RBH
Posts: 49 Joined: Sep. 2002
|
Two contiguous posts from the ISCID thread on the paper referenced above:Quote | Posted by RBH (Member # 380) on 23. May 2003, 18:17: Just to remind us all of the 'canonical' definitions of irreducible complexity, these are from ISCID's Encyclopedia:
Irreducible Complexity
Michael Behe's Original Definition: A single system composed of several well-matched, interacting parts that contribute to the basic function of the system, wherein the removal of any one of the parts causes the system to effectively cease functioning. (Darwin's Black Box, 39)
William Dembski's Enhanced Definition: A system performing a given basic function is irreducibly complex if it includes a set of well-matched, mutually interacting, nonarbitrarily individuated parts such that each part in the set is indispensable to maintaining the system's basic, and therefore original, function. The set of these indispensable parts is known as the irreducible core of the system. (No Free Lunch, 285)
[With a link to "irreducible core"] Irreducible Core
The parts of a complex system which are indispensable to the basic functioning of the system.
Michael Behe's "Evolutionary" Definition An irreducibly complex evolutionary pathway is one that contains one or more unselected steps (that is, one or more necessary-but-unselected mutations). The degree of irreducible complexity is the number of unselected steps in the pathway.
The first definition, Behe's original DBB formulation, is clearly an ahistorical one. There is no reference to the past or the pathway to the state of ICness so long as we interpret "basic function" to mean "current function" and assume that a system performs only one function or, if it performs more than one function, we can tell which is "basic." It is also the definition that specifies the operation necessary to classify a system as IC: the knockout procedure. "Interacting" can also be operationally determined by observing correlations between the behaviors of parts. The vagueness is in the term "well-matched." There is no way mentioned in the definition (nor elsewhere in DBB) for 'well-matchedness' to be measured. Hence operationally - that is, experimentally - we have only the knockout procedure and identifying interactions on which to determine IC or not-IC. On Behe's first definition, the programs that evolved to perform EQU meet the two operational criteria - knockout loss of function and interactions. Only the ill-defined "well-matched" stands between the programs and ICness.
Dembski's refinement of Behe's definition introduces two new elements: "basic, and therefore original, function" and "nonarbitrarily individuated parts." The first addition's reference to "original function" introduces history. In order to classify a system as IC we must know that the current function of some system was also its original function. The effect of this move is to definitionally eliminate cooption (which we know to be common in evolution) as a route to an IC system. Hence this definition is restricted to only those systems in which we know cooption did not play a role in the evolution of the system. This definition, in its reference to "irreducible core," preserves the knockout criterion.
The second addition in Dembski's definition is ambiguous. It is a negative prescription ('do not pick parts arbitrarily') but gives no guidance on what is non-arbitrary. In his NFL example of the flagellum, Dembski works with two levels. There's the 'parts of an outboard motor' level - power source, rotor, propellor - and the level of calculation - proteins. There is no clear justification for which level of parts to use for what part (!) of the definition; the choice seems to be arbitrary.
The programs that evolved to perform EQU do not meet Dembski's definition of ICness, since the final function performed by those programs is not the "basic, and therefore original" function. They coopted other functions. While some of those precursor functions are also performed by the final programs, other precursors were sometimes lost along the way. Hence the "original" functions were not always present in the final program.
Behe's "evolutionary" definition also invokes history. It requires that we know the complete pathway by which a candidate IC system evolved, so we can count the number of "unselected steps." This is also interesting for introducing the notion that "irreducible complexity" can take on values other than 0 or 1: "The degree of irreducible complexity is the number of unselected steps in the pathway."
By this "evolutionary" definition the programs that evolved to perform EQU are IC to some degree, since every step on the path to the programs that performed EQU was not "selected." In fact, some steps in at least some of the lineages leading to the final programs were deleterious and hence were selectively disadvantageous - there was selective pressure against them. Hence they display some degree of irreducible complexity.
Thus depending on the definition one chooses, the programs are IC, not IC, or IC to some degree, and we have no guidance in deciding which it is. Therefore unless and until Behe/Dembski, et al settle what IC means, it is useless from the point of view of doing meaningful research.
RBH |
and
Quote | Posted by Argon (Member # 276) on 24. May 2003, 12:06: RBH writes: Dembski's refinement of Behe's definition introduces two new elements: "basic, and therefore original, function" and "nonarbitrarily individuated parts." The first addition's reference to "original function" introduces history. In order to classify a system as IC we must know that the current function of some system was also its original function. The effect of this move is to definitionally eliminate cooption (which we know to be common in evolution) as a route to an IC system. Hence this definition is restricted to only those systems in which we know cooption did not play a role in the evolution of the system. This definition, in its reference to "irreducible core," preserves the knockout criterion.
You know, this definition would probably remove things like blood clotting, a large chunk of the immune system, and maybe even parts of the flagellum from the list of IC systems. For example, the blood clotting cascade is composed of numerous proteases that bear striking similarities to other proteases that are both ancestral to the clotting system and which have different "functions" in the cell. Thus the "original function" of most of the components had nothing to do with clotting. Ditto with the immune system. The current functions of flagellar components are mostly propulsion and cell adhesion, but parts of this system might have originated from a protein translocation system or pore. And so the "original functions" of the flagellar system components might not have been the same. How does one actually determine such things in ancient, ubiquitous systems that have undergone strong selection before diversification?
Obviously (to biochemists at least), it's a practical impossibility to be sure of the "original" function of any component. I think Dembski has a Platonic and "separate creation" view of organisms and biology. Ernst Mayr knocks that viewpoint down in a series of books.
In ruling out the possibility of co-opting other components Dembski seems to convert the IC definition into the truism that unevolvable IC systems are unevolvable. After all, does Dembski think that cells acquire large stretches of DNA that spontaneously appear out of nowhere and have no past? Yes, if one wanted to describe a system that had no past as "IC", one could do it. But then that definition would have little to do with the mechanisms by which evolution operates and would thus be orthogonal to the important question of evolvability.
RBH also wrote Thus depending on the definition one chooses, the programs are IC, not IC, or IC to some degree, and we have no guidance in deciding which it is. Therefore unless and until Behe/Dembski, et al settle what IC means, it is useless from the point of view of doing meaningful research.
Why should they be the final arbiters of what is and isn't IC? Behe spent plenty of time writing his book and developing his ideas. His whole thesis fundamentally relies upon the abilty to properly identify IC systems. Dembski also had a long time to develop a mathematical "model" of ICness. They have both observed and participated in many discussions about these definitions and the problems associated with their various criteria. They have also given many talks on the subject. Since both men understand the crucial importance of having useable, workable guidelines when performing research, particularly in a new area, I have little doubt that they would have let all these years go by without already presenting all that they could on this subject about the clarifications and important distinctions.
IC was first defined in Behe's book, DBB. As RBH mentions, it was an ahistorical definition. All subsequent changes made by Behe and Dembski require historical knowledge about a system and thus substantially change the nature of evaluation. I cannot see any good reason why they should bear the "IC" moniker without a subheading to indicate the additional criteria that have been met. For example, it can sometimes be simple to apply Behe's original, operational criteria to determine whether a system is IC. But what this does not tell us is whether such a system was evolvable. Thus we potentially have two classes of IC systems: evolvable or unevolvable. Once a system is determined to be IC, we can now apply additional tests to determine the subclass to which it belongs. Until then it should be placed in a third subclass: "evolvability unknown". Here is how I see the current organization:
Class: IC * Evolvability status: * Unknown * Evolvable * Criteria type 1: Intermediate steps reproduced. * Criteria type 2: Similarity to other evolvable systems. * Criteria type 3: etc.... * Unevolvable * Criteria type 1: Requires too many "lucky" neutral mutations. (from Behe IC v2) * Criteria type 2: No possible ancestors. (modification of Dembski IC v2) * Criteria type 3: Intelligent designer observed. * etc...
Personally, I'd prefer that the "IC" label be dropped from the subsequent "redefinitions". I think clever people could invent other, more appropriately descriptive labels that would reflect the actual criteria being applied.
[ 24. May 2003, 12:09: Message edited by: Argon ]
|
-------------- "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.
|