Imagine a coin-tossing game. On each turn, players toss a fair coin 500 times. As they do so, they record all runs of heads, so that if they toss H T T H H H T H T T H H H H T T T, they will record: 1, 3, 1, 4, representing the number of heads in each run.
At the end of each round, each player computes the product of their runs-of-heads. The person with the highest product wins.
In addition, there is a House jackpot. Any person whose product exceeds 1060 wins the House jackpot.
There are 2500 possible runs of coin-tosses. However, I’m not sure exactly how many of that vast number of possible series would give a product exceeding 1060. However, if some bright mathematician can work it out for me, we can work out whether a series whose product exceeds 1060 has CSI. My ballpark estimate says it has.
That means, clearly, that if we randomly generate many series of 500 coin-tosses, it is exceedingly unlikely, in the history of the universe, that we will get a product that exceeds 1060.
However, starting with a randomly generated population of, say 100 series, I propose to subject them to random point mutations and natural selection, whereby I will cull the 50 series with the lowest products, and produce “offspring”, with random point mutations from each of the survivors, and repeat this over many generations.
I’ve already reliably got to products exceeding 1058, but it’s possible that I may have got stuck in a local maximum.
However, before I go further: would an ID proponent like to tell me whether, if I succeed in hitting the jackpot, I have satisfactorily refuted Dembski’s case? And would a mathematician like to check the jackpot?
I’ve done it in MatLab, and will post the script below. Sorry I don’t speak anything more geek-friendly than MatLab (well, a little Java, but MatLab is way easier for this).
Can you explain, William? I don’t get it.
A few years ago, when the crowd at UD couldn’t make such a program work, I wrote one for my HP50G graphing calculator just to be contrary. And it has a few extra features as well.
Big programs and fancy platforms are not needed for these demonstrations.
sorry. guilty as charged. feel free to move to WJM’s topic, or the sandbox, or wherever you feel appropriate.
If my wife said something like this I’d say, “Yes, dear. I was wrong.”
Coming up on my 42 anniversary.
Uh, it was that Charles guy, I think.
who considers that the universe itself was designed, ab initio, with the capacity to produce these patterns.,
Count me in the second group. It’s ol’ fashioned Catholic theology.
The current scientific debate about the mechanisms at work in evolution requires theological comment insofar as it sometimes implies a misunderstanding of the nature of divine causality. Many neo-Darwinian scientists, as well as some of their critics, have concluded that, if evolution is a radically contingent materialistic process driven by natural selection and random genetic variation, then there can be no place in it for divine providential causality. A growing body of scientific critics of neo-Darwinism point to evidence of design (e.g., biological structures that exhibit specified complexity) that, in their view, cannot be explained in terms of a purely contingent process and that neo-Darwinians have ignored or misinterpreted. The nub of this currently lively disagreement involves scientific observation and generalization concerning whether the available data support inferences of design or chance, and cannot be settled by theology. But it is important to note that, according to the Catholic understanding of divine causality, true contingency in the created order is not incompatible with a purposeful divine providence. Divine causality and created causality radically differ in kind and not only in degree. Thus, even the outcome of a truly contingent natural process can nonetheless fall within God’s providential plan for creation.
I’ve implemented the following (1+1)-ES (I think it’s called), which means that competition is between just one parent and one offspring per generation. It’s in (hopefully portable) C, with a mutation rate of .004 and tie goes to the offspring. The latter is crucial — it seems that neutral mutations are important for balancing out the runs of heads. For example, if you have an unbalanced sequence like:
THHHTHHHHTHHHHHT
a neutral mutation can get you to
THHHHTHHHTHHHHHT
which can then get you to
THHHHTHHHHTHHHHT
It takes between 1 million and 3 million generations, which should take a minute or less on a normal PC.
Here‘s the code.
Chrome helpfully offers to translate this into English.
Fortunately C is my second language. And my favorite.
I’ve been bit too busy to come up with several examples for which this little program could be a model; but I should point out that my initial comment that the products of the numbers of heads in each partition could represent energy accumulation was not correct. Energy adds; not multiplies. (I knew that.)
But the series of heads and tails could represent a two state system in which each head could represent an energetic state and each tail could represent a ground state. They could be two-state atoms or spin states. If there are, say, N atoms each capable of holding a unit of energy, then the number of accessible states for a total of E units of energy is just the number of combinations of N things taken E at a time or,
Ω = N!/((N – E)! E!).
The entropy, S, is the natural log of Ω multiplied by Planck’s constant, which we can set to 1 for purposes of illustration (its value is an accident of the independent histories of the temperature and energy scales anyway).
The reciprocal of the temperature is the rate of change of entropy with respect to the corresponding change in energy. So if we take N to be large, and each unit of energy to be 1, then the change in entropy with respect to energy, or reciprocal of the temperature, is just the discrete derivative of S with respect to E,
1/T = ln((N-E)/(E+1)),
where E goes from 0 to N-1 units.
For those interested, try looking at the temperature as a function of E. You may find it interesting.
Also note that the entropy goes from 0 when all atoms are in the ground state, through a maximum when half the atoms are in an excited state, and back to zero again when all atoms are in the excited state. This is while energy is being added. And it behaves the same way when energy is drained out of the system.
So this little system that Elizabeth is using could be something like an array of two state atoms embedded in a crystalline lattice in which every 5th atom is attached to another atom or molecule that inhibits the transition of that two-state atom to an excited state.
However, that product of the numbers of heads (numbers of excitations) would have to be a feature of interest (“information?”) to someone not particularly concerned with the physics.
I am utterly lost as to what the point of this exercise is, or how it relates to Dembski’s jabberwockies. I do understand that we have here a point mutation exercise coupled with animal husbandry and herd management from a vindictive deity that slays all the ‘wrong’ children. Without fail, and at every harvest season. And that we’re trying to get over 1 in a chi squared test of… no idea how many independent variables.
Which shouldn’t seem to matter much given that evolutionary processes are defined as being non-independent tests.
So here’s the question: How much of the hash that headlines this is from Dembski’s examples and how much is just a misunderstanding of GAs and their relation to biology?
madbat089,
3) maximizes the product of H-runs = fitness function
4) liz’s coin toss string = maximizes the product of H-runs
// 3 and 4 are both true if, liz’s coin toss string = fitness function. The pattern of {H,T} is not the fitness function (it is the result), so 3 is accurate not 4.
For 2) we can recover the sequence by referring to the function, and identify the function by referring to the sequence, but we can’t recover liz’s sequence with, “maximizes the product of H”. The only specifications are {1 & 2}. {3} is an accurate description, but it is not specified, as there are many possible fitness functions that will maximize H. 4) is simply false.
Did you understand anything I wrote? (2) and (3) are in no sense equivalent, whereas (2) and (4) are equivalent, where the description after *=* is a property of the entity described in front of *=*. Furthermore, if *=* is taken to mean *results in* (which is a ready and widespread meaning of *=*), both (2) and (4) have meaning, are true, and are equivalent.
Woah. False. We cannot assign a DNA sequence by knowing the function of the corresponding protein. Neither can we identify the function of a string of DNA by knowing its sequence. Scientists really really wish they could….
Sorry, I don’t understand what you are saying here. *maximizes the product of H-runs* is the fitness-relevant trait. The trait is specified, according to Dembski, when only a tiny subset of your possible population of strings exhibits this function. That is given for this trait. That’s why Liz chose it.
Also, there is an unknown number of possible DNA sequences that code for the function *interacts with the glucocorticoid receptor*.
madbat089,
Hand this list to a scientist, and ask them to recover the sequence:
1) 10 1’s
2) regulate filament length mice
3) maximizes the product of H-runs
[A]
1) 1111111111
2) KYTSPVDMLGVVLAKKC QALVSDADYRNYLHQW TCLPDQNDVIQAK KVYELQSENMYKSD
3) …
Can you show any evidence that verifies that ‘the designer’ actually did “design the correct genetic programming/ genetic algorithm and the software does the rest”? What happened to side loading/intervention?
“No intervention required if the right program was written. But then the question would be- can such a program be written?”
Since you and other IDists claim that a program was written, it’s up to you to verify it.
“And BTW, right now we are saying that materialism’s science claims are meaningless.”
No matter how much you say that, you won’t get anywhere with science unless you show convincing, positive, testable evidence of whichever version of immaterial ID you’re promoting.
Are you saying that the universe, or living things, are computers?
Specification: the Pattern that Signifies Intelligence, in which he gives a formula for computing a value “chi” from a pattern, that if the chi of a pattern exceeds unity, then we must infer “Intelligence”, not “chance”. My point is that Natural Selection is also a candidate generator of such a pattern, and that Dembski’s chi>1 indicates is some kind of selection process, which could include “intelligence” but which also includes, for example, natural selection.
Dembski’s “chi” is not the “chi” of “chi square”.
You could regard my exercises as an analogy for of “animal husbandry” in which a farmer only breeds from the virtual organisms with the largest product, just as children have sunflower growing competitions and only plant seeds from the largest sunflowers. Or, as an analogy for a vindictive deity. But Darwin’s point was that “selection” also occures “naturally” – i.e. it does not require an intentional culler. The environment itself will do the job. So, equally, you can regard my exercise as an analogy for a population of critters trying to thrive on limited resources, in which those best at accessing those resources will be the ones most likely to raise viable offspring. The point is that there is no difference between the selection mechanisms in the three cases – only the intention differs, and the intention merely affects the criterion used to select, not the fact of selection itself.
Could you explain what you mean by this?
I don’t know what “the hash that headlines this” means. In that piece by Dembski he does not refer to biology explicitly, but attempts to make a general probabilistic claim about the origin of patterns. My point is that, given a system of self-replicators that replicate with variance, and some criterion that means that some variance have a better chance of replicating than others, the patterns that emerge will reliable exhibit chi>1.
That does not, of course, refute the the claim that only an Intelligent Agent could have produced self-replicators, but that is not the claim Dembski is making. Elsewhere he specifically claims that Darwinian mechanisms (for whic self-replication with variance is a prerequisite) cannot generate Specified Complexity in which chi>1. I have shown that starting with patterns in which chi is well below 1 (by definition, as my starting patterns are indeed randomly generated by a virtual coin-tosser), Natural selection can reliably lift them to chi>1.
Now, Dembski is aware of this kind of rebuttal, as it has been made many times. His response is not (that I am aware of) that we are assuming self-replication which itself requires intelligence, but that all selection systems involve the “smuggling in” of information in the form of a fitness function.
My response is that no “smuggling” is required. As I say above, sure, a fitness function can be provided by an intelligent agent, me, in this case, a sunflower-growing child in another. But an intelligent agent is not necessary – limited resources will do the trick.
Now, ID proponents might argue that an Intelligent agent was required to create a world with limited resources. Fine. But that wouldn’t distinguish ID from TE, and many ID proponents, including Dembski, are extremely opposed to TE!
And in any case, it is not true that I have provided the solution to the problem. I have not “smuggled in” the optimum answer to the “problem” I set my virtual organism. I merely gave them the problem. The evolving population has to find a tiny subset of optimal answers. Interestingly, it has difficulty finding the optimum answer, because it turns out that, given only point mutations, that optimum answer is quite “IC”. In other words, the fitness space is really quite unconnected at the top end – the tiny subset of answers in which chi>1 are quite widely separated in “search space”.
No matter. The population can find answers within it, given Natural Selection. Without Natural selection, my program could run on until the End Times and never get there.
Then why do you rely on what science provides in your everyday life?
He’s desperately trying to ignore your demonstration and tell himself you made it up out of whole clothe
Would it not be more apposite to ask: How is it that science can provide what we rely on in our everyday life?
That’s not a problem, Mike. Take the logarithm of the product of the number of heads in each run and you’ve got an additive energy. A head run with n heads adds energy −log(n). We can use base-10 logs for an easier comparison with Elizabeth’s original definition.
States with lowest energy have lone tails separated by head runs. The lowest energy states found by DiEb (4 3-runs and 97 4-runs) have E = −4 log 3 −97 log 4 = −60.3. Their entropy is S = ln[101!/(4!97!)] = 15.2.
One can do calculations along these lines to estimate the temperature in the target space.
Let’s stick with DiEb‘s sequences containing mostly 4-runs and a few 3-runs. These dominate the states of lowest energy.
14 3-runs and 89 4-runs:
E = −14 log 3 −89 log 4 = −60.263.
S = ln[103!/(89!14!)] = 38.77.
19 3-runs and 85 4-runs:
E = −14 log 3 −89 log 4 = −60.240.
S = ln[104!/(85!19!)] = 47.15
So replacing 4 4-runs with 5 3-runs increases energy by ΔE = 4 log 4 – 5 log 3 = 0.0023 and entropy by ΔS = ln[104(89!/85!)(15!/19!)] = 8.38. From these we obtain the temperature:
T = ΔE/ΔS = 2.7×10^{−4}.
Single-coin mutations are going to be rather ineffective at this temperature. For instance, when a 5-run is converted into two 2-runs (by changing the middle H into a T), the energy rises by log 5 − 2 log 2 = 0.097. The probability of acceptance of this move is exp(−ΔE/T) = exp(−359) = 10^{−156}.
Why don’t you, guys, move to WJM’s thread to discuss worldview questions?
Yes, I’ll do some moving. Hold on to your hats.
What does the topic of this thread have to do with anything ID claims?
No, what is important is whether I have in fact proven Dembski wrong.
This is verifiable independently of what I believe, because it is a matter of logic, something you yourself regard as a valid arbiter.
If you do not consider I have proven Dembski wrong, can you please point to the flaw in my logic. I you think that all that matters is what we happen to believe, then I disagree 🙂 I don’t think postmodernism applies to science or math.
Well perhaps you should take it up with Dembski- and I have exposed the flaw in your logic.
I have also tried to get others involved but they already understand that trying to correct you is not worth the effort.
So all I have left to say is you need to take it up with Dembski. I know if I was as sure as you are I wouldn’t hesitate.
One more thing:
“I beseech you, in the bowels of Christ, think it possible that you may be mistaken.”
Which may be meaningless as “Christ” doesn’t mean a thing to either one of us… 😉
It’s a quotation from Cromwell, named “Cromwell’s Rule”by the statistician Dennis Lindlay: “avoid using prior probabilities of 0 or 1, except when applied to statements that are logically true or false.”
LoL! I know Cromwell- I was just trying to say that YOU need to heed the quote wrt CSI…
I am sorry. I am a scientist, and I don’t even understand the meaning of your new #2?
In a few short sentences, expose the flaw in the logic.
I’ll then try if I can, to explain to you why you’re wrong.
Indeed. And so does Dembski. But I have provided my math, and my script, and you can check it.
junkdnaforlife:
We are clearly talking past each other. Let’s try this from a different angle:
According to Dembski, we cannot conclude that just because a pattern is one of a vast number of possible patterns that it was designed; to conclude design, it has to be one of a small subset of those patterns that conforms to some kind of *specification*. As a vast pool of possible patterns, Dembski chooses strings of coin tosses. As a specification that will only be a property of a very small subset, he chooses *compressibility* (ease of description). If, according to the relationship between the number of patterns exhibiting the specification and the number of possible patterns in the pool, the number of opportunities for them to happen at least once in the history of the universe is too low (=falls below Dembski’s probability threshold), he claims that we can reject non-design.
In other words, Dembski chose *compressibility* because it is one of the properties that coin toss strings can have that conform to his criterion for specification.
Lizzie uses, equivalent to Dembski, a string of coin tosses. She then chooses a property (*product of H-runs above a specific threshold*) that only a very small subset of the vast number of possible patterns exhibits. By Dembski’s criterion for a specification (according to the relationship between the number of patterns exhibiting the specification and the number of possible patterns in the pool, the number of opportunities for them to happen falls below Dembski’s threshold), that’s a specification. Thus, Lizzie’s simulation is valid, adequate test of Dembski’s claim.
The other interpretation of the silence of the “others” is, of course, that they can see that Elizabeth is right! – as she transparently is.There is no flaw in her logic.
Not that they’ll ever admit it, even to themselves…
What is strange is Dembski’s claim that Darwinian processes are inadequate to account for CSI.
If he was claiming that Darwinian processes must be initiated by an intelligent agent, that would be a different argument (readily rebutted, but my demonstration would not be a rebuttal).
So here, he actually gives, as the specification the “evolutionary pathway that brings about the pattern”: “a bacterial flagellum”. Mine is defined as: “a sequence of coin-tosses in which the product of the runs-of-heads is very large”. And we’ve calculated the search space for such sequences above a certain threshold, and confirmed that the “probabilistic resources”. I don’t know why he thinks that his H “takes into account Darwinian and other material mechanisms”, but he does. However, it is clear, from my example, that “Darwinian and other material mechanisms” are capable of hitting the target pattern. And it is also clear, from the fitness plots, that some of the intermediate patterns in the evolutionary pathway in my example are “IC” – they are not reached by consistently advantageous steps and frequently involve deleterious steps. The AVIDA examples show the same thing.
Dembski is simply wrong in his claim. His only defense, that I am aware of, is this story about “smuggling in” information via the fitness function.
If any ID proponents would like to discuss this, I’d be delighted to hear from them 🙂
This reminds me of an old joke in which God tells scientists “No, no, no. You go get your own dirt!”
Seriously, Dembski’s argument that information is smuggled through the fitness function is trivially true. The information content cannot increase if the fitness landscape is entirely flat (turning an evolutionary search into a purely random walk) or extremely rugged (algorithm trapped at a local minimum). So evolutionary theory begins with an assumption that fitness landscapes are neither too flat nor too rugged. At that point, Dembski says, roughly, “OK, but where did the smooth landscape come from?” This, in a nutshell, is what makes ID unscientific.
Like any scientific theory, evolutionary biology is concerned with a narrow set of questions. It does not have a goal of finding the ultimate answer to life, the Universe, and everything. Creationism, on the other hand, is interested in finding God in the gaps of scientific knowledge. Once you explain how CSI emerges through mutation and natural selection, creationists move the goal posts to the origin of fitness landscapes. They could not care less about CSI. Ho hum.
Liz,
I really like the product-of-runs-of-heads problem as an easily understood example of optimization. Did you come up with it yourself, or did you learn it from somewhere?
I came up with it myself 🙂 Someone else may have come up with it independently though, I don’t know. I was trying to think of something like DNA, but with two “nucleotides” rather than four (to be more analogous to coin-tossing, and thus directly comparable to Dembski’s coin-toss examples) and in which the sequence would “code” for something, as a sequence of codons does in DNA. And which was numeric so it could be easily implemented!
I thought of it on my bike. I get all my best ideas on my bike. It’s one of the reasons I cycle to work (it’s also a lovely route – along a riverbank, by a wooded escarpment full of interesting wildlife).
This is not too far from another line I was thinking about, namely a set of N Einstein oscillators.
In this case Ω = (N-1+E)!/((N-1)!E!)).
We get this because, for N oscillators, there are N-1 partitions distributed among E units of energy so we want the number of combinations of N-1+E things taken E at a time.
This is an interesting contrast to the two-state system because
ΔS/ΔE = ln((N+E)/(E+1)), where E goes from 0 to infinity.
i.e., with a plus sign in the numerator instead of a minus.
However this can’t apply to Elizabeth’s model because the number of oscillators is changing depending on how the energies are distributed and how many partitions are adjacent. The reason for this is because of the fixed number of heads and tales. That is a weird constraint.
The counting is a little funnier because we arbitrarily don’t multiply the zero-energy states along with the nonzero-energy states. And of course, the logarithm of a zero-energy state is negative infinity. I suppose we could count a single head as a zero energy state (log(1) = 0), but zero heads is a tail, and for every additional adjacent tail, we loose an oscillator. And we have to find all combinations consistent with a given total energy. So the counting gets pretty messy.
At first I was thinking about a set of “four-state” oscillators, but every string of tails just eliminates oscillators. So with the moving around of partitions, and with many adjacent partitions, not only do the number of oscillators keep changing, but for any given energy, we have to find all possible combinations of states consistent with that energy. And we do that for a changing number of oscillators as well as counting all combinations that produce the same total energy. And we have to allow for the fact that the energy saturates at 4 for any particular oscillator.
That was the part I have not had time to think about (I’m laid up with a nasty virus, had to cancel a cross-country trip, and I have a backlog of tasks piling up).
I don’t think taking samples from DiEb’s runs is going to give us the actual number of combinations of energy state for a given total energy. That has to be calculated.
It is not a usual kind of system, but it is certainly not unusual for a system to have microscopic states become available and unavailable as conditions change. It just makes things harder to calculate analytically.
No worries, Mike. Statistical mechanics is sufficiently versatile to cope with constraints. Working with a microcanonical ensemble (constraint of fixed energy) is a pain in the butt, so we switch to a canonical one. Working with a fixed number of particles is also difficult, so we go around that constraint by switching to the grand canonical ensemble and work at a fixed chemical potential instead of a fixed particle number.
Similar tricks can be used in this problem. The sequence-length constraint can be handled by adding a “tension” term to the energy. The energy of an n-run changes from −log n to −log n + σ(n+1), where σ is a Lagrange multiplier taking care of the length constraint. I’ll see if this program can be pushed all the way. Back-of-the-envelope calculations show that it might work.
I think DiEb’s estimate is right. 3-runs are the lowest-energy excitations. Here is a quick table of energy costs per excitation, with the length constraint taken into account:
2-runs: −log 2 + (3/5) log 4 = 0.0602
3-runs: −log 3 + (4/5) log 4 = 0.0045.
5-runs: −log 5 + (6/5) log 4 = 0.0235.
6-runs: −log 5 + (6/5) log 4 = 0.0647.
3-runs are by far the cheapest and are the only ones at low temperatures. We might have to take into account 5-runs, but they are few and far between. With the ground-state energy of −60.308, you can add at most 0.308 to it before crossing the threshold of −60. That energy window allows for a max number of 68 3-runs and only 13 5-runs. 2 and 6-runs are even more costly and you can add at most 5 of those.
To get a rough estimate, we can say that any combination of 3 and 4-runs is admissible. There are between 100 and 114 runs in total (depending on the exact number of 3 and 4-runs). There are roughly 2^114 = 2.1×10^34 such states, not far from DiEb’s number.
Although the numerical values are fine, there is a typo in the last line. Should be
2-runs: −log 2 + (3/5) log 4 = 0.0602
3-runs: −log 3 + (4/5) log 4 = 0.0045.
5-runs: −log 5 + (6/5) log 4 = 0.0235.
6-runs: −log 6 + (7/5) log 4 = 0.0647.
(Forgot to edit a copied and pasted expression.)
Elizabeth:
So yes, if by ‘natural selection’ you mean ‘unnatural selection’ then by all means everything is as you say, because you say it. Which is all a bit Humpty Dumpty, but I’ll join it by saying that you’re proven your point. So equally, we can all regard that you did not.
But if saying anything is to include and mean its opposite as well… Then we’re right back to the notion of non-Contradiction that brought me here in the first place and we’ve made absolutely no progress in the matter. So topically, on point, and with the preferred argumentation presented to me: You failed to refute Dembski. Because you say you did.
Or we can try working with actual definitions rather than ever-shifting sands of sophistry:
1. What is ‘unnatural selection’? If your toy-Darwin suffices then my cigarette lighter is natural climate variation.
2. What is Dembski’s ‘chi’? It is not ‘chi squared’ but almost surely (It’s Dembski) a renamed common concept.
No, what Elizabeth offered you was a choice of “analogies”.
An analogy is not real. It is an alternate “description” that is generated by a speaker to illustrate the “description” of something unrelated but similar, in the hopes that it will aid in understanding a point made by the speaker.
In your case, the misunderstanding seems to go deeper as the popular analogy, “The glass is half-empty / The glass is half-full” are not opposites.
Elizabeth said: “You could regard my exercises as an analogy for … a vindictive deity. … So, equally, you can regard my exercise as an analogy for a population of critters trying to thrive on limited resources, in which those best at accessing those resources will be the ones most likely to raise viable offspring.
Analogy: “resemblance in some particulars between things otherwise unlike”
– Merriam-Webster
It seems you fundamentally misunderstood Elizabeth when you thought her saying *x is a good analogy for y, but also for z, because all three of them result in the reproduction of only a certain proportion of the population with a certain distribution of traits* was her saying *y means x, and z means x, too*. In fact, I’ve never even heard of a definition of analogy that would support that kind of interpretation. How did you come up with that?
Oh I think it shouldn’t be an excruciatingly difficult problem. There might be a partition function that would work. I’m just too wiped out to think about it right now. All I want to do is sleep.
Give it a shot. I think we’re on the same page.
No, “natural selection” doesn’t mean “unnatural selection” but both mean selection. Artificial selection is what we call it when the fitness criteria are an “artifice” – ie. decided on by an intelligent agent. Natural selection is when the fitness criteria are simply those implicit in the hazards and resources of the natural world.
But as far as my exercise goes, it doesn’t matter, because the actual mechanism is the same in both cases.
No, you have misunderstood. A black cat is not the same as a white cat, but if what we are interested in is their cattiness, not their colour, then it doesn’t matter whether we talk about a black one or a white one. What we have in my exercise is a systematic fitness criterion. The fact that it was invented by me doesn’t mean it isn’t analogous to a naturally presented one, for example limited resources.
That’s exactly what we are doing, and I have defined what I’m doing with great precision. So has Dembski.
My model will serve as an analog for either natural or unnatural selection. Darwin’s insight was to see that both worked in exactly the same way – some environmental criterion determines what features in an organism will promote reproductive success. If that “environmental criterion” is a child who wants to win the tallest sunflower contest, then we call it “artifical selection”; if it is solar gain, then we call it “natural selection”. But the mechanism in both cases – and the result in this case – is the same – selection for taller flowers.
Well, read the paper. He gives the equation and explains its derivation quite clearly. I have used his definition, and plugged in relevant values from my exercise.
It turns out that the number of generations required is non-linearly dependent on the population size. With a population of 1000, I only got to 9.2x10e59 after 300,000 generations. Bumping the population size up to 10,000 reliably generates a solution over 10e60 in around 700 generations. My code and results are available here.
I do hope the ID proponents appreciate that crossover (SEX!) generates CSI even faster than asexual evolutionary mechanisms.
And not just CSI either!
… or so I’m told….
Patrick,
Patrick you say on your site:
“That being said, based on Dr. Liddle’s interpretation of Dembski’s paper, this exercise clearly demonstrates that a simple subset of known evolutionary mechanisms is more than capable of generating sequences of bits that exhibit what he seems to be claiming is CSI.”
// Based on Dr. Liddle’s interpretation indeed. You provide a string, I’ve taken the first two lines of it and included them in the following three examples to examine specificity. (However, clipping a fragment may not be representative of the specificity of a string, as we could toss a coin a billion times and find some clusters of patterns, which if clipped out and isolated would not be representative.) But for the purpose of this, the first two lines of your example seemed to be a fair representation of the string you posted.
(1) 3 1’s then 3 0’s repeat to 96
111000 111000 111000 111000 111000 111000 111000 111000 111000 111000 111000 111000 111000 111000 111000 111000
(2) Regulates thin filament length in mice
KYTSPVDMLG VVLAKKCQAL VSDADYRNYL HQWTCLPDQN DVIQAKKVYE LQSENMYKSD LEWLRGIGWS PLGSLEAEKN
KRASEIISEK KYRQPPDRNK FTSIPDAMDI VLAKTNAKNR
(3) 4 1’s then 0, 3 1’s then 0 twice, 4 1’s then 0 twice, five 1’s, 0 then 4 1’s three times, 0 then 5 1’s, 0 then 3 1’s three times, 0 then four 1’s twice, 0 then five 1’s, 0 then 4 1’s five times
11110 11101110 1111011110 11111 011110111101111 011111 011101110111 0111101111 011111 0111101111011110111101111
// I will leave it for the reader to decide which string is Patricks’s, as well as which would be considered examples of analogous specificity.
// 2 is a fragment of a string length of 1358, and if we wish to complete the string the description will not change. 3 is also a fragment, however if we complete the string the description will increase.
Would you care to comment on this post of mine? Why do you think Lizzie’s string is not specified in Dembksi’s sense of specification?