Subjects: Evolutionary computation. Information technology–Mathematics.
… the authors establish that their mathematical analysis of search applies to models of evolution.
I have all sorts of fancy stuff to say about the new book by Marks, Dembski, and Ewert. But I wonder whether I should say anything fancy at all. There is a ginormous flaw in evolutionary informatics, quite easy to see when it’s pointed out to you. The authors develop mathematical analysis of apples, and then apply it to oranges. You need not know what apples and oranges are to see that the authors have got some explaining to do. When applying the analysis to an orange, they must identify their assumptions about apples, and show that the assumptions hold also for the orange. Otherwise the results are meaningless.
The authors have proved that there is “conservation of information” in search for a solution to a problem. I have simplified, generalized, and trivialized their results. I have also explained that their measure of “information” is actually a measure of performance. But I see now that the technical points really do not matter. What matters is that the authors have never identified, let alone justified, the assumptions of the math in their studies of evolutionary models.a They have measured “information” in models, and made a big deal of it because “information” is conserved in search for a solution to a problem. What does search for a solution to a problem have to do with modeling of evolution? Search me. In the absence of a demonstration that their “conservation of information” math applies to a model of evolution, their measurement of “information” means nothing. It especially does not mean that the evolutionary process in the model is intelligently designed by the modeler.1
I was going to post an explanation of why the analysis of search does not apply to modeling of evolution. But I realized that it would give the impression that the burden is on me to show that the authors have misapplied the analysis.2 As soon as I raise objections, the “Charles Ingram of active information” will try to turn the issue into what I have said. The issue is what he and his coauthors have never bothered to say, from 2009 to the present. As I indicated above, they must start by stating the assumptions of the math. Then they must establish that the assumptions hold for a particular model that they address. Every one of you recognizes this as a correct description of how mathematical analysis works. I suspect that the authors recognize that they cannot deliver. In the book, they work hard at fostering the misconception that an evolutionary model is essentially the same as an evolutionary search. As I explained in a sidebar to the Evo-Info series, the two are definitely not the same. Most readers will swallow the false conflation, however, and consequently will be incapable of conceiving that analysis of an evolutionary model as search needs justification.
The premise of evolutionary informatics is that evolution requires information. Until the authors demonstrate that the “conservation of information” results for search apply to models of evolution, Introduction to Evolutionary Informatics will be worthless.
1 Joe Felsenstein came up with a striking demonstration that design is not required for “information.” In his GUC Bug model (presented in a post coauthored by me), genotypes are randomly associated with fitnesses. There obviously is no design in the fitness landscape, and yet we measured a substantial quantity of “information” in the model. The “Charles Ingram of active information” twice feigned a response, first ignoring our model entirely, and then silently changing both our model and his measure of active information.
2 Actually, I have already explained why the “conservation of information” math does not apply to models of evolution, including Joe’s GUC Bug. I recently wrote a much shorter and much sweeter explanation, to be posted in my own sweet time.
a ETA: Marks et al. measure the “information” of models developed by others. Basically, they claim to show that evolutionary processes succeed in solving problems only because the modelers supply the processes with information. In Chapter 1, freely available online, they write, “Our work was initially motivated by attempts of others to describe Darwinian evolution by computer simulation or mathematical models. The authors of these papers purport that their work relates to biological evolution. We show repeatedly that the proposed models all require inclusion of significant knowledge about the problem being solved. If a goal of a model is specified in advance, that’s not Darwinian evolution: it’s intelligent design. So ironically, these models of evolution purported to demonstrate Darwinian evolution necessitate an intelligent designer. The programmer’s contribution to success, dubbed active information, is measured in bits.” If you wonder Success at what? then you are on the right track.
The Series
Evo-Info review: Do not buy the book until…
Evo-Info 1: Engineering analysis construed as metaphysics
Evo-Info 2: Teaser for algorithmic specified complexity
Evo-Info sidebar: Conservation of performance in search
Evo-Info 3: Evolution is not search
Evo-Info 4: Non-conservation of algorithmic specified complexity
Evo-Info 4 addendum
Because it can’t speak, Mung.
They do.
It’s hard to even guess what the hell you mean by “ask”. Do you mean we literally speak to some organism, or a stretch of DNA, to get it’s effect on reproductive success?
For all I know astronomers don’t go weighing celestial bodies. They calculate their mass by observing how they move in space.
Orbital models are vastly different, you can just assign any mass to any number of planets in the model, and place them wherever you want, and then simulate their movement.
Why aren’t phoodoo and Mung pestering astronomers at some other forum? Somebody needs to let them know you can’t have orbits without a programmer’s input. If you need to locate celestial bodies in a computer model to simulate their orbits, who put the actual celestial bodies where they are?
Astronomy is full of materialistic scientists that don’t want you to know about these obvious facts. Put on your Sunday tin foil hats!
Yes. The allele makes it to the host organism’s offspring. In that sense, the allele has offspring.
Some alleles make it into more descendants than others. So they have higher fitness than others.
Fuck I never thought about it like that. I’m now aware that all programs programmed by humans, were programmed by humans. FUCK. Time to get sandals and prayer mats guys!
Tom raises an interesting point that the concept of fitness to someone doing evolutionary computation is not the same as an evolutionary biologist’s. Which is true – people doing evolutionary computation are, when it comes to evolutionary theory, laymen (me too, of course). They may have a view more informed by popular presentations and the way they deal with selection and fitness, which goes for Creationists too.
Rumraket and I are trying to relate the evolutionary biology concepts to a digital setting – prompted by this idea that evolutionary computation is nothing to do with evolution. I mean, why would people care? Oh yeah, evolution. Must be defeated in its many halls. There’s a payoff, apparently.
Where it becomes weird is the sight of Mung wiping his eyes in thrilled mirth at some phoodoo-inanity that an organism that has not reproduced yet has not yet ‘got any fitness’. I mean, I laugh too, but not for the same reason. They understand evolution perfectly well, thank you very much. Yet the simple idea that a measurement can be made whose value can’t be anticipated – in any other arena, a totally unexceptionable notion – becomes an opportunity for extensive thigh-slapping. .
You are speaking in generalities about how many GAs are used in practice, rather than how some particular GA can be made and run. And it’s not that I disagree, the concept of a “useful” behavior for some simulated entity is defined beforehand in most GAs(most, not all). I already affirmed all this back in this post.
A counterexample to your generalization about GAs here is Avida. The Avida program can be set up without there being any imposed rewards (such as resources or being copied to the next generation) to certain behavior. Nothing has been pre-determined by the programmers to be “what we’re looking for” or to be a “useful quality”, yet the program will manifest evolution with natural selection and differential reproductive success will spontaneously emerge.
This is why Avida is argued to constitute an actual example of evolution, rather than a simulation of it.
But you literally just said it is the other way around in nature.
Your own words “That differs entirely from the concept of fitness in nature. The concept of fitness in nature works as the exact opposite, we don’t look for that quality, we look at existence, and then consider the quality.”
Which raises the question why the hell IDists believe what they do when nature apparently works in the opposite way. It is obvious you’re really just engaged in question-begging reasoning.
Yes, exactly.
I honestly don’t really care how people who program GAs define fitness in their program. There is a definition of fitness from biology, and regardless of how some programmer has decided to define it in code by literally naming some function as “set_fitness” or whatever, the biological definition can still be used to make sense of what happens in GAs. And as such, what happens in some GAs or how GAs work can be thought of as analogous enough to the real world biological evolution for it to be informative about the process.
If there is something we are in doubt or want to know about, regarding real evolution, we can some times test it in a simulation. Or better yet, we can literally make another instance of evolution and run it on a computer.
Denying microevolution, essentially. If [offstage implementation of fitness differential] causes one allele to enter a mean 1.005 offspring to the other’s 1.000, this can have no causal effect on relative frequencies, because … ?
It’s worth noting that the windmill example is not just an abstract example.
What human designer would have created this: https://www.youtube.com/watch?v=VHLI343ssug
It’s also interesting that the physics in the model was wrong, the viscosity of air was incorrectly specified which meant that the physical versions did not work as well as they might. It’s almost as if “fine tuning” is irrelevant in such processes, a way will be found but it’ll be a different way depending on the actual values.
And there’s always this classic: http://boxcar2d.com/ where fitness is simply distance traveled/time. Simple and easy. And that leads to the more complex quickly: https://www.youtube.com/watch?v=HgWQ-gPIvt4
Right, the concept of fitness, THAT I TOTALLY DISAGREE WITH!
We don’t just happen to exist in nature, and by good fortune it has quality. That to me is total bullshit. There is no way the human brain just happen to come about in nature, and nature found it to be a useful mistake.
That you don’t even get that IDist don’t agree with your materialist explanations for the existence of useful complexity shows just how poor your understanding of the problem is, and should at least give you pause to rein in your own self-righteous assurances of your position. If you were self-aware enough that is.
phoodoo,
Argument by “that’s bullshit”. Fair play to you for being honest.
Feel free to provide your alternative.
First those IDists would have to understand those explanations!
But as I said, I’m open to alternative explanations, even non-materialist ones. But for some reason those that claim to possess such explanations never seem to be in a position to actually elucidate them.
phoodoo,
Is your position then that ID is a non-materialist explanation? My understanding is that ID takes no position on the identity of the designer. Presumably for you the designer of ID is god?
I’m glad that you’ve found another way to just assert in a caricature form what it is you don’t believe. Not that it’s an argument, it’s just the act of venting an opinion with a bit of scorn. Well done. I wish all science was that easy.
Sorry, but what we’re missing here is WHY there “is no way” the human brain evolved. If all you’re going to come up with here is just another way of putting together a sentence full of worlds like accident, errror, mistake and the like, don’t bother. The volume of the vocabulary you use to describe what it is you don’t believe, doesn’t at all tell us what is really the problem.
Nor does it actually constitute an argument, that you can stuff a sentence full of words like ‘accident’, ‘error’, ‘mistake’, ‘unintended’, ‘copying catastrophe’, ‘unguided’, “just happened” or whatever else mocking words you are emotionally compelled to signal your disapproval with.
Because that is all you ever do, phoodoo. You inform us of your OPINION with words you have chosen for their emotional content. You never actually get around to ARGUING, as in showing using valid logic, WHY the evolution of some particular entity, should be implausible.
Saying “There is no way the human brain just happen to come about in nature, and nature found it to be a useful mistake” is merely and nothing but an expression of an opinion. Nothing more. There is no why there.
WHY are you so convinced that there is “no way” the human brain “just happened to come about in nature, and nature to found it to be a useful mistake”?
That whole sentence is what you use to refer to the process of evolution. Another good indication that this whole topic is to you ruled almost entirely by emotion. You can barely get yourself to even use the word, you’d rather transform it into a caricature you can flesh out in a much longer sentence.
We get it, you don’t believe it and you are compelled to mock and caricature it at every opportunity. You’ve been doing it for years.
You never get to the WHY, though. Will that day ever be forthcoming?
I have never not got that. What I’ve been hoping for was something more than opinion. Finding new and ever more disapproval-signaling ways of telling us THAT you don’t believe it, has no more worth, than if I were to tell you that I DO believe it. Neither of those would tell us which one of us is right. Or why we believe as we do.
Because no matter what target you set in the weasel, you only get wind-mills! never brains!11!!1one!!
LMFAO
LoL.
What he said.
So biological evolution works and intelligent designers such as ID proposes are not needed to find things that are sufficiently good? I feel like we’re getting somewhere!
Yeah, that doesn’t apply to Avida.
Nor should it. Let’s see if I can finish (kind of) what I started this morning…
All, not most. You seem to be using genetic algorithms as an umbrella term for computer programs that have something or another to do with evolution. Not even the umbrella term evolutionary computation (covering GAs) covers that much. Evolutionary computation is a general approach to optimization (with search for a categorically good solution as a special case). Evolution is not an optimization process. An analysis of the performance of an evolutionary computation in generating a solution to a given problem is not applicable to models of evolution. When a modeler specifies an evolutionary process along with an event that tends to occur in the process, s/he is not indicating that the process generates a solution to a prespecified problem. The modeler provides an account of how the event may occur, and is in fact saying that the event will not occur unless the circumstances are right. The modeler does not use the evolutionary process to generate a solution to a given problem.
(I have a reason or two or three for highlighting, at the top of the post, the cataloging information from the Library of Congress.)
You’re not saying that search and modeling are the same. But… Given that evolutionary informatics is misapplication of engineering analysis (applicable to GAs) to scientific models of evolution (including Avida-based models), I have to play language cop here.
A common characterization is, I’m sure you know, “open-ended evolution.” Marks et al. do not use the term, but they certainly do recognize the difference between Tierra and the Avida-based EQU study. (As I noted way up the thread, they refer to the particular study as Avida. They give no hint that there have been various studies of open-ended evolution since Tierra.) Even when the Avida environment is customized to provide a model of “The Evolutionary Origin of Complex Features,” it’s wrong to interpret what the investigator does as design of an evolutionary process to generate a solution to a problem (determined independently of the process).
[I’m cutting this off without herding my catlike thoughts into the corral. I guess I’m saying that what looks like evolutionary search to the Two and a Half Engineers is not evolutionary search: the modeler specifies the event (Avidians calculating EQU) jointly with the process. The ostensible problem is not specified independently of the evolutionary process, i.e., it’s not given.]
I don’t recall that Pennock restricted the claim to open-ended evolution on the platform. The key point, I think, is that an Avidian is a digital organism that acquires by metabolism the energy required to reproduce itself. So, irrespective of whether the environment is customized for some particular study (e.g., origin of EQU), evolution occurs in lineages of reproducing organisms.
And also this bit he said?
No generations in the Avida EQU experiment. No assignment of fitness (though fitness can be calculated).
Do you believe that last paragraph, Mung?
If one can get past one’s Dawkinsophobia, one might be interested in his chapter 10, ‘An Agony In Five Fits’, in The Extended Phenotype*** in which he discusses the many opportunities for talking-past-each-other available in use of the term, as we are ably demonstrating here. It’s rather like ‘random’. One Of Those Words.
*** I hope there’s no copyright issue in linking this.
Alright, I understand now. And you’re right, I did use the term GA as an umbrella term referring to basically any software that incorporates some aspect of an evolutionary process.
I understand and I agree the distinction is important. I did understand that there was something very different between Avida and lots of other GAs I know of, but I was not aware that technically Avida isn’t actually a GA.
Of course you weren’t aware. Because this is what you do, talk out of your ass, and then tell those who don’t agree with you that they don’t know what they are talking about.
You might want to check your ego with the coat clerk.
I’m happy that you’ve found this opportunity to gloat. I don’t know why though. I was corrected on something and accepted the correction. And it has zero relation to the fact that you’re still spewing nothing but nonsense regarding the concept of fitness as applied to GAs or other forms of evolution-related software.
When you’re done maybe you can get back to answering this post?
Rumraket,
Yes, a theme emerges. Make no mistakes while phoodoo is about. Soon as someone else has pointed it out, he’ll be straight on it!
Fits Marks, Dembski, and Ewert perfectly. But they, unlike Rumraket, never retract.
Can I buy the book yet?
I missed an opportunity to gloat?
It’s also not a simulation. It’s a platform for running simulations. 🙂
If I say I do can I later recant?
I think it’s a bit sloppy to talk about Avida if what you really mean is a particular experiment run using the Avida platform, but I think it’s a forgivable transgression.
If you’re talking about the math I haven’t a clue.
But I can tell you this, the experiment to derive EQU from NAND sure as hell wasn’t designed to fail. You have to agree that it fell within the search space with a reasonable probability of being located. IOW, for that particular experiment we are in fact talking about a targeted search.
So why couldn’t they have used a NOR gate? keiths?
ETA: When I claim that ‘Avida is rigged’ that’s what I am referring to.
I just have to laugh at how many people have been suckered by this statement.
If you know you can get from A to Z, then it’s just a matter of finding a path from A to Z. (Let’s ignore the question of whether it’s possible to get from A to Z.) And finding a path from A to Z is a SEARCH problem. FFS.
Can I repeat myself?
https://en.wikipedia.org/wiki/NAND_gate
Anyone but Rumraket and Allan Miller think they pulled the NAND gate out of a hat? That’s design, not, “it just happened, that’s all.”
Me neither.
Can you give us an example of “watching a genotype”?
So, Allan, how would you rate the fitness of your “get_fitness” function?
For example, knowledge about NAND gates and what can be constructed from them as opposed to other available alternatives that were deliberately (as in, by design) excluded.
That’s right. Which makes Joe’s reply to phoodoo a non-sequitur. Deceptive even. Did Joe know what he was doing? I believe so. Joe isn’t dumb.
You provided us with a table of alleged “fitnesses.” You claimed they were fitnesses of “genotypes”, but that hardly make sense.
If they weren’t the fitnesses of alleles then I retract what I wrote. I associate the notation you used with alleles and not genotypes.
Because you have an opinion f yourself that is far too inflated and an opinion of others who disagree with you that is far too dismissive.
Tom’s not telling you anything that you haven’t heard before.
Do you dispute their Table 6.3? Do you dispute they are referring to a specific paper?
Well, first you look at, say, a finger. And if it exist, that is good, so you give it perhaps a 1. Then you look at something else, like maybe an elbow, and if there is none, that is less fit, so you give it perhaps a zero. But if you have like 7 elbows, that is much more fit, so that would get a high number, a very high number, because it is much more than average (unless it is a spider, which is still good, but not great).
But, it does get a little more confusing, now that we think about it, because fish bring the averages down. Way down. And sea urchins, they are making all of us less fit. No, make that they are making us more fit, because any elbow is better than a sea urchins.
But my stinger average is anemic-dam you urchin!
You’re in danger of helping me figure out how to express myself more clearly, rather than repeat myself. One acknowledgement from me in an OP, and you’re dead (virtual) meat at UD. 🙂
Lenski et al. used Avida to provide a model of “The Evolutionary Origin of Complex Features” (the title actually does mean something), not to solve a given problem. If the default instruction set had included NOR instead of NAND, then they would have addressed the origin of XOR (at least 5 NOR operations) instead of EQU (at least 5 NAND operations).
This illustrates how the event specified in the model depends on the evolutionary process specified in the model. The set of all Avidians performing the XOR function [oops — it’s the EQU function when NAND is in the instruction set] is not specified independently of the evolutionary process. That is, Lenski et al. did not start out with a problem to solve, and then design an Avida environment to generate a solution.
I’ve just explained why we are not talking about a targeted search. The event that you’re mistaking for a target is not specified independently of the process in which it tends to occur. The model is an account of how a complex function can emerge from simpler functions. It is anything but a claim that “evolution can solve every problem.” EQU-performing Avidians will not emerge unless the conditions are right. Lenski et al. are not saying that the conditions necessarily exist. They are saying that when the conditions exist, EQU tends to emerge. Again, from their abstract:
Marks et al. write something like “Avida is trying to solve a moderately hard problem” in the preface. And you’re evidently thinking the same way. I’m not blaming you for getting that impression. The people who use the efficacy of evolutionary computation in solving problems as evidence that evolution “works” are similarly confused. And I used to be one of them.
What is the definition of complex?
Do you not understand that I quoted from that specific paper, and highlighted text to call your attention to what was most important?
Do you not recognize what I emphasized in my quotation of Marks et al. above as a summary of their Table 6.3?
Do you not feel obligated, if you suspect that my sinfulness has blinded me to the genius of Marks et al., to use the link I provided, and have a look for yourself? You don’t need to read the paper to see that Table 1 (screen shot attached) resembles their Table 6.3. You need only read the caption to see that
Now reread the first quote. Is your impulse to look for a way to make Marks et al. right, no matter that they are obviously and egregiously wrong? Do you think it was clever for them to say absolutely nothing about energy in Avida (except when they quoted Lenski et al. referring to energy)? Do you think they did better in their published paper, “Evolutionary Synthesis of Nand Logic: Dissecting a Digital Organism“?
Read the paper, and then ask for help if you need it. Then again, you can look at the table above. The NAND complexity of a Boolean function (a concept that existed long before Avida came along) is the number of NAND operations required to compute the function [i.e., when no operation but NAND is available]. The computational merit of a function is 2 raised to the NAND complexity of the function.
Honestly, I don’t understand people arguing about stuff like this for years, and not going to primary sources. Getting whatever you can from a paper is better than ignoring it entirely. Papers written for scientists in general, as is the paper by Lenski et al., are usually quite readable.
Tom English,
No Tom, I don’t need to read some paper, and go down some crazy conspiracy theory trail of yours, to find out what YOU mean when YOU use the word complexity. You didn’t quote anything, you said “complex function can emerge from simpler function.”
So when YOU said that what did YOU mean by complex? If you need to talk about NAND tables and Boolean functions to explain what you mean by the word complexity, I don’t really see how you have any justification for making your statement that “complex functions can emerge from simpler ones” when you are not talking about any real world concepts, and are simpler sputtering obfuscation.
I can answer that. Where do you draw the line between simple and complex? It’s always going to be an arbitrary definition to say that something is complex rather than simple. There isn’t some universal standard of complex handed down from from on high. That means it’s up to the authors who use the term to define what they mean by it.
In general, since there isn’t some universally applicable definition (such as, it is complex if it has 13 parts, and simple if it only has 12), what you can instead says is that a more complex thing has more interacting parts, and/or parts that interact in more ways, than a less complex thing.
This will make sense of the Lenski paper too. When they say that they are explaining the origin of “complexity”, they’re not claiming that there is some universal cut-off where things become “complex” and they were not before. All they mean is that the system in question comes to have a greater number of interacting parts that come together to work, than it had in the beginning. They then try to identify the cause of this phenomenon. They do call the EQU function a “complex feature”(and it is indeed the particular one they are interested in how evolves), but also “the most complex function”, implying it’s not an absolute measure.
The issue is obviously what Lenski et al. mean, not what I mean.
I quoted plenty, and not so many comments ago.
So you have no frame of reference here, Donny. You’re like a child who wanders into the middle of a movie and wants to know…
I’m intrigued by this idea that NAND and boolean functions are not real-world concepts but just obfuscations.