RSS 2.0 Feed

» Welcome Guest Log In :: Register

  Topic: The Death of Irreducible Complexity, Archive of ARN postings< Next Oldest | Next Newest >  

Posts: 49
Joined: Sep. 2002

(Permalink) Posted: Mar. 27 2005,21:01   

Since ARN's Moderation is twitchy and its archived threads occasionally disappear from view, I''m archiving two posts I made there on irreducible complexity.  This is the first.

In another thread jon_e provided a link to a recent paper by Dembski revisiting irreducible complexity.  jon_e was making the point that "irreducible complexity" is alive and well in ID.  I had previously scanned the paper but had not read it carefully.  Looking again at it tonight, I see that Dembski has made a significant change in Behe's original conception of irreducible complexity, a change that eviscerates the utility of "irreducible complexity."  Rather than being alive and well, in the light of Dembski's new paper irreducible complexity is dead on arrival.

To realize the nature of the change, it's first necessary to know what an "operational definition" is.  Very briefly, an operational definition is a description of the procedures (operations) used to measure the value of a variable.  So, for example, an operational definition of "temperature" is a description of how temperature is measured -- the apparatus used, conditions that apply, and steps performed in making the measurement.  The Methods section of research papers contain explicit or implicit operational definitions of the variables under study.

With respect to any system, "irreducible complexity" is a variable that takes one of two values, 1 or 0 -- present or absent, true or false.  So an operational definition of irreducible complexity is a description of the steps carried out to determine whether a given system is or is not IC.  In Behe's original conception, the IC value for a system is assigned to be "1" (true) if the loss of any part/element/component prevents the system from performing the primary function that it performs when it is whole -- a 'knock-out' operation -- and "0" (false) otherwise.  So Dembski wrote in 1998, two years after DBB    
Central to his [Behe's] argument is his notion of irreducible complexity. A system is irreducibly complex if it consists of several interrelated parts so that removing even one part completely destroys the system’s function.
Also, whether a biochemical system is irreducibly complex is a fully empirical question: Individually knock out each protein constituting a biochemical system to determine whether function is lost. If so, we are dealing with an irreducibly complex system. Experiments of this sort are routine in biology.
The operation specified for determining IC is knock out a part and see if the system still works: that's the operational definition of "irreducible complexity".

In the recent paper referenced by jon_e, though, Dembski adds another operation to the procedure used to determine the value taken by IC:    
Thus, removing parts, even a single part, from the irreducible core results in complete loss of the system’s basic function. Nevertheless, to determine whether a system is irreducibly complex, it is not enough simply to identify those parts whose removal renders the basic function unrecoverable from the remaining parts. To be sure, identifying such indispensable parts is an important step for determining irreducible complexity in practice. But it is not sufficient. Additionally, we need to establish that no simpler system achieves the same basic function.  (Emphasis added)
To determine whether a system is irreducibly complex therefore employs two approaches: (1) An empirical analysis of the system that by removing parts (individually and in groups) and then by rearranging and adapting remaining parts determines whether the basic function can be recovered among those remaining parts. (2) A conceptual analysis of the system, and specifically of those parts whose removal renders the basic function unrecoverable, to demonstrate that no system with (substantially) fewer parts exhibits the basic function. (Emphases added)
That last criterion is an IC killer, at least empirically.  One must show that no system that is simpler than the system under analysis can perform the function performed by the system under analysis.  (I'm leaving aside the other change, the "rearranging and adapting remaining parts" addition to the original knockout operation.  That change also raises problems for determining the value taken by IC.)

Note carefully that it's not sufficient to show that some subsystem of the system under analysis can't perform the function; according to Dembski it is necessary to show that no simpler system can perform it, regardless of whether that simpler system resembles the system under analysis or not.

It might be thought that I'm giving Dembski's words an uncharitable reading, but that's belied by Dembski's own example, sandwiched between the two quotations above:    
Consider, for instance, a three-legged stool. Suppose the stool’s basic function is to provide a seat by means of a raised platform. In that case each of the legs is indispensable for achieving this basic function (remove any leg and the basic function can’t be recovered among the remaining parts). Nevertheless, because it’s possible for a much simpler system to exhibit this basic function (for example, a solid block), the three-legged stool is not irreducibly complex.
Please pause and think about that for a moment.  

Now continue reading.

On Behe's original operational definition, that three-legged stool is irreducibly complex: remove any of the four components (three legs and the seat -- Dembski forgot to mention the seat) and it can no longer function "to provide a seat by means of a raised platform", and so on the original operational definition the stool is irreducibly complex.  But under Dembski's revised operational definition, a three-legged stool is not irreducibly complex because some simpler system (that does not contain any of the parts of the original stool) can perform that same function.

As a result, in order to show that a system is IC, intelligent design "theorists" must show not only that the system fails to perform its function when any part is removed, but they must also show that no other simpler system can perform that function.  That is, they must establish a universal negative.  And (ask your friendly neighborhood logician) it is impossible to establish a universal negative.  (Hint: black swans.)  Dembski is back in the inductive soup.  On Dembski's new operational definition, not even Behe's mousetrap is irreducibly complex!

In my less-than-humble opinion, in revising its operational definition Dembski has thoroughly gutted the notion of irreducible complexity.


Edited by RBH on Mar. 27 2005,21:12

"There are only two ways we know of to make extremely complicated things, one is by engineering, and the other is evolution. And of the two, evolution will make the more complex." - Danny Hillis.


Posts: 49
Joined: Sep. 2002

(Permalink) Posted: Mar. 27 2005,21:06   

This is the second of the archived posts.

I remarked above that "rearranging and adapting remaining parts" was still lurking.  Let me bring it out of the shadows.

Besides 'no other simpler system", Dembski's second addition to the operational definition of irreducible complexity is
... by rearranging and adapting remaining parts determine ... whether the basic function can be recovered among those remaining parts.
 As Dembski describes it, we have knocked out a part and found that the system's "basic function" is gone.  Now we must establish that the remaining parts cannot be rearranged and/or adapted to restore the basic function.  I've been trying to think of a good analogy, and I think I have it: a simple household mousetrap.

Behe argues in Darwin's Black Box that a common mousetrap is irreducibly complex on his original definition.  It consists of five parts (hammer, spring, catch, holding bar, and platform).  Remove any one of them and the basic function of the mousetrap is gone.  While critics have argued that a mousetrap is not irreducibly complex, Behe has vigorously defended it, mostly by arguing that rearrangements and adaptations are necessary to get the simpler traps the critics described.  Poor Behe, betrayed by his comrade in arms.

Now, alas, the mousetrap falls prey to Dembski's second test, for it is clear that one can remove any one of the five parts and with some modest rearranging and adapting of the remaining parts recover the basic function.  See here for animations of mousetraps employing one, two, three, and four of the parts of the complete mousetrap.

The only component common to all of the reduced and adapted mousetraps is a piece of wire.  Any of the other parts can be eliminated and with suitable rearrangements and adaptations we can recover the basic function.  Now consider testing an intact mousetrap to see if it's IC.  In the first step of our analysis we knock out any component.  As the illustrations show, we can "rearrange and adapt" the remaining parts to recover the basic function.  Knock out any one of the parts and the remainder are sufficient, with some rearranging and adapting, to serve the basic function.  (Some of the "adaptations" -- or perhaps they're "rearrangements" -- would actually eliminate some of the remaining parts.)  The simpler system may not function real well, perhaps, but of course evolution doesn't need a whole lot of relative advantage to build on.  And the criterion for IC is elimination of the "basic function," not merely diminution or attenuation.

What does this mean?  It means that by Dembski's operational criteria for irreducible complexity, even Behe's mousetrap, the iconic example of irreducible complexity, is not IC.  

I'll say it again: Dembski has eviscerated irreducible complexity as an empirical marker of design, and in the process has cut his colleague off at the knees.


Edited by RBH on Mar. 27 2005,21:11

"There are only two ways we know of to make extremely complicated things, one is by engineering, and the other is evolution. And of the two, evolution will make the more complex." - Danny Hillis.


Posts: 49
Joined: Sep. 2002

(Permalink) Posted: Mar. 27 2005,21:10   

Also for the record, this is Dembski's sole response:
IC will be around for a long time. The nonsimplifiability criterion that I introduce is not nearly as onerous as RBH makes out. True, nonsimplifiability, as I apply it to IC, says that no simplification is possible for a system performing the basic function. But basic function, as I define it, also includes the way in which the function is performed. Thus, it is no simplification of the bacterial flagellum to substitute a paddle, say, that doesn't spin, that propels the bacterium through its watery environment, and that is simpler. Any simplification of the bacterial flagellum would have to be a bidirectional motor-driven propeller. If there's a concession in my treatment of IC with the nonsimplfiability criterion, it is more than made up for in requiring IC only for irreducible cores. Irreducible cores extend IC to many systems that Behe's original definition did not cover. It is often easy to show that cores are nonsimplfiable even if the apparatus as a whole isn't.

"There are only two ways we know of to make extremely complicated things, one is by engineering, and the other is evolution. And of the two, evolution will make the more complex." - Danny Hillis.


Posts: 5
Joined: Mar. 2005

(Permalink) Posted: Mar. 27 2005,21:35   

I hope no one minds if I throw in a something a bit more concrete -- I find this example rather interesting.  One of Michael Behe's examples of irreducible complexity is the eukaryotes' axoneme.  As such, I think you may take an interest in a recent paper (2004) by David R. Mitchell at the Department of Cell and Developmental Biology, SUNY Upstate Medical University in Syracuse, New York.  He argues that gliding motors which rely upon undulation would have preceded bending motors, and that such gliding motors -- although far less efficient than bending motors -- would not have any special requirements in terms of their structure (e.g., the 9+2 geometry of the axoneme, the doublets of microtubules, the radial spokes, the central pair of microtubules) but that each of these innovations would have improved either the strength or control over the flagellum.

The article is entitled:

"Speculations on the evolution of 9+2 organelles and the role of central pair microtubules"

I have found several links for David R. Mitchell's article:

Speculations on the evolution of 9+2 organells and the role of central pair microtubules (pdf)

Speculations on the evolution of 9+2 organelles and the role of central pair microtubules (pdf)

Speculations on the evolution of 9+2 organelles and the role of central pair microtubules (html)

Lynn Margulis (a leading proponent of the view that symbiosis is the source of much of the novelty in the generation of new species) had argued that the axonemes (as well as the mitochondria and chloroplasts) were originally bacteria which combined with the eukyrotes through symbiosis.  Apparently she got the axonemes wrong -- although she was right about the mitochondria (whose closest known living relative is the rickettsia bacteria responsible for typhoid) and the chloroplast (which was originally a species of cyanobacteria -- clumps of which are called blue-green algae).


Posts: 37
Joined: June 2004

(Permalink) Posted: Mar. 28 2005,14:31   

Dembski's final response is odd.  Why wouldn't the tripod structure of a three-legged stool be valid as a "way in which the function is performed" ?  

It seems like he is trying to grope his way to a definition of IC that really does rule out evolution but it doesn't look likely that he will succeed without making it impractical to determine if a functionis "IC" under that definition or not.

Which would put IC in the same boat with CSI (or SHI as I prefer to call it)


Posts: 4
Joined: June 2005

(Permalink) Posted: Feb. 16 2006,01:14   

A non-sequitor, and year-old news, but the topic heading seems the ideal place and I've never noticed anyone refer to the item before.

Check out:

for a short article on the genetic 'etymology' of snake venom.  Highly complex but far from irreducible, analysis suggests that it is, genetically speaking, a brew of secretions originating in several organs.  How clever of the snake.  In many Australian Aboriginal legends, the Creator has the form of a snake (at least they can prove the snake exists).

  5 replies since Mar. 27 2005,21:01 < Next Oldest | Next Newest >  


Track this topic Email this topic Print this topic

[ Read the Board Rules ] | [Useful Links] | [Evolving Designs]