|
Author
|
Topic: Dembski on functional subsystems in "Uncommon Dissent"
|
yersinia
Member
Member # 324
|
posted 22. July 2003 23:28
Dr. Dembski has recently put up his introduction to a soon-to-be published anthology of Darwinism skeptics: Uncommon Dissent.
Therein, he writes of the Lenski et al. (2003) simulation of the evolution of a complex, multiple-parts-required system,
quote:
This paper describes a computer simulation and thus contains no actual biology. Go to the discussion section, and you'll read: "Some readers might suggest that we 'stacked the deck' by studying the evolution of a complex feature that could be built on simpler functions that were also useful. However, that is precisely what evolutionary theory requires...." In other words, the computer programmers artificially built into the simulation what they thought evolution needed in order to make it work. The validity of this study therefore depends crucially on whether the simulation maps faithfully onto biological reality.
Unfortunately, it does not, and the study therefore doesn't prove a thing about real-life biological evolution. By requiring of their simulation that complex features exhibiting complex functions can always be decomposed into simpler features exhibiting simpler functions, the authors of this article begged precisely the point at issue with irreducible complexity in real-life biological systems. There is no evidence that real-life irreducibly complex biochemical machines can be decomposed this way.
We will leave aside the irony of Dembski (author of the tornado-in-a-junkyard strawman for flagellar evolution, the closest thing to actual biology in No Free Lunch) accusing someone of not having any "actual biology" in their study. The question that Dembski raises here is, can "complex features exhibiting complex functions" be "decomposed into simpler features exhibiting simpler functions"? Dembski says that the simulation's validity "depends crucially" on this assumption, and further argues that Lenski et al.'s simulation is invalid because "There is no evidence that real-life irreducibly complex biochemical machines can be decomposed this way."
But, in responding to Ken Miller just a few months ago, Dembski conceded exactly this assumption about irreducibly complex systems: the Icon of ID, the flagellum, can in fact be decomposed into several subsystems with independent functions:
quote:
To this let me add: A system is irreducibly complex in Behe's sense if all its parts are indispensable to preserving the system's basic function. That an irreducibly complex system may have subsystems that have functions of their own (functions distinct from that of the original system) is therefore allowed in the definition. It seems that Miller is unclear about the distinction between a definition and an argument. Irreducible complexity is a well-defined notion that is appropriately and ascertainably applied to the bacterial flagellum. Miller's concern ultimately seems not over the definition but over its use as an argument to rebut Darwinism. Miller's point here generally is that if subsystems can be found with functions of their own (perforce different from that of the original system since otherwise the original system would not be irreducibly complex), then those subsystems and their functions can be grist for selection's mill and underwrite a Darwinian account of how the original system arose.
Source: Still Spinning Just Fine: A Response to Ken Miller
(Dembski then goes on to argue that the fact of functional subsystems doesn't disprove Behe's arguement, but that is not the point Dembski is contesting in his Uncommon Dissent critique of Lenski)
If that weren't enough, Dembski has been even clearer elsewhere:
quote:
You've charged me with moving the goalposts and adjusting the definition of irreducible complexity because I require of evolutionary biologists to "connect the dots" in a causally convincing way. The dots here are functional precursors that could conceivably have evolved into the final system of interest. You state that previously I claimed that the dots couldn't exist because they wouldn't be functional. Please show me in Michael Behe's writings or my own where we deny that IC systems can be made up of subsystems that can be functional in their own right. The point is not whether subsystems can be functional on their own but whether they can exhibit the same function in the same manner as the system in question. You misrepresent our position.
Source: ISCID thread, page 2: Organisms using GAs vs. Organisms being built by GAs
And he even says in one place that such subsystems "always" occur:
quote:
But any reasonably complicated machine always includes subsystems that perform functions distinct from the original machine.
Source: Evolution's Logic of Credulity: An Unfettered Response to Allen Orr
In other words, Dembski has already admitted that his own argument against the Lenski paper is wrong. He has *already* admitted in print, repeatedly, that "complex features exhibiting complex functions can always be decomposed into simpler features exhibiting simpler functions." And yet he criticizes Lenski for their assumption:
quote:
...the authors of this article begged precisely the point at issue with irreducible complexity in real-life biological systems. There is no evidence that real-life irreducibly complex biochemical machines can be decomposed this way.
...even though it is -- even according to Dembski himself -- biologically *realistic* and therefore "actual biology".
Dembski spends most of his introduction to Uncommon Dissent accusing scientists of adhering to evolutionary theory for all kinds of ideological reasons, rather than evidence. Perhaps Dembski would be more successful in convincing them if he spent more time making sure his own arguments were self-consistent (let alone well-supported by evidence, a topic for another thread), and less time calling scientists names.
[ 22. July 2003, 23:42: Message edited by: yersinia ]
IP: Logged
|
|
Pim van Meurs
Member
Member # 541
|
posted 23. July 2003 13:28
Thanks Yersinia, an interesting collation of Dembski's statements.
I am reading through the Introduction myself and am somewhat surprised by its tone. But perhaps I do not understand what the targeted audience may be.
What I found interesting is the statement that "This paper describes a computer simulation and thus contains no actual biology" which seems to apply equally well, or better, to for instance Dembski's design inference or No Free Lunch, arguably theoretical attempts to provide for a foundation to infer/detect intelligent design without much of an application to biology either.
To continue with the introduction, I find it fascinating how such terms as myths and (perceived) motives are being used and even abused. It would have been helpful to have a more indepth analysis by Dembski or others of the work of Lenski for instance.
IP: Logged
|
|
Nel
Member
Member # 614
|
posted 23. July 2003 18:36
Well this is ironic cosidering Nic's closing statements. What Dembski is referring to here:
quote:
Unfortunately, it does not, and the study therefore doesn't prove a thing about real-life biological evolution. By requiring of their simulation that complex features exhibiting complex functions can always be decomposed into simpler features exhibiting simpler functions, the authors of this article begged precisely the point at issue with irreducible complexity in real-life biological systems.
is absolutely consistent with his statements to the effect that IC systems can have sub-complexes with a different function. The problem is that, with IC systems like the bacterial flagellum, you don't have simpler alternative functions all the way up to the flagellum. All you have is a 10 part system that is itself IC. In the Lenski experiment, many steps towards EQU was rewarded.They had to reward many specific intermediate steps leading up to the EQU. However, with the flagellum, there is no alternative function within the subset of the export machine (a 6-part system). Likewise, there is no alternative function from the C-ring to the rod.
[ 23. July 2003, 18:43: Message edited by: Nelson-Alonso ]
IP: Logged
|
|
yersinia
Member
Member # 324
|
posted 23. July 2003 20:08
Nelson writes,
quote:
What Dembski is referring to here...is absolutely consistent with his statements to the effect that IC systems can have sub-complexes with a different function. The problem is that, with IC systems like the bacterial flagellum, you don't have simpler alternative functions all the way up to the flagellum.
He didn't say "simpler alternative functions all the way up to" something, he said that the simulation assumes that "complex features exhibiting complex functions" that can "be decomposed into simpler features exhibiting simpler functions".
quote:
In the Lenski experiment, many steps towards EQU was [sic] rewarded. They had to reward many specific intermediate steps leading up to the EQU.
Nope, they just rewarded several simpler systems with different functions (seven systems other than EQU). Specifically, these were the rewards:
NOT -- 2
NAND -- 2
AND -- 4
OR_N -- 4
OR -- 8
AND_N -- 8
NOR -- 16
XOR -- 16
EQU -- 32
Source: Lenski et al. 2003, Table 1
yersinia
PS: Let's not turn this into another flagellum thread; we've discussed it to death elsewhere (links here and here) and I'm happy for people read those rather than re-hashing it again.
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 23. July 2003 20:47
Since Nelson posted the same remarks on both ARN and here, I'll post similarly.
Nelson Alonso wrote quote:
In the Lenski experiment, many steps towards EQU was (sic) rewarded.They had to reward many specific intermediate steps leading up to the EQU.
Wrong. Nelson should read the paper more carefully. "Each step" was not necessarily rewarded, and rewarding no "specific" step was required. In 36 control conditions of 10 runs each they dropped one (all of the 'simpler functions, one at a time) or a pair (all possible pairs, in fact) of the simpler functions, and lineages capable of performing the input-output mapping that corresponds to EQU evolved in all of them. Only when no intermediates at all were available did it not evolve. The paper included many more runs in control conditions than in the main 50-run experimental condition that people focus on, but the control conditions are important in interpreting what happened in the study. They rewarded as few as 6 simpler functions, not "many," and no "specific" function or pair of those functions was critical to evolve lineages capable of performing EQU. Evolution (and cooption) made do with what was available.
My own pilot work over the past several weeks with the installation of Avida 1.6.0 on a Beowulf cluster and Ver 1.3.0 on a PC network suggests that even fewer intermediates may be necessary, and that the 'slope' of the fitness function need not be as steep as that of Lenski, et al. Thus the evolution of irreducibly complex assembly language programs that perform the input-output mapping corresponding to EQU are not dependent on any specific set of intermediates nor on rewarding "many" intermediates on some steeply stepped fitness function.
RBH
IP: Logged
|
|
Nel
Member
Member # 614
|
posted 23. July 2003 20:50
Nic writes:
quote:
he said that the simulation assumes that "complex features exhibiting complex functions" that can "be decomposed into simpler features exhibiting simpler functions".
Nic, thats saying the same thing twice. If you reward many steps leading up to the function you want then you have "decomposed" it "into simpler features exhibiting simpler functions". Unfortunately thats not even true for the flagellum even taking the export machine into account. You can only give it two rewards.
Nic writes:
quote:
Nope, they just rewarded several simpler systems with different functions (seven systems other than EQU). Specifically, these were the rewards:
You say "nope" but then you confirm what I said. They have to give many rewards to systems with alternative functions. Compare that with the two (and the rest is just pure chance) and you have yourself a tornado in a junk yard. Each logical function is given a reward, EQU is given the greatest reward because it is the most complex function.
IP: Logged
|
|
Nel
Member
Member # 614
|
posted 23. July 2003 20:52
RBH writes:
quote:
Wrong. Nelson should read the paper more carefully. "Each step" was not necessarily rewarded, and rewarding no "specific" step was required.
Actually, "each step" in the context of the logical function that was performed, was rewarded and as the functions increased in complexity, the reward was greater. If there was no reward, the logical function that they performed would not persist. Being able to perform NAND is a specific step that is required and needs to be rewarded. But with the flagellum you can only give 2 rewards, the rest is just pure chance.
[ 23. July 2003, 20:55: Message edited by: Nelson-Alonso ]
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 23. July 2003 21:09
Nelson Alonso wrote quote:
Actually, "each step" in the context of the logical function that was performed, was rewarded and as the functions increased in complexity, the reward was greater. If there was no reward, the logical function that they performed would not persist. Being able to perform NAND is a specific step that is required and needs to be rewarded.
In fact that's not the case. One can delete the reward for performing the input-output mapping corresponding to NAND and lineages capable of performing the mapping corresponding to EQU still evolved. See Lenski, et al., pp. 142-3, "Different environments": quote:
Evidently, neither any particular simpler function nor any pairwise combination of functions was required to evolve this complex feature. (p. 143)
RBH
Aadded in edit: In fact, at least one of the Lenski, et al, runs (#106) that evolved to perform the EQU input-output mapping could not perform NAND!
[ 23. July 2003, 21:22: Message edited by: RBH ]
IP: Logged
|
|
Nel
Member
Member # 614
|
posted 23. July 2003 21:18
RBH,
Obviously you can have different combinations of functions and not all of them are particularly required (which once again shows how plastic EQU is and therefore irrelevant to things like flagella). However, for some organisms NAND was a required function, although others did not need it because they went through another route. In all cases, many simpler functions need to be rewarded. Show me a case where only two simpler functions were rewarded and yet they still obtained EQU. Also, show me a bacterial flagellum where evolving MotAB is not necessarily required step in getting a bi-directional motor.
Actually considering how much more complex the flagellum is from a program operation like EQU, it obvious that comparing the amount of simpler functions needed to obtain flagellum is going to be greater in comparison.
[ 23. July 2003, 21:38: Message edited by: Nelson-Alonso ]
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 23. July 2003 21:42
Nelson Alonso wrote quote:
Obviously you can have different combinations of functions and not all of them are particularly required (which once again shows how plastic EQU is and therefore irrelevant to things like flagella). However, for some organisms NAND was a required function, although others did not need it because they went through another route. In all cases, many simpler functions need to be rewarded. Show me a case where only two simpler functions were rewarded and yet they still obtained EQU.
First, I think it shows how powerful evolution is. Take away what you first said was necessary (NAND) and it still 'figures out' a way of evolving complex features. Now you say there are other routes. In any high-dimensioned space there are almost inevitably multiple pathways, and what evolution does is 'find' those pathways.
Sure, take away enough intermediates and the EQU input-output mapping doesn't evolve - Lenski, et al. showed that in one of their control conditions. But that merely says that on a flat zero-fitness landscape stuff doesn't evolve. Gee. A non-uniform non-zero-fitness landscape is necessary for selection to occur! As I said somewhere else, that's finally catching up with Darwin. Only 150 more years to go!
I'll leave the flagellum argument to others - as yersinia said, that's been done, and by people better qualified to do it than I. But I do know that your descriptions of what was in the Lenski, et al., experimental design have not been complete or fully accurate, and thus your dismissal of it does not have force.
I can't show you the two-intermediate case (yet) - I'm still dealing with installation bugs. But as and if I have time I'll look at it. I know it happens with four intermediates, none of them NAND.
RBH
IP: Logged
|
|
Nel
Member
Member # 614
|
posted 23. July 2003 21:46
RBH writes:
quote:
. In any high-dimensioned space there are almost inevitably multiple pathways, and what evolution does is 'find' those pathways, often in ways we cannot anticipate and after the fact have great difficulty reconstructing except when we have the full record of the evolutionary path, as we do in Avida.
Since there are so many ways, and you are comparing this with flagellum, show me 10 such pathways for the bacterial flagellum.
quote:
Sure, take away enough intermediates and the EQU input-output mapping doesn't evolve - Lenski, et al. showed that in one of their control conditions.
But that is precisely the problem that IC points to. The analysis of EQU was predicted to occur by Behe himself when he said:
quote:
Even if a system is ireducibly complex (and thus cannot have been produced directly), however, one can not definitely rule out the possibility of an indirect, circutious route. As the complexity of an interacting system increases, though, the likelihood of such an indirect route drops precipitously. And as the number of unexplained, irreducibly complex biological systems increases, our confidence that Darwin's criterion of failure has been met skyrockets toward the maximum that science allows.
Here is the problem that Dembski is referring to and which the Lenski analysis does not address. With bacterial flagella, the problem is "As the complexity of an interacting system increases, though, the likelihood of such an indirect route drops precipitously."
IC systems like bacterial flagella are cases where it is near flat zero-fitness from the first protein of the export machine to the 10th part, and then again from the c-ring to the rod. 4 or 6 rewarded functions may not likely to be enough. Especially since program code is so different from molecular machines.
[ 23. July 2003, 21:48: Message edited by: Nelson-Alonso ]
IP: Logged
|
|
yersinia
Member
Member # 324
|
posted 23. July 2003 21:54
Nelson quotes Behe on indirect pathways:
quote:
Even if a system is ireducibly complex (and thus cannot have been produced directly), however, one can not definitely rule out the possibility of an indirect, circutious route. As the complexity of an interacting system increases, though, the likelihood of such an indirect route drops precipitously. And as the number of unexplained, irreducibly complex biological systems increases, our confidence that Darwin's criterion of failure has been met skyrockets toward the maximum that science allows.
But, Nelson, this is mere assertion on Behe's part. His book is all about how direct pathways can't produce IC. He doesn't, say, spend a chapter somewhere documenting that indirect pathways are unlikely, he just asserts it in that one quote.
Dembski concedes in many places that IC systems have functional subsystems. But when he is critisizing Lenski, he criticizes them for assuming that there are selectable functional subsystems. You haven't resolved this contradiction.
If you want to talk about the flagellum more, please start another thread. That's not the topic here.
IP: Logged
|
|
Nel
Member
Member # 614
|
posted 23. July 2003 22:03
Nic writes:
quote:
But, Nelson, this is mere assertion on Behe's part. His book is all about how direct pathways can't produce IC. He doesn't, say, spend a chapter somewhere documenting that indirect pathways are unlikely, he just asserts it in that one quote.
This is not an assertion, it's actually quite intuitive. If you don't have many intermediates to reward in an evolutionary pathway, then most of the pathway is likely to be just pure chance, and pure chance drops the probability. Thats why natural selection was proposed in the first place.
Nic writes:
quote:
Dembski concedes in many places that IC systems have functional subsystems. But when he is critisizing Lenski, he criticizes them for assuming that there are selectable functional subsystems. You haven't resolved this contradiction.
Sure I have, you havn't addressed my counter-argument. Dembski is referring in that quote, which you emphasized yourself,
quote:
By requiring of their simulation that complex features exhibiting complex functions can always be decomposed into simpler features exhibiting simpler functions
But if you can only decompose it to very few simpler functions, then you aren't really negating the fact that you are ushering in too many chance events. In fact, as RBH notes, EQU didn't evolve when you took away too many intermediates. Thats the problem that IC points to.
Nic writes:
quote:
If you want to talk about the flagellum more, please start another thread. That's not the topic here.
It is a good example of what this topic addresses and I don't need to start another thread for it.
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 23. July 2003 22:26
Nelson Alonso wrote quote:
Since there are so many ways, and you are comparing this with flagellum, show me 10 such pathways for the bacterial flagellum.
In no discussion of the Lenski, et al., paper have I made any comparison with the flagellum: I'm not qualified to do that. All my remarks in that vein have been directed at the concept of "irreducible complexity" in general and the conditions under which IC objects, features, or processes could evolve.
Nelson's remarks contain an interesting leap of faith. He says quote:
IC systems like bacterial flagella are cases where it is near flat zero-fitness from the first protein of the export machine to the 10th part, and then again from the c-ring to the rod. Thats the problem.
In fact, that's the question: Is the fitness function flat? I am not an authority on the flagellum (though I have perforce read more than I ever wanted to about it in these discussions) so I won't comment specifically on it. But I will claim that to not know what (or whether) precursors (at whatever level of analysis) might have been available for cooption to form a bacterial flagellum is not to show that the fitness function is/was flat and zero. As analyses of the Lenski, et al., lineages show, and as analyses of the devices evolved in the hardware evolution research literature show, it is very very difficult to reconstruct an evolutionary pathway looking at just the end product. Even when one knows that an entity evolved (because it did so in one's computer or on one's field-programmable gate array), and even when the entity is relatively simple, the entity itself contains only partial and incomplete information about its history. Further, even understanding how it does what it does can be very difficult. Those considerations should induce just a dab of caution in those who say of the flagellum, 'Well, we can't see how it could have evolved, so therefore it couldn't have.'
The claim that the fitness landscape is flat from here to there, when in fact all one really knows is that one doesn't know what the topography is, is a pure argument from ignorance: I don't know what its topography is, therefore it has this specific topography. That's a strong claim to make on what is no more than a lack of knowledge. But again, I'll leave the flagellum arguments to others. I'm here just interested in an accurate portrayal of the Lenski, et al., work. And that's been rare in the criticisms of it.
RBH
IP: Logged
|
|
Nel
Member
Member # 614
|
posted 23. July 2003 22:40
RBH writes:
quote:
All my remarks in that vein have been directed at the concept of "irreducible complexity" in general and the conditions under which IC objects, features, or processes could evolve.
Then you are not saying anything differently from what Behe said in the very first printed version of the IC thesis as applied to modern ID theory. IC systems that have many simpler functions can evolve (and is thus simple itself). Complex systems like flagella are harder, because of what Dembski calls the "sea of non-functionality".
RBH :
quote:
As analyses of the Lenski, et al., lineages show, and as analyses of the devices evolved in the hardware evolution research literature show, it is very very difficult to reconstruct an evolutionary pathway looking at just the end product.
Not true. In fact, Lenski et. al. had written their own version of EQU which is much simpler than most that actually evolved. The same is true for Biology. It is simple to envision the stepwise evolution of hemoglobin because it has so many functional intermediates, even simpler to envision the evolution of a two-part protein system that is IC. However, the reason why it is hard to envision the stepwise evolution of things like flagella is because it is so complex with very little, if any, simpler alternative functions.
Really your reply here is just one big fat promisory note, which is nothing different from all the other promisory notes we get with other complex biological systems.
[ 23. July 2003, 22:49: Message edited by: Nelson-Alonso ]
IP: Logged
|
|
|
Topic Closed
Printer-friendly view of this topic
