Two forthcoming peer-reviewed pro-ID articles in the math/eng literature
| January 20, 2009 | Posted by William Dembski under Informatics, Intelligent Design |
The publications page at EvoInfo.org has just been updated. Two forthcoming peer-reviewed articles that Robert Marks and I did are now up online (both should be published later this year).*
——————————————————-
“Conservation of Information in Search: Measuring the Cost of Success”
William A. Dembski and Robert J. Marks II
Abstract: Conservation of information theorems indicate that any search algorithm performs on average as well as random search without replacement unless it takes advantage of problem-specific information about the search target or the search-space structure. Combinatorics shows that even a moderately sized search requires problem-specific information to be successful. Three measures to characterize the information required for successful search are (1) endogenous information, which measures the difficulty of finding a target using random search; (2) exogenous information, which measures the difficulty that remains in finding a target once a search takes advantage of problem-specific information; and (3) active information, which, as the difference between endogenous and exogenous information, measures the contribution of problem-specific information for successfully finding a target. This paper develops a methodology based on these information measures to gauge the effectiveness with which problem-specific information facilitates successful search. It then applies this methodology to various search tools widely used in evolutionary search.
[ pdf draft ]
——————————————————-
“The Search for a Search: Measuring the Information Cost of Higher Level Search”
William A. Dembski and Robert J. Marks II
Abstract: Many searches are needle-in-the-haystack problems, looking for small targets in large spaces. In such cases, blind search can stand no hope of success. Success, instead, requires an assisted search. But whence the assistance required for a search to be successful? To pose the question this way suggests that successful searches do not emerge spontaneously but need themselves to be discovered via a search. The question then naturally arises whether such a higher-level “search for a search” is any easier than the original search. We prove two results: (1) The Horizontal No Free Lunch Theorem, which shows that average relative performance of searches never exceeds unassisted or blind searches. (2) The Vertical No Free Lunch Theorem, which shows that the difficulty of searching for a successful search increases exponentially compared to the difficulty of the original search.
[ pdf draft ]
—————
*For obvious reasons I’m not sharing the names of the publications until the articles are actually in print.
278 Responses to Two forthcoming peer-reviewed pro-ID articles in the math/eng literature
Leave a Reply
You must be logged in to post a comment.
jerry [269] I see you’ve found some of the problems you run into when trying to use the species concept in an evolved world. “Species” is an idea straight from the ancient Babylonian world and its Old Testament variations. An Intelligent Designer God designs and builds animals and plants separately and they grow separately and you expect to always be able to clearly separate different species from each other. But in an evolved world, a single population starts to split and you have two populations that share some attributes and not others and you have real trouble separating one from the other. Hence we have the problem of teosinte and corn and wolves and chihuahuas. Are they a single species or two? Scientists wrestled with this idea for quite a long while before finally, rather grudgingly, settling on the ability to cross fertilize as the definition of species. Our official, man-made definition of species is any group of animals that can mate and produce fertile offspring. Under evolutionary conditions, this marks the point where some key genes have diverged enough so they don’t match up when mated and the offspring either dies or some key part of it’s reproduction process is inoperative. It works in the sense that from that point on the two populations are forever separated and will always evolve separately, but, as you’ve shown, that definition has a lot of problems. Another problem is that it’s useless with non-sexually reproducing organisms and that includes the vast majority of all living creatures, which are single celled. You and I, for instance, have more bacterial cells in us (mostly in the digestive tract) than human cells.
Under evolution, there’s just no hard and fast division between two populations that are evolving apart short of reproductive incompatability, even though one is very much needed. For instance, how do you call it when two plants have 99.9999+% identical genomes, but one is a clump of grass and one is corn? Or one is a chihuahua and one is a wolf? You can say they’re both the same species, but if you plant teosinte instead of corn, your family is going to starve and if a chihuahua goes for your throat it’s very annoying, but if a wolf goes for your throat, you’re dead. I guess one way to know if ID or evolution is right is to ask how easy it is to separate species and so far the data points entirely towards evolution.
I’m afraid teosinte is NOT mutated corn. We have archeological specimens of teosinte plants and pollen going back a long, long ways, but corn doesn’t even start to appear until about 9,000 years ago and it’s a very crude, teosinte-like version of corn. Incidently, all of the early differences are due to natural selection. Humans changed the teosinte environment in a big way when they started eating it and planting it. Once people started saving their best plants for seed and eating the lesser plants, it put natural selection on steroids and the plan transformed from teosinte, a barely edible grass, into corn, a highly desireable and nutritious plant to humans, in a few thousand years.
“When we refer to no natural examples of FCSI ever being formed by nature we do not include life since life is the topic under discussion.” My mind just boggles at this. When we refer to no natural examples of FCSI ever being formed by nature, we do not include life, although life is a part of nature, since life is the topic under discussion.” Since the question is, “Does life produce FCSI?” you seem to have constructed an impregnable position. A veritable Maginot line of logic.
“… the changes we have seen are all trivial in terms of genomic change and no complex new functional capabilities have ever been demonstrated.”
It’s certainly true that the changes we see in teosinte to corn are “trivial in terms of genomic change”. There are about 2 billion DNA base-pairs in the corn genome and if we have three point mutations, that’s only 3/2,000,000,000ths of the corn genome. And yet just those three changes make the plant taller and narrower and concentrate the ears near the stalk, the ears get larger and the coating on the kernels gets softer. “Trivial” changes in the genome can obviously make very significant changes in the phenotype and it’s the phenotype that feeds you or rips your throat out.
As for “no complex new functional capabilities have ever been demonstrated”, I’d say that just those three mutations produce functional changes that are complex enough to turn a barely edible weed into something that can start to feed the entire New World.
How many base-pairs do you think mutated to turn teosinte into corn? I’m betting it’s way under a thousand, which, as you say, is “trivial in terms of genomic change”.
I’m glad you brought up bat sonar. I read something about that a few years ago. This is from memory and I don’t have a citation handy, but since it’s the topic under discussion I don’t see why I have to provide one. Don’t want to beg the question, after all. For starters, you and I can already do a crude form of echolocation. If you shout, “Hello!” and hear a fainter “Hello!” a few seconds later, you know you’re some distance from a large reflecting object, such as a cliff. If you say, “Hi!” and hear an almost immediate series of echos, you’re in a gymnasium or concert hall sized room. If you say, “Hi!” and instantly get a load of echos in return, you’re in a small room. If you hear a lot of high frequencies in the echo, the surfaces are hard, if not they’re soft. You can tell direction pretty well too, by using the already existing ability of your ears and brain to tell what direction a sound is coming from. Blind people get very good at this, by the way.
Now if you’re a small flying mammal hunting insects in dim light, what kind of changes can help you find more insects via echo-location? Well, for starters you can shout, “Hi!” louder. This will detect insects farther away. You can make your shouts shorter, too, which makes your distance measurements more accurate. Or, like some bats, you can “chirp” where you send out a loud cry that sweeps from a low to high frequency. That helps you sort out the mass of echos that come back to you. If you hear low and high sounds, you’ll know that the low pitched sounds are coming from a greater distance away because they were uttered first.
Better ears will help too. If the shape of your outer ears changes, they can make sounds coming from different directions have phase relationships that increase your ability to tell direction. Your face can change too, to make those phase relationships clearer. (That’s why bats tend to have faces like gargoyles – it helps them determine directions.) If you develop a really, really loud chirp or yelp, you can have a problem with your voice damaging your hearing. (I’ve heard some bats have such loud calls that they can stun insects with them.) Contracting muscles in the middle ear while you shout can clamp the auditory bones, protecting them from damage. I’m sure there are many more ways to improve echo location.
Now can you tell us which, if any, of these changes are impossible to produce? And remember before answering that all of them and more exist in contemporary bats and other echo-locators.
djmullen,
I am stopping this conversation. Some others can waste their time with you. You do not know what ID is about and seem only to impose your own ill conceived perceptions. So you can have the last word till you start dealing with reality and what people say.
For example, I know all the problems with the definition of species you bring up and they have been discussed here before and I have read about the problem in pro-Darwinian and anti ID books as well as with pro ID books and discussions and so it is a non issue. Wolves and dogs theoretically parted millions of years ago but still remain the same species by biological textbooks. My example of the Chihuahua and the wolf was meant to point out the absurdity of your example.
I just pointed out that it is hard to use an example of one variant of a species morphing into another variant of the same species as proof of Darwinian macroevolution. Especially when it looks like the mechanism is probably artificial selection and not natural selection. And if it was by natural selection, ID accepts that example with open arms. So what point are you making by first choosing it and then defending it. And you defend it badly because no one here said teosinte came from corn when it was obvious that it is the reverse. It seems the indigenous people of the Americas started the process several thousand years ago probably first by accident and then maybe they got smart. So what is your point. Your discussion seems pointless. Don’t bring up trivial examples that ID is wholly in sync with and expect us to take you seriously. Nothing you said so far has any merit except maybe in your own mind. Adios.
Jerry #271#
“A more interesting question seems to be can any change or any distribution of changes ever lead to a protein reaching one of these other isolated protein islands of foldable proteins?”
This has been investigated and there are cases where very few changes are necessary to connect two different islands of foldable proteins. One reported artificial example for a small protein (56 amino acids) showed that only seven mutations where necessary to reach a different stable! fold and a new functionality (Proc. Natl. Acad. Sci. USA, Vol. 104, No. 29 pages 11963-11968, Orban, Bryan & coworkers).
An even more extrem example is the so called Arc-repressor protein where two mutations are necessary to change the fold and a single muation results in a protein existing in an equilibrium between two different folds. (Sauer & coworkers, Nature structural biology 2000, Vol. 7, page 1129-1132)
For folds of RNA there is one example where an RNA can adopt two totally different folds associated with two totally different functions.
Thus, the islands of possible protein folds are not necessarily isolated. Furthermore, one should not forget that not all functional proteins must be folded to function but many fold only upon interaction with a ligand. This means that even partially unfolded proteins might not be non-functional.
The cited papers should be open access since they are more then a year old.
Jerry @271
Excellent question. This should get the conversation back on topic.
If the probability distribution of solutions generated by MET mechanisms is not uniform then the NFL theorems and Dr. Dembski’s two new papers are not applicable, as written, to biological systems.
In fact, MET predicts and observations confirm that the solutions generated by MET mechanisms do not form a uniform distribution, very far from it in fact. Further, the search space itself is not uniform — there are clusters of viability.
Given that, Dr. Dembski’s papers cannot be said to be “pro ID” without further work to tie his results specifically to real world biological systems.
The uniform probability distribution is also important in the calculation of CSI, but the discussion above has already explained that.
JJ
rna,
These are interesting examples. And should be brought up at the appropriate time in the new Durston thread. If you do not, then I will. It could mean that I have misinterpreted what Durston has said, always a possibility, or there could be some other explanation. Read Durston’s comments at
http://www.uncommondescent.com.....ent-303307
Thank you for the information.
Jerry [273] “I just pointed out that it is hard to use an example of one variant of a species morphing into another variant of the same species as proof of Darwinian macroevolution.”
Questions of species don’t change the fact that three single point mutations caused gross changes in the body plan of teosinte AND a new tissue in the covering of the kernels. Do micro evolution three times and presto, you’ve got macro evolution.
“…no one here said teosinte came from corn when it was obvious that it is the reverse.”
I misinterpreted your statement in [269]: “The example of teosinte to corn is an example of devolution since the gene pool of teosinte probably contains the gene pool of corn except for a mutation or two.” But, since this is now the topic of discussion, it would be begging the question to discuss it.
JayM 270,
I thought I’d check back on this. You’ve missed my points completely and are reduced to repeating yourself. The only way you can continue your assertion of “uniform probability distributions are not valid for biology” is to ignore specific cases where they are and then refer back to cases where they are not. Humorously, you then say that I haven’t addressed your points yet I did not attempt to address them in the first place and took pains to explain why: they are valid within their limited scope as I’ve already said multiple times! If you don’t get it at this point I don’t know what else to say. Others are understanding my meaning. Although, it’s obvious you haven’t read No Free Lunch or other explanations of calculating CSI (or FCSI) since Dembski did exactly what you demanded in #255 and again in #270. You also made a gross error by referring to the entire genome instead of the specified sequence. So at this point I’m forced to conclude that you do not comprehend what you’re attempting to criticize.
Now my copy of No Free Lunch is somewhere in storage. Dembski spent pages going over the flagellum, but the result was 10^-1170 or around 3800+ informational bits (no calculator on hand so that # may be way off). As I said in 267 “the Explanatory Filter can take multiple types of inputs (which also makes it susceptible to GIGO and thus falsification). Two are (a) the encoded digital object and (b) hypothetical indirect pathways that lead to said objects.” Now if MET mechanisms can indeed traverse an indirect pathway to the flagellum then Dembski’s calculations amount to Garbage In.
So, essentially what you’re demanding of Dembski is that he accept what is in question as already verified: to assume that an indirect pathway for every object exists when doing calculations. That premise would make Dembski’s position untenable. But since no one can even come up with a potential indirect pathway then how can we calculate otherwise?
BTW, based upon the flagellum research of the time Dembski ran his calculations based upon the assumed IC core of 30 proteins out of the total 42. It’s possible that less are required for the IC core, but even if halved that would not substantially change the outcome since it’d still be well over 500 informational bits. This is also why Dembski and Behe’s positions are inter-dependent. As Behe acknowledged long ago IC as an indicator for intelligent agency is only sustainable if all potential indirect pathways are infeasible, which is where Dembski’s work comes in. And Behe is trying to research the limits of MET mechanisms, which would validate Dembski’s assumption. But like you I prefer Behe’s line of inquiry since it’s based on things more tangible and easier to comprehend.
On a side note, here’s one foresighted mechanism. And the specific ID-compatible hypotheses I was referring to were variants of front-loading and other hypotheses about foresighted mechanisms (viral, epigenetics, etc.). I cannot help it if you are ignorant about the full breadth of ID thought but please do not attempt to argue with me on this basis alone.