# Gpuccio’s Theory of Intelligent Design

Gpuccio has made a series of comments at Uncommon Descent and I thought they could form the basis of an opening post. The comments following were copied and pasted from Gpuccio’s comments starting here

To onlooker and to all those who have followed thi discussion:

I will try to express again the procedure to evaluate dFSCI and infer design, referring specifically to Lizzies “experiment”. I will try also to clarify, while I do that, some side aspects that are probably not obvious to all.

Moreover, I will do that a step at a time, in as many posts as nevessary.

Creating CSI with NS
Posted on March 14, 2012 by Elizabeth
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)

Now, some premises:

a) dFSI is a very clear concept, but it can be expressed in two different ways: as a numeric value (the ratio between target space and search space, expressed in bits a la Shannon; let’s call that simply dFSI; or as a cathegorical value (present or absent), derived by comparing the value obtained that way with some pre define threshold; let’s call that simply dFSCI. I will be specially careful to use the correct acronyms in the following discussion, to avoid confusion.

b) To be able to discuss Lizzie’s example, let’s suppose that we know the ratio of the target space to the search space in this case, and let’s say that the ratio is 2^-180, and therefore the functional complexity for the string as it is would be 180 bits.

c) Let’s say that an algorithm exists that can compute a string whose product exceeds 10^60 in a reasonable time.

If these premises are clear, we can go on.

Now, a very important point. To go on with a realistic process of design inference based on the concept of functionally specified information, we need a few things clearly definied in any particulare example:

1) The System

This is very important. We must clearly define the system for which we are making the evaluation. There are different kinds of systems. The whole universe. Our planet. A lb flask. They are different, and we must tailor our reasoning to the system we are considering.

For Lizzie’s experiment, I propose to define the system as a computer or informational system of any kind that can produce random 500 bits strings at a certain rate. For the experiment to be valid to test a design inference, some further properties are needes:

1a) The starting system must be completely “blind” to the specific experiment we will make. IOWs, we must be sure that no added information is present in the system in relation to the specific experiment. That is easily realized by having the system assembled by someone who does not know what kind of experiment we are going to make. IOWs, the programmer of the informational system just needs to know that we need random 500 bits string, but he must be completely blind to why we need them. So, we are sure that the system generates truly random outputs.

1b) Obviously, an operator must be able to interact with the system, and must be able to do two different things:

- To input his personal solution, derived from his presonal intelligent computations, so that it appears to us observers exactly like any other string randomly generated by the system.

- To input in the system any string that works as an executable program, whose existence will not be known to us observers.

OK?

2) The Time Span:

That is very important too. There are different Time Spans in different contexts. The whole life of the universe. The life of our planet. The years in Lenski’s experiment.

I will define the Time Span very simply, as the time from Time 0, which is when the System comes into existence, to Time X, which is the time at which we observe for the first time the candidate designed object.

For Lizzie’s experiment, it is the time from Time 0 when the specific informational system is assembled, or started, to time X, when it outputs a valid solution. Let’s say, for instance, that it is 10 days.

OK?

3) The specified function

That is easy. It can be any function objectively defined, and objectively assessable in a digital string. For Lizzies, experiment, the specified function will be:

Any string of 500 bits where the product calculated as described exceeds 10^60

OK?

4) The target space / search space ratio, expressed in bits a la Shannon. Here, the search space is 500 bits. I have no idea how big the target space is, and apparently neither does Elizabeth. But we both have faith that a good mathemathician can compute it. In the meantime, I am assuming, just for discussion, that the target space if 320 bits big, so that the ratio is 180 bits, as proposed in the premises.

Be careful: this is not yet the final dFSI for the observed string, but it is a first evaluation of its higher threshold. Indeed, a purely random System can generate such a specified string with a probability of 1:2^180. Other considerations can certainly lower that value, but not increase it. IOWs, a string with that specification cannot have more than 180 bits of functional complexity.

OK?

We must observe, in the System, an Object at time X that was not present, at least in its present arrangement, at time 0.

The Observed Object must comply with the Specified Function. In our experiment, it will be a string with the defined property, that is outputted by the System at time X.

Therefore, we have already assessed that the Observed Object is specified for the function we defined.

OK?

6) The Appropiate Threshold

That is necesary to transorm our numeric measure of dFSI into a cathegorical value (present / absent) of dFSCI.

In what sense the threshold has to be “appropriate”? That will be clear, if we consider the purpose of dFSCI, which is to reject the null hypothesis if a random generation of the Oberved Object in the System.

As a preliminary, we have to evaluate the Probabilistic Resources of the system, which can be easily defined as the number of random states generated by the System in the Time Span. So, if our System generates 10^20 randoms trings per day, in 10 days it will generate 10^21 random strings, that is about 70 bits.

The Threshold, to be appropiate, must be of many orders of magnitude higher than the probabilistic resources of the System, so that the null hypothesis may be safely rejected. In this particular case, let’s go on with a threshold of 150 bits, certainly too big, just to be on the safe side.

7) The evaluation of known deterministic explanations

That is where most people (on the other side, at TSZ) seem to become “confused”.

First of all, let’s clarify that we have the duty to evaluate any possible deterministic mechanism that is known or proposed.

As a first hypothesis, let’s consider the case in which the mechanism is part of the System, from the start. IOWs the mechanism must be in the System at time 0. If it comes into existence after that time because of the deterministic evolution of the system itself, then we can treat the whole process as a deterministic mechanism present in the System at time 0, and nothing changes.

I will treat separately the case where the mechanism appears in the system as a random result in the System itself.

Now, first of all, have we any reason here to think that a deterministic explanation of the Observed Object can exist? Yes, we have indeed, because the nature itself of the specified function is mathemathical and algorithmic (the product of the sequences of heads must exceed 10^60). That is exactly the kind of result that can usually be obtained by a deterministic computation.

But, as we said, our System at time 0 was completely blind to the specific problem and definition posed by Lizzie. Therefore, we can be safely certain that the system in itself contains not special algorithm to compute that specific solution. Arguing that the solution could be generated by the basic laws physics is not a valid alternative (I know, some darwinist at TSZ will probably argue exactly that, but out of respect for my intelligence I will not discuss that possibility).

So, we can more than reasonably exclude a deterministic explanation of that kind for our Observed Object in our System.

7) The evaluation of known deterministic explanations (part two)

But there is another possibility that we have the duty to evaluate. What if a very simple algorithm arose in the System by random variation)? What if that very simple algorithm can output the correct solution deterministically?

That is a possibility, although a very ulikely one. So, let’s consider it.

First of all, let’s find some real algorithm that can compute a solution in reasonable time (let’s say less than the Time Span).

I don’t know if such an algorithm exists. Im my premise c) at post #682 I assumed that it exists. Therefore, let’s imagine that we have the algorithm, and that we have done our best to ensure that it is the simplest algorithm that can do the job (it is not important to prove that mathemathically: it’s enough that it is the best result of the work of all our mathemathician friends or enemies; IOWs, the best empirically known algorithm at present).

Now we have the algorithm, and the algorithm must obviously be in the form of a string of bits that, if present in the System, wil compute the solution. IOWs, it must be the string corresponding to an executable program appropriate for the System, and that does the job.

We can obviously compute the dFSI for that string. Why do we do that?

It’s simple. We have now two different scanrios where the Observed Object could have been generated by RV:

7a) The Observed Object was generated by the random variation in the System directly.

7b) The Observed Object was computed deterministically by the algorithm, which was generated by the random variation in the System.

We have no idea of which of the two is true, just as we have no idea if the string was designed. But we can compute probabilities.

So, we compute the dFSI of the algorithm string. Now there are two possibilities:

- The dFSI for the algorithm string is higher than the tentative dFSI we already computed for the solution string (higher than 180 bits). That is by far the most likely scenarion, probably the only possible one. In this case, the tentative value of dFSI for the solution string, 180 bits, is also the final dFSI for it. As our threshold is 150 bits, we infer design for the string.

- The dFSI for the algorithm string is lower than the tentative dFSI we already computed for the solution string (lower than 180 bits). There are again two possibilities. If it is however higher than 150 bits, we infer design just the same. If it is lower than 150 bits, we state that it is not possible to infer design for the solution string.

Why? Because a purely random pathway exists (through the random generation of the algorithm) that will lead deterministically to the generation of the solution string, with a total probability of the whole process which is higher than our threshold (lower than 150 bits).

OK?

8) Final considerations

So, some simple answers to possible questions:

8a) Was the string designed?

A: We infer design for it, or we infer it not. In science, we never know the final truth.

8b) What if the operator inputted the string directly?

A: Then the string is designed by definition (a conscious intelligent being produced it). If we inferred design, our inference is a true positive. If we did not infer design, our inference is a false negative.

8c) What if the operator inputted the algorithm string, and not the solution string?

A: Nothing changes. The string is designed however, because it is the result of the input of a conscious intelligetn operator, although an indirect input. Again, if we inferred design, our inference is a true positive. If we did not infer design, our inference is a false negative. IOWs, our inference is completely independent from how the designer designed the string (directly or indirectly)

8d: What if we do not realize that an algorithm exists, and the algorithm exists and is less complex than the string, and less complex than the threshold?

A: As alreday said, we would infer design, at least until we are made aware of the existence of such an algorithm. If the string really originated randomly througha random emergence of the algorithm, that would be a false positive.

But, for that to really happen, many things must become true, and not only “possible”:

a) We must not recognize the obvious algorithmic nature of that particular specified function.

b) An algorithm must really exist that computes the solution and that, when expressed as an executable program for the System, has a complexity lower than 150 bits.

I an absolutely confident that such a scenario can never be real, ans so I believe that our empirical specificity of 100% will be always confirmed.

Anyways, the moment that anyone shows tha algorithm with those properties, the deign inference for that Object is falsified, and we have to assert that we cannot infer design for it. This new assertion can be either a false negative or a true negative, depending on wheterh the solution string was really designed (directly or indirectly) or not (randomly generated).

That’s all, for the moment.

AF adds “This was done in haste. Any comments regarding errors and ommissions will be appreciated.”

## 263 thoughts on “Gpuccio’s Theory of Intelligent Design”

1. Crocodiles managed to find their way to isolated islands in the Pacific. One should be careful with one’s choice of metaphor. In the duplicated gene scenario, the original unmodified gene provides the floatation while the copy drifts. And selection isn’t specific in its demand for new function. It doesn’t have targets.

2. gpuccio: “4) The ability to positively select and expand each new string according to the calculated product if, and only if, the product is higher than 10^60.

Let’s remember that, apparently, the probability to get a string whose product is higher than 10^60 is 1:2^394 (information kindly provided by olegt). “

Why don’t you just replace Elizabeth’s whole simulation with a (2^394) sided die?

3. those with higher fitness are positively selected and those with lower are negatively selected.

Just to clarify this, as I can see it being misrepresented, more conventionally, these would be termed beneficial and deleterious.

Positive and negative are frequently used in relation to the causal implementation of the fitness differential – ‘positive selection’ may be seen as a preservative cause of selection upon a particular allele, ‘negative’ as an eliminatory one. But they are really inseperable, the yin and yang of a fitness differential. You can’t do one without doing the other, because ultimately what counts is the effect on allele frequency – the proportions in the population.  If I decide to keep the 60% of my record collection I like best, or throw away the 40% I like least, it amounts to the same thing – a ‘population’ enriched in the qualities preserved and/or removed by this ‘selective agent’ (who happens to be capable of making an active choice, but that’s by no means an essential qualification). Similarly, a bird missing moths due to camouflage would have exactly the same effect as one that can see them all clearly but just ‘prefers’ more brightly coloured ones.

At any point in an evolutionary series, the terms ‘positive’ and ‘negative’, as I used them above, relate to the gradient caused by present differential fitness, not the causal reason there is a differential at all, which may involve either preservation of the ‘favoured’ fraction’ or elimination of the ‘disfavoured’, and amount to the same thing.

4. gpuccio: Not only the Weasel phrase, but the whole text of Hamlet, if that text is alredy written in the algorithm. Is that new dFSCI? Obviously not.

In an evolutionary algorithm, Hamlet would represent all the multidimensional mountains and valleys that make up the fitness landscape.  And yes, a map contains a lot of specified information.

gpuccio: It was easier to just print the output directly from the program.

Yes, it would be, but it wouldn’t demonstrate the ability of the evolutionary algorithm to navigate the landscape. Nor does being able to navigate Hamlet demonstrate that the biological realm is such a navigable landscape.

5. 909 gpuccio September 30, 2012 at 12:36 am

So, let’s go back to Lizzie’s algorithm. The important point now is: the algorithm can measure a product for each new string, but it must not be able to select for that product unless and until it is higher than 10^60. Indeed, that product level is our threshold for the new biochemical function to appear. And only if the new biochemical function appears, it can be positively selected and expanded by NS.

When the FAIL is so strong, one facepalm is not enough.

I hope this comment will stimulate gpuccio as some sort of negative selection.

6. gpuccio: “I have clearly shown that Lizzie’s algorithm is an “implementation” of IS, and not a “model” of NS. That is the important point. “

What you have shown is that you are attempting to redefine both the term “evolution” and the term “algorithm”.

An algorithm need not have any knowledge of its parameters at all.

Here is an algorithm to sum two numbers: A + B = C.

There are no numerical values at all yet it is clear how the algorithm works and it is also clear that no predefined “intelligent” solution is “inside” the algorithm.

Here’s Elizabeth’s algorithm embedded into a callable function that we invoke:

Lizzie_MutateAndSelect( Population, Size, ptrFunction_Environment );

Note that the “Lizzie” “algorithm” itself knows no specifics at all.

It just processes elements of a population, whatever that population may be.

We could also use the “Lizzie” algorithm to perform “Methinks it is like a weasel”, without changing the algorithm but simply changing the “environment”.

7. Mung: Here’s a model: Mendel’s Accountant “Mendel’s Accountant (MENDEL) is an advanced numerical simulation program for modeling genetic change over time…” Has anyone on your side of the aisle that you know of created something similar?

Yes. More particularly, the flaw in the Mendel’s Accountant has been determined. The calculation of “working fitness” is broke. From Mendel’s Accountant:

do i=1,total_offspring
work_fitness(i) = work_fitness(i)/(randomnum(1) + 1.d-15)
end do

The effect of ‘divide by random’ is to eliminate the vast majority of the signal from genetic or phylogenetic fitness.

If you look at through the Evolutionary Computation thread, there is a great deal of discussion and results from independent tests, including our own Mendel’s Bookkeeper.

8. What Sanford did with his Mendel’s Accountant was to try and ‘prove’ his preconceived result. He didn’t look at his own work skeptically, and when it gave the results he was looking for, his work was done.

What we attempted with Mendel’s Bookkeeper was to create a simulation to help determine under what circumstances evolutionary algorithms run to extinction, and when they don’t, especially an exploration of the critical point between the two.

9. GA Models are invalid except when Creationists build them to disprove evolution or promote some End-is-Nigh tooth-gnashing?

Incidentally, I downloaded Mendel’s Accountant, and it brought with it a parasite, a browser hijacker. That does not invalidate it, just a warning.

10. As keiths has already noted, gpuccio can’t stake out a position and defend it. Therefore it seems pointless to try and explain the flaws in his current argument because tomorrow he will abandon it and make up another excuse. There is some entertainment value in this chase as his arguments shift away from Dembski’s puffed-up math and toward the good old tornado-in-the-junkyard territory.

So, strictly for entertainment purposes, I will point out where he is wrong (again). I’m just curious where he will dart next.

His latest argument boils down to placing restrictions on fitness function. In his understanding, it should be zero almost everywhere (a sea of junk) except on a small island of functionality. As you cross from the sea to the island, the fitness function jumps discontinuously (a cliff). Once you get to the island, you can walk to its top point by following the gradient, which is smooth. Natural selection will work on the island, but it can’t get you to the island.

It’s a nice theory, gpuccio, but it doesn’t work in practice. You can see what a fitness landscape looks like when biologists actually try to measure it in some real systems. Here is a figure from a PLoS ONE article depicting such a landscape (on a log vertical scale) inferred from experimental data. It looks nothing like your theory. Here is how the authors summarize it:

According to the estimated parameters, the landscape was plotted as a smooth surface up to a relative fitness of 0.4 of the global peak, whereas the landscape had a highly rugged surface with many local peaks above this relative fitness value.

Y. Hayashi et al., “Experimental rugged fitness landscape in protein sequence space,” PLoS ONE 1, e96 (2006). doi:10.1371/journal.pone.0000096

There is no level surface of strictly dysfunctional junk on that map. Most proteins are not very functional, but some are less dysfunctional than others, and they win in a competition judged by natural selection. When you start in a random place on the map, you have an easy time getting to islands of proteins that are much better than the starting point. There the landscape gets rugged and it’s not easy to jump from peak to peak.

Note also that the map has more than one island of high functionality.

This is a picture that is almost diametrically opposite to your theory. Read the article. Think it through. Get back to the drawing board.

11. gpuccio, once you knee-jerk reaction subsides, let us know what you think of their fitness landscape that rises continuously at low levels of fitness. How does that square with your theory? What are your next steps?

12. I have had a number of opportunities over the years to watch live examples in which an ID/creationist bends and breaks scientific concepts to agree with sectarian beliefs. It is an agonizing process of word-gaming as every word is reworked repeatedly to mean something completely different. New words are often invented on the spot.

The end result might turn out to be a statement of a scientific concept that looks superficially similar to the real concept; but to the ID/creationists, the words all have different meanings. So the ID/creationist declares agreement with the science and claims to love science; but the “science” that the ID/creationist loves has nothing to do with reality.

The website of AiG has many videos in which AiG “scientists” are following exactly this bending and breaking routine. They are simultaneously hilarious and nauseating to watch.

13. GP

1) Anything that is selected must be able to influence reproductive success.

2) New functions that can influence reproductive success in a complex replicator are usually very complex, because they have to integrate themselves in a complex, integrated scenario (the replicator itself and its metabolism).

3) In biology, this is made worse by the observation that new biochemical functions are usually supported by new sequences and structures, unrelated to what already exists.

4) Complex functions are not decostructible into simpler, small, functional selectable steps. There is no logical argument for that to be generally true, and there are not empirical examples that this is generally possible, least of all in biology.

5) Therefore, NS cannot generate new dFSCI. This is an intrinsic limit of the process.

On the other end, if we use an implementation of IS that creates selectable intermediates at each single bit variation, and therefore a continuous functional space to some target, we can easily reach that target. Saying that this is a model of NS, and that therefore we can draw conclusions about NS and biology from that, is simply a lie.

Ah, good old civility! People are being dishonest, now?

I honestly think that a model of NS is any process that has replication and a consistent differential probability that some forms of the replicated entities will be replicated compared to others in the current population. You are saying it is not a model of NS because biology is a bit complicated, and (you say) novelty is nearly always accompanied by a saltational leap in the dark, because (you say) reality does not possess explorable paths between one function and another, and (you say) complex functions are irreducibly complex, not just now, but at all points in their history. Which is all possible, but a tad vague and waffly. There is plenty that evolution cannot explore, for those and many more reasons. So? It can still trickle through the cracks that it can explore. It boils down to your conviction that the space around every current ‘function’ is absolutely sealed up tight. Nothing could have got there, and it can’t go anywere else. Or just some current functions. In which case, which? It’s protein baraminology.

Actually, Lizzie hamstrung herself slightly by excluding some ‘biological reality’: recombination. This process has a mahoosive impact upon the ability of a ‘search’ to explore protein space. One could make it more ‘biological’ yet by introducing a great deal more ‘lethal’ space, and a more rugged landscape, but still the combination of fragments from different strings is a powerful means of widening the net of the ‘search’.

The proportion of times such leaps in the dark land on functional space is entirely dependent on the density of function, which you cannot intuit by reading Hamlet.

14. 4) Complex functions are not decostructible into simpler, small, functional selectable steps. There is no logical argument for that to be generally true, and there are not empirical examples that this is generally possible, least of all in biology.

5) Therefore, NS cannot generate new dFSCI. This is an intrinsic limit of the process.

This is the hill GP has chosen to die on? Really? Really, a bare-faced assertion that unguided natural selection intrinsically is impotent in the real world? Really, no natural selection at all? That’s it? That’s GP’s ID position? Well, that’s logically compatible with certain types of christian (and muslim, I think) theology which claim that every single tiniest happenstance is directed by god as part of god’s plan. In which case, I want to meet that god face to face to spit in its eye for its “intelligent design” of E coli 0157, which tortured our son nearly to death. Fuck that god, or “designer”, or whatever GP wants to imagine it to be.

15. It’s pitiful, isn’t it? After all the hand-waving about ‘dFSCI’ and ‘intelligent selection’, gpuccio’s ‘theory’ boils down to nothing more than this:

Natural selection cannot generate complex functions. Why? Because the functional areas of the fitness landscape don’t logically have to be connected. Since they don’t have to be connected, I’m entitled to assume that they never are connected. Natural selection cannot bridge the gaps that I’m assuming are there. QED.

16. It would be possible to derive a ‘random’ way of deriving the fitness function for a GA, rather than just ‘intelligently’ choosing it. Take the depth of snow in your yard on December 14th last, add the license number of the next vehicle to pass, the number of times the word “obvioulsy” appears in UD comments … then simply evaluate strings according to their approach to this peak, and don’t copy (breed from) the worst.

17. The travelling salesman problem has a randomly generated fitness landscape. There is no necessity for a connectsble fitness gradient.

18. Hey, that’s cute.  I never would have guessed that anyone could make that typo even once by accident, much less repeatedly …

Obvioulsy, it’s an “intelligently designed” spelling improvement! ;)

19. This is the same claim that KF always makes – that all biological functions are isolated islands surrounded by oceans of non-function. A claim that is not backed up with any evidence, and contradicted by plenty.

If I understand some of GP’s demands correctly, one of the things he is asking for (apart from an UN-navigiable landscape!) is an environment with intrinsic selection – in other words differential survival has to be an emergent property of the agents behaviour in an environment and not something that is explicitly coded for.

Perhaps he imagines fitness functions as being analogous to some no-corporeal entity that floats around, using its intelligence to select agents that it likes … ?

I have a half baked idea for minimal a simulation with implicit selection, if I get time I’ll try and write some code.

20. All hail the flagellum! And its relative, the Type III Secretory System! They live on neighbouring islands, surrounded by Irreducible Sea, and enhance bacterial abilities to mess with us. [And cloroquine resistance ... plasmids as vectors of antibiotic resistance ... ]

21. I think once you’ve decided that evolution can’t get to here, you can just dispense with all simplifications that try and analyse how evolution could have got to here. It can’t because ‘evolution does not work like that’. Evolution in or towards modern, complex, integrated organisms cannot be desconstructed because those organisms themselves cannot be deconstructed. Which is fallacious reasoning.

Computational resources do not permit a full simulation of organisms and environments. One might imagine that it should include both a genotype and a phenotype. The genotype should code for proteins and RNA under gene expression control, which should fold according to chemical law and catalyse reactions in a realistic manner, and interact in virtual cells. Perhaps multicellular phenotypes should build up, and interact with a coevolutionary environment of other replicators and …

Hang on, though .. why? All the issues GP points out relate to phenotype. Phenotype is an arena of complex interaction between different parts of the genotype and with the environment … but still, you don’t have to model it explicitly (unless it is relevant to your study). All that phenotype is, in the evolutionary world, is the proxy for genotype with which selective agents interact. Because the currency of evolution is DNA (even when ‘epigenetics’ is invoked), that is all that passes down multiple generations. All you need is a means of ‘evaluating’ the digital genotype. Selection ‘evaluates’ via the genotype’s expressed tastiness, or ease of capture, or something. There is no doubt that many traits are multigenic, and interaction is complex (you think biologists may have missed that?). So you can include whatever you decide is phenotypically important in the matter of epistatic and pleiotropic interaction – but you can shortcut computing phenotypic expression, and just model gene interactions directly.

But it still boils down to DNA, which is just a long string of bases. However intricate the interaction of parts of these strings in life, the replication process ‘flattens’ the complexities and simply copies the whole lot. Likewise, death removes the whole lot.

So perhaps what GP could do, rather than hand-waving concerns, is to build a simple model and then ‘complexify’ it. Build in the biologically real things that he believes model-builders omit unjustifiably, and demonstrate how they sap the system of evolutionary power (if that is what he thinks will happen in complex, more ‘real’ digital organisms).

22. Behe’s argument boils down to the claim that multiple, co-dependent mutations are extremely unlikely. Which at face value is true. All versions of ID depend on this reasoning, because all versions of ID boil down to gaps arguments, with design as the fall through when evolution can’t be described in pathetic detail.

As soon as someone demonstrates that four co-depenndent mutations can occur in a tightly controlled lab experiment with a tiny population, the argument shifts from the mutation count to the magnitude of the functional gain. Nevermind that the original goalpost involved mutation count. I suppose the next edition of Behe’s Edge will correct this. Along with correcting the argument that gain of function is rare.

My point would be that because ID is based entirely on gaps, its entire argument consists in goalpost shiftng.

The flagellum was once comprised of modules that required all to be present in order to have any selectable function. But there are approximately two dozen different flagella incorporating various subsets of the rotary version’s components.

23. Zachriel: In an evolutionary algorithm, Hamlet would represent all the multidimensional mountains and valleys that make up the fitness landscape. And yes, a map contains a lot of specified information.

gpuccio: could you please explain how your concept of multidimensional fitness landscape can help find the right functional sequence for a new protein domain with a new protein structure and a new biochemical function, for instance a new enxymatic activity, from unrelated DNA strings?

As we said above, the algorithm only demonstrates that *some* complex landscapes are navigable (sometimes in surprising ways), not that the biological realm is navigable. However, it does address a source of specified information. The complexity of the genome can increase due to its interaction with the environment. There’s no inherent barrier as suggested by your comments.

24. Er, Gregor’s Bookkeeper.

25. gpuccio: could you please explain how your concept of multidimensional fitness landscape can help find the right functional sequence for a new protein domain with a new protein structure and a new biochemical function, for instance a new enxymatic activity, from unrelated DNA strings?

The algorithm also allows exploration of how evolutionary algorithms can evolve entirely new sequences by mixing and matching existing components. This also shows why evolutionary search is much, much faster than random assembly—basic structures have multiple uses in completely different contexts. (e.g. the “ing” in “king” can also act to turn a verb into a gerand or participle, “saying”).

26. gpuccio: could you please explain how your concept of multidimensional fitness landscape can help find the right functional sequence for a new protein domain with a new protein structure and a new biochemical function, for instance a new enxymatic activity, from unrelated DNA strings?

Keep in mind that proteins are far more flexible than words. Many folds are made up of just a few amino acids, with the rest of the protein just a scaffold. Multiple folds can often perform the same function. Many proteins have multiple domains and multiple functions, and rearranging these domains can create novel structures. There is significant phylogenetic evidence of domain evolution including domain rearrangements.

Yang & Bourne, The Evolutionary History of Protein Domains Viewed by Species Phylogeny, PLOS One 2009.

27. I do wonder what GP has in mind when he says “new protein domain” or “new biochemical function”? Does he have one that he considers ‘new’, and definitively inaccessible by the probabilistic resources available to any ancestors – something that can be investigated, rather than his personal, very general assumptions about the structure of protein space and its distribution of function?

The only proteins we need to concern ourselves with are the ones that exist, and their accessibility from other points in the space that (on the evolutionary assumption) were occupied by ancestral sequences with similar or other functions. So it would help to have a concrete example, rather than this ‘function is universally restricted to tiny, widely-separated islands’ nonsense, which is empirically demonstrated to be untrue.

28. We went over the origin of protein domains a couple years ago on Mark Frank’s blog. Gpuccio bailed out when it was pointed out that a significant percentage of random folds have useful biological function (something that junk DNA lacks). Useful to the point of being able to rescue otherwise disabled bacteria.

The key element in this study is that the rescuing sequences have no sequences in common with the original, knocked out sequence. The haystack is jam packed with needles.

GP has also managed to maintain silence on the Lenski experiment, through an 800 post thread, on how a designer would replicate or improve the first 20,000 generations, in which critical, but non-adaptive, mutations accumulated.

Now that it has absolutely been demonstrated that neutral mutations can accumulate and enable later functional mutations, and now that it has absolutely been established that new functionality can arise without weakening existing function, I would expect some apologies from those who have been arguing otherwise.

29. a significant percentage of random folds have useful biological function (something that junk DNA lacks). Useful to the point of being able to rescue otherwise disabled bacteria.

That would be this I keep blethering on about?

The essence of proteins is modularity. And most structural modules have many ways of being formed. And because recombination of DNA fragments is an almost inevitable consequence of trying to keep hold of all this fine-yarn knitting, modules get swapped within the same protein and between proteins with great ease. It’s not just whole genes that are duplicated but fragments too (indeed, nothing demarcates a gene boundary to mitosis/meiosis). This also serves to make much wider leaps around protein space than a simplistic point-mutation model, mental or computational, may suggest. If space is widely explored from current position, and the generality of space of a type readily reached from there is significantly function-rich, it’s hard to see where GP’s assumed barriers lie, other than good old IC, which is more to do with the interaction of multiple genes than the ‘exploratory unlikelihood’ of single ones.

30. Allan,

Your link is broken due to a point mutation which took you off the island of functionality.

31. Your link is broken due to a point mutation which took you off the island of functionality.

Oh, bugger! I’ve posted that link about 4 times and screwed it at least twice. I spend most of my life trying not to get washed off the island of functionality.

32. Mung: The components are instances of some protein domain.

Some components are simpler motifs or repeated sections. Novel folds can be found in random sequences, and a majority of folds appear to have diverged from a small number of ancestors.

Zachriel: Keep in mind that proteins are far more flexible than words.

Mung: That’s debatable, and probably irrelevant.

We were referring to fragility. Words are much more likely to be broken by a substitution.

Mung: Who cares if they can be re-arranged?

Scientists trying to reconstruct their natural history. And, for the purposes of this discussion, it shows how and why evolutionary processes can be so effective at finding functional proteins.

33. Mung: Generating CSI with Faulty Logic

I take a fair coin and toss it 5 times: HTTHT
That’s one of the possible set 2^5 sequences (1/32).
Now I copy that sequence, and modify one position at random.
Say the first position: TTTHT
Is that sequence a member of the original set of 2^5 sequences?

Yes.

Mung: What is the relevance, if any, for calculating CSI?

Do tell.

34. Who cares if they can be re-arranged?

It matters for one’s inferences about the searchability of protein space (it’s not really a ‘search’, but it still finds stuff!). A sequence that can only be amended by point mutation finds more of the wider space closed off to it, since it can only probe very local regions, and especially if on an adaptive peak is more likely to be surrounded by detrimental sequence and less likely to encounter ‘new function’, however that may be defined. Even if stepwise paths are available, their exploration is slow. Recombinations probe further faster, allowing access to regions that could not be traversed stepwise due to lack of a non-detrimental path, and shaking them out of local maxima.

An important, quite subtle point about recombination is that it enables huge leaps to be taken with some care. If a protein suffered an insertion of 30 completely random bases, it is much more likely to be disabled than if a 30-base fragment of a different, working protein were inserted. The individual elements of a recombination event have been through the filter of natural selection, and will be inherently less disruptive than some ‘purely random’ sequence of the same size, even if they come from a protein with a completely different ‘function’.

Behe’s CCC calculation suffers from a lack of consideration of recombination too, incidentally. It’s a very important, sometimes underappreciated evolutionary force. Which one may argue is itself a Design feature, but one can’t hold that position and simultaneously dismiss its evolutionary capacity.

Lots more stuff to sneer at if you Google ‘protein domain evolution’.

35. gpuccio: (Thank you, Zachriel, for providing me a quick refernce to one of my favourite papers. I am afraid, however, that you have not really understood what it mean. It is not about “evolution” of the domains, but rather about their emergence in natural history. I quote:

“Notwithstanding, these data suggest that a large proportion of protein domains were invented in the root or after the separation of the three major superkingdoms but before the further differentiation of each lineage. When tracing outward along the tree from the root, the number of novel domains invented at each node decreases”. Emphasis mine.)

Right. So, according to the paper, 800+ domains were ‘invented’ *after* the divergence of the superkingdoms. It’s appears to be the usual evolutionary pattern of adaptive radiation followed by increasing specialization.

gpuccio: What you say about recombination as sense, and I can agree. But I don’t think it can solve the fundamental problems about completely new information. Anyway, it can be reasonable to try to evaluate the real powers of recombination on some empirical basis, but it is certainly true, as you say, that it is a “sometimes underappreciated evolutionary force”. Underappreciated and not much supported by evidence, although often generically invoked.

Keep in mind that your “don’t think” encompasses all evolutionary algorithms. Evolutionary algorithms, such as Word Mutagenation, can show you how and why recombination is such a powerful force for novelty.

36. Zachriel: And, for the purposes of this discussion, it shows how and why evolutionary processes can be so effective at finding functional proteins.

Joe: Intelligently designed evolutionary processes or blind watchmaker evolutionary processes? Or is cowardly equivocation still the best you and your ilk can provide?

In this case, we’re concerned with typical evolutionary algorithms based on  random mutation and recombination.

37. kairosfocus: Are you and/or anyone of your acquaintance taking me up on the offer of an up to 6,000 word essay for UD presenting your view and empirically warranted grounding of the blind watchmaker thesis style account of origins?

Try Darwin’s Origin of Species (1859). It’s a bit dated and longer than 6,000 words, (the 6th edition is 190,000 words), but Darwin considered it just a long abstract, and it still makes for a powerful argument.

38. Joe: Biased towards a goal, which means you are talking about Intelligently designed evolutionary processes.

Or navigate a fitness landscape (which may or may not be dynamic), which is sufficient to understand certain basics of the process, such as the ability of recombination to create novelty.

Zachriel: Try Darwin’s Origin of Species (1859). It’s a bit dated and longer than 6,000 words, (the 6th edition is 190,000 words), but Darwin considered it just a long abstract, and it still makes for a powerful argument.

Joe: But the evidence gathered since then has not borne out his “powerful argument”, which makes it impotent.

There is considerable evidence, much of it in Origin of Species, that shows that natural selection can be a mechanism of adaptation. Artificial selection shows that there are selectable intermediaries between quite different forms. Peter and Rosemary Grant’s work on finches in the Galápagos Islands shows how this works in nature.

39. Joe: Artificial selection is NOT natural selection-

No, it’s not. However, it does show that selectable intermediaries exist, and that selection for a simple trait, such as size, will result in multiple genetic and physiological changes.

Joe: NS requires that the change be random/ due to chance and no one ahs demonstrated that wrt finches

Natural selection works on existing variations. The change isn’t random, but due to changes in the environment.

Joe: Please define your use of “fitness” and also how it is you determined that recombination is a blind watchmaker process.

In an evolutionary algorithm, the fitness landscape is explicitly defined, and recombination is random.

40. gpuccio,

It’s taken me a couple of days to catch up on the discussion that took place over the weekend. Good stuff all around. I don’t intend to restate what’s already been presented so eloquently by many of the TSZ participants, but there are a couple of points I’d like to address. The first is this comment of yours, which you repeated in one or two other forms:

Any fitness function in ant GA is intelligent selection, and in no way it models NS.

Others here have noted that you are confusing the model and what is being modeled. Labeling all GA fitness functions as “active selection” misses the point of what the fitness function is modeling. We observe differential reproductive success in real biological systems. We further observe that this differential is correlated with the varying ability of individual organisms to utilize the resources of the environment. All the fitness function is modeling is that environment. It’s obviously an extremely simple model, relative to real world environments, with far fewer dimensions, but it serves the purpose of simulating a situation in which phenotypic variation affects reproductive success.

This is where it is important to avoid mixing levels of abstraction. The specifics of how organisms are rated relative to each other, what threshold is used to stop the run, and the goal of the programmer are immaterial. The essence of the model is to determine if certain observed mechanisms operating an environment where different genomes result in different rates of reproductive success will result in populations with certain characteristics. It turns out that they do.

Of course it is important to understand the limitations of a model. It may be that some important aspect of reality isn’t included and that aspect might invalidate conclusions drawn from the model. However, based on simulations like those done for Lizzie’s CSI problem and biological experiments such as Lenski’s, it does appear that the mechanisms of the modern synthesis are quite capable of generating high functional complexity, by your definition.

I realize that you don’t call it dFSCI because you require dFSCI to not have a “deterministic” explanation by definition, but that just further demonstrates that dFSCI is a measure of knowledge (or ignorance) not of any objective characteristic.

41. gpuccio,

There is only one way to model NS in an informational environment, like a PC or a PC network. I have proposed it many times, but darwinists seem to be horrified by the simple thought of it. However, I will rpeat it here for you:

1) Take an informational environment, set up in a completely blind way in respect to the experiment we are going to make. IOWs, the people who set up the envirnment must not know how we will use it. However, the environment, like all natural informational environment, must have specific resources, like RAM, mass memory, procedures, and so on.

2) Introduce a computer virus in the environment, that can replicate and use the resouces of the environment. It should also include a mechanism of random variation to its code, like random errors when it copies itself (the mechanism can be tweakable to allow for testing different rates of random variation).

3) And then, just wait.

That’s it. No active measurement of fucntion, no active expansion of the measured function. After all, we are modeling NS, not IS.

It sounds like you would consider an evolvable Corewars style system to be a real model of natural selection, essentially a simulation where, like in the real world, the only goal is survival and reproduction. That might be an interesting programming challenge.

I’m curious, how would you measure functional complexity in such an environment? Would it simply be the length in bits of the digital organisms? If an organism with sufficient functional complexity to meet your dFSCI threshold were to appear, would you consider it to have dFSCI or would the fact that it arose through evolutionary mechanisms, which might even be tracked mutation by mutation, mean that the dFSCI medal could never be earned?

42. gpuccio,

NS occurs only in replicators who use environmental resources. It is NA only if all the observed effect is determined by modifications in the fundamental functions of the replicators (metabolism, survival and replication itself) so that differential reproduction is observed.

Can you see that this is not what is happening in Lizzie’s algorithm?

a) There are no autonomous replicators, and therefore no use of envoronmental resources.

The genomes are modeled as replicators. Are you claiming that only actual running programs would be a valid model? If so, why?

b) The function is predefined, and has nothing to do with replication or with using environmental resource.

That’s not the purpose of the fitness function. As I tried to explain in a previous comment, you’re mixing the implementation details with the purpose of the model.

One could consider the environment to be “predefined” in the real world, in the sense that it exists before each generation of a population. It nonetheless provides the criteria by which organisms are evaluated with respect to each other.

c) The function is measured, with utmost precision, by the algorithm.

Well, in most GAs reproduction is stochastic, just as in real environments. I’m not sure what the precision of the fitness function has to do with the usefulness of the model.

d) The function is actively rewardes (either negatively or positively) by the algorithm, according to its measurement, although it generates no difference in reproducing fitness.

This demonstrates some confusion about the purpose of the fitness function. The fitness function is part of the model of differential reproductive success. Along with the selection mechanism, it determines reproductive success. The overall process, in most GAs, is stochastic.

So, Lizzie’s algorithm implements Intelligent Selection, and not Natural Selection.

A distinction without a difference. The model shows that the mechanisms of the modern synthesis are quite capable of generating functional complexity in excess of that required by your dFSCI.

43. onlooker quotes gpuccio (above) and asks:

The genomes are modeled as replicators. Are you claiming that only actual running programs would be a valid model? If so, why?

onlooker,

That quote is from September 29th, when gpuccio still believed that

as I said, all forms of “fitness functions” are IS. But you can indirectly observe NS in true informational replicators, as I suggested in the computer-virus experiment.

It’s yet another symptom of his inability to distinguish the model from the thing being modeled. There is an enormous difference between his computer virus experiment, which is a real Darwinian process that happens to take place on a computer, versus a model of a Darwinian process, in which replicators are modeled, variation is modeled, and differential reproduction is modeled (via a fitness function).

The criticisms seem to have finally sunk in, and gpuccio changed his position on September 30th (without acknowledging that he was doing so, as usual, until I called him on it). He now accepts that fitness functions are okay:

I will show first why Lizzie’s GA is an example of IS, and not of NS. I will then show how it should be modified to drop IS and generically “model” NS…

The important point now is: the algorithm can measure a product for each new string, but it must not be able to select for that product unless and until it is higher than 10^60…

I think we should ask gpuccio to assign version numbers to his arguments. Otherwise it’s impossible to keep track of all the flip-flops.

44. Joe: NS requires that the change be random/ due to chance

WTF?? No it doesn’t. NS could operate perfectly well on variations introduced by a designer and biased in any way.

45. It is pretty clear that ID/creationists over at UD are still trying to word-game their “arguments” against evolution; that word-gaming process has been getting increasingly contorted ever since the 1970s.  We are seeing the same thing at AiG and ICR.

Their “arguments” have indeed evolved – from creation science to cdesign proponentsists to ID – but the most fundamental feature of that evolutionary process has been its strict avoidance of getting the real science right.

46. Joe: Nature doesn’t select and natural selection could never produce a toy poodle even given selectable intemediates.

That’s the point, of course. There are selectable intermediaries between wolves and toy poodles, and selection for very general traits (size, curly hair, docility) can result in the evolution of complex genetic and physiological changes.

Joe: So, yes there needs to be variation and it needs to be random/ ie a chance event.

There has to be variation for natural selection to work, but it doesn’t have to be the result of a chance event. It may already exist in the population as part of the inherited variation. It could be inserted into the genome by magic, and natural selection would still work. (However, for many characters, they form a standard distribution consistent with random variables.)

Joe: The variation is the change and according to the modern synthesis is entirely by chance. Changes due to the environment would be directed changes ala Dr Spetner’s built-in responses to environmental cues”- IOW more evolution by design.

You seem to be confusing the sources of variation with natural selection, such as when you said “NS requires that the change be random/ due to chance”. Natural selection acts on existing variations, from whatever source.

47. Mung: Where did the protein domains come from that are required for recombination?

Some novel protein domains are available to completely random processes. However, the natural history is not well-documented.

48. GP:

What you say about recombination as sense, and I can agree. But I don’t think it can solve the fundamental problems about completely new information. Anyway, it can be reasonable to try to evaluate the real powers of recombination on some empirical basis, but it is certainly true, as you say, that it is a “sometimes underappreciated evolutionary force”. Underappreciated and not much supported by evidence, although often generically invoked.

I’m not sure what you think isn’t supported by evidence – nor what you really mean by ‘completely new information’. The front end of gene A attaching to the back end of gene B is ‘completely new information’, even though the partners pre-existed. And the catalytic activity of the new combination is not compelled to be that of either source, still less some halfway hybrid between them.

Recombination clearly happens, by several different mechanisms throughout the living world, including us every time we make a gamete, as crossover does not know where the genes are. Where within a gene and alignment is good, we simply swap front and back of the same gene, but nonetheless this introduces a rate change in exploration of space (or it does modelled in bloody GAs, anyway!***). Behe’s CCC argument relies on serial mutation 1 then 2 or 2 then 1, with no benefit till both occur in the same individual. Calculations show it to be of low (though not vanishing) probability. But since 1 and 2 must necessarily be at different positions in the gene, recombination can occur between them, increasing the chance substantially, even though recombination will cause occasional loss of 1-2 links.

Other mechanisms clearly occur that cause fragments to be moved greater distances in the genome. The existence of long areas of sequence identity (or close enough to be revealed by statistical test) in different genes – the very thing that enables us to declare homology of a ‘domain’ – is regarded as evidence of the within-genome common descent of that sequence by duplication, which necessarily involves a recombination event. Other explanations for that homology are pretty ad hoc – one could infer that it was moved (or tooled in situ) by a Designer – but what would distinguish identity from such a source from that caused by known mechanisms of recombination?

*** If you are right about GAs, it is one hell of a coincidence that a method of exploring certain kinds of digital space using only the biological observables of differentials in birth and death, mutation and (optionally) recombination should have such power that they are popular tools in engineering and maths, as well as biological applications unrelated to modelling evolutionary mechanism (eg phylogeny) … and yet you think the algorithm has NO power in the very realm that inspired it – biology? And despite working fine in other statistical fields, according to some anti-common-descenters in ID their use in tree-building leads inevitably to false phylogenies …! Every time they are applied to the biology that inspired them, they apparently fall to bits. One hell of a coincidence.

49. Joe: Evolution by design.

Nevertheless, it demonstrates the existence of selectable intermediaries.

Zachriel: There has to be variation for natural selection to work, but it doesn’t have to be the result of a chance event.

Joe: It cannot be planned/ directed and still be natural selection.

The source of variation doesn’t have to random for natural selection to still occur. Darwin posited a non-random theory of Pangenesis, for instance. You’re confusing two different processes, the sources of variation and selection.

50. It sounds like you would consider an evolvable Corewars style system to be a real model of natural selection, essentially a simulation where, like in the real world, the only goal is survival and reproduction.

What he wants is something like Thomas Ray’s Tierra, as we discussed on Mark Frank’s blog lo these many months agao.

I’m curious, how would you measure functional complexity in such an environment?

Excellent question. I wish you luck in getting a straight answer. Intelligent Design Creationists in general and UD denizens in particular are not known for their willingness to make testable claims.

51. Joe: IOW it demonstrates the severe limits of natural selection.

Just so we’re clear, you agree that there are selectable intermediaries between wolves and toy poodles?

Joe: If the source of variation is planned then it cannot be natural selection, by definition.

That is false. Again, you are confusing two different processes; the sources of variation and natural selection. For instance, if a genetically modified organism escapes into the natural environment, it will be subject to natural selection just like any other phenotype.

Joe: Darwin always referred to variation by chance.

That is also false. Darwin proposed a non-random source of variation called Pangenesis, a speculative theory which included Lamarckian inheritance of acquired traits.

52. Once again, if anyone has ever produced a creationist critique of evolution as understood by science rather than as misrepresented by creationists, nobody here has ever found it. ALL critiques are of straw men.

And so conversations with creationists consist often of “this is false, here’s the correction” repeated until the creationist drops the subject, often with insults.

53. Mung: GA’s do not use only differentials in birth and death, mutation and (optionally) recombination.

Differential refers to differences due to relative fitness, usually defined by a fitness function or map.

54. It’s slightly surprising how many people are willing to judge the efficacy of GA’s without being able to write one, even at the specification or pseudocode level. This really ought to be a prerequisite to discussing evolution.

55. Mung is under the impression that a GA has to be seeded with potential solutions:

For example, potential solutions must be encoded into a “chromosome.”

No, Mung, potential solutions do not have to be encoded into a “chromosome”. That is optional. You can start with a purely random “genome”, as Lizzie does in her program.

56. Mung writes:

For example, potential solutions must be encoded into a “chromosome.”

…and tries to support his statement by saying:

There is at least the possibility that a solution will be found among the first 100 randomly generated genomes [in Lizzie's program], though she doesn’t actually check to see if that is the case.

Suppose Lizzie’s program initialized the genomes to all 0′s. Then there would be no potential solutions among the initial genomes, yet the program would still converge to a solution.

Mung’s statement is incorrect. It is not mandatory to encode potential solutions into the genome.

A few minutes after posting his initial comment, Mung seems to realize that he’s overstepped, and softens his claim. This time, instead of claiming that potential solutions must be encoded into the genome, he backtracks and links to a site that merely explains that you have to pick the encoding scheme:

The most critical problem in applying a genetic algorithm is in finding a suitable encoding of the examples in the problem domain to a chromosome.   [emphasis is Mung's]

Well, duh. Of course you have to have an encoding. It’s a computer program! Complaining about that is as silly as complaining about this: You give Johnny a fiendishly difficult 12th degree polynomial equation to solve. He asks if the solutions are numbers. Before you can stop her, your friend tells Johnny that yes, the solutions are numbers. You complain bitterly, saying that she has given the answer(s) away.

Telling Johnny that the answers are numbers doesn’t give away the solutions. Selecting an encoding for a GA doesn’t give away the solutions, either.

Maybe someone over there at TSZ will be kind to you before you put your foot in it any more than you already have.

I’m sorry, Mung, could you repeat that? I think you have your foot in your mouth.

P.S. Please remind your buddy Upright Biped that Reciprocating Bill has some questions for him, and that Allan and I have refuted the latest version of his “semiotic argument for ID”.

57. Also natural selection is supposed to be blind and mindless. And that cannot be with directed mutations.

God, Joe – Variation and Selection are two different things! How long have you been discussing evolution, now? Exlicitly chosen mutations can still be filtered by the blind and mindless process (which could not be otherwise, unless there is also an Intelligent Selector with a population-wide overview) of one type leaving leaving more or fewer offspring than another.

58. Zachriel: Some novel protein domains are available to completely random processes.

Mung: Novel. Would that be like, new?

Maybe you can Allan can talk: (link to this:)

I do wonder what GP has in mind when he says “new protein domain” or “new biochemical function”? Does he have one that he considers ‘new’, and definitively inaccessible by the probabilistic resources available to any ancestors – something that can be investigated, rather than his personal, very general assumptions about the structure of protein space and its distribution of function?

The only proteins we need to concern ourselves with are the ones that exist, and their accessibility from other points in the space that (on the evolutionary assumption) were occupied by ancestral sequences with similar or other functions. So it would help to have a concrete example, rather than this ‘function is universally restricted to tiny, widely-separated islands’ nonsense, which is empirically demonstrated to be untrue.

Note that I am asking GP what he considers ‘new’, not denying that anything in biology can ever be considered such. It’s a word we can choose to apply if we wish, and since GP does, I’d like to know what specific example he has in mind, rather than casting the word across the entirety of life because it must apply to something, somewhere.

My point about accessibility in ‘space’ was that, barring a very few bases, every DNA base in existence today has apparently been template-copied from another. This is the mechanism that probes protein space, randomly, and detects ‘novelty’ within it if you wish to call it that. There appears to be no significant mechanism to introduce new DNA sequence other than through template copying and fragment shifting, outwith ID. There is no separate process that assembles new base sequences out of thin air, rather than from the various mutational processes acting upon existing sequence.

There is no doubt that ‘new’ folds, and ‘new’ function, must be capable of arising. But that ‘newness’ is something that we make a call on, not evolution. There is not a different mechanism dealing with ‘newness’ vs that dealing with ‘oldness’.

59. GP, responding to my point on Behe’s omission of recombination.

I think you are wrong here. [...]

The real point is that, while your discourse about recombination can make some sense in the recombination of functional elements, it is of no importance in the case of individual mutations that have no function until they conflate in a more complex output. The important point is: a recombination can certainly join two mutations, but it can join any set of two mutations with the same probability, unless we can show that some mutations, and in particular those that are necessary for the future function, recombine more frequently than others. IOWs recombination in this case does not alter the probabilistic scenario.

This is an error often made by many darwinists. [...]

This is an error often made by Creationists! Without a better grasp of population genetics, and an apparent confusion about the separate roles of phenotype and genotype in evolution, you simply hand-wave away the probabilistic relevance of recombination. It’s only important when genes recombine, but not when subunits do? How come there are so many common modules, then? Not design, surely? The first, maybe, but what about all the children?

On the CCC, it’s still a probabilistic case for A-B, but you have to include all the contributors to the probability of that combination, otherwise you’re just weighting the game to win a point. Behe makes an argument based upon serial probability of double mutation. The independent mutations A and B have an individual probability of arising, and the second mutation has a similarly small probability of arising wiithin the A- or B-bearing subset of the population. The result is a very small probabilistic product of the two. But given the nonzero probability of A-only and B-only existing (which must be the case in order to have a nonzero serial probability), there is a further probability to be taken into consideration – you don’t have to wait for A to add the B mutation, or B to add the A; there can be recombination between members of the A- and B- populations. This probability is multiplicative, akin to the counter-intuitive ‘birthday’ probability, and gives a substantial boost to the probability that A-B will arise in the population.

Crossover and other recombinational mechanisms massively increase the power and reduce the search time of GAs. Even on ‘single-gene’ models – ‘subgene’ swapping has a real effect. This is a fact, which you could establish for yourself by writing a GA.

You seem to think that gene disruption – a phenotypic effect -  puts a damper on ‘real’ evolution. And of course it does. But to what extent? And is that extent greater than the disruption caused by new combinations of separate genes? You can’t just say “yes!”. If genes were just evaluated as ‘raw’ DNA, their evolution would be closer to that exhibited in the simplest GAs. Now, you could model that phenotypic constraint. You could introduce constraints such that recombinational products were typically less fit, particularly where involving subunits. Then, unsurprisingly, the power of recombination on that GA would be diminished. And if recombination were under genetic control, it too would be selected against. So the question is: how biologically realistic is the extreme version of that constraint? Has it just been imported because you don’t like the implications of the less-constrained model?

I say again: within-protein and between-protein recombinations happen, as demonstrated by sequence data. So while some dampening is introduced by phenotype, it is not so extreme as to forbid recombination (else there would be selection against ALL recombination, since the mechanisms don’t ‘know’ where the genes start and finish). And domains can be deconstructed. Four amino acids will make a turn of a helix. Duplicate that ‘proto-domain’ a few times and you have an extended helix, 50, 100 bases long … and the ID-er comes along and declares that the domain is irreducible complex – for, if you remove it from the modern protein, or even chop it back to 4 bases, it ceases to work!

60. Yeah, he’s pretty confused on that point.

Mung: There is at least the possibility that a solution will be found among the first 100 randomly generated genomes, though she doesn’t actually check to see if that is the case.

Yes, that the nature of randomness and its fit to a landscape. Think about it. Some random sequences will inevitably fit better than others.

61. gpuccio: It is possible that certain recombinations are favoured versus others by the structure itself of the genome, for instance whole gene or whole exon recombinations could be favoured versus purely random ones. There can be genomic sites that make recombination more likely. All that could increment the power of recombination, but should be explained as an adaptive mechanism already present in the existing genome.

That isn’t necessary to show that recombination is a powerful mechanism for generating novelty. It depends on the fitness landscape, of course. In word-space, the “ing” in “king” can recombine to create gerunds or participles, such as “saying”. Word-space is full of such simple motifs that readily combine to create new words. Similarly, in protein-space, simple motifs are often repeated, and recombination between sequences that exhibit such motifs are much more likely to generate workable proteins.

gpuccio: This is an error often made by many darwinists. A random effect does not change the probabilities of a specific output, unless we can demonstrate some explicit connection between the effect and the output.

That’s the error of IDists. They assume that evolutionary processes are no better than random assembly—but that’s simply not the case. If you recombine workable protein sequences, you are much more likely to find a new workable protein sequence than random assembly alone.

gpuccio: NS is selection based only on fitness/survival advantage of the replicator. The selected function is one and only one, and it cannot be any other. Moreover, the advantage (or disdvantage, in negative selection) must be big enough to result in true expansion of the mutated clone and in true fixation of the acquired variation. IOWs, NS is not flexible (it selects only for a very tiny subset of possible useful functions) and is not poweful at all (it cannot measure its target function if it is too weak).

Natural selection is based on the reproductive fitness of the replicator. There can be many functions that accomplish this aim, so if longer legs provide an advantage, then it can be subject to natural selection. In the abstract, this is done with a fitness landscape, but more detailed simulations are possible. As for the size of the advantage, that is also easily simulated.

62. potential solutions must be encoded into a “chromosome.”

All ‘chromosomes’ – start, intermediate and solution – must exist in the space-of-all-possible-strings available to the GA! Kind of axiomatic. In the case of evolution, the space-of-all-possible-chromosomes contains all-real-chromosomes. It’s the space of all AT/CG/GC and TA pair strings. So you couldn’t not start with a ‘potential solution’ (‘solution’ in evolution being a fitter genome for now).

In a GA you could start with a string of length zero, since (unlike biology), the copy method is not part of the heritable string. One would, of course, need a method which could add bits to such an empty string, and a fitness function that did not regard such strings as inviable. In a sense, an evolutionary GA could be regarded as examining the behaviour of ‘extra’ DNA, tacked onto a taken-for-granted replicative core. I know ID-ers don’t like taking the OoL for granted, but it is simply not part of evolution.

The start string does not have to be a solution, but a path needs to exist that allows it to become one according to the variational methods incorporated and the probabilistic resources available. In any GA you have no idea if such a path exists, nor what the actual solution will be – kind of the point of doing it, to see.

63. Zachriel: Some novel protein domains are available to completely random processes. However, the natural history is not well-documented.

gpuccio: What do you mean? To what are you referring here?

Random sequences can form active proteins (Keefe & Szostak, 2001). The origin of the original protein domains is still largely conjectural. However, random sequences are fairly rich in active proteins. By the way, if you want to find a needle in a haystack, try sitting on it.

Zachriel: Keep in mind that your “don’t think” encompasses all evolutionary algorithms. Evolutionary algorithms, such as Word Mutagenation, can show you how and why recombination is such a powerful force for novelty.

gpuccio: Can you give us the code? Can we discuss the oracles in it?

The algorithm is very simple. The landscape is the dictionary of valid words. The population is composed of sequences of letters that form words. The algorithm randomly mutates and recombines these sequences of letters. If they form a word, they enter the population. If they do not form a word, they do not enter the population. So, if the population includes “king” and “say”, they might evolve in the next generation to form “hay” (mutation) and “saying” (recombination).

A couple of insights: It is possible to evolve long words much faster than random assembly. Recombination is essential to this process.

64. There can be genomic sites that make recombination more likely. All that could increment the power of recombination, but should be explained as an adaptive mechanism already present in the existing genome.

This is simply contradictory. Shuffling bits and pieces of protein is an adaptive mechanism because it increases the power of module shuffling, which is a disruptive mechanism and has limited power of evolutionary exploration? Make your mind up!

The bottom line point to bear in mind is that recombination (distinct from exon shuffling) is blind to gene expression. Totally. So it has nothing to ‘go on’ to establish what would be a legitimate swap and what would not. It is variable across genome length, for sure, for many reasons both ‘active’ and ‘passive’, but it is not attracted by regions that could do with a bit of a shake-up so much as repelled by those which would be better without.

There are many different kinds of recombination, and I don’t know how much benefit there is in lumping them all together as ‘adaptive’ – which must involve direct genetic control with a fitness effect on the genes mediating that control to be meaningful. Recombination due to viruses, transposons, damage misrepair, ectopic misalignment in meiosis – these are no more obviously adaptive in themselves than point mutation. But, nonetheless, all recombinations, whether adaptive or not, still promote much wider exploration of protein space than you started off allowing for – but this is not always a good thing. Such exploration is not to the benefit of any individual organism, or most genes. It’s just something that happens, and organisms adapt if that-which-happens throws up a beneficial combination – one more source of the spectrum-of-variation on which NS works both positively and negatively.

65. Joe: In what way is Lamarkian inheritence non-random?

Because the posited source of variation is not random with respect to fitness, but are advantages acquired by the parent through use that are passed down to the children. So, if the parent uses a certain muscle a lot, then the child will be born with a larger muscle.

Joe: For example, if a man loses his arm in an accident, an acquired trait, that would be random.

A mouse losing a tail is not a heritable trait. (Weismann, 1899).

Joe: GAs are a DESIGN mechanism, period.

So are weather simulations and calculations of planetary orbits.

Joe: Natural selection requires the fitness be due to heritable random variation(s)

We already pointed to a simple counterexample. If a genetically modified organism enters the natural environment, it would be subject to natural selection. For that matter, so would a domestic dog entering the wild, à la The Call of the Wild.

66. Joe: In what way is Lamarkian inheritence non-random?

Because the posited source of variation is not random with respect to fitness, but are advantages acquired by the parent through use that are passed down to their children. So, if the parent uses a certain muscle a lot, then the child will be born with a larger muscle.

Joe: For example, if a man loses his arm in an accident, an acquired trait, that would be random.

A mouse losing a tail is not a heritable trait. (Weismann, 1899).

Joe: GAs are a DESIGN mechanism, period.

So are weather simulations and calculations of planetary orbits.

Joe: Natural selection requires the fitness be due to heritable random variation(s)

We already pointed to a simple counterexample. If a genetically modified organism enters the natural environment, it would be subject to natural selection. For that matter, so would a domestic dog entering the wild, à la The Call of the Wild.

67. Weird problem. Editing results in a “spam” error, and the comment disappears.

68. Joe: Natural selection requires the fitness be due to heritable random variation(s)

Natural selection doesn’t give a damn how the variations were generated, nor what people variously mean when they stick ‘random’ in a sentence.

It’s also debateable whether the variation needs strictly to be heritable, although obviously you are’t going to get any evolutionary change if it isn’t. Heritability isn’t a boolean; it’s a continuum (unless you argue that 0 and non-0 are boolean, which is true but less informative). Heritability influences the coupling between the drive of NS and the wheels of evolutionary change. Anything over 0% means that phenotypic sorting has the power to influence genotype frequencies.

69. gpuccio: But he never analyzed the original random sequences, which were selected for a mere very week ability to bind ATP, and then intelligently engineered into the final protein.

That’s right. They form compact three-dimensional structures, i.e. folds, capable of enzymatic activity. That’s what we were talking about.

gpuccio: Oh, yes. The algorithm is very simple. And it has a whole dictionary as a oracle! Simple indeed.

That’s right. The algorithm is very simple. The landscape, however, is highly complex and specified. Yet, the simple algorithm can navigate the complex landscape billions of times faster than random trial.

gpuccio: And how does the algorithm know that a word was formed? Ah, I forgot! The dictionary.

That’s right. It’s no different than comparing a sequence to a vast library of possible proteins.

gpuccio: And I suppose that the dictionary is essential to appreciate the successes of recombination.

Not at all. It’s just a single example. You “didn’t think” recombination would produce different results than mutation or even random trial, and your objection was very general. Hence, a general example is sufficient for you to see why recombination is an essential evolutionary mechanism.

Again, because words share many of the same motifs, recombining words has a much higher likelihood of producing a new word than random assembly. Similarly with proteins, which also exhibit motifs.

70. Joe: Differential reproduction due to heritable random variation (mutation)= natural selection

The “random” is extraneous. Nor is mutation the only source of variation. Natural selection can occur when there are existing variations in a population, regardless of whether there is a source for novel variations.

71. I said randomly generated genomes. I’d say the chance that Lizzie’s program generated 100 strings of all 0′s at random are about the same as her generating CSI.

Nice try, Mung, but your own words betray you:

There is more to a GA exploring certain kinds of digital space than differences due to relative fitness (usually defined by a fitness function or map), mutation and (optionally) recombination.

For example, potential solutions must be encoded into a “chromosome.”

Encoding potential solutions into a chromosome implies there is a problem to be solved.

Information about which potential solutions are more likely to solve the problem must be implemented. [emphasis mine]

The bolded statements are both wrong, for the reason I already gave:

Suppose Lizzie’s program initialized the genomes to all 0′s. Then there would be no potential solutions among the initial genomes, yet the program would still converge to a solution.

This shows that you don’t need to encode potential solutions into the chromosome, and you don’t need to implement “information about which potential solutions are more likely to solve the problem.”

P.S. I think you’re being unnecessarily modest by refusing to nominate yourself for your own Junk for Brains Award. You’ve earned it, Mung.

P.P.S. Please remind your buddy Upright Biped that Reciprocating Bill has some questions for him, and that Allan and I have refuted the latest version of his “semiotic argument for ID”.  Upright seems to be afraid of defending his argument, as usual.

72. Mung: Today’s Junk for Brains winner is Zachriel, who chide’s ID’ers for “assuming that evolutionary processes are no better than random assembly” while appealing to random assembly by a random process such as recombination.

Mung: I guess Zachriel can speak for himself, but randomness (as stochasticity) is of course part of the evolutionary process – recombination and mutation for sure, but stochastic influences on allele frequency changes come into it as well. The mistake would be to assume that this means that evolutionary ‘search’ is simply a matter of pulling sequences out of a metaphorical bag until, with a probability of 1 in n (the number of possible sequences) the desired sequence is hit. A random walk including random recombinations of existing sequence isn’t the same as random ‘assembly’ repeatedly from scratch.

If you don’t make that mistake, well done you, but many people, from Fred Hoyle onwards, have.

73. How many times do you need to point out that evolution doesn’t test the universe of sequences, but just the next door neighborhood?

How many times do you need to point out that the mere existence of alleles refutes the vertical cliff charicature of sequence landscape?

Recombination is a larger scale mutation than base point change, but it is still a walk in the neighborhood.

74.  Zachriel: The “random” is extraneous.

Joe: Only to equivocators, like yourself.

Hmm. You provided this definition on your own blog: “Natural selection is the result of differences in survival and reproduction among individuals of a population that vary in one or more heritable traits.”

Natural selection occurs on existing variation. Consider a population of moths, some of which are white, and others are black. Natural selection might occur if white moths are preferentially eaten by birds, leaving the black moths to leave offspring. This is true regardless of the original source of the variation.

Zachriel: Natural selection can occur when there are existing variations in a population.

Joe: Yes, it can.

There you are then.

75. Mung: Zachriel, who chide’s ID’ers for “assuming that evolutionary processes are no better than random assembly” while appealing to random assembly by a random process such as recombination.

We admit, our language was poorly chosen. Thought it was clear in context, but you may not have followed the entire thread. By random assembly, we meant where each sequence is completely randomized with respect to previous sequences. Now that the point has been clarified, perhaps you would like to respond to our actual point.

Z: That’s the error of IDists. They assume that evolutionary processes are no better than searching completely randomized sequences—but that’s simply not the case. If you recombine workable protein sequences, you are much more likely to find a new workable protein sequence than searching completely randomized sequences.

76. Mung: There is at least the possibility that a solution will be found among the first 100 randomly generated genomes, though she doesn’t actually check to see if that is the case.

Zachriel: Yes, that the nature of randomness and its fit to a landscape.

Mung: iow, I’m right. you know it. But you don’t have the guts to tell keiths.

There doesn’t seem to be any disagreement between our position and keiths’. There’s apparently some confusion on your use of the word “solution”.

Mung: For example, potential solutions must be encoded into a “chromosome.”

Your statement appears to imply that someone is “encoding” solutions into the “chromosome”. If the sequences are randomized, assuming we are fitting solutions to a landscape of some sort, then some may naturally fit better, albeit probably poorly, than others.

Perhaps you are referring to the nature of the landscape. Some landscapes are such that evolutionary algorithms don’t work well, or don’t work at all. That depends on the specifics, of course, but nearly all the objections raised by kairosfocus, gpuccio, Mung and others are so general as to apply to all landscapes. For instance, recombination is a very powerful mechanism across many rugged landscapes, and evolutionary algorithms work millions of times faster for such landscapes than simply choosing random sequences one after another.

Mung: Every chromosome generated by the GA is a potential solution. Else what is the point of generating them?

Of course they are *potential* solutions, though solution may or may not be a single entity, but a best fit, for instance.

Allan Miller: In a GA you could start with a string of length zero

Mung: What would a string of length zero consist of?

∅. You sound like someone who was just told about the existence of zero. Word Mutagenation usually starts with the single-letter word “O”.
http://www.zachriel.com/mutagenation/Sea.asp

77. Mung: Today’s Junk for Brains winner is Zachriel

We really appreciate the honor, but we have already won the coveted Once Twice Thrice Quadrice Banned Award.

78. gpuccio,

It is completely wrong to model NS using IS, because they have different form and power.

As I said, you help me to refine my concepts, and I appreciate that.

Before someone states that I am changing arguments, I would suggest that you read again my original definitions of IS and NS, from which this statement can very clearly be derived:

“d) NS is different from IS (intelligent selection, but only in one sense, and in power:

d1) Intelligent selection (IS) is any form of selection where a conscious intelligent designer defines a function, wants to develop it, and arranges the system to that purpose. RV is used to create new arrangements, where the desired function is measured, with the maximum possible sensitivity, and artificial selection is implemented on the base of the measured function. Intelligent selection is very powerful and flexible (whatever Petruska may think). It can select for any measurable function, and develop it in relatively short times.

d2) NS is selection based only on fitness/survival advantage of the replicator. The selected function is one and only one, and it cannot be any other. Moreover, the advantage (or disdvantage, in negative selection) must be big enough to result in true expansion of the mutated clone and in true fixation of the acquired variation. IOWs, NS is not flexible (it selects only for a very tiny subset of possible useful functions) and is not poweful at all (it cannot measure its target function if it is too weak).

This seems to be getting to the essence of our disagreement, especially when combined with your following comment:

A distinction without a difference. The model shows that the mechanisms of the modern synthesis are quite capable of generating functional complexity in excess of that required by your dFSCI.

This is exactly the type of wrong statement that has prompted me to analyze in detail this issue. Have you read my post #910 in the old thread? Please, refer to it for any following discussion on this.

Yes, I re-read your 910 where you discuss what level of functionality is selectable. I find your thresholds to be arbitrarily selected, but that’s not relative to the essential difference I think we’re finding.

What I see is you focusing on the details of how a model is implemented rather than on the concepts being modeled. Yes, in many GAs the environment is modeled via a fitness function that is designed to accomplish some goal and the threshold for terminating the simulation is set (independently of the fitness function and other components of the GA) to recognize when that goal has been reached. None of that changes the fact that the GA is a model of an observed, natural process.

It doesn’t matter if you label the model “intelligent selection” to somehow distinguish it from “natural selection”, what matters is that the pertinent mechanisms of the model are the same as those we observe in real biological systems.

Heritable variation with differential reproductive success does, demonstrably, generate large amounts of functional complexity, according to your own definition. The only reason not to consider the results of these mechanisms of the modern synthesis to have dFSCI is because you define dFSCI in terms of knowledge about the provenance of the results and you define those mechanisms as “deterministic”.

If you disagree, and I suspect you do, please explain why your distinction between “intelligent” and “natural” selection has any bearing on what is being modeled rather than the details of how the model is implemented.

79. gpuccio,

I spent some time yesterday looking through the Tierra project and it does appear to meet your criteria for what you think is a proper model of natural selection. Before I go further with it, I would like a clarification from you.

I’m curious, how would you measure functional complexity in such an environment? Would it simply be the length in bits of the digital organisms? If an organism with sufficient functional complexity to meet your dFSCI threshold were to appear, would you consider it to have dFSCI or would the fact that it arose through evolutionary mechanisms, which might even be tracked mutation by mutation, mean that the dFSCI medal could never be earned?

It’s easy. I would proceed like Lenski. I would “freeze” (copy) the virus periodically to examine its code. If and when any functional string of code expresses a new function that helps the virus to reproduce, and therefore partially or totally replace the simpler version, then it will be easy enough to evaluate the funtional complexity of that new string of code, with the ususal methods detailed at the beginning of your thread at TSZ.

What “usual methods”? How, exactly, would you compute the functional complexity of a digital organism in Tierra? Patrick noted in the threads he referenced on Mark Frank’s blog that Tierra results in a number of different reproductive strategies, including parasitism and hyper-parasitism. What is the functional complexity of those organisms?

80. During the first 20,000 generations in the Lenski experiment, mutations occurred that were neutral with regard to citrate metabolism, but which turned out to be crucial after subsequent mutations.

How does the intelligent selector identify  and promote precursor changes?

81. gpuccio: d1) Intelligent selection (IS) is any form of selection where a conscious intelligent designer defines a function, wants to develop it, and arranges the system to that purpose.

A clarification please. What do you mean by “system”? If you mean the entire simulation, then obviously that would preclude any and all simulations.

82. Mung

What would a string of length zero consist of?

I presume you mean ‘what’s the real-world equivalent?’ rather than ‘how would you code it?’, but just in case, an example of a string of length zero in VBA would be one which returned zero to the VBA LEN function. In COBOL, a group with a next level OCCURS DEPENDING ON X where X is set to zero. I’m sure many languages offer the same kind of thing. The null, the empty set, the nothing. While the ‘replication’ function of biological replicators is a vital part of the string, that role is taken by the copy method in a GA, so the strings themselves don’t actually need to consist of anything at the start. The point of bringing them up is to point out that such strings are not likely to be ‘solutions’ to any worthwhile GA, so you aren’t necessarily ‘pre-seeding’ the population with anything.

So consider the zero-length digital organism as the absolute minimal replicator common to all GAs. As long as a method exists that occasionally adds random bits to a string, something will soon emerge, and variations between these ‘non-null’ bit-strings can be evaluated by the selection module. A set of strings of length zero evidently cannot vary, but they can still ‘compete’ via drift. You can still replicate and remove strings of length zero from a population.

But you could learn something about evolution by observing the behaviour of such populations, particularly inevitable coalescence of ancestry, before building up to something more elaborate. That’s the whole point of modelling, to reduce to essentials then reconstruct. The behaviour of finite replicating populations with no selection, mutation or recombination tells you a lot about the role of replication.

(I do know that such organisms do not actually exist…).

Lizzie could easily have started from a string of length zero. If no possibility of extension existed, it would not work. But given a proportion of mutations that add bases (just like reality), the nature of her fitness function would mean that strings would simply extend forever. But if a ‘length cap’ were also part of the fitness function – the longer a string, the more likely to break and die, say – then provided it was not so punitive as to to disallow 500-bit strings, a 500-bit string with product > 10^60 could easily be generated, even from a null string. More generally, the GA would be expected to converge on the highest product available to strings of a length at or just below a ‘breakiness threshold’ – an ‘optimal’ length where the reward for higher products is counterbalanced by the penalty for length.

Try it.

83. To onlooker (at TSZ):

Any fitness function in any GA is intelligent selection, and in no way it models NS.

Please, do not consider any more that statement. Keiths is right, it was a wrong generalization.

Thank you for saying that. I appreciate your willingness to admit error.

d2) NS is selection based only on fitness/survival advantage of the replicator. The selected function is one and only one, and it cannot be any other… IOWs, NS is not flexible (it selects only for a very tiny subset of possible useful functions)…

Fitness functions measure and reward fitness, by definition. But there are many, many, fitness functions, not just one. We have to select the right fitness function for the application.

A fitness function that measures and rewards whiteness is fine if we are modeling a scenario in which whiteness contributes to survival and reproduction, as it does in the evolution of arctic hares. Not so much if we are modeling the evolution of tortoises, or running a GA that optimizes antenna designs.

A fitness function that rewards shell strength is fine if we are modeling a scenario in which shell strength contributes to survival and reproduction, as it does in the evolution of tortoises. Not so much if we are modeling the evolution of arctic hares.

A fitness function that rewards “product of run lengths in a sequence of 1′s and 0′s”, as in Lizzie’s example, is perfectly acceptable if we are modeling a hypothetical world in which “product of run lengths” contributes to survival and reproduction. Not so much if “primeness of run lengths” is the true criterion.

These are all Darwinian processes. The point of Lizzie’s example is to show how a Darwinian process can solve a problem (and generate dFSCI) without any information from the fitness function other than “better” or “worse”. Real world Darwinian evolution can also solve problems (and generate dFSCI) without any information from the environment other than “better” (you survived and produced lots of viable offspring) and “worse” (you died early or failed to reproduce for some other reason).

Your claim that there can be “one and only one [fitness function], and it cannot be any other” is false. There are many possible fitness functions. Some of them lead to the production of dFSCI, others don’t. The question is not whether it is legitimate to use other fitness functions. The question is whether a particular fitness function is legitimate for the scenario being modeled.

84. keiths: Suppose Lizzie’s program initialized the genomes to all 0?s. Then there would be no potential solutions among the initial genomes

Mung: That is false. Again, you demonstrate that you don’t understand what is being discussed. They would still be a potential solution. Just not a good solution. Just not an actual solution. A string of 500 0′s is still in the search space.

But I was reminded of a challenge I had issued. That challenge consisted in setting all strings to the same initial value, rather than having them randomly generated.

So please, have Lizzie initialize all her starting population of strings to all 0′s. By all means. Let’s see how well it performs then.

Mung – I thought you’d written your own version that took 10 seconds? Surely you could try the amendment yourself.

Here’s my prediction: a uniform starting population of all-0′s will be little different from a completely variable one in its ability to search the space. They will initially all be the same, and all products will be zero, therefore the fitness function will have nothing to select on, therefore the population will simply ‘drift’, replicating the same monotonous point in space. But if mutation occurs, variation will arise, the fitness function gains traction, and the ‘random (stochastic) walk’ has taken its first baby steps.

As I have said elsewhere, you could start with one digital ‘organism’ of bit-length zero and still find the peak, provided the mutation method includes something that can increase the number of bits in a string – a biologically ‘real’ amendment.

This is the relevance of ‘being able to write a GA’ to evolution. Or more so, to run one and play with it and see what happens when you fiddle with the various subroutines – the selection, mutation, recombination methods – and their parameters. They are all at least intended to duplicate the ‘real’ processes of the evolutionary synthesis. If they don’t, you need to be able to explain to the profs using them why they are barking up the wrong tree – and you can’t do that if you don’t even know what you are talking about. Give it a whirl – twiddle the knobs; turn them up, down or completely off – it won’t bite you.

85. So please, have Lizzie initialize all her starting population of strings to all 0′s. By all means. Let’s see how well it performs then.

Mung – I thought you’d written your own version that took 10 seconds? Surely you could try the amendment yourself. Here’s my prediction: a uniform starting population of all-0′s will be little different from a completely variable one in its ability to search the space. They will initially all be the same, and all products will be zero, therefore the fitness function will have nothing to select on, therefore the population will simply ‘drift’, replicating the same monotonous point in space. But if mutation occurs, variation will arise, and the fitness function gains traction.

On the off chance that Mung doesn’t want to try this with his own code for some reason, I just ran the test. If you get him to make a prediction, I’ll share my results and see who of the two of you is closest.

86. Mung:

Again, you demonstrate that you don’t understand what is being discussed. They would still be a potential solution. Just not a good solution. Just not an actual solution.

A string of 500 0′s cannot be a solution. Something that cannot be a solution is not a “potential solution.”

A string of 500 0′s can be mutated over time into a solution, but we know without a doubt that a string of 500 0′s is not a solution. It’s not a “good solution.” It’s not an “actual solution.” It’s not a “potential solution”. The only kind of solution it is is a “non-solution.”

A string of 500 0′s is still in the search space.

Obvioulsy [heh], and if that’s all you meant by “potential solution” then I would have no objection. However, you clearly think that information has to be smuggled into the initial genomes, as your full statement reveals:

There is more to a GA exploring certain kinds of digital space than differences due to relative fitness (usually defined by a fitness function or map), mutation and (optionally) recombination.

For example, potential solutions must be encoded into a “chromosome.”

Encoding potential solutions into a chromosome implies there is a problem to be solved.

Information about which potential solutions are more likely to solve the problem must be implemented. [emphasis mine]

This is reinforced by your challenge:

But I was reminded of a challenge I had issued. That challenge consisted in setting all strings to the same initial value, rather than having them randomly generated.

So please, have Lizzie initialize all her starting population of strings to all 0′s. By all means. Let’s see how well it performs then.

I’m very curious. Why do you think it would be a problem if the initial genomes were set to all 0′s?

Have you thought this through?

87. Mung: Darwinian evolution does not need “the right fitness landscape” to work. (What would a “wrong” fitness landscape look like?)

The vast majority of conceivable landscapes are not amenable to evolutionary algorithms, such as highly chaotic or random landscapes. Landscapes that are amenable to evolutionary algorithms usually exhibit local structure. Indeed, some IDers argue that protein fitness landscapes are too rugged for evolution to be effective.

88. Mung: GAs work with a coding of the parameter set, not the parameters themselves.

Well, yes. That’s the genetic part of genetic algorithms, which are a subset of evolutionary algorithms. So? What did you think it meant?

89. Mung,

You issued this challenge:

But I was reminded of a challenge I had issued. That challenge consisted in setting all strings to the same initial value, rather than having them randomly generated.

So please, have Lizzie initialize all her starting population of strings to all 0′s. By all means. Let’s see how well it performs then.

If, as you claim, you don’t think that Lizzie smuggled information into the genomes by initializing them randomly, then what is the point of your challenge?

After you answer, go ahead and take your version of Lizzie’s program, set the initial genomes to all 0′s, and let us know what happens. We’ll see how it compares to Patrick’s results.

It should be amusing.

P.S. I see you’re also confused about fitness landscapes. Here’s a thought: Wouldn’t it make sense to learn about evolution and GAs before condescending to people who actually understand them?

90. Mung: “However I freely admit it is not an exact duplicate of Lizzie’s program just written in another language, I rather attempted to capture the “spirit” of what she built. “

You don’t understand what Elizabeth is trying to do, do you?

Instead of coding right away, why don’t you just describe, in English, what you think you are attempting to do with your exercise.

91. Weclome DrBot

Apologies. Your comment, like everyone’s first comment, was held in moderation and I only just spotted it. Where’s Lizzie!

92. Mung quotes Allan Miller:

While the ‘replication’ function of biological replicators is a vital part of the string, that role is taken by the copy method in a GA, so the strings themselves don’t actually need to consist of anything at the start. The point of bringing them up is to point out that such strings are not likely to be ‘solutions’ to any worthwhile GA, so you aren’t necessarily ‘pre-seeding’ the population with anything.

So consider the zero-length digital organism as the absolute minimal replicator common to all GAs. As long as a method exists that occasionally adds random bits to a string, something will soon emerge, and variations between these ‘non-null’ bit-strings can be evaluated by the selection module. A set of strings of length zero evidently cannot vary, but they can still ‘compete’ via drift. You can still replicate and remove strings of length zero from a population.     [Emphasis Mung's]

Mung, the GA expert, then presumes to ‘correct’ Allan by quoting from a book:

: Introduction to Evolutionary Computing

The choice of representation forms an important distinguishing feature between different streams of evolutionary computing. From this perspective GAs and ES can be distinguished from (historical) EP and GP according to the data structure used to represent individuals. In the first group this data structure is linear, and it’s length is fixed, that is, it does not change during a run of the algorithm.    [Emphasis Mung's]

Mung apparently hopes that none of us have heard of Google. Google ‘GA variable length chromosome’ and you get page after page of links to authoritative descriptions and discussions of GAs that fit the bill.

Oops.

93. You’re right, Toronto. Mung has no clue.

Two days ago, he announced his version of Lizzie’s program with typical Mungian bombast, even grandly labeling his program ‘Mung World’:

Mung World

ok, so I created my own version of Lizzie’s program.

took less than 10 seconds
1522 generations

What’s the big deal?

Allan and Patrick called Mung’s bluff, asking him to respond to his own programming challenge. Suddenly the bluster turned into excuses, apologies and requests for advice:

Mung World

I tossed my program together in a short evening. I am actually rather pleased with it, I even managed to make it object-oriented (for the most part).

However I freely admit it is not an exact duplicate of Lizzie’s program just written in another language, I rather attempted to capture the “spirit” of what she built.

It’s a bit rough around some of the edges, but I would like suggestions on how it can be improved.

I call my digital organisms LiddleLizzards, in honor of Elizabeth.

Here’s my LiddleLizzard class. I think the first thing that can use improvement is the mutate method, it’s pretty rough. .

# mutates this chromosome
def mutate
chromosome[rand(500)-1] = ’1′
end

All I do here is set one position in the chromosome to a ’1′. If it’s a zero it gets changed, if it’s a ’1′ it’s like a neutral mutation. I don’t know what that cashes out to in terms of a mutation rate, if someone wants to tell me.

Some potential modifications:

1. Set the chosen locus to either a zero or a one, that would not be too difficult to code.

2. Explicitly set the mutation rate.

3. Create a Mutation object that is passed in when the digital organism is created that encapsulates it’s mutation parameters.

4. Pass in the length of the string to generate rather than hard-coding it in a constant.

Honest evaluation, criticism, and suggestions for improvement are welcomed. You can leave comments at that link as well.

I had a look at the code for Mung’s LiddleLizzard class. It’s atrocious, and it bears no resemblance to Lizzie’s program.

His ‘mutate’ method sets a random bit in the chromosome to 1. It never sets bits to 0. That’s right — Mung’s program latches! KF will be apoplectic.

Mung’s fitness function looks for the longest sequence of consecutive 1′s in the chromosome. The length of that longest sequence is the fitness value. That’s it. No kidding.

That’s just the class definition. I’d hate to see the rest of the code.

Either Mung has absolutely no idea what Lizzie’s program does, or he doesn’t know to code. Or both.

94. Darwinian evolution does not need “the right fitness landscape” to work. (What would a “wrong” fitness landscape look like?)

Your problem, keiths (and apparently the problem of a few others over there at TSZ), is that you don’t know what a fitness landscape represents.

We’ll need to clarify what is meant by “landscape.” To me a fitness landscape isn’t something that is there waiting to be discovered (or climbed, ala Mount Improbable), it’s something that is created as populations evolve.

I see you’re also confused about fitness landscapes. Here’s a thought: Wouldn’t it make sense to learn about evolution and GAs before condescending to people who actually understand them?

It wasn’t entirely clear what sort of landscape(s) he was talking about, So I decided to wait and find out. You, otoh, plow ahead unabated.

95. Mung:

Think about what a fitness landscape for a 64 bit encryption key would look like – you have 18,446,744,073,709,551,615 possible key values and only one of them is right.

The gives a binary fitness result of either 0 or 1. All a GA will do with a landscape like this is drift – any mutation that doesn’t produce the correct key value will result in a fitness of zero. Until the result is found (and the program terminates) all population members will have the same zero fitness value, all will reproduce with equal probability, so there will be no differential reproduction.  All you get is random drift – you might as well use a random search because in a search space like this a GA will do no better than just random sampling.

96. Mung – here is a fitness landscape for a 10 bit key (1024 possible values) – visualised in two dimensions.

00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000100000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000

What is the most effective method of navigating this fitness landscape?

97. I think we should all encourage Mung in his endeavor. He is at least trying to understand what Lizzie and a few others here did. That’s pretty rare at UD.

Mung, you do need to modify the mutation function to switch a bit randomly  in either direction.

I am also not sure how your fitness is defined. Not knowing Ruby, I can’t parse this piece of code

# calculates the “fitness” of this chromosome
def fitness
score = 0
chromosome.scan(/1*/).each do |str|
score = str.length if str.length > score
end
score
end

98. Mung: We’ll need to clarify what is meant by “landscape.”

A fitness landscape represents relative reproductive fitness. It can represent a real-life situation, such as protein function, or be an abstraction.

Mung: To me a fitness landscape isn’t something that is there waiting to be discovered (or climbed, ala Mount Improbable), it’s something that is created as populations evolve.

Certainly, real biological landscapes change, but static landscapes are often sufficient for particular studies, such as protein evolution. Keep in mind that even though the fitness landscape may be static in a simulation, the total environment changes due to competition with neighbors. In any case, it makes sense to start with static landscapes, but the basic behavior is often similar with dynamic landscapes.

Mung: The way I understand Zachriel’s argument is that he is appealing to a bag or assortment of pre-existing components (aka protein domains) that can be used in proteins and that their availability for use somehow lends less of a random character to the process (making a functional protein more likely) even though the main proposed mechanism for this shuffling is recombination, itself a random process and the protein domains themselves also arose largely as a result of a random process (perhaps “guided” by “natural selection”).

Not quite. The assumption is that splice and insertion points on existing replicators are random, not along functional divisions. Even then, it’s easy to show that the chance of successful results can be millions or billions of times greater than using randomized sequences.

Mung: Personally I have no conflict with regular repeated processes going on inside living organisms because to me that smacks of teleology.

There are many types of recombination (e.g. sexual), however, we’re concerned with random processes.

Mung: “In the first group this data structure is linear, and it’s length is fixed, that is, it does not change during a run of the algorithm.”

Yes, that’s a common type of genetic algorithm, but certainly not the only one, otherwise, you couldn’t simulate genomes that change size during evolution.