refer you to Hype theory which those of us who have worked a few years in high tech use to filter out garbage claims.

TabAtkins said...

If I'm reading you correctly, you are making the following claim:

To get a GA to produce a correct answer, one must first design that answer into the program.

I refer you to the following link as my rebuttal:
http://www-cse.uta.edu/~cook/ai1/lectures/applets/gatsp/TSP.html

In case my succinct rebuttal was too complex, I shall explain it. The Traveling Salesman problem lies in the NP-complete domain. What this means is that any attempt to find an exact answer will take an exponential amount of time. What this means is that above a certain point, it's essentially impossible to actually find an exact solution, because it simply takes too much computational resources.

This also happens to be nearly impossible for humans to discover. Point someone at a picture of 50 dots, and tell them to connect them all, making a single path with the shortest possible distance, and they won't be able to. We won't even be able to get close most of the time.

On the other hand, if you can accept that the answer you'll receive is merely very close to the correct answer, a GA can solve the problem in a very, very short amount of time, as demonstrated amply by the link above.

Note that this is actually because GAs utilize randomness, or non-determinism, in their design. NP stands for Nondeterministic Polynomial time, which means that they are solvable easily (in polynomial time) using methods that employ randomness, but exact solutions are currently thought to be exponential time.

Game playing, as noted, is similar. LOOK at the example in my previous post that I pointed to specifically, where they evolved a Master-level checkers playing program. *Read* it, it's very short. The *only* information the designers put into the design of the program was a "piece differential" element in the final output. However, their solution provably outperforms a program that uses piece-differential by itself.

That does *not* sound like the designers put in information about how to win at checkers. If you actually knew what you were talking about, you'd realize that.

References, sir, are the antidote to nonsense. Otherwise we're just having a pissing contest. I can show that I'm right, over and over again. Look at the links I provided - they're fairly definitive. You, on the other hand, are simply asserting things without any support.

Looney said...

I am well aware of the traveling salesman problem. Solving games is not the same as designing structures. If we look at something like quadratic functions, GA is awful. These occur all the time in real life.

Also, the amount of intelligent design that goes into encoding a traveling salesman problem and defining the objectives, etc., etc., is what really counts. Usually there are a large number of solution methods that work once the intelligent design work is done. Sometimes a problem lends itself to one method or another. Usually, this is not the case.

In real life, engineers don't choose the problems we must going to solve. The problems confront us regardless. It isn't academia.

TabAtkins said...

Re: Quadratic functions
Your point is? I'm not sure exactly what you're referring to by "quadratic functions", but it doesn't actually matter. GAs are not a panacea that can solve every problem effectively. As I said before, they are very good at solving non-linear problems with a large number of variables - problems which are *very* difficult for humans to solve effectively.

As well, when possible one should always try to find an analytical solution, if only because GAs are *very* computationally intensive. Analytical solutions will invariably be much less expensive to compute. When the problem is just too complex for that sort of thing, though, GAs are very effective.

Re:"Also, the amount of intelligent design that goes into encoding a traveling salesman problem and defining the objectives, etc., etc., is what really counts."

Uh, yeah. Duh. You have to define the problem properly. Otherwise, the GA will likely be solving a different problem than the one you *thought* you were telling it to. There is a funny example of researchers who were experimenting with GAs using a programmable hardware array. They wanted it to assemble hardware components in such a way as to produce an oscillating electrical signal. When it succeeded, they found that it had not done what they wanted, but had instead assembled a radio receiver that amplified an oscillating signal from a nearby piece of equipment!

If you define your problem badly, the algorithm will work badly. This is true no matter what sort of algorithm you use. However, defining the objectives of a problem and finding a good way to represent it is not the same as solving the problem (or else they wouldn't need the GA at all). I can easily provide a way of encoding the TS problem, and the objective (or fitness criterion) is trivial - complete solutions with a shorter length are more fit. But I cannot actually solve the travelling salesman problem myself. The two things are completely different problems with completely different difficulties.

Re:"Usually there are a large number of solution methods that work once the intelligent design work is done. Sometimes a problem lends itself to one method or another. Usually, this is not the case."
Non-sequitur. This has nothing to do with whether or not GAs are effective. It is simply expressing an obvious (but irrelevant to the discussion) truth - that different problems may be best solved with different solution methods.

Re:"In real life, engineers don't choose the problems we must going to solve. The problems confront us regardless. It isn't academia."
Non-sequitur. Again, you are expressing an obvious and irrelevant truth. Does this statement in any way imply that GAs are bad at solving appropriate problems? Nope. What it is implying is that I somehow believe that GAs are the be-all and end-all of solution methods, which is a statement that I never made nor implied in any way (unless you can provide quotes that show otherwise).

Looney said...

Jackalmage, we seem to have an agreement that GA is useful for some problems, but not all. The problem is that for GA to have been the author of life, it must be efficent on all problems, regardless of the setup. In real life, how we set up the problem is critical to the success of the optimization.

Quadratic functions occur all over engineering: Choosing the right size for a support, the angle between limbs, the size of a leaf, the amount of protein to produce ... Typically it is an optimum from many variables.

TabAtkins said...

Re:"The problem is that for GA to have been the author of life, it must be efficent on all problems, regardless of the setup. In real life, how we set up the problem is critical to the success of the optimization."

Ah, no. Not even close. GA just needs to be good at the one problem that life itself is trying to solve: finding a decent optimum among an extremely large number of variables that all interact in a nonlinear fashion.

This just happens to be precisely the sort of problem that GAs are good at.

GA has the added benefit of requiring only two things to operate: reproduction with mutation, and selection. The first is happily provided by a number of physical systems (life included), and can be theorized to have appeared in early molecules, and the second is provided by the simple fact that resources are finite, and things which are better at gathering those resources will necessarily take them from things which are worse at it.

What this means is that GA in life may itself by a 'good enough' solution - it's so easy to implement that it's difficult *not* to stumble across it. Plus, though it may not be efficient across all domains, it *will* find a decent answer eventually, which is sufficient when you don't have any other method.

Re: "Quadratic functions occur all over engineering: Choosing the right size for a support, the angle between limbs, the size of a leaf, the amount of protein to produce ... Typically it is an optimum from many variables."
All right, I'll bite. Would you point to some evidence that GAs are bad at solving this sort of problem?

Re: "Jackalmage, we seem to have an agreement that GA is useful for some problems, but not all. "
If you've changed your mind significantly from your comments earlier today and yesterday, then I suppose we have. However, I quote you as saying,
"All of the intelligent design work for non-genetic methods must be done to make a genetic method work. I made a number of posts on this already. The genetic method will burn the most cpu time and give you the worst answer, but they can't work at all without an intelligent designer. Ditto for genetics in life. The order of convergence on genetic algorithms is similar to a random search."

That seems pretty clear to me - you're stating that GAs are crappy and of little or no use to anyone. Do you admit that this statement was wrong? If so, what precisely made you change your mind? If not, then no, we haven't come to any such agreement, and I regard any attempt to insinuate that we have as an attempt to move the goalposts. By that I mean that once someone proves you wrong on one point, you simply back off and say, "Well, yeah, but I'm still right about this, and that makes my whole point still right."

I would like to think that such deception is below you. Prove that I'm wrong for thinking otherwise?

TabAtkins said...

I must apologize, but as it's 10pm here and that means my work shift is over, I'm going to have to log off for the night. Once I head home, it's straight for bed or else the wife gets angry. I'll bookmark the page and continue any discussion tomorrow, however.

Doppelganger said...

Loony,

I am puzzled by something. Someone else has already mentioned this, but it is getting somewhat irritating - you do not seem interested in responding to anything anyone actually writes, or at best you pick out some minutiae and focus on that.

If you cannot provide viable comments, just say so, but do not carry on as if you are genuinily trying to have a discourse here, because you clearly are not.

You have thus far conflated human ID with BIG ID and the origin of life with evolution. You have denigrated those in academia and misrepresented Theo Dob's famous statement. You have declared that anything not in recorded histroy is shady as far as you are concerned.

Did I miss anything?

Doppelganger said...

I will refer you to Hype theory which those of us who have worked a few years in high tech use to filter out garbage claims


You know what I use to filter out garbage claims?

A sampling:

1. The use of analogies as evidence
2. Repeated use of ill-defined terms despite frequent requests for clarification or correction
3. Conflation of disparate topics
4. Inability to remain on topic

etc.
etc.
etc.

Doppelganger said...

In real life, engineers don't choose the problems we must going to solve. The problems confront us regardless. It isn't academia.

I've never really understood this hatred towards academia that so many creationists seem to have. Perhaps it is because those in academia are not right-wing fundy bots? I don't know. But it is truly asinine to think that those in academia are out fo touch or inexperienced. I cannot speak for engineering faculty, but in those terrible 'soft' sciences like biology and geology, not only do we have to have 'done' to get a job in the first place, we have to continue to 'do' for the most part all the way through our careers (unless one is a lecturer or the like). This air of superiority that some in the private sector seem to exude is premised not in any objective reality, but in fantasy driven prejudice.

And sadly, these folks do not seem able to realize this.

Looney said...

I documented the reason for the anti-intellectualism on a number of earlier post. (Speaking as an intellectual and the son of an engineering professor.)

Let's just say that if a really smart person studies science, moves on into engineering, and then joins the management team at Enron, we presume he is corrupt until proven otherwise. If the same person finishes a Ph.D. and becomes a professor, will his moral character be any different?

You think it is an irrational disregard. I think of it as no more fairy tales.

Rev. BigDumbChimp said...

Let's just say that if a really smart person studies science, moves on into engineering, and then joins the management team at Enron, we presume he is corrupt until proven otherwise.

Holy shit. That's a huge non-sequitur and a prime example of sophistry. Taking a specific inflaming incident (Enron) and then blowing that out as some unrelated comparison to encompass all PhDs all the while trying to confuse the actual points being made. Sheesh.

I'm going to re-interpret your claim to not just focus on Enron and focus on all in corporate positions. You may assume that all in these positions are corrupt I however do not unless they give me reason to do so.

Wow.

Smokey said...

"... but in those terrible 'soft' sciences like biology and geology, not only do we have to have 'done' to get a job in the first place, we have to continue to 'do' for the most part all the way through our careers..."

This ties in with Looney's massive ignorance of the theories he thinks he's demolishing. Note that ID proponents produce no new data, and pretend that scientific controversies consist of dueling essays, instead of the reality of appeals to data. Scientists have to produce, and reviews or essays don't count for squat if one is trying to get a grant in biology.

Unknown said...

Looney is certainly loony, but I question the rest of you.

Gentlemen, what is the point of discussing issues with someone who can't grasp those issues?

Would you engage in an educated discussion with a drunk or a deranged street person? Of course not, because it's a waste of time. Looney is delusional, and word wars with him are pointless.

Discussing reality with someone who has proven they cannot grasp reality is an absurd endeavor. Looney isn't ever going to get it. An intelligent discussion with a nut is not an intelligent discussion.

There have been a great many brilliant responses to Looney.

But you are singing to the deaf.

Unknown said...

Upon further thought, I suspect Looney is an off-duty auto mechanic in a tavern with WiFi. I think you gentlemen are being scammed.

What does Occam's Razor tell you about this scenario? Looney is an expert in several fields; he's familiar with every example and author. He's got a fancy government job using evolutionary computing, but doesn't believe in evolution. He's a scientist who demonstrates little understanding of science.

He's highly educated, but can't spell, read or think. He ignores the valid points you make. Some of the things he says are provably wrong or just plain goofy.

I smell a scam. A half-smart drunk could whip up non-sequiturs out of Google searches all night long.

"In Internet terminology, a troll is someone who intentionally posts ... inflammatory messages about sensitive topics ... to bait users into responding." -- from the "troll" Wiki page

Doppelganger said...

Well, you bring up some good points, but your characterizations - which I agree with - can, unfortunately, be accurately applied to a large number of anti-evolutionists with engineering backgrounds.

I suspect Looney is what he says he is - an engineer who is a creationist. Creationism seems to have that affect on otherwise rational, intelligent folks.