A peer-reviewed article that supports ID … or something else (part 2)

| 212 Comments

By Joe Felsenstein, http://www.gs.washington.edu/faculty/felsenstein.htm

In a previous thread here, and in other blogs, there have been many people arguing that William Dembski and Robert Marks’s recent “pro-ID” paper isn’t really pro-ID, that it is equally compatible with theistic evolution or even nontheistic evolution. William Dembski has now replied at his Uncommon Descent blog to these comments.

He argues that

The key contention of ID is that design in nature, and in biology in particular, is detectable. Evolutionary informatics, by looking at the information requirements of evolutionary processes, points to information sources beyond evolution and thus, indirectly, to a designer.

and

Theistic evolution, by contrast, accepts the Darwinian view that Darwinian processes generate the information required for biological complexity internally, without any outside source of information. The results by Marks and me are showing that this cannot be the case.

Dembski and Marks’s argument is (in effect) that smoothness of the adaptive landscape means that information has been built into the situation, and that natural selection does not create new information, but instead transfers this existing information into the genome. To Dembski, the Designer acts by creating this information.

There is no requirement that this creation of information happen multiple times. A Designer (or just the laws of physics) could set up the world so that it is one in which adaptive surfaces are smooth enough that natural selection succeeds in bringing about adaptation. That setting-up could have happened back before the first living organisms existed.

Should other supporters of ID be happy with such a picture? It certainly does not argue for the fixity of species, or against large-scale evolutionary change. But I suspect that many theistic evolutionists would find it consistent with their views.

Evolutionary biologists may prefer a different definition. Intelligent Design only differs from existing theories on evolution if it involves a Designer who intervenes at least once after the origin of life. If ID advocates want to argue that there is something wrong with evolutionary biology, they should put forward a theory that makes some different prediction about what happens during evolution after that origin.

Dembski draws the distinction as involving where the information comes from. Evolutionary biologists will probably prefer to focus on whether there is evidence for interventions by a Designer.

212 Comments

The key contention of ID is that design in nature, and in biology in particular, is detectable.

So friggin’ detect it, already! Geez…

Joe wrote: “William Dembski has now replied at his Uncommon Descent blog to these comments.

It should be noted that there were no comments allowed on Dembski’s reply posted last night. And he cut off comments after only nine comments on his previous blog entry on this topic on August 19.

Are we allowed to know who were the peers who purportedly reviewed this article? Or is this another Meyer/Sternberg scam?

I am not sure that fully understand all of this. But it seems clear that the adaptive landscape is not all that smooth. Many creatures have gone extinct. While we have seen a great deal of adaptation and an increase in variety of living creatures, it is by no means true that every branch has survived.

George writes…

I am not sure that fully understand all of this. But it seems clear that the adaptive landscape is not all that smooth.

Yes. In fact, the “adaptive landscape” is full of discontinuities, many of them way too steep to climb.

Rats make it to an island, and all the ground-nesting birds go extinct.

A virus suddenly mutates and becomes virulent, and Tasmanian Devils are pushed to the brink.

And don’t even start the dinosaurs on how bad a big meteorite can ruin your whole day.

Things like this have happened over, and over, and over and over during the entire history of Earth.

The “adaptive landscape” is cratered and pockmarked and strewn with cliffs.

Still, as long as some part of the landscape has a climbable slope, evolution works there, and that’s enough.

I think that you have highlighted Dembski’s issue. I’m not sure that he understands it or is being purposefully oblique. He wants to say that evolution is impossible and that we all came about by a number of discrete events (poofs) brought about by God. However, all he is showing is that the fitness landscape is amenable to evolution.

I see other people say that one characteristic of committed creationists is the inability to understand analogies and models. I wonder if this is Dembskis problem rather than outright dishonesty.

William Dembski wrote:

The key contention of ID is that design in nature, and in biology in particular, is detectable.

And since there is no testable mechanism for how this design is introduced, let alone how to quantify it, it falls outside of the realm of science. Methodological supernaturalism is what really distinguishes ID.

Ok, here is the apparently relevant passage. How relevant is questionable because quantum mechanical systems can jump from state to state, yet follow a distribution of probabilities determined by solutions of the Schrödinger equation for the problem in question.

However, whether the system is classical or quantum mechanical, the most important cases we are dealing with in complex evolving systems are non-linear in their behaviors.

This means that systems can “snap” into and out of various states. Oscillating systems can flip among modes with some of the smallest perturbations. This is also why different levels of complexity in systems are governed and dominated by different emergent phenomena.

COI does not preclude better-than-average performance on a specific problem class [8], [18], [20], [31], [35], [43]. For instance, when choosing the best k of n features for a classification problem, hill-climbing approaches are much less effective than branch-and-bound approaches [25]. Hill-climbing algorithms require a smooth fitness surface uncharacteristic of the feature-selection problem. This simple insight when smooth fitness functions are appropriate constitutes knowledge about the search solution; moreover, utilizing this knowledge injects information into the search. Likewise, the ability to eliminate large regions of the search space in branch-and-bound algorithms supplies problem-specific information. In these examples, the information source is obvious. In other instances, it can be more subtle.

COI stands for Conservation of Information (whatever that means).

So it is not clear to me what D&M are arguing here. But it is true that, at the level of complex organic systems existing within the energy ranges of liquid water, the potential wells among which a system slips and slides are shallow and close together.

The electrical forces in within solids, liquids, and organic systems arise primarily because of Van der Waals forces. These govern the behaviors of membranes and other subsystem parts of a complex organic system

Van der Waals potentials arise because of non-linear interactions between atoms and molecules in close proximity. They are shallow and smooth. Anyone interested can look them up in any good physics text; I won’t eat up space elaborating on them here.

For living organisms in a changing environment, changes in that environment have frequently been small enough that the variability in a population of organisms overlaps the range of shift in the “potential wells” from the old environment to the new. So the mean of the population characteristics can easily slip to a nearby state.

If that shift falls outside the range of variability, extinction is usually the result.

So I don’t see what point D&M are trying to make other than to disparage the use of our knowledge about Nature, or that many search algorithms in computer programs make use of the fact that solutions are “nearby” and that potential wells are indeed smooth.

But that doesn’t have to be the case. We can put in whatever we know about Nature. It’s not illegal nor is it “faking” or putting the answer into the program ahead of time (if that is the argument they are attempting to set up).

The better we understand Nature, and the more we can incorporate this understanding into our computer programs, the more those programs mimic what we see in Nature. It is one of the ways we validate our understanding.

Now if Dembski can get one of his programs to detect a designer, he will have demonstrated his understanding of deities. Or will he? Did he properly initialize his variables?

I don’t think I will hold my breath.

Michael J said:

I think that you have highlighted Dembski’s issue. I’m not sure that he understands it or is being purposefully oblique. He wants to say that evolution is impossible and that we all came about by a number of discrete events (poofs) brought about by God. However, all he is showing is that the fitness landscape is amenable to evolution.

I see other people say that one characteristic of committed creationists is the inability to understand analogies and models. I wonder if this is Dembskis problem rather than outright dishonesty.

I could also be the old Gish Gallop, mud wrestle, stir-the-pot, shtick to keep attention focused on himself. I would not be surprised if he began claiming his paper is being widely cited in the scientific community.

ID/creationists have something like a 40 year history of leveraging “respectability” from the legitimate scientific community by constantly engaging in public and prolonged mud wrestling. It apparently says to the rube following that their dear leaders are important members of the scientific community and that ID is a serious contender that frightens the hell out of the “establishment”.

I think that is where this may be going.

George said: Many creatures have gone extinct.

It would be more accurate to say most creatures have gone extinct. In fact, somewhere between 99.9% and 99.99+% of creatures (species, actually, of course) have gone extinct. Evolution is extravagantly wasteful. (Which puts theistic evolutionists into a bind, because their god is so unlovingly inefficient - but even then not as genocidally cruel as the God of Noah’s Flood, but that’s another story.)

Bill wrote:

“The key contention of ID is that design in nature, and in biology in particular, is detectable. Evolutionary informatics, by looking at the information requirements of evolutionary processes, points to information sources beyond evolution and thus, indirectly, to a designer.”

RIght, but see the thing is that your paper doesn’t do this. All the paper shows is that evolution will not occur without an environment, a trivial result at best. It also seems to show that evolution can occur in the absence of strong selection, but that doesn’t preclude evolution from occuring.

The paper most cetrtainly does not point to any information sources beyond evolution and thus it does not point to a designer. Even if it did point to some other information source, that hardly settles the question of what that source might be.

The funny thing is that if this guy ever does succeed in proving the existence of God, no real person of faith would be happy about it.

Is “Conservation of Information” an actual, well-established principle, or is this just more wankery from WAD?

Paul I think your assertion about wastefulness is wrong. The important thing philosophically would seem to be the survival of life, not which particular form it happened to be in. Would the world somehow be more perfect if all (or most) species were still alive each in small populations (abiotic limitations of the biosphere)?

Dave C said:

Is “Conservation of Information” an actual, well-established principle, or is this just more wankery from WAD?

It would be the latter. Dembski made up the term but it has no use in the scientific community.

You can even get some idea of the issues involved in its use when even Dembski uses examples in this paper under discussion in which he calculates what the “information” would be in English.

What would that calculation be in German? How about Chinese? Dolphin? Chicken?

Conserved quantities in physics do not depend on the language.

Even worse has been the conflation of “information” with entropy and then claiming that “information” is conserved. Entropy is not conserved.

Conflation and confusion are pretty much the game with ID.

Mike Elzinga said:

Dave C said:

Is “Conservation of Information” an actual, well-established principle, or is this just more wankery from WAD?

It would be the latter. Dembski made up the term but it has no use in the scientific community.

You can even get some idea of the issues involved in its use when even Dembski uses examples in this paper under discussion in which he calculates what the “information” would be in English.

What would that calculation be in German? How about Chinese? Dolphin? Chicken?

Conserved quantities in physics do not depend on the language.

Even worse has been the conflation of “information” with entropy and then claiming that “information” is conserved. Entropy is not conserved.

Conflation and confusion are pretty much the game with ID.

Thanks, Mike. I figured as much, but I wasn’t sure.

James F said:

William Dembski wrote:

The key contention of ID is that design in nature, and in biology in particular, is detectable.

And since there is no testable mechanism for how this design is introduced, let alone how to quantify it, it falls outside of the realm of science. Methodological supernaturalism is what really distinguishes ID.

I think it’s more like a priori supernaturalism. After all, they’re mostly looking at these things as “how can I turn this into support for the Designer that I know exists?” Usually they fail even at that and try to turn it into “how does this not support evolution?”

Dave C said:

Is “Conservation of Information” an actual, well-established principle, or is this just more wankery from WAD?

Wankery. There hasn’t been anything like “conservation of information” rigorously used in the sciences, not insignificantly because there are lots of definitions for “information” out there. It seems to me that if you go into the lab with this supposed “Law” of Conservation of Information in your mental toolkit when designing experiments, like the Conservation of Momentum in physics, you’re just assuming from the start that evolution doesn’t produce any information (what happens to be his conclusion here, shockingly enough). That’s what Conservation means, isn’t it?

George:

I am not sure that fully understand all of this. But it seems clear that the adaptive landscape is not all that smooth. Many creatures have gone extinct.

According to J. Coyne in his recent book, “Why evolution is true”, that is the case.

In fact most species go extinct without leaving descendants. 8 of 10 die out eventually. 1 leaves a descendant species 1 splits into two species

The number of species is roughly a steady state on short time scales.

DS said:

RIght, but see the thing is that your paper doesn’t do this. All the paper shows is that evolution will not occur without an environment, a trivial result at best. It also seems to show that evolution can occur in the absence of strong selection, but that doesn’t preclude evolution from occuring.

The paper most cetrtainly does not point to any information sources beyond evolution and thus it does not point to a designer. Even if it did point to some other information source, that hardly settles the question of what that source might be.

The funny thing is that if this guy ever does succeed in proving the existence of God, no real person of faith would be happy about it.

I think this is hitting around the hidden agenda in the D&M paper.

Because of the number of times they repeat the point in their paper, it seems to me that D&M would like to claim that solutions cannot occur without the input of “intelligence”.

This point is made repeatedly by making calculations of the probability of landing on a solution by purely random means and the increased probability of landing on a solution when “information about the target”, as they like to put it, is included in the search algorithm.

It’s a not-so-subtle conflation of “Nature cannot find solutions without the input of intelligence” with “computer programs cannot find solutions without the input of intelligence”.

But as I have said before, this paper repeatedly (and deliberately, I claim) mischaracterizes how computers are being used in scientific research.

Therefore, whatever conclusion they want their following make, does not follow from this mischaracterization of science.

It’s the same ID/creationist shtick all over again.

Dembski:

Evolutionary informatics, by looking at the information requirements of evolutionary processes, points to information sources beyond evolution and thus, indirectly, to a designer.

and

Dembski:

Theistic evoluton.…snip.…required for biological complexity internally, without any outside source of information. The results by Marks and me are showing that this cannot be the case.

Dembski seems to be making claims beyond his paper. The information source beyond “evolution” is the environment aka fitness landscape. How the Designer, jesus, sneaks in here is not explained. Not seeing how the environment or fitness landscape can be confused with a deity.

I’m really going to have to say it. This claim looks like a combination of bafflegab and lying. Nothing Dembski hasn’t been doing for 20 years.

Again, this is in reference to an artificial and controlled environment, hence the submission of the paper to an engineering journal. The “peer review” is for the validity of the paper on engineering, not biology. The distinctions of the “landscape” are artificial and–for the needs of information retrieval–contrived. Dr. Dembski may argue all he wishes that information existed before it was information (“smoothness”), but now we are back to the chicken and the egg.

Out of curiosity I looked up “Conservation of Information.” It’s something Dr. Dembski created as part of his “Specified Complexity,” (i.e. some things are so complex they call for a Creator). He has an article about it in Wikipedia.

This Specified Complexity was thoroughly trashed in the Dover trial, where the plaintiffs demonstrated that no matter how complex something was in nature, it was still an example of evolution in action.

Because of the number of times they repeat the point in their paper, it seems to me that D&M would like to claim that solutions cannot occur without the input of “intelligence”.

One could also point out that nature isn’t actually looking for solutions to problems. People sometimes describe is that way, but that’s a loose analogy, and like any analogy it fails if taken too far. When a person sets out to solve a problem, the problem was known in advance.

Evolution doesn’t have any particular problem in “mind” to be solved; even if a species does evolve a trait that saves it from a threat, the new trait might not have any resemblance to what a person might think of.

Henry

Paul Burnett said:

Joe wrote: “William Dembski has now replied at his Uncommon Descent blog to these comments.

It should be noted that there were no comments allowed on Dembski’s reply posted last night. And he cut off comments after only nine comments on his previous blog entry on this topic on August 19.

Are we allowed to know who were the peers who purportedly reviewed this article? Or is this another Meyer/Sternberg scam?

You can always query the editor of the IEEE Transactions journal in which this was published, though I doubt you’ll get an answer.

But conservation of information (COI), as noted above, is likely Dembski’s own jargon, like Behe’s irreducible complexity notion, that is meaningless except to him because he “defined(?)” it, like his and other creationists’ nonsense terms.

Finally, Dembski, et. al have yet to come up with the ultimate definitive test that tells us what/who the designer is. He and his fellow dishonesty institute colleagues have openly stated it’s the christian deity, but it could just as well be zeus, the FSM, allah, or a fish. So all this is really just rubbish that he proposes.

raven said:

Dembski seems to be making claims beyond his paper. The information source beyond “evolution” is the environment aka fitness landscape. How the Designer, jesus, sneaks in here is not explained. Not seeing how the environment or fitness landscape can be confused with a deity.

In a real sense, evolution by natural selection is an algorithm for transferring ‘information’ from an environment to the genome of a species. The distribution of alleles in a population changes as a function of changes in the environment. Someone – Dawkins, maybe? – suggested that the genome of a species is a palimpsest recording the prior selective environments that species’ ancestors lived in.

Hm. For “algorithm” above, read “process.”

James F said:

William Dembski wrote:

The key contention of ID is that design in nature, and in biology in particular, is detectable.

And since there is no testable mechanism for how this design is introduced, let alone how to quantify it, it falls outside of the realm of science.

My understanding of how Dembski detects design is that he does this:

1. He shows that a genotype achieves a much higher fitness than could be accounted for by purely random mutational processes (without natural selection acting), using a criterion that this high a fitness could not be achieved even once in the Universe over all the time available. This is quite easy: organisms achieve fitness this high all the time.

2. Then he announces that design has been detected.

But wait, you cry, what about natural selection? Can’t it be the reason the genotype is so well adapted? Then Dembski invokes his Law Of Conservation of Complex Specified Information which basically says natural selection can’t increase the fitness this much. (Basically what he calls Specified Information is a proxy for having a high fitness in these arguments). Therefore, when you see nonrandomly high fitness it must be design.

How could he have done that? Well, because his Law actually doesn’t do that. It doesn’t work to conserve the same quantity (fitness) but changes the definition of the specification in midstream. Plus it violates some of its own stated logical conditions. Which means it is both wrong and irrelevant (a combination hard to achieve!).

If you want to see the argument laid out more extensively, see [warning: shameless self-promotion] my article in Reports of the National Center for Science Education, available at that link. There I cite the people who have come up with these criticisms of his work.

Mike Elzinga said:

So it is not clear to me what D&M are arguing here. But it is true that, at the level of complex organic systems existing within the energy ranges of liquid water, the potential wells among which a system slips and slides are shallow and close together.

For living organisms in a changing environment, changes in that environment have frequently been small enough that the variability in a population of organisms overlaps the range of shift in the “potential wells” from the old environment to the new. So the mean of the population characteristics can easily slip to a nearby state.

I am not sure how closely adaptive peaks can correspond to the potential wells in molecular systems. Generally what we are talking about when we use the phrase is plots of population fitness against gene frequencies. If you have two genes, each with two alleles, and it happens that genotypes AABB and aabb have high fitness, with the other double homozygotes AAbb and aaBB having lower fitness, and the various heterozygotes in between, then when you plot population mean fitness against the gene frequencies of alleles A and a you get two peaks with a valley in between.

But for each genotype all this comes from lots of lots of molecules bouncing in and out of potential wells, so I am not sure I can get the one surface from the other. Perhaps I’m being unimaginative.

Joe Felsenstein said: then when you plot population mean fitness against the gene frequencies of alleles A and a you get two peaks with a valley in between.

Oops, that should have been A and B.

From the viewpoint of algorithmic information theory, evolutionary algorithms work because the information content of the fitness landscape is low. If the fitness function is very weakly correlated, i.e., approximating the Pure Noise condition, then hill-climbing or search-by-mutation-and-selection won’t do much better than random sampling. But a function of this type has high Kolmogorov-Chaitin information: random sequences are incompressible. A fitness function which is random in the Martin-Loef sense is exactly the kind of landscape which stymies a search algorithm.

Dembski and Marks are basically doing a kludgy and vague version of fitness landscape analysis. Their “active information” (which I’m not sure should even be called “information”, being the difference between surprisals) is a functional of the algorithm being evaluated and the fitness function which is being optimized. It mixes up the “hardness” of the fitness function with the complexity of the algorithm used to optimize it; Dembski and Marks then reify this number into something which can be “hidden” or “smuggled”.

Joe Felsenstein said:

I am not sure how closely adaptive peaks can correspond to the potential wells in molecular systems. Generally what we are talking about when we use the phrase is plots of population fitness against gene frequencies. If you have two genes, each with two alleles, and it happens that genotypes AABB and aabb have high fitness, with the other double homozygotes AAbb and aaBB having lower fitness, and the various heterozygotes in between, then when you plot population mean fitness against the gene frequencies of alleles A and a you get two peaks with a valley in between.

But for each genotype all this comes from lots of lots of molecules bouncing in and out of potential wells, so I am not sure I can get the one surface from the other. Perhaps I’m being unimaginative.

I am talking about this in the language of a physicist because the notions of work, energy, and entropy as well as “information’ have come up in a way that has lead to many issues of conflation and confusion (as evidenced, for example, by one of the posters who insists that “work”, and he emphasized in the physics sense, is required to transfer “information” from the environment into the genome).

Biological systems have reached such a level of complexity that those emergent properties that are more appropriate to describe them have no meaning at the levels below where we can clearly talk about potential wells and the more microscopic phenomena that are best described in the language of physics.

However, as in all complex systems, those higher levels are what we in the physics community often refer to a “phenomenological manifestations of the underlying physics”. Other terminology, from statistical mechanics for example, refers to them as macroscopic states that are consistent with underlying microscopic states.

It is perfectly rational and appropriate to use those “upper level” or macroscopic properties of a system to describe the system. But it is also important to be aware that the underlying physics and chemistry are not being violated by what the entire system is doing within its environment. I am sure every biologist is at least subliminally aware of this and would not disagree.

But in the context of these arguments with ID/creationists, language really gets screwed up, and we in the physics community are seeing physics concepts being bent beyond recognition. Just in this discussion alone, the concept of “hill-climbing” has been subtly conflated with potential hills in physics (i.e., work being required to go up them). We don’t blame the biologists; we blame the obfuscators from the ID/creationist community.

So those “fitness functions” are appropriate, but what is going on beneath the surface is chemistry and physics. Turned upside down, those fitness peaks are a manifestation of underlying molecular subsystems relaxing into myriads of potential wells in addition to the actual forces acting on the systems as a whole and moving them toward different phenotypic configurations from generation to generation. This is not Lamarckism, its selection and drift.

Energy is flowing “downhill”, the atoms and molecules of the system are slipping in and out of potential wells, and entropy is increasing globally. There is no clear relationship between the entropy of any part of a complex living system and the complexity of the system itself. All of these points have been misconstrued by the ID/creationist community, and defenders of science have often unwittingly picked up the bad language.

In short, I as a physicist am not criticizing the concepts that the biologists use to describe biological systems. They are entirely appropriate.

However, as a physicist who has had considerable experience with complex condensed matter systems, I see the day coming when physicists, chemists, and biologists will need to be far more consciously aware of the connections among their respective disciplines in living systems. Ultimately, the concepts among the various levels of complexity will have to be consistent. We all hope that someday, we can trace the properties at high levels to those at lower levels.

I am more than happy to be corrected if I misconstrue any concepts of the biologists (I hope I haven’t mangled any biology too badly).

And I would hope we in the physics community can offer some clarification on what is happening in these biological systems in terms of the underlying physics.

This selective/adaptive landscape stuff is fine, but I think people sometimes lose sight of the trees for the wood. Is Dembski really saying that the “Designer” (PICC) has to SPECIFY a selective landscape, such that that for gazelles to avoid cheetahs, one useful strategy might be that they could be FASTER? Or for cheetahs to catch gazelles, THEY need to be faster in turn?

Gee, thanks Mr Pixie - that is most helpful. Thankyou for defining that information-laden selective landscape that permits such a counter-intuitive and para-Darwinian survival strategy to be available to the respective genepools of gazelledom and cheetahdom.

Yet, ALL of the selective landscape effectively boils down to such local conditions - which may change from time to time, and at the end of the day, all that is simply mathematical. Perhaps Dembski also thinks that the Pixie can change the Mandelbrot set, or re-define Pi.

In my younger years, inanity irritated me, but now I appreciate that some people can elevate it to an art form of the highest virtuosity.

Ray Martinez said: Your penchant for defining “mutability” to be “variation” is again noted. Your argument ASSUMES its conclusion each step of the way: “variation means mutability has occurred.” This is called “begging the question.”

No, its called paying attention to what words actually mean. Immutable means not able to be changed. You seem to be claiming that species are immutable because the mechanism of change is divine action. That’s just a meaningless claim.

But the heart of your claim seems to be that variation within a species can occur but cannot lead to speciation. This is factually incorrect, as speciation has been directly observed. Again, you can claim some divine mechanism for this observed speciation if you want, but by doing so you’re accepting the observed fact that speciation occurs; therefore, species aren’t immutable.

Now on a personal note, I have to say that I find your hypothesis (change occurring via divine action) theologically appealing. That would mean that Lenski and other biologists can force God to perform miracles on schedule simply by repeating their experiments, and that idea tickles me.

eric said:

That would mean that Lenski and other biologists can force God to perform miracles on schedule simply by repeating their experiments, and that idea tickles me.

And what a relief to know that if God didn’t jump through the hoop we set, well, it was just because God works in mysterious ways and does what he/she/it thinks best for us.

About this Entry

This page contains a single entry by Guest Contributor published on August 24, 2009 4:32 PM.

Iguana iguana was the previous entry in this blog.

Evolution of the appendix? is the next entry in this blog.

Find recent content on the main index or look in the archives to find all content.

Categories

Archives

Author Archives

Powered by Movable Type 4.381

Site Meter