“Natural Selection’s [too limited] Reach”

I hadn’t realised that the Biologic Instute had a diary/blog.

February’s article is by Ann Gauger, although it consists largely of quotations from Douglas Axe, and is called Natural Selection’s Reach.

What continues to astonish me about ID proponents is just how ignorant they are of evolutionary theory.  Ann Gauger starts by setting out to address a reader’s query:

A reader wrote us recently to ask why natural selection can’t extract enough information from the fitness landscape to explain complex features.

Well, obviously I dispute the premise of the question, but assuming that the reader and Gauger share the view that natural selection’s “reach” is too limited to “explain complex features”, let’s see how Gauger explains her stance.  She starts by rightly saying that:

It all depends on what you think the fitness landscape looks like

and then lets Axe go on to explain further.  Unfortunately, Axe appears to think it looks like this:

with complex features situated on the high distant peaks, and natural selection only able to move a step at a time, and only upwards.

If this were actually a realistic model of the fitness landscape, of course, natural selection would be doomed.  But what Axe has presented is, at most, a two-dimensional map.  This would be what the landscape would look like if organisms could only vary along two dimensions, say, size and colour. When the population had optimised its size and colour, it would be stuck on the top of one of those little mud ridges, with nowhere to go but down.

Does Axe really think that there are only two dimensions to the fitness landscape of biological organisms?  Or has it simply not occurred to him to extend his metaphor to the harder-to-envisage realm of multidimensional fitness space?

The “limited reach” of natural selection is not the problem – yes, populations generally can only “move” over the fitness landscape step by tiny step. But that doesn’t matter if the landscape consists of many ramps along many dimensions, and the more dimensions there are, the more ramped routes there will be, thus vastly increasing the accessibility of those distant peaks.

Gauger concludes her piece:

What is the mutational reach of natural selection in general? It’s very short. For bacteria, our work suggests the reach is only a few mutations at a time.  That’s not enough to get a genuinely new function for a protein, let alone a new pathway made up of a handful of proteins, or a metabolism made up of hundreds of proteins. For larger multicellular organisms like us, with slower generation times and smaller populations, the problem gets worse. Much worse.

 

So unless someone paved a highway to Mt. Whitney that went uphill every step of the way, Darwin’s engine would never get out of Death Valley. But a paved highway isn’t evolution, it’s design.

Nope. A route that goes uphill (or at least not drastically downhill – drift helps too) every step of the way is highly probable in a multidimensional fitness landscape, and a multidimensional fitness landscape doesn’t have to be designed.  You just need an environment in which there are lots of resources that are exploitable by the kinds of variation possible to biological organisms. And given that we know (and both Axe and Gauger must know, being biologists) that similar phenotypes have similar genotypes, the landscape is not only full of climbing ramps, but is also “paved” – smoothed by the similarities between the phenotypic consequence of similar genotypes.

Honestly, it’s not that I’m “skeptical” of ID – I have just yet to read a pro-ID article, even by the academics in the field, that doesn’t make rookie errors about evolutionary science.  Oh, except for Todd Wood, and he relies on faith, not science, for his belief.

94 thoughts on ““Natural Selection’s [too limited] Reach”

  1. For bacteria, our work suggests the reach is only a few mutations at a time.

    Remarkably busy designer then!

    So is their version of ID “constant intervention” I wonder or just dropping in new components complete in new generations?

    They make claims but don’t explore the absurdities that result from them.

    So unless someone paved a highway to Mt. Whitney that went uphill every step of the way

    But that’s their claim, no? If not that, then what? Presumably they are saying that nothing evolved(s) and everything is created in-place in-eco system and is “as is”.

    I wonder when they think this all happened? And again, remarkably busy old designer they seem to be making the case for.

  2. I am astonished that they still say that

    natural selection can’t extract enough information from the fitness landscape to explain complex features.

    Of course William Dembski gave a mathematical proof of this. But then it collapsed in a heap — it isn’t proven at all. But still many posters and commenters at sites like UD repeat the conclusion. If pressed on what the proof is, they sometimes retreat to saying that it is “an empirical generalization”. Or they say that the information isn’t generated by natural selection, it is already there, front-loaded, in the fitness surface. Which ducks the question and (quietly) admits that if the fitness surface is smooth enough, natural selection can work just fine. Gauger’s statement (one she cites approvingly as said to her by questioners) is that natural selection “can’t extract enough information from the fitness landscape” which sounds like it says that even when the fitness surface is fairly smooth somehow natural selection does not work.

  3. Well, I’m not bothered by how busy the designer had to be. It’s the invoking of a designer to solve a problem that doesn’t actually exist that I take issue with.

  4. For bacteria, our work suggests the reach is only a few mutations at a time.

    Extrapolating from bacteria to the capacities of the entire biological world should be done with caution. It is true that the same is done with the Lenski setup as positive evidence for ‘blind’ evolution’s search capacity, but when people try and scale up Lenski’s rate of change to a typical eukaryote, or extend prokaryotic limitations to a non-prokaryote system, they are pulling a fast one.

    Tying in with points I made in the ‘Phinehas’ thread, what seems to ramp up the capacity of evolutionary ‘search’ is the capacity of genes from different organisms to recombine. In the necessarily restrictive world of Lenski, vertical descent is pretty much the only game in town, because cells aren’t neighbours for long enough, or different enough, to swap genes in any way that will make a difference. When you are a tiny prokaryote, subject to all manner of mechanistic constraints, you very rapidly get to the top of your local optimum. Vertically, your own genome is the sole source of novelty, and you have already pretty thoroughly explored the point-mutational neighbourhood, and various rearrangements thereof.

    Simple LGT mechanisms allow some ‘sharing’ of … uh … information. Antibiotic resistance is the classic example, but an entire genome’s worth of sequences is being worked upon in parallel, and beneficial traits can ‘infect’ the population from anywhere around it, with some independence from whole-genome neutralisation of fitness effects that exerts a drag on vertical mechanisms.

    Added to that, endosymbiosis brings in a whole genome’s worth of sequence, initially encapsulated in the endosymbiont, but gradually transferred to the ‘executive’ host, leaving a small cadre that is better synthesised on site. The driver favouring this transfer would possibly be the minimisation of genetic conflict between two complete genomes fighting over the wheel. But it gives a massive injection of genetic novelty, as well as an ‘adaptive’ boost in terms of the endosymbiont’s actions (mitochondria, chloroplasts, perhaps peroxisomes and hydrogenosomes).

    The crossover of meiosis brings another mechanism by which ‘parallel processing’ by an entire population can network any beneficial sequence. That is not the reason for crossover in meiosis, but it is a massively important side-effect. The ‘discoverer’ of a particular beneficial sequence becomes, through Natural Selection and iterative recombination, the ancestor of that locus in every individual of a future population, its next-door neighbours also likely to be the selected survivors of the various competitions in which they conferred their own benefit.

  5. Joe Felsenstein:
    Which ducks the question and (quietly) admits that if the fitness surface is smooth enough, natural selection can work just fine.

    It doesn’t even have to be that smooth. Recombination can often overcome most regional fitness peaks. A multidimensional landscape just has to be ‘comprehensible’ rather than chaotic.

  6. There are certainly many unsolved problems in evolution – how simple were the simplest Darwinian-capable self-replicators? How did the genetic code emerge? How did cell nuclei arise? How did multicellularity get started?

    And IDists might be perfectly entitled to raise these specific problems. But the bizarre thing is how regularly they raise issues that simply aren’t problems at all, for what seems to be culpable ignorance of evolutionary theory. I was looking at Axe’s “Stylus” paper the other day. He clearly has no clue about how to model the kind of fitness landscape properties we know that fitness landscape of biological organisms to possess. Why not? And Dembski is the same (yeah, I was browsing the meagre files of BioComplexity…), with his claims that evolutionary algorithms “smuggle in” the solution as part of the fitness function, or at least the directions for finding a solution.

    Well, inasmuch as a GA is a tool for finding a solution to a specific problem, obviously the problem is “smuggled in” (in plain sight) – but what if there are billions of problems, and the GA finds solution to a subset of them? Do we say the solution, or the problem, has been “smuggled in”?

    Or perhaps he means (though he does not say so) that the variance-producing mechanisms are part of the solution? Well, they certainly affect the topography of the landscape, but with a rich enough landscape, and enough sources of variance, solutions to survival will be found, without anything being “smuggled in” other than the ability to self-replicate with variance, which is the prerequisite.

    And if that’s their beef, why bang on about fitness landscapes at all?

  7. “What continues to astonish me about ID proponents is just how ignorant they are of evolutionary theory.”

    What astonishes ID proponents is the ignorance of ID theory displayed by it’s critics.

    “I have just yet to read a pro-ID article, even by the academics in the field, that doesn’t make rookie errors about evolutionary science.”

    You’ve read Dembski’s article on Specification. How does he make rookie errors about evolutionary science (whatever that is) in that article?

  8. “Oh, except for Todd Wood, and he relies on faith, not science, for his belief.”

    Todd would is not an ID proponent. And he relies on faith and not science for his belief about what?

    “…academics in the field…”

    Academics in what field? ID theory or “evolutionary science” (whatever that is).

  9. Mung: “What astonishes ID proponents is the ignorance of ID theory displayed by it’s critics. ”

    “ID theory” consists solely of “theory of evolution denial”.

  10. Mung:
    “What continues to astonish me about ID proponents is just how ignorant they are of evolutionary theory.”

    What astonishes ID proponents is the ignorance of ID theory displayed by it’s critics.

    “I have just yet to read a pro-ID article, even by the academics in the field, that doesn’t make rookie errors about evolutionary science.”

    You’ve read Dembski’s article on Specification. How does he make rookie errors about evolutionary science (whatever that is) in that article?

    How would you possibly be able to recognize an error, even if it were shown to you, in a theory you admit knowing nothing about?

    About the only solid notion of ID I’ve found is “that’s not it” repeated ad nauseum to every critic.

  11. Joe Felsenstein,

    Its a good point of common reasoning that natural selection working upon original primitive elements could not produce the fantastic glory of biology.

    yet before all these ideas of what natural selection could or couldn’t do it is the main point that its all speculation.
    There is no scientific methodology behind these evolutionary options for creationists to get our claws into.
    Speculation is speculation.
    Saying selection could do this and that is not scientific biological evidence that it could do this or that or did.
    Some ID folk then try to put math behind the unlikelyness of selection working upon a simple foundation.
    Yet its still all about musings of past and gone events and processes.

  12. Mung:
    “Oh, except for Todd Wood, and he relies on faith, not science, for his belief.”

    Todd would* is not an ID proponent.

    Well, he believes the universe was designed, specifically by the Abrahamic God. He doesn’t think much of the ID arguments, but considers the conclusion correct.

    And he relies on faith and not science for his belief about what?

    Belief that the universe was designed.

    “…academics in the field…”

    Academics in what field? ID theory or “evolutionary science” (whatever that is).

    ID theory.

    *Hey, another homophonic typist! Me too! Come and have a beer!

  13. Robert Byers:
    Joe Felsenstein,

    Its a good point of common reasoning that natural selection working upon original primitive elements could not produce the fantastic glory of biology.

    yet before all these ideas of what natural selection could or couldn’t do it is the main point that its all speculation.
    There is no scientific methodology behind these evolutionary options for creationists to get our claws into.

    There is plenty of scientific methodology, Robert! Oodles of it!

    Speculation is speculation.
    Saying selection could do this and that is not scientific biological evidence that it could do this or that or did.

    But there is plenty of biological evidence, in lab, field, and from artificial selection. We can actually observe natural selection in real time, and do, and yes, it does “do this or that”, i.e. result in populations of individuals better fitted to their current environment.

    Some ID folk then try to put math behind the unlikelyness of selection working upon a simple foundation.
    Yet its still all about musings of past and gone events and processes.

    But the past leaves evidence, Robert. Otherwise we would never discover anything. And certainly would never be able to convict anyone of a crime.

  14. Mung: You’ve read Dembski’s article on Specification. How does he make rookie errors about evolutionary science (whatever that is) in that article?

    Yes, indeed “whatever that is”.

    There are many critiques of Dembski’s work out there, most of them remain unanswered. This is why his work has gone nowhere – anyone who stumbles on it quickly finds the rebuttals.

    http://www.talkreason.org/articles/dembski.cfm
    http://www.csicop.org/si/show/presentation_without_arguments_dembski_disappoints/
    http://www.talkreason.org/articles/inference.cfm
    http://www.talkreason.org/perakm/Demb_Dream_inSkeptic.htm
    http://www.talkdesign.org/cs/theft_over_toil

    But I’ll leave the last word to Dembski

    I’ve pretty much dispensed with the EF. It suggests that chance, necessity, and design are mutually exclusive. They are not. Straight CSI is clearer as a criterion for design detection.

  15. Eric Anderson responds on UD (do come over, Eric, I don’t bite :)), but unfortunately seems to have missed a rather crucial part of my post:

    OK, I finally made my way over to TSZ to see what these “rookie mistakes” were that Lizzie is making much of. I have not looked at all the posts, but I did read one about Axe and Gauger and here is the “mistake” they make:

    Without getting all technical, Axe and Gauger apparently feel that the fitness landscape is punctuated by peaks and valleys, with the result that organisms are likely to get stuck at a fitness peak. This is the egregious error Lizzie sees. She acknowledges that natural selection would be “doomed” under such a scenario.

    Now, this is not a new concept with Axe and Gauger. Indeed, many people have argued that the fitness landscape is as described above.

    Lizzie argues, however, that this is not so. The landscape, according to her is like this:

    And given that we know (and both Axe and Gauger must know, being biologists) that similar phenotypes have similar genotypes, the landscape is not only full of climbing ramps, but is also “paved” – smoothed by the similarities between the phenotypic consequence of similar genotypes.
    (emphasis added)

    Now this is certainly a thoroughly Darwinian view of things harking all the way back to Sir. Charles. Darwin felt that there was an unbroken continuum — slight successive changes — such that the fitness landscape would be essentially flat, like colors merging almost imperceptibly on a color chart.

    Let me repeat, for Eric’s benefit, if he is reading this:

    Does Axe really think that there are only two dimensions to the fitness landscape of biological organisms? Or has it simply not occurred to him to extend his metaphor to the harder-to-envisage realm of multidimensional fitness space?

    The “limited reach” of natural selection is not the problem – yes, populations generally can only “move” over the fitness landscape step by tiny step. But that doesn’t matter if the landscape consists of many ramps along many dimensions, and the more dimensions there are, the more ramped routes there will be, thus vastly increasing the accessibility of those distant peaks.

    And, Eric, far from:

    Unfortunately for Darwin, and unfortunately for Lizzie, that is not at all what we see in nature.

    it is eactly what we see in nature – multiple fitness dimensions along which populations can move. And that is what potentially provides the ramps between peaks that Gauger, Axe, and now Eric, believe are missing. But all three of them, surely, know that there are more than two dimensions along which a real biological population can move. Do they not grasp the math that shows that as dimensionality increases, so the number of potential routes between one peak and another increases? Just as going north may lead you to a peak, going west may take you along a ridge to a higher peak, except that that we are talking hyperplanes here, not your easy-to-visualise 2D compass points. Come on, Gauger, Axe, Eric – do the math!

    We see breaks and leaps and uniqueness and discontinuity. We in fact see organisms occupying disparate points on the map.

    It is possible — logically possible, that is — that these points we currently see are indeed joined together by smoothed, paved, easy-to-access pathways. But that is an assumption, not an empirical fact,

    It’s an empirical fact that there are many dimensions along which a population can move in the fitness landscape, which is another way of saying there are countless ways to skin a cat – to become better adapted to an environment, or to a neighbouring niche in that environment, not just two. If a population has optimised the size and colour of its individuals, so that it is sitting on peak on a 2D map, what’s to stop it optimising its speed, and so ascending a ramp to another a peak along that third dimension? And having achieved that summit, what about optimising tooth size, or digestive capability, or reaction speed, or…? All those features are dimensions of the fitness landscape, and potentially connected by ramps. And what’s to stop a population sitting on a peak, while a sub-population explores different peaks? That’s speciation. Evolution is the story of a vast-dimensioned fitness landscape, full of peaks in one dimension that are potentially connected by ramps in another, with populations sitting on peaks while others move to others. And the height of the peak is not a measure of complexity but of fitness to that population’s environment. We do not sit on a higher peak than, say bacteria. We have are simply on a different one, having got there by (putatively) traversing a very different set of ramps.

    This is why we have to be careful of too-concrete metaphors. The fitness “landscape” is a good one, but if it fools people into thinking that the biological fitness landscape is plotted on a 2D map, then it is seriously misleading, and apparently misleads even intelligent people like Gauger, Axe and Anderson into false conclusions about the limits of evolution. As I said, evolution’s reach is no problem as long as there are ramps, and the fact that organisms can differ in countless ways tells us that there are countless dimensions in which a population may find an upward slope.

    and those frequent oases in the organism’s long and directionless path to that distant and dusty peak are more a mirage imagined by the desperate traveler than an actual point of refuge. Furthermore, there are good reasons to think that many points on the map — from individual biochemical systems to whole organisms — are in fact steep peaks that cannot be so easily traversed.

    No, there are not “good reasons”. There is some poor reasoning that leads to this conclusions, but Irreducible Complexity turns out to be a) not a theoretical bar (because, arches) and b) not a bar anyway (turns out evolutionary processes can result in IC features (because, AVIDA).

    Thus, at most what Lizzie could have objectively said would have been that some people disagree with the idea of a discontinuous landscape and argue for a more level landscape. Fine. But that certainly doesn’t mean that Axe and Gauger have committed some rookie mistake in their description of the landscape.

    Yes, it does, because they consider only a small-dimensioned landscape. If they were aware that their analogy was too-simple, and that by adding countless dimensions on top of their two, the landscape must become more traversable, they should have said. If they weren’t, it was a major mistake.

    Further, Lizzie’s assertion that the landscape is smooth, and therefore easily traversable by natural selection, is nothing more than that — an assertion. It certainly has not been demonstrated.

    There is another issue here. Eric ignored my first, namely the high-dimensionality of the biological fitness landscape, but it is true that one could have a high dimensioned landscape that was nonethe less rugged. For example, it might be possible, having reached an optimum colour and size, for a population to be better off if it had teeth (increase in fitness along a new dimension) but be faced with a valley between it and the tooth-mountain. So the second issue is: how rugged is the multi-dimensioned fitness landscape? Eric says:

    Moreover, not only is a smooth fitness landscape not an empirical fact, it is really just a restatement of the theory: namely, that slight, successive changes can lead to just about anything.

    It thus commits the error of assuming the very thing that needs to be demonstrated. And that is something we (rookies and all) quite accurately understand about evolutionary theory.

    Well, no. Again, I think the photographic metaphor has got in the way of comprehension. A “smooth fitness landscape” in practice means a scenario in which a population of self-replicators that replicate with heritable variation in reproductive success (I’ll start abbreviating that to STRHVRS, I think) have the property that novel variants tend to be similar to the parent variant. A scenario in which the STRHVRS did not have this property (A child that differed from its parent more likely to be radically different than slightly different) might still be one with a high-dimensioned fitness landscape, but the landscape would be impossibly rugged. Many combinations of heritable characteristics would result in fitness, but those combinations would be almost impossible to reach by natural selection.

    But it is an empirical fact that biology is like the first scenario (children tend to resemble their parents), not the second. This means that it is empirical fact that the fitness landscape is smooth, just as it is an empirical fact that it is high-dimensioned. In other words there are multiple peaks, along multiple dimensions, connected by smooth ramps.

    If this is the result of design, all the designer had to do is to ensure that the initial population of STRHVRSs had the property of having offspring that varied only slightly from their parents, and that a large range of phenotypic variations would eventually be possible. If ID proponents want to call that “front-loading”, then that’s fine – we still don’t have an OOL hypothesis that yet satisfactory accounts for the simplest STRHVRSs, and therefore we don’t know whether that property was intrinsic, or spookily unlikely. But then let Darwin of the hook – his theory works just fine, given the necessary starting conditions.

    Of course there is still the odd chasm in the perceived fitness landscape – the origins of sexual reproduction, or multi-cellularity, for example, or the origins of DNA itself – but it’s not as though evolutionary science is short of ideas on those scores, nor evidence to support some of them, and if ID is going to cling to the rare chasm in the perceived fitness landscape, then we are back to ID-of-the-shrinking-gaps.

    In the mean time, people like Axe and Gauger would do well to become either more informed, or more honest (depending where the problem lies) about what we don’t just know empirically, but what is blindingly obvious – that biological organisms can vary along many dimensions, and that their offspring tend to resemble their parents, and that therefore the fitness landscape is not low-dimensioned and rugged, but high dimensioned and smooth.

  16. Mung:
    “What continues to astonish me about ID proponents is just how ignorant they are of evolutionary theory.”

    What astonishes ID proponents is the ignorance of ID theory displayed by it’s critics.

    Well, there’s more than one version of “ID theory”, and each one has different flaws, from Dembski’s on the one hand, to Granville Sewell’s on another. And then there’s Upright Biped’s. And I get a bit sick of being told I don’t understand ID theory when I criticise one version by a proponent of another. For instance, Dembski’s case is that Darwinian evolution can’t account for the generation of complexity [on edit] what he calls Specified Complexity. But when that (highly flawed) case is dismantled, frequently someone shifts the goal posts and says “but you don’t understand ID – ID is about how life/the universe/existence must have been designed”. I understand Dembski’s argument pretty well – and I can see (and have frequently pointed out) what’s wrong with it. Others have made the points more eloquently. Ditto with Granville Sewell.

    If neither Dembski nor Sewell’s idea comprise “ID theory” – what does?

    “I have just yet to read a pro-ID article, even by the academics in the field, that doesn’t make rookie errors about evolutionary science.”

    You’ve read Dembski’s article on Specification. How does he make rookie errors about evolutionary science (whatever that is) in that article?

    Probably time for (yet another) thread on this. But that particular paper doesn’t really address evolutionary science at all, except in passing. Dembski acknowledges that to do his proposed calculation (of CSI) he’d have to know the probability distribution for an event under the evolutionary hypothesis. But he doesn’t even hazard such a probability distribution. So it’s true that that particular paper doesn’t make a “rookie error” about evolution per se (though his entire argument falls down if he can’t produce a probability distribution under his Fisherian null, for which he needs the probability distribution under the darwinian hypothesis, and he simply makes no attempt to do so in that paper). Relevant Dembski writings include No Free Lunch, and his papers with Robert Marks. Those do make rookie errors, which Dembski has actually acknowledged (well, not that they are errors, but that Darwinian “search” does actually work,even when the solution is not programmed in as part of the fitness function).

  17. Lizzie,

    Isn’t the Axe argument about the probability of an extant protein evolving into another extant protein? If I haven’t mischaracterized this it seems to be a rather stupid argument.

  18. petrushka:
    Lizzie,

    Isn’t the Axe argument about the probability of an extant protein evolving into another extant protein? If I haven’t mischaracterized this it seems to be a rather stupid argument.

    I think so, yes, but I was responding to that specific article by Gauger. Which is misleading to the point of falsfication, on basic, highly pertinent facts.

    But if Axe’s case is that protein evolution,specifically, is rugged, then that is probably worth a thread of its own. Is he saying that novel variants of protein-coding sequences do not tend to produce similar proteins to their parent sequence?

    Facts would seem to be against him.

  19. OMagain has listed a whole series of refutations of Dembski’s CSI argument. I must (of course) add my own article:

    ncse.com/rncse/27/3-4/has-natural-selection-been-refuted-arguments-william-dembski

    Taken together, these articles show that Dembski’s conservation theorem, his Law of Conservation of Complex Specified Information, is both unproven and it is not formulated in a way that would show that natural selection cannot work to put CSI into the genome. Dembski’s other argument, that the No Free Lunch Theorem prevents natural selection from working better than pure mutation, has also (see same articles) been shown to be a misapplication of the NFLT.

    So when ID advocates say that the observation of CSI in a genome shows the presence of Design, they have no basis for such a statement. The only defense of their statements that has been presented (Dembski and Marks’s Search for a Search argument) does not refute natural selection. Instead it argues about Design being needed to create a universe in which natural selection would in fact work.

    So where is the “ID theory” that validates use of CSI to detect Design?

  20. For those following along at UD: yes, I should have said “specified complexity” above. My claim remains true with the emendation, now made. Specified complexity, as defined by Dembski, can be increased by means of a Darwinian algorithm, as many, including myself, have shown, and as Dembski agrees.

    Dembski however, claims that such algorithms “smuggle in” information. Not in any sense that affects the validity of Darwinian biological evolution, they don’t.

  21. Joe Felsenstein,

    Instead it argues about Design being needed to create a universe in which natural selection would in fact work.

    Ah, you have finally explained what search for a search is about. It is a restatement of the Gospel of John in the language of information theory.

    I wondered what his point might be.

  22. Lizzie,

    Yup. And that refutes the ID types who cite CSI as an indicator that natural selection cannot have made the genome as well-adapted as it is.

    petrushka,

    (Taking your Gospel revelation as a sly quote of Dembski) the difficulty Dembski and Marks face with their S4S argument is that smoothness of fitness landscapes might come by Design, but it also seems predicted by the weakness of long-range interactions in physics, so a gene acting in my toe is unlikely to affect my vision as well.

  23. Maybe, but genes affecting the pigment of your toe can affect your vision. Part of the high dimensional spaghetti code problem faced by the designer.

    One of the reasons I argue that design is impossible., except by evolution.

  24. petrushka:
    Maybe, but genes affecting the pigment of your toe can affect your vision. Part of the high dimensional spaghetti code problem faced by the designer.

    One of the reasons I argue that design is impossible., except by evolution.

    Just to let you know I appreciate your continued hammering of this very important point!

  25. KF notes the definition of CSI:

    [[Specified complexity can be defined:] “. . . since a universal probability bound of 1 [[chance] in 10^150 corresponds to a universal complexity bound of 500 bits of information, [[the cluster] (T, E) constitutes CSI because T [[ effectively the target hot zone in the field of possibilities] subsumes E [[ effectively the observed event from that field], T is detachable from E, and and T measures at least 500 bits of information . . . ”

    He adds

    At this stage, there is no excuse for EL’s strawman misrepresentation. That is evidence of UN-reasonableness, and of speaking with willful disregard to truth that one knows or should know, hoping to profit from the misrepresentation.

    In what way does anyone profit from misrepresentation?

    I’ve often asked KF what biological process is in action when the “needle in a haystack the size of the cosmos” analogy is referenced. To my knowledge he’s yet to connect those dots.

    KF refers us to

    On the design inference and explanatory filter, cf. 101

    Which concludes:

    But instead, we have excellent, empirically based reason to infer that the best explanation for the FSCO/I in body plans, including the first, is design.,

    Where the FSCO definition is given as:

    i] If we find ourselves in a practical cosmos of 10^57 atoms — our solar system . . . check,

    ii] where also, we see that something has an index of being highly contingent I, a measure of information-storing or carrying capacity,

    iii] where we may provide a reasonable value for this in bits,

    iv] and as well, we can identify that the observed outcome is from a narrow, independently describable scope T within a much larger configuration space set by I, i.e. W.

    v] then we may infer that E is or is not best explained on design according as I is greater or less than the scope 500 bits.

    On that basis EL has “willful disregard to truth” yet KF can claim design in biology on the basis of ASCII messages.

    However, if we came to the same string later and saw that the coins somehow now had the bit pattern of the ASCII codes for the first 143 or so characters of this post, we would have excellent reason to infer that an intelligent designer, using choice contingency, had intelligently reconfigured the coins. that is because, using the same I = 1,000 capacity value, S is now 1, and so Chi_500 = 500 bits beyond the solar system threshold. If the 10^57 or so atoms of our solar system, for its lifespan, were to be converted into coins and tables etc, and tossed at an impossibly fast rate, it would be impossible to sample enough of the possibilities space W to have confidence that something from so unrepresentative a zone T, could reasonably be explained on chance.

    Therefore the flagellum was designed?

    Oh, rather the chances of the flagellum component pieces coming together randomly are so low that design seem logical in light of that.

    I suppose I should congratulate KF for making the best he can out of such slim pickings.

    “A thing that you are not claiming happened is so unlikely to happen that even you’ll agree with me that it’s unlikely to have happened”.

    I’ve talked to KF about this many times before but he’s never changed his opinion. But then again, neither have I. But if he was right he’d be able to school me on why his “haystack” is relevant. That’s my take on it anyway.

  26. petrushka:
    It made the list at UD of the stupidist arguments againsr ID.

    Well, it’s not an argument against an omnipotent designer who can poof anything into existence at will, but if we are arguing from analogy from human design, if a human had to design something like a biological organism, the obvious thing to do would be to use some kind of evolutionary algorithm, as indeed, breeders have done for millenia, but also as we develop things like like pattern recognition and discriminator machines. Hard for a designer, easy for evolution.

  27. I do find it odd, the mindset that, following an unresolved disagreement, sees only two alternatives – that the other has willfully refused “correction”, or is too stupid to understand. The third – that KF himself might have made an error that he has yet to spot – never seems to cross his mind.

    Obviously, we hold to views because we think them right, and if we still hold them after a discussion, then it is because we have not found our respective partner in discussion persuasive. Most people at least allow for the possibility that they have not been persuasive because they failed to see a flaw in their own argument.

    I certainly do, and I often find some. But for KF, the only options on the table seem to be that the other person is either stupid or willfully resistance to correction. Being mistaken himself never seems to cross KF’s mind. Perhaps he needs more beseeching.

  28. Lizzie,

    Evolution is the story of a vast-dimensioned fitness landscape, full of peaks in one dimension that are potentially connected by ramps in another, with populations sitting on peaks while others move to others. And the height of the peak is not a measure of complexity but of fitness to that population’s environment. We do not sit on a higher peak than, say bacteria. We have are simply on a different one, having got there by (putatively) traversing a very different set of ramps.

    The multidimensionality of the fitness landscape for a given species doesn’t even include the dynamic interactions with the landscapes of other species.

    The fact that a particular species occupies a peak (successfully produces viable offspring) may also be due to the fact that another species is successful in that same environment, having got there by an entirely different route.

    Symbiosis, for example, intertwines the fitness landscapes of two or more species; neither being able to survive without the other; plants and pollinating insects, for example.

  29. Eric again (do come over, Eric, this is kinda silly!):

    Lizzie’s comment @216 via Joe:

    There is some poor reasoning that leads to this conclusions, but Irreducible Complexity turns out to be a) not a theoretical bar (because, arches) and b) not a bar anyway (turns out evolutionary processes can result in IC features (because, AVIDA).

    Oh, boy, not the AVIDA bluff again. As it relates to irreducible complexity, AVIDA is an exercise in irrelevance. It assumes a relatively smooth, easy-to-traverse landscape.

    No, it doesn’t. In fact, with the parameters given in the paper, some features can only be reached by leaping a fitness chasm. That was The Whole Point.

    And wonder of wonders, it delivers on its underlying programming. But the very point in question is whether biology exists in a smooth, easy-to-traverse landscape, so assuming it certainly doesn’t demonstrate it.

    It neither assumed it, nor demonstrated it. What it demonstrated was that given a rugged landscape, that the population could still reach the summit.

    Indeed, the authors acknowledge (in the very same paper that made all the headlines) that without their step-by-step reward system carefully leading all the simplistic ‘creatures’ up Mount Improbable, that it doesn’t work.

    Eric has again revealed that he doesn’t understand evolutionary theory. Clearly evolving features in AVIDA are rewarded – because that is what natural selection is. In nature, as in AVIDA, fitness is fecundity. AVIDA is set up so that certain features result in greater fecundity, just as in nature. But to make it hard for themselves, the most fecundity-promoting features were ones that required many steps, including, it turned out, some deleterious steps, to evolve. If AVIDA had left out the fecundity parameters, and had no features conferring additional fecundity, then they wouldn’t have been modelling natural selection, they would have been modelling drift. And if you run AVIDA with a drift-only set of parameters, the most complex features, surprise surprise, don’t evolve.

    AVIDA most certainly is not a valid simulation of IC being created in the real world by natural processes.

    It is an absolutely valid refutation of Behe’s concept. Behe claimed that a thing wouldn’t evolve if the previous step (from a simpler parent) was non-advantageous. He then raised the bar and had orders of IC, in which a certain number of previous steps had to be non-advantageous. In AVIDA, the fancy function (EQU) often evolved after many non-advantageous steps, including substantially deleterious steps. Therefore Behe is falsified. His principle is incorrect.

    Perhaps, though, I’ve missed all the truly innovative functional, irreducibly-complex engineering systems AVIDA has designed in the real world over the past few years. After all, if it really works one would think it would have been put to good use — sheesh, it’s even free to use. Those of us who are involved in engineering would certainly prefer to just sit back and watch sports on TV while our “algorithms” come up with the next big innovation. What’s that? Crickets . . .

    Eric is “involved in engineering”, and doesn’t know that GAs can be used to solve real-life engineering problems?

  30. Mike Elzinga:
    Lizzie,

    The multidimensionality of the fitness landscape for a given species doesn’t even include the dynamic interactions with the landscapes of other species.

    The fact that a particular species occupies a peak (successfully produces viable offspring) may also be due to the fact that another species is successful in that same environment, having got there by an entirely different route.

    Symbiosis, for example, intertwines the fitness landscapes of two or more species; neither being able to survive without the other; plants and pollinating insects, for example.

    And then there’s the fact that the evolving population is itself part of the changing landscape – which can become an arms race.

  31. Petrushka’s point is also extremely important.

    Maybe, but genes affecting the pigment of your toe can affect your vision. Part of the high dimensional spaghetti code problem faced by the designer.

    One of the reasons I argue that design is impossible., except by evolution.

    It is also the case, when walking on a multidimensional fitness landscape, that a species is taking multidimensional steps. Each dimension isn’t sampled in succession.

  32. Elizabeth: “I have just yet to read a pro-ID article, even by the academics in the field, that doesn’t make rookie errors about evolutionary science.”

    Now that we’ve cleared up the matter of what you meant by “academics in the field” your claim is just not believable.

    There are numerous articles out there by academics in the field of ID Theory who don’t even mention “evolutionary science”, much less make rookie mistakes about it.

    Surely Dembski qualifies as an ‘academic in the field” of ID Theory.

    And we all know you’ve read his paper on Specification.

    I repeat, since you seem to have failed to address the question:

    “You’ve read Dembski’s article on Specification. How does he make rookie errors about evolutionary science (whatever that is) in that article?”

    Feel free to say he doesn’t make rookie mistakes about “evolutionary science” (whatever that is) in that article and retract your absurd (and false) claim.

  33. Elizabeth,

    Seriously, have you even read the Axe/Gauger book that you pretend to be refuting?

    Want me to send you a copy?

  34. Joe Felsenstein on March 31, 2013 at 2:24 pm said:

    “OMagain has listed a whole series of refutations of Dembski’s CSI argument. I must (of course) add my own article:”

    Glaringly absent from Joe’s “References” section is Dembski’s paper on Specification, which is the “pro-ID article” by “an academic in the field” that is of immediate relevance here.

    It’s a paper Elizabeth has read. It’s a paper which, according to Elizabeth, makes “rookie errors about evolutionary science.”

    Literature bluffs aside, do you have anything relevant to offer Joe?

  35. Elizabeth,

    Do you classify Todd Wood as “an ID proponent” and/or as an “academic in the field” of ID theory?

    My own personal experience is that he is anti-ID and can hardly be classified as an academic in the field of ID theory.

  36. Elizabeth:

    “There are certainly many unsolved problems in evolution – how simple were the simplest Darwinian-capable self-replicators? How did the genetic code emerge? How did cell nuclei arise? How did multicellularity get started?

    And IDists might be perfectly entitled to raise these specific problems.”

    And we do. But when we do, we are told we’re not entitled to do so.

  37. Elizabeth:

    “Well, inasmuch as a GA is a tool for finding a solution to a specific problem, obviously the problem is “smuggled in” (in plain sight) – but what if there are billions of problems, and the GA finds solution to a subset of them? Do we say the solution, or the problem, has been “smuggled in”? ”

    You wrote a GA, right? Didn’t you openly brag about the fact that it smuggled in the solution?

    So what is the subset of billions of problems that your GA finds a solution for?

    Doesn’t “to find a solution for” sound the least bit teleological to you?

  38. Seriously, Mung; there are some of us here who have been reading ID/creationist stuff for close to 50 years. We know your pseudo-science and your socio/political history far better than any of you do.

    Can you or any of your cohorts claim to have extended the same courtesy by reading any science texts for understanding instead of for quote-mining and misrepresenting?

    Where did your science education stop? From what we see of your responses, it has to have been before high school. Can you even pass the requirements for eighth grade science?

    We have already demonstrated – here and over at Panda’s Thumb – that not one of you can pass basic concept tests in science; nor can any of you do simple high school level calculations. How do you explain that?

    Do you know how to do anything but taunt and throw feces? Is that the life you are reduced to?

  39. And we do. But when we do, we are told we’re not entitled to do so.

    Boing! Incorrect assertion! Well, you may be able to scare up an instance where someone on the internet has said such, but really, all of these questions are questions anyone is perfectly entitled to ask. You’ll get a scientist’s best attempt at an answer but – because they are unsolved problems – it won’t be definitive. But ID doesn’t raise them in order to get that kind of answer, but to crow “you don’t really know, therefore Design”.

    Of course you’re entitled to probe questions with everyone else. If you can come up with something more definitive than the ‘scientists’, have at it. Don’t be such a martyr.

  40. Mung:
    Elizabeth,

    Seriously, have you even read the Axe/Gauger book that you pretend to be refuting?

    Want me to send you a copy?

    Mung, please read the rules of this site, here.. And then please read my OP. I am not “pretend[ing]” to refute any book by Axe/Gauger. I didn’t even know they had written a book together. The OP is a critique of a blog-post. In it, Gauger indicates that she, and Axe, think that the biological fitness landscape is low-dimensioned and rugged, and she bases her dismissal of evolutionary theory on the supposition that this is the case. It is not, as I have explained, and as many others have explained better. The natural fitness landscape is self-evidently high-dimensioned (we know that there are countless ways that organisms can vary), and is also self-evidently fairly smooth (variant offspring resemble their parents).

    Do you disagree?

    And if you don’t, do you think that Gauger does? Because if she doesn’t, then it is, at best, an extremely badly written article.

  41. Mung:
    Elizabeth: “I have just yet to read a pro-ID article, even by the academics in the field, that doesn’t make rookie errors about evolutionary science.”

    Now that we’ve cleared up the matter of what you meant by “academics in the field” your claim is just not believable.

    There are numerous articles out there by academics in the field of ID Theory who don’t even mention “evolutionary science”, much less make rookie mistakes about it.

    Can you give me a reference for one? Even Granville Sewell, whose rookie mistakes are in physics, mentions evolutionary science, and gets it wrong.

    Surely Dembski qualifies as an ‘academic in the field” of ID Theory.

    Yes. He is one of the people I was thinking about.

    And we all know you’ve read his paper on Specification.

    Indeed.

    I repeat, since you seem to have failed to address the question:

    “You’ve read Dembski’s article on Specification. How does he make rookie errors about evolutionary science (whatever that is) in that article?”

    Feel free to say he doesn’t make rookie mistakes about “evolutionary science” (whatever that is) in that article and retract your absurd (and false) claim.

    Dembski’s mistake about evolutionary theory in that paper is to indicate that he can subsume “Darwinian and other material mechanisms” into a “chance hypothesis” and treat it as a null.

    But I will modify my claim and say: I have yet to read an ID article about evolutionary theory that does not make rookie mistakes about evolutionary theory.

    Although same appears to be true about physics.

  42. Mung:
    Elizabeth,

    Do you classify Todd Wood as “an ID proponent” and/or as an “academic in the field” of ID theory?

    My own personal experience is that he is anti-ID and can hardly be classified as an academic in the field of ID theory.

    You are probably correct. He is an academic, and his research project is to discover how an Intelligent Designer (the Abrahamic God) made the world, not to prove that an Intelligent Designer did so. So, depending on how you define “ID theory” he may be beyond the pale. Shame. He’s a good man, is Todd.

  43. Mung: Doesn’t “to find a solution for” sound the least bit teleological to you?

    It is sometimes difficult to use language with it’s teleological biases in scientific communication without prompting exactly this.

    UD used to on a daily basis re-post scientific news and simply bold all the words that make it seem like they were talking about ID.

    And this is where much of the trouble begins.

    yes, perhaps “to find a solution for” sounds, to you, teleological , but what of it?
    Does that mean that GAs are? No.
    Does it mean that you’ll claim GAs are simply on the basis of a description of the? Absolutely.

    So play your word games Mung, when you get published with a critique based on them then the first round is on me!

  44. What Allan said.

    And if it’s happened at all, my bet is that it happened within an argument about how Darwinism is Wrong, and suddenly the ID proponent moved the goalposts to challenging OOL.

    But feel free to provide me with a counter-example.

    And as far as I personally am concerned, you are perfectly “entitled” to bring up OOL as evidence for a explanatory gap that might evince design. I wouldn’t find it very persuasive, however, as it is a rapidly shrinking gap.

  45. Mung:
    Elizabeth:

    “Well, inasmuch as a GA is a tool for finding a solution to a specific problem, obviously the problem is “smuggled in” (in plain sight) – but what if there are billions of problems, and the GA finds solution to a subset of them? Do we say the solution, or the problem, has been “smuggled in”? ”

    You wrote a GA, right? Didn’t you openly brag about the fact that it smuggled in the solution?

    So what is the subset of billions of problems that your GA finds a solution for?

    Doesn’t “to find a solution for” sound the least bit teleological to you?

    I’m not sure what you mean. But let me try to be more clear.

    In a toy/virtual/designed evolutionary model, the fitness function is usually a measure of how well the virtual critters perform one specified task or set of tasks (receive a radio signal; perform logic functions; discriminate between the brain scan of a person with schizophrenia and one without; optimally configure a gas distribution network; find the shortest delivery route etc). This would be analogous, in the natural world, to performance on some task that directly affected reproductive success, for example, moving quickly; storing energy; hiding from predators; etc); however, in our designed model we artificially link reproductive success to the task we want to be done. In other words, we design a world in which performing our task leads to reproductive success.

    However, in the natural world, any task that actually does lead to reproductive success will do the trick – no-one has to artificially link one to the other (although we can and do, as when we domesticate wild organisms through selective breeding, i.e. link performance of some task WE want the critters to do, like bearing lots of sweet flesh, or being sweet and bug-eyed, to reproductive succeess – by only breeding from those who do the task the best). Self-evidently, good camouflage, for a prey species, will lead to reproductive success, for example, as will efficient energy storage for a species in an environment where resource supply fluctuates.

    So no teleology is required when we talk about “solutions to a problem” in evolution. The problem, in all cases, designed or wild, is to reproduce efficiently within the current environment whether that environment is one that, by design, rewards behaviour that the designer wants to see, or is simply the environment the population, for realz, finds itself in.

    So the only information that is “smuggled in” to a GA (or the natural environment) are the problems that need to be solved for the critters to be able to reproduce with maximal efficiency. And they are not “smuggled” at all – they are hiding in plain sight, in the form of the fitness function, which, in the case of the natural world, is the natural environment.

    So if all Dembski is saying is that for things to evolve, they need to have a challenging environment provided for them – where is the ID argument? That a Designer made the world a tough place for his critters to evolve in? Or that he made critters in the first place?

    Neither is a critique of Darwin’s theory. The latter is metaphysics, the former is OOL.

  46. Mung:
    Joe Felsenstein on March 31, 2013 at 2:24 pm said:

    “OMagain has listed a whole series of refutations of Dembski’s CSI argument. I must (of course) add my own article:”

    Glaringly absent from Joe’s “References” section is Dembski’s paper on Specification, which is the “pro-ID article” by “an academic in the field” that is of immediate relevance here.

    It’s a paper Elizabeth has read. It’s a paper which, according to Elizabeth, makes “rookie errors about evolutionary science.”

    Literature bluffs aside, do you have anything relevant to offer Joe?

    Do you think that Dembski says something materially different or new in that (unpublished) paper than he says in No Free Lunch? If so, what is the key new point that Joe has left unaddressed?

  47. (Not an April Fools posting): I have read the Dembski article that Mung references. Mung says that it is “of immediate relevance”. It isn’t relevant to the issue I raised — namely whether there is any justification for using the presence of CSI as indicating Design:

    * Nowhere in the article does Dembski present any defense of his Law of Conservation of Complex Specified Information.

    * Does that mean that he no longer uses the LCCSI as his method of ruling out the role of natural selection in putting specified information into the genome? No, because in Addendum I he cites his previous books, including No Free Lunch, and makes clear that

    The changes in my account of these concepts here should be viewed as a simplification, clarification, extension, and refinement of my previous work, not as a radical departure from it.

    So we are still left without any defense of the LCCSI by ID types. In Specification when Dembski addresses Design Detection he describes it as eliminating “the chance hypothesis”. A little before that he describes that hypothesis (in his bacterial flagellum example) as “an evolutionary chance hypothesis that takes into account Darwinian and other material mechanisms”.

    Does he have any general method for ruling out that natural selection could put CSI into the genome, a method that works in all cases? The LCCSI was supposed to be that. But alas for Dembski’s argument, the LCCSI conservation law is dead, and neither Mung nor any ID type has breathed life into the corpse.

    And yet, when alive, it was the only justification for the statement that natural selection could not account for the presence of CSI in the genome.

    Mung needs to explain why the LCCSI works to eliminate the kind of “chance hypothesis” that includes natural selection. If Mung can’t, then he is backing a zombie method.

    Note: explain, not just cite some other work by Dembski and claim that it is “of immediate relevance”. Explaining why LCCSI works to rule out natural selection is the task of immediate relevance.

Leave a Reply