Two forthcoming peer-reviewed pro-ID articles in the math/eng literature

_{William Dembski

January 20, 2009

Informatics, Intelligent Design}

Share: Facebook; Twitter; LinkedIn; Flipboard; Print; Email

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.

Comments

But here you have slipped up and treated the abstract code as physical, and have gone from there to say that if scientists model genes as encoding proteins, then something must have done the encoding. ... In mainstream evolutionary theory, there is no encoder. Merely changing the bases in a chromosome is not encoding. Exchange of DNA among organisms, perhaps mediated by viruses, is not encoding.
Huh?? Your objection just confuses the issues. Unguided variation may not be an "encoder" (which I presume you're saying is inherently an intelligent agent) but changes in information must meet the coding scheme in order to be valid for transcription. Error correction just enforces this requirement of being encoded properly. And I'd like to see how you'd get from a basic chemical replicator to an information-based replicator with rules defining a coding scheme. In any case, this particular discussion already took place not too long ago in A Simple Gene Origination Calculation.Patrick_{January 26, 2009
January
01
Jan
26
26
2009
09:55 AM
9
09
55
AM
PDT}

Leaving aside the points related to OOL and the initial genetic toolkit, this is the main issue:
Once you have a population of imperfect replicators, djmullen is absolutely correct when he notes that there is no need to search a large genome space. ... That’s equivalent to just looking around the (very) local region of genome space for equivalent or better fitness. Over many generations, the genomic makeup of the population may change, even significantly. At no point, however, is there a need to search prohibitively large regions when starting from a viable point.
The unqualified assumption is that for every potential long-range target there is a series of fitness-improving functional (or at least non-deleterious) intermediates in the "(very) local region of genome space". ID proponents think that for most starting "viable points" in genome space that the search landscape could be likened to a valley surrounded by sheer cliffs. The valley consists of other viable points in local genome space, but is limited in scope. Some ID proponents believe these sheer cliffs can be traversed by intelligent mechanisms. Darwinists who think that "many micro-evolutionary events leads to macro-evolution" think there exists very narrow paths burrowing to other valleys, but even these are very difficult to traverse, and we're lacking evidence for their existence. Other Darwinists think there are mechanisms that allow the traversal of the cliffs to produce macro-evolution. And when it comes to long-range targets this leads to standing challenges like the flagellum. Although I'd say that gpuccio at #118 does a better job of describing the problem than I do in those previous discussions.Patrick_{January 26, 2009
January
01
Jan
26
26
2009
09:38 AM
9
09
38
AM
PDT}

CYJman: The problem is that ID can also invalidate the paper if you want it to. We know that evolutionary search spaces in biology are non-uniform so arguing that they couldn't have been 'found' just begs the answer that they were designed - which is exactly the position of many theistic evolutionists.Laminar_{January 26, 2009
January
01
Jan
26
26
2009
09:29 AM
9
09
29
AM
PDT}

Dembski and Marks can do fancy things with probabilities, but in the applications Dembski really cares about, he cannot get the probabilities he needs. I have always said that active information may have engineering applications.Sal Gal_{January 26, 2009
January
01
Jan
26
26
2009
09:10 AM
9
09
10
AM
PDT}

gpuccio, As Patrick points out, a code is an abstraction. All models are abstractions. You are generally very good about avoiding reification. But here you have slipped up and treated the abstract code as physical, and have gone from there to say that if scientists model genes as encoding proteins, then something must have done the encoding. Consider that you are doing with the genetic code what others do with fitness functions.
It could have been encoded through the amazing works of RV + NS, as darwinists believe, but encoded it is just the same.
This goes to the crux of the matter. In mainstream evolutionary theory, there is no encoder. Merely changing the bases in a chromosome is not encoding. Exchange of DNA among organisms, perhaps mediated by viruses, is not encoding. I commented on probability because you seem to regard complex biological structures and processes as improbable. Tell me if you in fact do not. I mentioned "wonderment" because your degree of wonder at complexity may be converted into a subjective (Bayesian) probability, but not an objective (frequentist) probability. No one can assign an objective probability to the proposition that our empirical observations of the universe may be accounted for strictly in terms of matter, energy, and their interactions. And this spells death for improbability-based arguments that some events have causes attributable to something other than interactions of matter and energy.Sal Gal_{January 26, 2009
January
01
Jan
26
26
2009
09:06 AM
9
09
06
AM
PDT}

Patrick #109:
As for the information/code and rocks discussion: a code is an abstraction. How is a rock in a stream an abstraction?
Thank you, that's helpful and simple. My question now becomes: where is the "abstraction" in DNA? I think we're projecting the abstraction onto it. Before we came along, there was just this mechanism -- if one series of base pairs was present, a cell developed in one way, if another base pair, another course of development. Grossly oversimplified, perhaps (probably), but I don't think it's fundamentally off base. Anyway, we looked at it and conceived of the "abstraction" of "instructions" or "code."pubdef_{January 26, 2009
January
01
Jan
26
26
2009
08:51 AM
8
08
51
AM
PDT}

Some people here seem to be asking, "How does this relate to life, evolution, and ID Theory?" Well, first, life is founded upon the processing of a system of signs/units (measurable in binary digits). The organization of these bits form patterns and a certain percentage of those patterns, when processed produce function which may aid in the survival of the system itself. Since the bits are subject to mutations and transfers of many types, they are subject to bit flips and operations. Thus, the patterns are subject to search (as defined in the first para above in my previous comment). This is why mutations can change one functional or non-functional pattern into another functioning or non-functioning pattern. When a functional pattern is generated, it can be said that that pattern has been found by searching through the possible patterns. Obviously evolution by natural selection provides the ratcheting filter whereby some patterns are kept and other are rejected. Now, "How does this relate to ID?" Since it is just as improbable to find the pattern which can be processed into function as it is improbable to find the match between search procedure and search space which will increase the probability of finding the functional pattern, then the probability of finding the set of laws and initial conditions to allow the structure of life and evolution from a uniform search space is just as improbable as finding the results of life and evolution. So, we can have an infinite regress of fortuitous and highly improbable matchings of search space to search procedure (active information) to ultimately increase the probability drastically of finding a system which can apply foresight (the brain), or we can hypothesize other options. The infinite regress seems too similar to "its turtles all the way down," and I am personally unaware of any evidence of a tested pre-universe evolutionary process, and so this provides no real explanation for highly improbable, functional results seen in life. "Evolution-did-it" doesn't cut it anymore. That evolution must now be explained since it is also just as difficult to find within a higher order search space as the human brain would be to find in a uniform search space -- that is, by chance. If we must ultimately explain everything in terms of law and chance, then it seems that the only real option, other than "turtles (evolutionary algorithms) all the way down" would be to simulate evolution generating itself from background noise (chance) and an arbitrary collection of laws (set of laws put together absent any consideration for future results – absent foresight). But is there another option? Of course, ID Theory says that there is. Systems which are capable of foresight (modeling the future and producing targets) can also increase the probability of finding a given pattern. Intelligent systems do this by applying foresight when matching the proper search procedure to search space in order to increase the probability of finding a given pattern. This has been observed and evidence provided by the NFLT shows that a simulation of evolution will not work unless "problem specific information" about the search space and target is incorporated into the behavior of the algorithm. Basically, this provides evidence that evolution can not even be simulated without some knowledge of the characteristics of the target being incorporated into the programming of the behavior of the interaction between search procedure and search space. So if future knowledge of the target is necessary to even simulate evolution, what happens to the hypothesis that evolution needs no foresight? Put all of this together and we see that the ID Hypothesis is valid and scientific since it is based on observation and is the only verified option that is provided by the math within these two papers. Furthermore, it is falsifiable by showing through testing and observation that just law and chance absent foresight can increase the probability of finding a given target. So far, no one has shown that to be the case and the NFLT provides evidence that it may be practically impossible.CJYman_{January 26, 2009
January
01
Jan
26
26
2009
07:46 AM
7
07
46
AM
PDT}

It seems that some people here are not quite understanding what a search is in relation to these papers. The word "search" in this context has no intrinsic teleology associated with it. It merely describes the flipping of bits (bit operations), while describing the search space in terms of bits. This is where the probability of a pattern comes in to play. The rest of this is kinda lengthy, so I'm gonna post the rest in sections. What is the probability that a given pattern (measured in bits) will be generated by unguided (random) bit operations? That is simply calculated as the pattern's probability -- assuming a uniform search space. Now, the point of these papers is to ask [and answer] the question: "how can the probability of generating a pattern be increased *and what is the probability of finding a way to increase the probability of finding that given pattern*?" Little bit of a tongue twister, but read it a couple times and it'll make sense. Well, basically, there can exist a search procedure (bit flipping operation) which rejects some bit operations/flips and keeps others. However, if the search space is uniform, this does not increase the probability for finding a given pattern. What needs to happen is that the search procedure needs to be matched with the proper search space that will allow the filter to actually improve the probability for finding a given pattern. This has already been proven in the NFLT. Dembski and Mark's paper picks up from there. What is the probability of matching a search space to a search procedure in order to increase the probability of finding a given pattern? They have merely shown that the probability of finding that match can be no less (and apparently increases exponentially with every higher level search) than the probability of finding the given pattern within a uniform search space in the first place. So appealing to a non-uniform search space and an evolutionary filter does not provide a solution to increasing the probability, since the probability (information) is merely moved back a step to the probability of finding the set of laws and initial conditions which will provide a non-uniform search space and ratcheting filter to increase the probability of finding the given pattern. Thus, according to the math provided by Dembski and Marks, it is just as improbable to find a given pattern (measured against a uniform search space) as it is to find the search procedure (and landscape) to increase the probability of finding that given pattern.CJYman_{January 26, 2009
January
01
Jan
26
26
2009
07:43 AM
7
07
43
AM
PDT}

gpuccio @118
So I ask you: how did these two proteins (and all the other hundreds of different proteins, with hundreds of different foldings, 3D structures and functions) arise?
I think you're asking a different question than djmullen was answering. ;-) There are two issues being conflated here. The first, which I think is what you're focusing on, is origin of life. That is, of course, a fascinating area of research. The second, which djmullen discussed so eloquently, is the result of evolutionary mechanisms once life, or at least populations of replicating entities, exists. This is where I have trouble seeing the connection between Dr. Dembski's two papers and ID. Once you have a population of imperfect replicators, djmullen is absolutely correct when he notes that there is no need to search a large genome space. Most replicators will have the same genetic makeup of their parent(s). Some will mutate but still maintain far more similarity than difference. That's equivalent to just looking around the (very) local region of genome space for equivalent or better fitness. Over many generations, the genomic makeup of the population may change, even significantly. At no point, however, is there a need to search prohibitively large regions when starting from a viable point. How to get to that point and the size of the connected, viable regions are interesting questions. JJJayM_{January 26, 2009
January
01
Jan
26
26
2009
07:37 AM
7
07
37
AM
PDT}

jerry[114], My family tree starts with a guy named Sven in the early 17th century. Doesn't get more Swedish than that. I need to visit you; here in San Antonio it hardly ever freezes and I would love to see some winter. And listen to sarcastic comments... Anyway, yes, I did see your attempt to answer yourself. I don't think we can sort this out until we learn how these results are intended to be relevant to ID/biology. The construction in the "search for a search" paper is very technical (I haven't read the other paper). I understand the technicalities but cannot see the connection to biology. Skal! POProf_P.Olofsson_{January 26, 2009
January
01
Jan
26
26
2009
06:54 AM
6
06
54
AM
PDT}

djmullen: Do you really mean what you are saying? Let's start with your (obviously imaginary) bacterium with only one protein. First of all, in case you forgot it, we still have to explain how he got that protein (that is, we still have to solve the problem of OOL). You will probably say that it originated from simpler proteins. But the smallest proteins which can fold and exercise an autonomous biochemical function as enzymes are usually 80-100 aminoacids long (one of the smallest reported subunits is 62). So, where are the simpler proteins (functional proteins) from which the actual ones we see today originated? But let's leave that alone, for now. You have your one protein bacterium. Now you state the most incredible thing: that when a bacterium duplicates its DNA, there is no search necessary. That is incredible, not because it is not true, but because it is so trivial that I really can't understand why you have to state it. Who has ever thought to deny that? It is obvious that DNA duplication is not a search! But let's go on. Now, let's compare your one protein bacterium with a (more realistic) 500 protein bacterium. And let's suppose that the one protein bacterium is the progenitor of the other one (being simpler, that's a fair supposition). Now just tell me: whence did all the other 499 proteins come? You say: it's easy, they came by evolution, just substituting an aminoacid at a time, and always remaining in the sweet spot of functional proteins. And I say, are you kidding? You are simply ignoring a very essential fact: the 500 proteins are completely different one from the other. Do you want an example? In the genome of E. coli, a well known bacterium, we can find at least 2018 proteins described, grouped in 525 groups of similar proteins. The biggest group includes 52 proteins, while the smallest groups include only two proteins each, and there are 291 of them. Now, let's take, randomly, two proteins from two of the smallest groups. One is called dinG, is described as a "Probable ATP-dependent helicase", and is 716 aminoacids long. The second is called clpA, is described as "ATP-dependent Clp protease ATP-binding subunit" (it is indeed a subunit of a more complex protein), and is 758 aminoacids long. Now, I have taken these two examples completely at random, from two of the 291 smaller groups (of two similar proteins) in the E. coli genome. I could have taken thousands of different pairs. I have just selected two proteins with approximately the same length, for a more clear discussion. Then I have blasted the two sequences with blastp, on the NCBI site. That software, which is routinely used for research, looks for possible alignments between two (or more) protein sequences. This is the result: Only four possible partial alignments were found. The best was 49 aminoacids long, and presented 15 identities (15/49). The second was 7/8. The third was 12/30, and the fourth 6/11. In other words, even considering all four possible alignments (which, obviously, are not really compatible one with the other), we get only 40 identities out of two sequences which are each more than 700 aminoacids long. So I ask you: how did these two proteins (and all the other hundreds of different proteins, with hundreds of different foldings, 3D structures and functions) arise? How were these two completely different functional sequences found, remaining always in your imaginary "sweet spot"? How can your concept of "no search" apply to that? Just answer that. And, as you seem to be familiar with combinatorial computing, I need not remind you that the combinatorial space of 700 aminoacid sequences is 20^700.gpuccio_{January 26, 2009
January
01
Jan
26
26
2009
06:39 AM
6
06
39
AM
PDT}

djmullen @116 That was an incredibly lucid explanation, thank you. Of course, it supports my personal contention that the "edge of evolution" (where have I heard that phrase before?) is a rich area for ID research, so I'm a friendly audience for you. While I would like to see more peer-reviewed papers supporting ID, I find your and Professor Olofsson's arguments persuasive. After re-reading Dr. Dembski's two papers, I can't honestly portray them as supportive of ID. There is a gap between the math and what we know about biology. I look forward to someone making the necessary connections. JJJayM_{January 26, 2009
January
01
Jan
26
26
2009
04:58 AM
4
04
58
AM
PDT}

gpuccio: I'll try to express myself better with an example. Let's use an extremely small bacteria, one with only a single protein gene. "Protein gene" refers to a portion of its DNA whose pattern of DNA bases specifes the pattern of amino acids in a protein. (To simplify the discussion, we'll assume that the stretch of protein gene DNA is continguous and codes for a single protein.) Further assume that this particular bacteria has been happily reproducing for some time and that its offspring are viable bacteria, also capable of reproducing, and that it wants to reproduce again. Where is the bacteria going to get the exact sequence of DNA bases it needs in its protein gene that will specify the correct sequence of amino acids to produce the needed protein? Dembski and Marks seem to think the bacteria has to find this sequence via some sort of a search process, which starts with every conceivable DNA sequence that will fit into the gene and rejects all but the single sequence of DNA bases that will produce the correct protein. At least that's what I gather from reading the words in the paper and ignoring the math, which is beyond me. This would be a gargantuan task. If the protein gene was only 400 base-pairs long, there would be 4^400 or 6.6 E+240 different combinations to try. The universe will burn to a cinder long before the bacteria even gets properly started with this task. What the bacteria does instead is much simpler - it merely makes an exact copy of its DNA, including the section that codes for that protein, and hands that copy down to its offspring. No searching is necessary. Since the bacteria has been using its DNA and the protein it encodes for to successfully reproduce, we know that the protein gene is a good one because successful reproduction is our definition of success. No searching whatsoever is necessary. Now let me show you when the bacteria actually does "search" the search space and how it greatly improves the odds of finding a workable protein gene by searching only areas of the search space that are "close" to itself. Let's use a larger bacteria, one with 500 protein genes. That means that 500 stretches of its DNA encode for proteins and we know that all of those proteins "work" because this bacteria is successfully reproducing. Suppose that a single base-pair in one of the 500 protein genes gets mutated during reproduction. That means that the other 499 protein genes are good ones because they didn't mutate and are identical to their parent's. So the bacteria isn't chosing a point in the search space at random, it's choosing a point that is so close to where it started from that 499 out of 500 protein genes are known to be correct. The odds of the new genome being servicable are obviously hugely more likely than if you choose a point in the search space at random, which would be done by mutating every single base pair. Think of evolution this way: three or four billion years ago, a sub-microscopic molecule was randomly put together that was able to reproduce itself before the same forces that assembled it tore it apart. It's offspring was also able to reproduce itself before it was destroyed. This puts the molecule squarely in the tiny section of the humongus search space that is able to successfully reproduce. The molecule doesn't have to search for the sweet spot, it's already in it, in a section that's likely enough so that natural forces could randomly assemble a molecule that's in it in a few hundred million years or less. All the molecule has to do to keep its offspring in the sweet spot is to make an exact duplicate of itself. No searching necessary. When the molecule is mutated, then the new, different offspring is tested to see if it's also in the sweet spot. It's tests itself by trying to stay intact and reproduce. Again, only nearby sections of the search space are tested. If the molecule is 100 atoms long and we only change a single atom, then 99% of the molecule is "known good" and we're only testing the effect of the single changed atom. This is much more likely to be successful than randomly assembling a 100 atom long molecule and seeing if it works. If it doesn't work, the new molecule is destroyed. But if the new pattern is good, then we now have TWO different species of molecules that successfully self-reproduce and evolution is off and running. This is how evolution works. Every viable species is always in the sweet spot. It is never necessary to search for a successful DNA pattern to hand down to the offspring, you merely duplicate your known good pattern. If that pattern mutates, you are only then searching the genome space for a viable DNA pattern and you are only searching that space in the immediate vicinity of your known good pattern. I honestly don't understand why Dembksi and Marks are so concerned with searching large spaces. That has nothing to do with evolution.djmullen_{January 26, 2009
January
01
Jan
26
26
2009
04:23 AM
4
04
23
AM
PDT}

Jerry Re your post #106. You finished with: This discussion has appeared several time in the last few weeks and the anti ID people think they have scored points by asking for a definition of information when the simplest definition from a dictionary will suffice. This makes a point which runs through much of the discussion. Words such as “information”, “symbol”, “code” and “meaning” are bandied about as though their use was obvious and unambiguous. But they are not. Even the link you provide has multiple different simple definitions of information. I wrote a detailed response to this. It is too long for a reasonable comment so I posted it hereMark Frank_{January 26, 2009
January
01
Jan
26
26
2009
02:23 AM
2
02
23
AM
PDT}

Prof_P.Olofsson, Like everyone from Sweden, i can trace my ancestry back to an Ericsson so I tell my kids they are descendants of Leif Ericsson. I am cool, it has only been above freezing once in the last 10 days where I live. You don't like my sarcasm about the anti ID comments? Some are pretty silly. Did you see my comment at #110 where I try to answer my own question. I am just trying to get some understanding of what all the technical stuff is about. Maybe you can help me.jerry_{January 25, 2009
January
01
Jan
25
25
2009
10:36 PM
10
10
36
PM
PDT}

jerry[97], And you're a descendant of vikings if I remember correctly! To answer your question, no, I am not ready to say that. I see your comment [9] now, I missed it before. I agree. I would also like to know how these results are supposed to be relevant to biology. The only logic I can imagine for the "search for a search" articel is that either the darwinian search algorithm is chosen according to the Kantorovich-Wasserstein probability distribution, or else there is support for ID. I could not argue for such a claim if my life depended on it. For starters, what is it supposed to mean that an organism "searches for a search"? Then, why does it have to search according to the aforementioned probability measure? What does that measure even mean intuitively? I have actually offered some constructive criticism about the math in the "search for a search" paper. As for applications, my point is that if somebody claims that an article is "pro-ID" then I think it is fair to ask how, when there are no pro-ID claims in the actual article. The fact that the authors are pro-ID is not enough for me. So I'm left to guess but I can't come up with anything reasonable that I can even argue against. In that sense, yes, I suppose I am not being constructive but I don't have much to be constructive about. Let your viking blood cool down now and be nice to Mark Frank! He's a very nice person.Prof_P.Olofsson_{January 25, 2009
January
01
Jan
25
25
2009
09:40 PM
9
09
40
PM
PDT}

pubdef et al., In the paper "The origin of biological information and the higher taxonomic categories", Stephen C. Meyer wrote:
Dembski (2002) has used the term “complex specified information” (CSI) as a synonym for “specified complexity” to help distinguish functional biological information from mere Shannon information--that is, specified complexity from mere complexity. This review will use this term as well.
see also
Biological specification always refers to function. An organism is a functional system comprising many functional subsystems. In virtue of their function, these systems embody patterns that are objectively given and can be identified independently of the systems that embody them. Hence these systems are specified in the same sense required by the complexity-specification criterion (see sections 1.3 and 2.5). The specification of organisms can be crashed out in any number of ways. Arno Wouters cashes it out globally in terms of the viability of whole organisms. Michael Behe cashes it out in terms of minimal function of biochemical systems.- Wm. Dembski page 148 of NFL
Joseph_{January 25, 2009
January
01
Jan
25
25
2009
03:11 PM
3
03
11
PM
PDT}

Gpuccio - I will leave the definition of code and symbol and concentrate on simulation. I ask that in a good simulation of it, the simulated NS derive from the fitness fucntion intrinsic in the digital environment where the simulation is run (such as the operating system, and any other software or hardware resource present in the environment), and not from an artificial fitness function introduced by the programmer of the simulation. At last I understand what you want from a simulation although I cannot see why you create this constraint. I don't think Tierra is going to satisfy you. It is hard from the available documentation to understand how Tierra decides which individuals to eliminate but I am willing to bet the programmer didn't just rely on Windows or Linux to do the elimination. There is a reference to code written to eliminate individuals based on their lack of reproductive success. I am guessing this means how many they spawn i.e. fecundity not fitness. But anyway it appears to be written for the purpose. But in any case stop and think. Suppose we do create a simulation where individuals reproduce and die wthout any code written to deliberately kill them off. i.e. it is just the hardware and operating system. The hardware and operating system are designed by someone - well a team of people. They didn't design it to eliminate individuals but it does the job and no doubt by inspecting the code it would be possible to deduce the algorithm it (accidentally) implements for eliminating individuals. Now suppose someone writes code deliberately that implements the same algorithm. They have now written a fitness function. What's the difference? Why do you accept it when the code was a by-product of the OS but not when it was written intentionally for that purpose implementing the identical algorithm?Mark Frank_{January 25, 2009
January
01
Jan
25
25
2009
02:27 PM
2
02
27
PM
PDT}

I have a question since I do not understand the basic argument of this thread. Is the question: Can sexual reproduction amongst members of a population with a specific gene pool find any possible combination of elements in that gene pool? And if so what are the limits for the gene pool? Now there are three sources for alleles or other genetic elements in the gene pool, those already currently in the gene pool in at least one of the members of the population of the gene pool, a mutation on a genomic element of a gamete of a member in the gene pool essentially adding a new element to the gene pool and then there is possible recombination of a DNA sequence during sexual reproduction that can also add something new to the gene pool. Not all combinations of these genomic elements exist in the population but it can be theorized that a currently non existing combination may have some reproductive benefits and would be selected if it showed up. And so the issue is how likely such a combination will show up in a future member of the population. Combinations not available because of the current makeup of the gene pool are just not possible unless there were infusions of new genetic material from some place. So is the question about how easy or how long it will take to produce a combination that has better selection possibilities and secondly whether that combination represents anything significantly new in terms of functional capabilities. In other words is a flying squirrel currently possible within the gene pool of the squirrel population? And if so how long or how difficult would it be to get to the genetic combination that allows a squirrel to fly. Or is the combination of changes in the genome to get there just so implausible or even impossible because no combination of genomic elements in the population could ever lead to it so that naturalistic methods have no chance what soever. And Dembski and Marks are trying to show that whatever combination of fortuitous events took place, it would never reach certain places even if it were theoretically possible. And the term "search" has no meaning because nothing is really searching and the term is just used to describe the results of possible naturalistic processes. Or are we just having intellectual fun with this discussion. I know Darwin just viewed the organism as infinitely malleable and the right combination of environment and luck would lead to almost anything you can imagine. I guess I am trying to answer by own question posed at #9.jerry_{January 25, 2009
January
01
Jan
25
25
2009
02:12 PM
2
02
12
PM
PDT}

As for the information/code and rocks discussion: a code is an abstraction. How is a rock in a stream an abstraction? The positioning of the sun and earth does not encode information in itself. The positioning of ink on a paper does, but the atomic properties of the ink do not inherently encode anything related to the information being outputted. There are real properties involved, which may influence the usage of a code, but the information content is not inherent to these real properties. A series of rocks encoding my name on a plane would be an abstraction that does not rely directly on the properties of the rocks, or streams, or whatever.Patrick_{January 25, 2009
January
01
Jan
25
25
2009
01:41 PM
1
01
41
PM
PDT}

If I breed pigeons I decide which ones survive and on what basis. Suppose I breed pigeons on the basis of speed and the result is a pigeon that has a radically different breast bone structure (I don’t design the breast bone structure - in fact I may not even know it exists). Would this not be an impressive demonstration of Darwinian mechanisms in action? But the selection mechanism (speed) is completely designed.
1. Depends on the information content for this "new" bone structure, does it not? But let's assume it's something radical (not minor) and ONLY Darwinian mechanisms are at work and functional intermediates are always 2-3 steps away. That would at least show that Darwinian mechanisms related to variation are at least up to the task for this one object. 2. As you said the long term goal for the selection mechanism is designed. This gets back to the funneling problem for natural conditions described in #76. Can you think of a hypothetical environment that could produce the same effect in pigeons? This brings to mind my comment on natural selection and squirrels. I could artificially breed squirrels and presumably I could eventually produce a flying squirrel. But what if I could not?Patrick_{January 25, 2009
January
01
Jan
25
25
2009
01:28 PM
1
01
28
PM
PDT}

Mark: # 92: "On the other hand they may depend on some kind of causal relationship other than an arbitary agreement - for example dole queues are a symbol of an economic depression. UCU causes the production of Serine. This is because of biochemistry - not some arbitrary agreement. Therefore it falls into the second category of symbol." No, UCU causes the production of Serine because of an agreement between the code in DNA and the code in the translation system. That agreement is not due to biochemistry, although obviosuly the single steps of the process of recognition (like the coupling of codon and anticodon in the tRNA) follow the laws of biochemistry. You say: "But that is only to say that UCU causes Serine in the context of the transalation system - given the presence of the translation system then there is every necessity that UCU leads to Serine." That's true. But there is no reason of necessity that the same codon (UCU) be connected to Serine both in the information embedded in DNA and in the recognizing codon in tRNA. The relationship between UCU and Serine has no biochemical cause, and is purely symbolical. And that relationship is the same in two completely different parts of the cell. That's all I am saying. For further elaboration on this, please ckeck my answer to Sal Gal at # 104.gpuccio_{January 25, 2009
January
01
Jan
25
25
2009
01:12 PM
1
01
12
PM
PDT}

pubdef, Information gets used in different ways. But we tend to use it in common language as data that mean something. And that is how I believe most use it here. So information is just a piece of data and each nucleotide is a data point. Search the internet for definitions of the word. For example, go here http://www.onelook.com/?w=information&ls=a Use for information such things as news; intelligence; words facts; data; learning; lore Each nucleotide in a DNA string is thus, a piece of information. Each molecule in rock is a piece of information. If one wants to go further down the structure, then be my guess. Each substructure down to the quarks will be pieces of information. Now the DNA and rock are also complex. So each is an example of complex information. However, some units of the information in the DNA specify something else, for example a gene specifies a protein and sometimes RNA. And some of these proteins and RNA have functions and some of these proteins and RNA work together as functional units or systems. So the information in the DNA is complex, specified and and the elements specified are functional. Life is the only place this appears in nature. It appears quite frequently with human intelligence. Now it is quite possible that not all the DNA in a genome is specified and functional and may be just junk but a large part is not. Hence DNA is functionally complex specified information or FCSI. This is what the whole debate is about. This discussion has appeared several time in the last few weeks and the anti ID people think they have scored points by asking for a definition of information when the simplest definition from a dictionary will suffice.jerry_{January 25, 2009
January
01
Jan
25
25
2009
01:06 PM
1
01
06
PM
PDT}

Mark: There is probably still much to say, but I will try to be brief. Please refer also to my other answers to others. #89: Obviously a simulation is not the real thing, because digital entities are not biological entities. But what I ask is that the same logical concept is tested in the simulation, which is believed to act in the real thing. As NS in the darwinian theory acts according to the fitness function of the environment, which is written by nobody, I ask that in a good simulation of it, the simulated NS derive from the fitness fucntion intrinsic in the digital environment where the simulation is run (such as the operating system, and any other software or hardware resource present in the environment), and not from an artificial fitness function introduced by the programmer of the simulation. That is not an absurd or unrealistic request, Indeed, as you can see, JayM has understood it very well, and pointed to an example (Tierra) where my requests seem to be satisfied. Therefore, it is possible to satisfy those requests, and I am convinced that such a premise guarantees a much better simulation of the "concept" of NS. I have no pretense that any digiatl simulation can really model the "substance" of biological reality. You say: "What does an “intrinsic” capacity to survive mean in this context? What is the “true” replicating ability as opposed to any other replicating ability?" It's very simple: I mean the natural capacity of the digital replicators to replicate and survive in the digital environment, not because of some judgement or measurement made on them by an artificial fitness function, but simply because their code can better utilize the resources in the digital environment. You say: "It is almost as if you want the environment and the die/survive mechanism to develop through evolution as well as the individuals that live in that environment." No, that was never my point. The environment is alredy set at the beginning, like any digital environment is. It could change, or not, but not "through evolution", because the environment in natural history is not supposed to change "through evolution". Similarly, the die/survive mechanism is initially set by the programmer, through his programming of digital replicators which can survive and replicate in that environment. That is done by the programmer, because we are not simulating OOL. But any successive variations in survival or death of modified replicators would happen as a consequence of what evolution has done, plus the intrinsic rules of the system. And no, I don't think that artificail selection is "an impressive demonstration of Darwinian mechanisms in action". Your example with pigeons, if it were true, would only demonstrate that a different breast bone structure can arise by RV in pigeons, and that an intelligent observer interested in speed can select it because it helps gain speed. But it tells nothing about NS. To show a complete darwinian mechanism, you should show that the different breast bone structure, after arising by RV, is selected and expanded in the population of pigeons through a reproductive advantage, be it due to speed or to any other associated function. Can you see the difference?gpuccio_{January 25, 2009
January
01
Jan
25
25
2009
01:03 PM
1
01
03
PM
PDT}

Sal Gal (#98 and 99): OK, in the genetic code we observe the decoding. But if something is decodedm it must have been encoded. I am not saying that in itself that proves that it was encoded by a designer. It could have been encoded through the amazing works of RV + NS, as darwinists believe, but encoded it is just the same. I don't want to start here an useless discussion about terms. I just think that you, and pubdef, are charging me of what I have never said. I am saying only two things: a) The genetic code is a symbolic code, in the sense that the information stored in DNA as protein coding genes can only be retrieved by means of a compelx system, the translation system, where exactly the same symbolic correspondences are embedded. And that correspondence is not in any way connected to biochemical laws, but only to a semantic connection between the stored information in DNA and the translating system in tRNAs. b) The information in protein coding genes is functional information, because it is perfectly apt to guide the synthesis of a perfectly functional protein. Please notice that the function is in no way present in the DNA sequence (DNA can never act as an enzyme), but arises only in the final protein, as a consequence of the information of the DNA. The possibility of errors in the process does not change anything of that. And I cannot understand to what you are referring when you speak of my "wonderment at the transformation of sequences of bases into proteins" and to some "argument from improbability" about that which should imply the whole universe and its Creator. I am not wondering at the transformation of anything. I am just saying a very simple thing, that the transformation of a sequence of bases into a protein can never happen because of natural biochemical laws. It requires a very complex system of biochemical machines. In the same way, a newspaper cannot come out of mechanical laws of the universe, if you have not the journalists, the printing machine, and so on. I am very susprised that I must discuss with you about those simple things. What has that to do with the universe, the Creator, and arguments from improbability?. If you think that it is an argument from improbability that a newspaper is the product of journalists and of a printing apparatus, well, I am all for arguments from improbability. The real question for the genetic code, and for the information in DNA, is: Why and how is it there? Why and how is it as it is? I don't think that this is a stupid question, or an "attempt to circumvent absurdity", or that trying to answer it strictly depends on our personal experience of the Creator (although I would probably agree with you that, in a sense, everything does). Darwinists have been trying to answer that question for decades, without directly drawing from their personal experience of the Creator, and I certainly don't blame them for that, even if I don't agree with their answers. ID is just trying to answer it in a different, and IMO much better, way.gpuccio_{January 25, 2009
January
01
Jan
25
25
2009
12:39 PM
12
12
39
PM
PDT}

# 97 pubdef I think you charge yourself unfairly. You asked what is it that makes a rock in a stream not a code, while DNA is a code, other than greater complexity. Gpuccio, being a gentleman, attempted to answer it, and this clarified what he thinks a code is. Seems like it was a good question to ask.Mark Frank_{January 25, 2009
January
01
Jan
25
25
2009
12:17 PM
12
12
17
PM
PDT}

pubdef (#97): "In other words, a physical interaction attains the status of “information” when it can give you “a specific useful information.”" I can't see your problem. The physical interaction of this answer to you post, (all the physical events which cause the bits of my answer to appear on your screen) has the status of information because you can red the words, and understand the meaning (although it does not seem to be particularly useful to you, but that's another story). A protein coding gene is information because it can instruct the translation system about the correct aminoacid sequence. If we cannot agree on such simple concepts, I really don't know what else to say. Again, I am not trying to imply "immediately", with that, that the sequence of nucleotides was in some way written by a designer. That is the conclusion of the whole ID theory, but I don't want in any way to imply that in my concept of code or of information. I am just saying that a protein coding gene contains in its organized sequence, not in its general biochemical structure, the information for a functional protein. That remains true even if that information was produced by unguided mechanisms like RV and NS, as darwinists believe. And the information is stored through a symbolic code. That is simply true. But I am afraid I cannot say it more clearly than I alredy have. You say: "As long as it appears to me that ID relies on how complex something is, I don’t think I’ll be impressed. In my limited experience, evolutionary biologists have been quite aware of how complex nature can be." But the main concept in ID is not complexity, but specification. Complexity is everywhere, but specification is characteristic of designed things. The problem is, some things are designed and specified, but they are simple. In that case, the specification we observe could in alternative be the product of a random process, and we cannot infer design with certainty, even if it was really the cause of what we observe. Those are the famous false negatives, which are part of the ID theory. That's where complexity becomes important. Associated to specification, it becomes the rule to infer design with only virtual (logical) possibility of false positives. But it's specification which is the real characteristic product of design, and not complexity.gpuccio_{January 25, 2009
January
01
Jan
25
25
2009
12:15 PM
12
12
15
PM
PDT}

JayM: I find your last post very balanced, and I don't think that you are confused about my points. In a sense, I think you have understood them very well, and that we agree on much. I will try to clarify what seems to be still not clear. 1) I am not so aware of the current status of GAs, so I apologize for not being able to enter into details. I will use your indications to deepen my knowledge, and maybe we can go further in the discussion. 2) I think it should be clear that I have no objection that well programmed GAs can achieve specific answers. In the same way, I have no objection that, if a GA really does not introduce active information, its results can be considered interesting data, and should be carefully evaluated. I remain of the idea, howevere, that the best simulation of the "concept" of NS is the kind I have suggested, and which seem to be implemented in tierra. In a sense, any artificail introduction of a fitness function could potentially introduce active information, and it could be difficult to demonstarte or understand how in each single case. That's why relying only on the spontaneous properties (or, if you want, fitness function) of an independent digital system seems the best solution. 3) I agree with you that Tierra, at least at first scrutiny, does what I have requested. I am happy with that, but obviously I will try to understand more about it, thanks to your new links. Just give me some time. But I can well accept that Tierra can be the kind of simulation which I have in mind. 4) I am in no way "adding additional constraints about new code and CSI". I have no objection to Tierra, if it really works as I understand. I have only said that its results seem to me, at ny current level of understanding, good simulations of microevolutionary events. I am adding no constraints about CSI: I just think that Tierra has not produced new CSI. I will be more explicit. I think that Tierra has produced some optimization of an existing code, within the range of a random search. That is interesting, and it is ceratinly important data, but it is not CSI. But I don't want to anticipate a discussion about that until I can personally understand the data from Tierra. I am interested essentially in two aspects: a) How much new functional complexity has been generated (that is, how many new functional bits). b) If new functional specifications have emerged (that is, for instance, new algorithms, and not only an optimization of the existing algorithm and code). That was the meaning of my expression "new code", but I understand that it probably was not precise enough. 5) I absolutely agree with you that there is extensive potential for ID research in all that. I am convinced that if ID theorists and darwinist theorists (and anybody else interested) could work together in mutual respect, instead of fighting, and compare their views in the field of active research and theoretical confrontation, much good would ensue for science.gpuccio_{January 25, 2009
January
01
Jan
25
25
2009
11:58 AM
11
11
58
AM
PDT}

gpuccio (and Bill Dembski), Your wonderment at the transformation of sequences of bases into proteins is not an indication of its improbability. You, an entity within the universe in which you and I agree that the process exists, cannot transport the two of us without and demonstrate to me the objective probability of a universe in which the process exists. Arguments from improbability are absurd attempts to circumvent absurdity. (What is the CSI of the semiotic agent in terms of which CSI is defined?) My knowledge of the Creator is experiential, not logical. I can tell other people steps that might lead them to similar experience, but I cannot speak sensibly of the experience itself.
For since in the wisdom of God the world through its wisdom did not know him, God was pleased through the foolishness of what was preached to save those who believe.1 Corinthians 1:21
This verse is not popular with preachers. It never grabbed me until I saw it in an apologetics presentation of Bob Marks. The title of the slide is "Faith before Science." Bob and I have different beliefs, but I greatly admire people who make such statements outright.Sal Gal_{January 25, 2009
January
01
Jan
25
25
2009
11:52 AM
11
11
52
AM
PDT}

gpuccio (and pubdef): In a coding system, as defined by humans, there are both encoder and decoder. We observe cells doing work to frame sequences of discrete objects and transform them into sequences of discrete objects of a fundamentally different kind. This looks like decoding, but with no empirical observation of an encoder, we are not justified in calling it that. Humans recognized the analogy to decoding, and began to perform encoding operations. The fact that there exist genetically engineered organisms decoding human transmissions implies neither that organisms typically decode nor that genetic information is generally due to human-like activity. In fact, if we adopt the perspective that imperfect replication of DNA is erroneous, then algorithmic information usually enters the genome by error.Sal Gal_{January 25, 2009
January
01
Jan
25
25
2009
11:52 AM
11
11
52
AM
PDT}

Prev 1 … 4 5 6 7 8 … 10 Next

You must be logged in to post a comment.

Leave a Reply