Thursday, September 01, 2005

Heddle's Twaddle

In a comment to his August 30th blog entry, pro-ID blogger David Heddle made the following observation:

I use genetic algorithms in my work. They are great at certain classes of optimization problems. They are nothing, however, like real life; they are not, at least at the moment, even realistic models of evolutionary adaptation. A GA for a circuit, for example, will find a nice solution, but it will be a Rube Goldberg machine with components that have no purpose. Biological systems, from anyone's perspective I would wager, are not Rube Goldberg's but elegant, um, designs.

Anyone's perspective? Gee, I could have sworn that just the other day I was reading the work of some biochemist or other who argued for the Rube Goldberg side of things:

Modern biochemists have discovered a number of Rube Goldberg-like systems as they probe the workings of life on the molecular scale. (p. 77)

Heck, Chapter four of the book this came from is entitled “Rube Goldberg in the Blood.” The book in question is Darwin's Black Box by some bloke named Michael Behe.

The fact is, Heddle would be hard-pressed to find a biologist who would describe biological systems as elegant designs, as opposed to Rube Goldberg machines. It is the nearly universal experience of anyone who has looked carefully at the inner workings of complex biological systems that they are invariably inefficient and, from an engineering standpoint, inelegant. They appear as if they were cobbled together from readily available parts and survived because they were just good enough to provide a survival advantage to the organisms who possessed them.

This is why Stephen Jay Gould talks about “the senseless signs of history” as being strong evidence for evolution. This is why vestigial structures feature so prominently in any discussion of evolution. And this is why ID folks spill so much ink desperately trying to explain away the problem of vastly suboptimal design.


At 8:57 AM, Blogger David said...

GAs produce extraneous components, often a nontrivial percentage of the whole, that are not vestigial—they never had a purpose, as can be determined by painstakingly examining all generations. The Rube-Goldberg aspect of biological systems is not like this at all and, has been demonstrated in a fair number of cases, to be “Rube-Goldberg of the gaps”. That is, at one time the appendix might have been the equivalent of an unused transistor in a GA designed circuit, but no more. That I can make such a statement also addresses the issue at hand, that GA’s are nothing like real life. It will never, ever turn out that an unused transistor in a GA circuit has a purpose I didn't expect. In biology, the track record for such assertions is poor. GA's are simple algorithms solving relatively simple problems. Biological systems are infinitely more complex. To say that GAs (with their designed fitness functions) and the kind of solutions they uncover offer any sort of support for evolution is ridiculous.

Thanks for the link.

At 9:19 AM, Anonymous Anonymous said...

I'd have a lot more respect for David Heddle if he simply admitted he was wrong, instead of trying to justify his false claim with more blather.

At 9:50 AM, Anonymous Tracy P. Hamilton said...

david said:
"The Rube-Goldberg aspect of biological systems is not like this at all and, has been demonstrated in a fair number of cases, to be “Rube-Goldberg of the gaps”. That is, at one time the appendix might have been the equivalent of an unused transistor in a GA designed circuit, but no more."

Heddle, you got that bass-ackwards. The appendix used to be useful, and is now worse than useless, because of its historical constraint. Learn some biology before you presume to lecture people who know more about it.

At 9:51 AM, Blogger Jason said...


Genetic algorithms were invented to apply biological wisdom to engineering problems. They were never intended to prove anything about evolution. However, they do show that search algorithms based on random variation and nonrandom selection can find solutions to problems that blind search would almost certainly not find. They are quite effective at refuting the claims of people who insist on describing evolution as a chance process. GA's do not prove evolution, but they do show that evolution is plausible.

You say that biological systems are complex, but that has no relevance to determining whether a genetic algorithm is likely to find them. What matters is the geometry of the fitness landscape and the number of points we are allowed to sample. If you're trying to decide whether natural selection is likely to find the vertebrate eye in the course of four billion years of evolution, you do not say, “Well, the eye is very complex, so no simple algorithm can find it.” Instead you ask, “Is there a smooth path of increasing fitness through genotype space, beginning from some reasonable starting point and ending with the eye,” and “Are we allowed to sample enough points to traverse the entire path.” If the answers to those questions is yes, then a genetic algorithm will find the eye with high probability.

It is the consensus view among the scientists who study these things that, for the eye, there is such a path and there has been enough time. Likewise for every other complex system that has ever been studied in detail. And certainly the ID folks have provided no reason at all for thinking that there isn't such a path. Instead they focus on irrelavancies like irreducible complexity, or they make bold mathematical pronouncements that can not be backed up with actual calculations.

The fact that your GA's produce solutions with truly uselss parts is simply an artifact of whatever fitness function you are using. They are not telling us something fundamental about what GA's are capable of. Apparently your function does not exact a very high price for hugely wasteful solutions. In nature, organisms inevitably do pay a high price for diverting resources into worthless structures. That is why we don't find organisms littered with parts that serve no function whatsoever.

What we do find, however, are a great many parts in biological systems that serve a function out of all proportion to their complexity (like the eyes of cave-dwelling rodents, ostrich wings, or the pelvic bones of snakes). We also find structures that by the standards of human engineers are terribly inefficient and inelegant. That is the norm, not the exception. In other words, they look like structures that were cobbled together by the mindless application of a simple algorithm, and not structures that were designed by an omnipotent designer.

I can't imagine what point you think you are making by all of that “Rube-Goldberg of the gaps” nonsense.

At 10:24 AM, Blogger David said...

The "Rube Goldberg of the gaps" is demonstrated quite nicely by Tracy Hamilton's comment:

"The appendix used to be useful, and is now worse than useless."

Here Hamilton (caustically and incorrectly, a bad combination) asserts that the appendix is useless, and I infer, therefore a GA-like Rube Goldberg artifact. Alas, the appendix is now known to have a function in the digestive and lymphatic systems.

I do not deny that GAs work (I use them.) I don't deny that they work based on evolutionary concepts. That is clear. I deny that their solutions, at least at the present, look anything at all like "real life."

At 11:29 AM, Blogger Jim said...

Heddle writes: "It will never, ever turn out that an unused transistor in a GA circuit has a purpose I didn't expect."

It reminded me of this interesting result: "Dr. Adrian Thompson has exploited this device, in conjunction with the principles of evolution, to produce a prototype voice-recognition circuit that can distinguish between and respond to spoken commands using only 37 logic gates - a task that would have been considered impossible for any human engineer. He generated random bit strings of 0s and 1s and used them as configurations for the FPGA, selecting the fittest individuals from each generation, reproducing and randomly mutating them, swapping sections of their code and passing them on to another round of selection. His goal was to evolve a device that could at first discriminate between tones of different frequencies (1 and 10 kilohertz), then distinguish between the spoken words 'go' and 'stop.'

This aim was achieved within 3000 generations, but the success was even greater than had been anticipated. The evolved system uses far fewer cells than anything a human engineer could have designed, and it does not even need the most critical component of human-built systems - a clock. How does it work? Thompson has no idea, though he has traced the input signal through a complex arrangement of feedback loops within the evolved circuit. In fact, out of the 37 logic gates the final product uses, five of them are not even connected to the rest of the circuit in any way - yet if their power supply is removed, the circuit stops working. It seems that evolution has exploited some subtle electromagnetic effect of these cells to come up with its solution, yet the exact workings of the complex and intricate evolved structure remain a mystery (Davidson 1997)." [emphasis added]

At 11:57 AM, Anonymous David Heddle said...


You might want add that, at least the last time I heard about it, Thompson's circuit was:

(a) not reliable when the programming was downloaded to other FPGAs. In other words, the GA found a solution that was very specific, exploiting subtle variances in the components on which it evolved. This, once again, is not like real life. The GA for humans works in an environment of great diversity.

Oh, and (b) (and this was fatal, though not really important for this discussion) operated only over a impractical temperature range (unlike human designed versions).

At 2:05 PM, Anonymous JY said...

So you admint your point (b) is irrelevant, which leaves you with (a) the supposed fact that the GA "for humans" (I'm not sure what exactly you mean by this -- at one point you use GA to mean the process by which solutions are found, but by talking about the "GA for humans" you seem to mean the solution itself) "works in an environment of great diversity". It's certainly true that evolution "works" (finds "solutions") in widely varying environments. But it certainly isn't true that the solutions it finds invariably "work" in diverse environments. In the case mentioned by Jim, the solution an evolutionary search process found was extremely sensitive to its environmental conditions. That seems to be very much "like life" to me. Indeed, humans are very sensitive to subtle variations in the components upon which they are built, just as the FPGA solution was.

At 2:40 PM, Anonymous Anonymous said...

Heddle just digs a deeper and deeper hole for himself. For every claim he's made, someone's posted a counterexample. Why doesn't he have the intellectual honesty to simply admit he was wrong?

At 3:12 PM, Anonymous David Heddle said...

Oh Mr. Anonymous, your comment reminds me of the chattel on Family Fued that shout "good answer" when someone on their team answers "root canal" to the question "name something that people take on vacation." Let's examine your claim:

Tracy Hamilton stated that the appendix was "worse than useless", and thus (presumably) an example of a GA-like artifact. This was wrong two ways:

(1) GAs produce non-vestigial unused artifacts. Even Hamilton allowed that the appendix was vestigial.
(2) But that is also wrong, because the appendix is, in fact, not useless.

Jason essentially agreed with me. He wrote "The fact that your GA's produce solutions with truly uselss parts is simply an artifact of whatever fitness function you are using." Leaving aside the fact that I was not discussing my GAs but those that design circuits he is correct—those GAs produce oddities because they are nothing, at least at present, like real life.

He again mentions some vestigial organs, etc., but once again I never said GA's are not like real life because they produce vestigial components, but rather they have components that were never used in any generation. It would be more like if all humans had a nonfunctioning spleen behind their left ankle. If that were the case, then GAs and real life would start to look similar.

Jim gave the example of Thompson's circuits which are fascinating and remarkable. Once again, however, they look nothing like real life. JY added (that might be Jim, I don't know) that the fact that they are fragile is life-like, but that is not true. The result of the circuit GA was a program that could be downloaded into another FPGA, where it promptly failed. If the result of our GA (i.e., our genes) were as fragile, we'd be extinct. The circuit GA program failed when downloaded into what for all purposes was the same environment, but apparently microscopic differences in the E&M characteristics were enough to do it in. In real life that would be like one twin living while the other dies, merely because the unfortunate twin stood two feet closer to the moon and was done in by tidal effects.

So tell me Mr. (or Ms.) Anonymous, where are the viable counter-examples?

Good Answer!

At 4:55 PM, Anonymous Tracy P. Hamilton said...

David heddle keeps digging...

I said:
"The appendix used to be useful, and is now worse than useless."

David said "Here Hamilton (caustically and incorrectly, a bad combination) asserts that the appendix is useless, and I infer, therefore a GA-like Rube Goldberg artifact."

No, you should infer that you have it bass-ackwards. You said
appendix once useless, now useful.
In fact it was once useful (in digestion), and now WORSE than useless.

This is independent of any GA argument you may have been struggling to formulate.

"Alas, the appendix is now known to have a function in the digestive and lymphatic systems."

Yeah, if "function in the lymphatic system" means be so prone to infection as to require significant lymphatic tissue there! So, tell us
the function of the appendix, and why the design MUST include the high risk of blockage and infection.
Peyer's patches seem to function quite well in the walls of the small
intestine without requiring dead ends that can easily get blocked and infected. Is the small intestine also part of the lymphatic system? The caecum in herbivores is where the appendix is in humans, and it harbors flora that excrete cellulase. What are the microorganisms in our appendix doing? Is the appendix there as a home for them to work in?

Your chance to shine, and show me how "incorrect" I am.

At 5:20 PM, Anonymous David Heddle said...

Ah, Hamilton, I see the problem, I wrote:

"That is, at one time the appendix might have been the equivalent of an unused transistor in a GA designed circuit, but no more."

which by your last comment claiming that I wrote:

"appendix once useless, now useful."

shows the source of the misunderstanding. My original comment did not mean (I'll take responsiblity although the context should have been clear) that the appendix was once actually useless and is now useful (the opposite of vestigial) but rather it was considered useless (and hence an example of a GA artifact) and is now known to be useful.

As for its usefulness, you can google as easy as I. You'll get many hits.

At 5:35 PM, Blogger Jason said...


I'm afraid the anonymous commenter who said you were digging a deeper hole for yourself had it right.

First, this all started because you said that from anyone's perspective biological systems are elegant designs and not Rube Goldberg machines. In fact, exactly the opposite is true. The strking thing about most complex biological systems is that they make little sense from an engineering standpoint. They are highly inefficient and inelegant. The ubiquity of vestigial structures is one example of that phenomenon.

Second, in your last comment you explicitly equated “vestigial” with useless. You argued that the appendix is not vestigial because it has some use. I'm afraid that's one of those standard creationist howlers that get people on my side of this so annoyed.

Vestigial and non-functional are totally different things. The eyes of cave dwelling rodents have a function. They plug holes in the rodent's head. But the eyes are no less vestigial for that. No engineer would build a sightless eye for the purpose of plugging holes in the head.

Here's Mark Ridley, from his textbook Evolution:

“Calling an organ vestigial does not necessarily mean it is functionless. Some vestigial organs may be truly functionless, but it is always difficult to confirm universal negative statements. Fossil whales called Basilosaurus, which lived 40 million years ago, had functional pelvic bones and may have used them when copulating; the vestigial pelvis of modern whales arguably is still needed to support the reproductive organs. However, that possibility does not count against the argument from homology. Why, if whales originated independently of other tetrapods, should they use bones that are adapted for limb articulation to support their reproductive organs? If they were truly independent, some other support would be used.”

Next, the reason evolution does not produce useless spleens behind the left ankles of organisms is that nature's fitness function exacts a high price for that level of inefficiency. Nature selects for reproductive advantage, and any creature diverting so many resources to totally worthless structures will inevitably be deficient in some other area. Hence, they get selected against.

I suspect that the circuits you were describing were being selected primarily for their functionality. The totally useless parts you refer to hung around because the fitness function being used was not sensitive enough to eliminate them. But it's not some fundamental limitation in GA's that they simply must produce totally worthless parts. The analogy between GA's and natural selection does not break down on this point. There are differences, certainly, between evolution by natural selection and the GA's used by human engineers. One of the biggest is that in GA's, the initial point of the search is usually chosen at random, whereas in evolution the intial point is given to us by the origin of life. This gives evolution a huge leg up in finding complex structures, however.

Finally, species in nature routinely go extinct because they are overspecialized to one environment. This is a basic principle of modern ecology. It's simply incredible that you would argue that a hypersensitivity to environmental conditions is something that distinguishes the products of GA's from the products of natural selection. The fact is that every species in nature, humans included, are incredibly sensitive to changes in their environment.

At 6:01 PM, Blogger David said...

Jason, I am convinced (mostly by tgibbs on my site) that on a microscopic level there are Rube-Goldberg like aspects in biological systems. I should have stuck to large scale structure--which is what you see in GAs. In GAs, the solutions are Rube Goldberg on all scales. That is, 5% useless components would not be unusual. When I look at a body, I see only fuctional components. If human bodies were designed by todays GA's, we'd look like monstrosities, or at least Picassos.

As for humans sensitive to their environment, of course you are correct, but that sensitivity is nothing like what was found in the GAs. You realize, of course, that the program was downloaded into an identical FPGA and it failed. Compared to that, humans can tolerate gigantic environmental shifts.

Maybe GAs will improve. But as they stand today, they do not mimic real life very well.

Your quote from Ridley is one of the reasons I so prefer physics, Chemistry, and math to biology. It sounds like plausible but also like an opinion.

At 1:33 AM, Blogger Jim said...

1. Mr. Heddle, (and JY, who isn't me), I brought up the example merely to show that engineers' expectations can be confounded by GA's (the whole list contains quite a few novel, genuinely unpredictable solutions to problems). That's all.

2. Rube Goldberg arguments are aesthetic, not logical; even from an engineering perspective, it is not always easy to demonstrate that a simpler solution to a problem exists. Hence Behe's backpedaling from "irreducible complexity" to "irreducible core complexity."

3. David wrote, "If human bodies were designed by todays GA's, we'd look like monstrosities, or at least Picassos." But this is again a purely aesthetic argument. Perhaps we do look like monsters to everyone but ourselves (remember the classic Twilight Zone episode?)

At 7:37 AM, Anonymous Anonymous said...

"GAs produce extraneous components, often a nontrivial percentage of the whole".

This is a invalid generalization argument since it is dependent on the fitness function you're employing. Introducing a cost in the fitness function for useless parameters does inevitably lead to a reduction of these. The fact that we see junk on DNA level and not to the same extent on organ level is just a matter of the fitness penalty associated with an extra useless organ as opposed to extra base pairs.

As for the circuit, the fact that it is not resistant to other environments is exactly what we would expect since the environment it evolved in is static as opposed to dynamic. It's called overfitting, and could easily be the case for humans if our environment was static with respect to a parameter and there was some fitness benfit to be gained from specializing.

Try making a GA with dynamic fitness landscape and penalties for junk. If you want to observe speciation as well I advice you to introduce a geographic measure, and create an incest parameter to prohibit procreation of species that are too divergent.

At 11:19 AM, Blogger Robert O'Brien said...


You previously threatened to delete the comments of another anonymous poster who was critical of you and those who agree with you, but I have yet to see you extend that policy to the anon critical of Dr. Heddle. Why is that? Perhaps you think "...consistency is the hobgoblin of small minds."

As for me, even though I do not care for Elsberry, I admonished Dave Scot (who appears to be missing some screws) for calling him a "fairy" and accusing him of lying on my blog.

At 12:47 PM, Blogger Jim said...

Robert O'Brien, that's "a foolish consistency." Makes a small, but appreciable, difference.

At 1:15 PM, Blogger Robert O'Brien said...


I am aware of the full quotation. The ellipses were intelligently designed.

At 3:58 PM, Blogger Jason said...


As I explained at the time, what angered me about the anonymous commenter in the previous post was (a) That he attacked me personally,(b) That the argument he was making was unusally stupid, and (c) That while he was attacking me personally and making stupid arguments all of his writing was so rude and juvenile that I didn't especially want to host it at my blog.

For all of that I did not delete his comments. I simply asked that he identify himself, which he decided not to do.

As I also explained at that time, I greatly prefer that people not leave anonymous comments. As long as the comments are tolerably productive, however, I don't make an issue of it. The commenter in the previous post stepped over the line, in my opinion. And since this is my blog, my opinion is the only one that matters.

At 4:03 AM, Anonymous oscoda said...

companies marketing mineral makeups and also get the best bargains in mineral makeup you can imagine,
find aout how to consolidate your students loans or just how to lower your actual rates.,
looking for breast enlargements? in Rochester,
homeopathy for eczema learn about it.,
Allergies, information about lipitor,
save big with great bargains in mineral makeup,

change edition interviewing motivational people preparing second

interviewing motivational people preparing second time

interviewing people motivational preparing for a second time

black mold exposure

black mold exposure symptoms

black mold symptoms of exposure

free job interview questions

free job interview answers

interview answers to get a job

lookfor hair styles for fine thin hair

search hair styles for fine thin hair

hair styles for fine thin hair

beach resort in the philippines

great beach resort in the philippines

luxury beach resort in the philippines
iron garden gates, here,
iron garden gates,
wrought iron garden gates
, here
wrought iron garden gates
You: The Owner's Manual: An Insider's Guide to the Body That Will Make You Healthier and Younger
eat eating mindless more than think we we why

texturizers here,
black hair texturizer,
find aout how care curly hair,
find about how to care curly hair,
care curly hair,
lipitor rash,
lipitor reactions,
new house ventura california,
the house new houston tx,
new house washington dc,
new house pa philadelphia,
san antonio tx house new,
house new pa philadelphia,
new house washington dc,
new house ventura california,
the house new houston tx,
house new san antonio tx,
the house new houston tx, that you are looking for,
new house ventura california, you need to buy,
new house washington dc,
house new pa philadelphia,
new house san antonio tx,

hair surgery transplant

air filter allergy

refurbished dell laptop computers

hair surgery transplant

air filter allergy

refurbished dell laptop computers

hair surgery transplant

air filter allergy

refurbished dell laptop computers

chocolate esophagus heartburn study

chocolate esophagus heartburn study
be informed,

digestion healing healthy heartburn natural preventing way

digestion healing healthy heartburn natural preventing way
sew skirts, 16simple styles you can make!,
sew what skirts 16 simple styles you,
rebates and discounts on sunsetter awnings,
sunsetter awnings discounts and rebates,
discount on sunsetter awnings

truck and bus tires 12r 22.5, get the best price,
tires truck and bus 12r 22.5 best price,
tires truck bus tires12r 22.5 best price,
plush car seat strap covers,
car seat strap covers,plush,
car seat strap, plush covers,
oscoda voip phone systems, the best!,
oscoda voip the phone system,
oscoda voip phone systems,
exterior iron gates,
oriental wrought iron gates,
powder coated iron garden fencing,

At 5:39 AM, Anonymous black mold said...

black mold exposure,
black mold symptoms of exposure,

wrought iron garden gates,
your next iron garden gates, here,

hair styles for fine thin hair,
search hair styles for fine thin hair,

night vision binoculars,
buy, night vision binoculars,

lipitor reactions,
lipitor reactions,

luxury beach resort in the philippines,
beach resort in the philippines,

homeopathy for baby eczema.,
homeopathy for baby eczema.,

save big with great mineral makeup bargains,
companies marketing mineral makeups,

prodam iphone praha,
Apple prodam iphone praha,

iphone clone cect manual,
manual for iphone clone cect,

fero 52 binoculars night vision,
fero 52 night vision,

best night vision binoculars,
buy, best night vision binoculars,

computer programs to make photo albums,
computer programs, make photo albums,


Post a Comment

<< Home