Uncommon Descent Serving The Intelligent Design Community

C’est la Avida


I recently posted a brief essay entitled “Beware of Question-Begging Computer Simulations” (linked below) in which I referenced an article by Eric Anderson. Since then Eric and I have corresponded by e-mail and he offers the following comments.


I was recently pointed to a discussion thread (www.uncommondescent.com/index.php/archives/802) on Uncommon Descent regarding my brief review (www.iscid.org/pcid/2005/4/2/anderson_bits_bytes_biology.php) of the program known as “Avida.” Initiated by Gil Dodgen with very kind and no doubt undeserved accolades, the thread contains a number of interesting remarks and counterpoints. Having waded through comment number 27, I came to the tentative conclusion that j was on the right track and that I need not provide any further input. However, the subsequent post managed to convince j that j’s own reading of the material was inaccurate, and things went a bit sideways from there to the end of the thread.

I am unfortunately not possessed of adequate time to respond to each post individually, nor to follow up on this thread for that matter, but Gil has kindly agreed to let me open with observations and then pose a pair of questions for consideration.

First, the observations:

Valerie, who is, by my impression, an intelligent and gifted debater, leads the charge with the following: “Eric Anderson’s essay is fraught with problems. The biggest is that he seems to misunderstand the concept of irreducible complexity as propounded by Michael Behe and William Dembski.” This is no small allegation. After all, if I have misunderstood what Behe and Dembski are talking about, then although not irrelevant by definition, my views at least stand in opposition to the primary individuals who have championed and brought to the masses the very notion of irreducible complexity.

The theme proceeds: “To Anderson, any system which can be approached by a ‘cumulative pathway’ is not irreducibly complex.” This is a remarkable accusation. Valerie has apparently either missed or misunderstood my footnote 6, which references my detailed analysis (www.iscid.org/pcid/2004/3/1/anderson_ic_reduced.php) of Dembski’s and Behe’s “Irreducible Complexity Revisited.” Though admittedly painful in length, my paper details several issues surrounding irreducible complexity and includes a detailed discussion of the possibility of a cumulative pathway. Indeed, it is in that paper that I coined the term “cumulative irreducible complexity” and juxtaposed the concept against what I term “per se irreducible complexity.” One is free, of course, to object to my analysis and question my conclusions in “Irreducible Complexity Revisited,” but the allegation that Anderson does not allow for a cumulative pathway to irreducible complexity will not stick.

Fair enough, Anderson, you may understand that a cumulative pathway is possible, but why did you footnote a definition that excludes such a cumulative pathway, and does not this demonstrate that you do not understand Behe’s and Dembski’s views on the matter?

I think not. First (and it really is an aside in terms of my critique of Avida), my point regarding Avida is not that a particular definition of irreducible complexity is at question, as I clearly indicated in the forgotten footnote. My point is that Avida adopts as its programming functionality the very concepts in question: that a cumulative pathway exists, it is a relatively easy pathway, there are regular rewards for incremental advancement along the pathway, and so forth. The entire exercise is one of question begging, and Avida cannot be saved by nuanced quibbles over the definition of irreducible complexity. That is precisely why I utilized a simple demarcation, footnoted the fact that I was doing so and that such an approach was sufficient for an analysis of Avida, and then moved on.

With regard to Behe and Dembski’s views, their general approach has been to conclude as an initial matter that a cumulative pathway is so unlikely as to be briskly dismissed without extensive discussion. One may uncharitably characterize their approach as “assuming” that a cumulative pathway does not exist, but I believe it is more accurately characterized as a preliminary conclusion, based on a review of the probabilities. Again, their paper, “Irreducible Complexity Revisited” is worth reviewing in this regard.

Additionally, Valerie, who references Behe as someone who considers a cumulative pathway to be a live possibility, must have missed Behe’s overall point. Valerie quotes the following passage from Behe: “Demonstration that a system is irreducibly complex is not a proof that there is absolutely no gradual route to its production. Although an irreducibly complex system can’t be produced directly, one can’t definitively rule out the possibility of an indirect, circuitous route.”

With the above quote, Valerie attempts to show that Behe considers cumulative pathways to be a live possibility and that my understanding of Behe’s position is in dire need of improvement. Yet Valerie either misunderstands or misrepresents Behe’s position. Specifically, Valerie fails to quote the next two sentences from the relevant passage, which demonstrate that the previous quote is the foil for Behe’s actual views: “However, as the complexity of an interacting system increases, 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.”

Allow me to paraphrase the complete thought: it is possible, as a matter of sheer logic, that an indirect route to the irreducible core exists, but it is not a realistic probability and will not be discussed further. Indeed, if one objectively reviews Behe’s essay (www.arn.org/docs/behe/mb_idfrombiochemistry.htm), it becomes clear, both from the immediate context and the positioning within the essay, that the last sentence quoted above is Behe’s preliminary conclusion about the likelihood of an indirect route. Indeed, he does not discuss it further in that essay. It is worth pointing out that Behe made the same argument in “Darwin’s Black Box,” and with very nearly the same words. In quoting the first half of the relevant paragraph from Behe, Valerie has parroted the passage but misunderstood the message.

Dembski has addressed in excellent detail his views on the matter in “Evolution’s Logic of Credulity: An Unfettered Response to Allen Orr” (www.arn.org/docs/dembski/wd_logic_credulity.htm), which should be required reading for all inclined to fancy that they possess a special ability to perceive unguided evolutionary pathways through the hazy mists of time.

While Behe and Dembski may be faulted for not attacking the cumulative pathway in a more fulsome manner, I consider this to be principally because they view a cumulative pathway as utterly unlikely. And as long as we are on the subject of sheer possibilities, it is also possible, of course, that I am ignorant of Behe’s and Dembski’s position on probabilities, their insights on irreducible complexity and their debate on design in general, but in the present matter I believe my analysis stands.

Now, a pair of questions to consider for all who are interested in the concept of irreducible complexity:

1. What is the difference between the quantity and quality of the complex specified information that is required to bring about an irreducibly complex core via a cumulative pathway as opposed to an abrupt pathway? (Note that I am not talking about whether one believes in their heart of hearts that a cumulative pathway is possible and that a selection mechanism will drive the process; I am simply talking about the kind and amount of information required.)

2. What is the basis of the oft-cited distinction between an “improved” function and a “different” function and how can we draw a principled distinction?

While not desiring to limit responses, I would suggest that these are inquiries that call for careful and contemplative consideration.

Eric Anderson

Eric: Thanks. With your current explication, it all makes sense to me now. [You implicitly confirm that you use the word "cumulative" differently from Dembski, as I had surmised. In my opinion, I think this should be explicitly stated in your essays. A large percentage of people who read them will also have read Demski's "IC Revisited" and be similarly confused. (You published both "IC Reduced" and "Bits, Bytes, and Biology" after Dembski published "IC Revisited.")] Note: In my previous comment, I more accurately should have said: "(I assumed that when you said “no cumulative pathway,” you meant “no feasible [rather than known] cumulative pathway.”) " j
Thanks, j, for your excellent thoughts and careful analysis. J wrote: “This seemed to indicate that you had a clear understanding of IC as Behe and Dembski use the term. (I assumed that when you said ‘no cumulative pathway,’ you meant ‘no known cumulative pathway.’)” I meant nearly that, but not quite. If we are talking about sheer logical possibility, then Behe, Dembski, you, I and everyone else of reason will acknowledge that as a matter of pure logic, there may be a cumulative pathway. That, of course, is not where the interesting part of the discussion lies, and so Behe in Darwin’s Black Box, for example, quickly points out that while there is a possibility as a matter of logic, as a practical matter there is not a live possibility, and then he moves on with other points. Thus, in my view, Behe and Dembski acknowledge the possibility of a cumulative pathway to irreducible complexity, but do not, as a rule, consider it to be a live possibility worthy of extensive consideration. You quote me as saying: “…if a program were written that had no possible cumulative pathway, then the writers of that program could be fairly accused of assuming up front that the complex feature was irreducibly complex. Thus, evolutionary algorithms seem to be between a rock and a hard spot: assume a cumulative pathway and then you are unable to challenge irreducible complexity; assume there is no cumulative pathway and then you are unable to support irreducible complexity.” My purpose in the above passage is to point out the difficulty of creating a program that mirrors biotic reality. In particular, we do not presently know whether there is a viable cumulative pathway to any irreducibly complex system existing in nature. As a result, the programmers are forced to make an assumption, and it is a whopper of an assumption: you either assume that there is a cumulative pathway and program accordingly, or assume there is not and program accordingly. Either way, you have just taken the program down the assumed path – a real world example of the mighty power of suggestion. This problem is not limited to Avida; this is the quandary we are in with all attempts to write programs that demonstrate this or that pathway to current organisms or functions. Incidentally, in making the above point about this programming challenge, I have contented myself to a particular kind of irreducible complexity, which in my other paper I term “per se irreducible complexity.” I do not equate irreducible complexity with unevolvability. This is partly because I do not view irreducible complexity as having a single meaning. And that is precisely why I belabored several pages to develop a basis for distinguishing between what I term cumulative irreducible complexity and per se irreducible complexity. I use the terms to describe the definitional criteria. Either form of irreducible complexity may be “evolvable,” again as a matter of sheer logic and with a healthy dose of sheer luck, but that is a separate inquiry. You wrote: “This would seem to equate IC and unevolvability (”no possible cumulative pathway”) . . .” Again, for purposes of analyzing Avida’s incongruity, it matters not what kind of irreducible complexity we use. You then quote me from Irreducible Complexity Reduced, and Bill Dembski from Irreducible Complexity Revisited. You well note Dembski’s city example, and use it to contrast his views with mine. J wrote: “This difference leads to a different definition of the term ‘irreducible complexity’ when they are contrasted. Behe and Dembski don’t equate ‘irreducible complexity’ with ‘unevolvability.’ (They allow the possibility of indirect pathways to contrived forms of IC, but argue against the practicality of such pathways for real-world examples of IC.)” I also do not equate irreducible complexity with unevolvability. For the most part, I am not even talking about whether a system is evolvable or not, as that is a secondary inquiry beyond the initial definitional aspects. My view is that no form of irreducible complexity is likely to come about through unguided natural processes, so in that sense, I have a very similar view as Behe and Dembski: again, there is sheer logical possibility, but not a live practical possibility. Finally, ah yes, the ol’ city example. I happen to disagree with the thrust of Bill’s city example, which is part of the impetus for me to examine the definitional aspects of irreducible complexity in much more detail and propose that we need to distinguish between cumulative irreducible complexity and per se irreducible complexity. You will note, of course, that I spend considerable time discussing the city example in my paper. All of this is of course irrelevant to the critique of Avida, which can be critiqued on the basic flaw of having incorporated as premises the very conclusions it is trying to reach. Rather than getting hung up on a particular type of irreducible complexity, defined and used in my critique for convenience’ sake, we ought to turn our thoughts to the more interesting nuances about irreducible complexity itself – nuances that your careful thoughts are bringing to the surface. Am I on the right track in my critique of Irreducible Complexity Revisited? Should we be rethinking this whole question of “improved” function versus “different” function (or as you say, “direct” versus “indirect”)? I think that largely depends on our answers to the two questions I posed above, namely: 1. What is the difference between the quantity and quality of the complex specified information that is required to bring about an irreducibly complex core via a cumulative pathway as opposed to an abrupt pathway? 2. What is the basis of the oft-cited distinction between an “improved” function and a “different” function and how can we draw a principled distinction? ----------------- Thanks again for sharing your view and your thoughtful analysis. I will unfortunately be out of the country for a while, so will not be able to follow up on this for some time, but hopefully the above will provide a sense as to where I am coming from. Eric Anderson Eric Anderson
I just read the comment that Anderson was talking about and then looked back to what the person "J" was referring to. I then looked at the canyon example provided by Valerie and have the following comments. To understand this someone will have to read the last 10 comments from the previous thread starting with Valerie's canyon analogy. The canyon example is the strongest example against Neo Darwinism that I can think of but is used by Valerie to undermine IC. There are back packers on the other side of a seemingly impassable canyon but we do not know how they got there, only that they are there. There must a path because they are there since helicopters are not allowed. Just because we cannot find a path for the backpacker it does not mean it doesn't exist. Thus, a path or way must exist. This metaphor of a canyon for evolution is then used as an argument against IC. But it is bogus because it reveals the basic weakness of the Neo Darwinian argument. For every back packer who reached the other side there are thousands of back packers dead in the canyon who supposedly never made it to the other side and these dead backpackers should indicate the route that they were taking when they died. Some should be very close to the other side when they died but their campsites should be discernable along the way. A single backpacker is not possible unless Valerie is espousing miracles. At each of the thousand of campsites the remains of the other backpackers must remain. Now all these backpackers died a long time ago so while there are scattered campsites in the canyon none link up very closely with the successful backpacker. All the relevant campsites have disappeared. But why are there no more backpackers traversing the canyon heading for one of the various ledges on the other side. It seems today few if any are making the trip. There should be thousands if not millions of campsites for those backpackers who should be currently making the trip. But there are none. All the campsites are from a long time ago. In other words there should be thousands of campsites today in the canyon near each other and which represent those who decided to go a little further but then decided to stay awile. Sort of like settlers in the 1800's as they made it across the US. So the problem is we never see these any of the other backpackers or their campsites. Somehow they all disappeared. And there are no current ones. Obviously we are referring to the fossil record for those backpackers who did it a long time ago and tht all the others that originally started out are now gone. But why are there no current ones setting up camp along the way or why are there no species that are morphing into other species today. We should have a living evolution as well as one that only happened in the distant pass. Similarly, the computer program that ends up on the other side of the canyon must have had several camp sites along the way, each one viable for a long time in case it did not want to go further. It may take a thousand iterations to get there but there had to be a lot of good alternatives along the way. If Avida got to the other side or if it gets there every time then where are those campsites in the canyon where it spent each iteration Each one should have a fully functional camp site for the trip and each should have been a practical alternative for the long term or else the backpackers would have died out. So I am saying if a computer program that mimics evolution gets to pay dirt, it only can do so if there are thousands of potential long term resting spots along the way. If it cannot produce any of them then the program is nonsense. This is all obvious to most of us but the Darwinist play games thinking up lots of clever ways that it might have happened differently while never showing one proven path. Dave Scott had a similar comment to mine but I wanted to emphasize the lack of current evolution and in fact Valerie's metaphor actually undermines Neo Darwinism. To me the lack of a living evolution is the biggest nail in the coffin for Darwinism. Why has it stopped. John Davidson had one reason but I think there may be another. That is it never happened in the first place. jerry

Avida suffers from the same flaw that RM+NS suffers from. In that way it's a perfect model of NeoDarwinian Evolution. The flaw:

Demonstrate the laughably simple and extrapolate to the hideously complex.

Avida evolves a CISC instruction from RISC instructions and claims this proves that a bacteria can acquire a flagellum bit by bit. NDE shows that finch beaks can vary in size, moth wings vary in color, through RM+NS, then claim this proves that bacteria can gradually change into baboons.

It is absolutely safe to say that, if you meet somebody who claims to believe these glib extrapolations, that person is ignorant, stupid or insane (or wicked, but I'd rather not consider that). :lol:

[Cut to the chase: Everything hinges on the words in boldface below.] Hello Eric: I'm glad that you had the opportunity to respond to the discussion of your writings that transpired on this blog. I had read "Bits, Bytes, and Biology" sometime in 2005; if I had spotted any glaring errors at that time I didn't remember doing so. But since Valerie had raised questions about it, I began to re-read it. I saw footnote 6:
As used in this essay “irreducible complexity” matches the irreducible complexity initially proposed by Behe and refined by Dembski, meaning that there is no cumulative pathway to such complexity. I have shown in my article “Irreducible Complexity Reduced: An Integrated Approach to the Complexity Space,” (see www.evolutiondebate.info) that this is not a logical requirement of the concept of irreducible complexity, and I term Behe’s and Dembski’s approach “per se irreducible complexity.” Avida targets this per se irreducible complexity, and it is sufficient for our present purposes.
This seemed to indicate that you had a clear understanding of IC as Behe and Dembski use the term. (I assumed that when you said "no cumulative pathway," you meant "no known cumulative pathway.") I scanned the rest of the essay, and quickly posted my comment #27, consisting primarily of statements by Behe and Dembski on the subjects of the IC evolvability, and Avida/IC, respectively. (I did not read any of your essay "IC Reduced" at that time.) However, Valerie's reply in comment #28 caused me to re-read your essay more carefully, and when I did, I discovered the following:
Evolutionary Assumptions [of the Avida research in question]... 1. There is a cumulative pathway to complexity. ...if a program were written that had no possible cumulative pathway, then the writers of that program could be fairly accused of assuming up front that the complex feature was irreducibly complex. Thus, evolutionary algorithms seem to be between a rock and a hard spot: assume a cumulative pathway and then you are unable to challenge irreducible complexity; assume there is no cumulative pathway and then you are unable to support irreducible complexity.
This would seem to equate IC and unevolvability ("no possible cumulative pathway") -- the intermediate functions used in the Avida research were distinct from EQU. Then I read your definition of "cumulative complexity" in "IC Reduced":
In order for our system A-E to be cumulatively complex, it is necessary to show a continuous pathway from a simple system to our more complex system. For example, if each of systems A, A-B, A-C and A-D exhibit functionality in their own right, then A-E may be cumulatively complex. (Anderson, "IC Reduced", p. 9)
Hence, you appear to use the word "cumulative" to mean "gradual (by any pathway)". But Dembski uses it to mean "gradual and direct":
Irreducible complexity differs sharply from another form of complexity that may be called cumulative complexity. A system is cumulatively complex if the parts of the system can be arranged sequentially so that the successive removal of parts never leads to the complete loss of function. An example of a cumulatively complex system is a city. It is possible successively to remove people and services from a city until one is down to a tiny village—all without losing the sense of community, which in this case constitutes the city’s basic function. (Dembski, "IC Revisted", pp. 5-6)
This difference leads to a different definition of the term "irreducible complexity" when they are contrasted. Behe and Dembski don't equate "irreducible complexity" with "unevolvability." (They allow the possibility of indirect pathways to contrived forms of IC, but argue against the practicality of such pathways for real-world examples of IC.) Try as I might to see how your understanding could be compatible with Behe and Dembski's, I was simply unable to refute Valerie's argument; I conceded her point. I assumed that everyone agreed -- especially Gil, Salvador, and DonaldM -- since there were no subsequent comments made in your defense. I'm glad to see Gil following up on this. I know that DaveScot has at least temporarily banned Valerie from commenting ("for boring the moderator with argumentum ad populum", "the torrid pace of [her] commenting", and "[making] no attempt [to learn] stuff [that's] not hard to learn" -- see www.uncommondescent.com/index.php/archives/886 ). However, I do hope that "Sunshine" (see www.uncommondescent.com/index.php/archives/884 comment #42) will have the opportunity to weigh in on this, too. Regards, j
1) Quantity and Quality together quickly exhaust the resources for chance probability. A cumulatieve pathway to Quantity of csi COULD be explained by the fact that after each event, the probability of the next event is the same as the previous event. One could say, "OK, this really was just that one chance in a trillion." But Quality of CSI cannot be explained in the same way. We may see what appears to be the Virgin Mary in the water marks left by a leaky roof. We are free to attribute the image to chance or not as we chose. But if we see there the detailed image of the Mona Lisa, complete with color, texture, brush marks and Da Vinci's signature, an accumulation of chance events as the cause is obliterated. 2) As evidence for random design one way vs. intelligent design the other way, I say the distinction between improved function and different function is imaginary. Calling one system an "improvment" over another is no proof of RMNS. Though the systems may have similarities, they are ever two separate systems. The eye of a squid and the eye of a man are two systems. Each is suited to a larger system; neither is an improvement over the other. Red Reader

Leave a Reply