|
Author
|
Topic: A sequence of tests for IC
|
Rex Kerr
Member
Member # 632
|
posted 19. March 2003 00:52
quote:
I've never heard of a knockout restoring function.
This happens fairly frequently in regulatory cascades, and the second knockout is called a "suppressor" of the first mutation. Suppressor screens are done all the time, and very often give interesting results.
As a first approximation, one can probably ignore suppressors, but to have much confidence in one's results they really need to be considered. I think it is entirely reasonable to try to get tests for various definitions of IC without worrying about suppressors, and then refining the tests from that point.
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 19. March 2003 02:28
Alonso said:
quote:
quote:
quote:
The case where removing D from the system, and function F* is preserved, yet singly removing any of the remainder of the parts destroys function was not described.
In fact, this was described. I mentioned that if we remove D from the system, and function F* is preserved, then we are left with system A,B,C where D was a redundant part. We then remove A from the system, and if function F* is preserved then we have system B,C where A and D were redundant. We then remove B and if function F* is preserved then we have the system C and therefore it is not irreducibly complex.
In case I was not clear enough, let me describe this case again. It was not the case in which we have functionality all the way down to a single “part”. Rather it is the simple case of an apparently IC “core” of A,B,C. Remove D, and the function still occurs. But (with D removed) remove either A, or remove B, or remove C, and the function stops. I assume this case is an IC core of A,B,C. This was not described in your test -- you might want to add this kind of case.
I know that your “test” was somewhat descriptive rather than exhaustive, but there is a difference from the “irreducible” core being a singleton (and thus trivial and not “complex”) and the irreducible core being smaller than the initial list of parts. I understand that the triple case A,B,C as “core” could be viewed as simply the equivalent of the starting case -- but then there is no difference between definition MP-1 and MP-2 in that case. The MP-2 added the notion of finding the “core”, and I was suggesting that the “test” should include procedure to locate the “core”. And that the procedure should give a case where the core was complex, rather than trivial. Your example of finding an irreducible core was not IC, and thus your test did not include the case of finding a core that was smaller than the whole, and was “complex”, and thus the case was IC.
Let me propose a version that contains this issue, and you tell me if it captures things correctly. I tentatively gave it a label, but this will be withdrawn if it is not agreed to capture an aspect of the definition of MP-2 correctly.
quote:
(tentative) MP-2-Test-2:
Structural Defintion
Start with system X with parts A,B,C,D that have function F*.
To test whether this is an IC system using MP-2 I remove either A, B, C, or D from the system. If the function F* is lost then the system is irreducibly complex according to MP-2.
If we remove D from the system, and function F* is preserved, then we are left with system A,B,C where D was a redundant part. (Try all other possible cases of redundant parts after done examining this case.)
With D removed, then remove either A, B, or C from the system. If function F* is lost then we have an irreducibly complex “core” of A,B,C according to MP-2.
However if we then remove A from the system, and if function F* is preserved then we have system B,C where A and D were redundant. If we then remove B and if function F* is preserved then we have the system C and therefore it is not irreducibly complex according to definition MP-2. Any reduction to a singleton part preserving function F* means that the system is not “IC” due to lack of complexity of the singleton “core.”
--
Alonso, I suspect you misread my third case, let me put it in words:
Specified function occurs when A,B,C,D are present.
Specified function occurs when B,C,D are present.
Specified function occurs when A,C,D are present.
Specified function occurs when A,B,D are present.
Specified function occurs when A,B,C are present.
However with any 2 or fewer parts, the specified function does not occur.
Is this IC according to MP-2? Should one test for this case. (I realize that any particular “test” won’t necessarily test for all possible matches to the definition. And of course one should not label such a test as though it did test for all possible combinations that match the definition, the test is not simply following the terms of the definition but is a more extended procedure to identify possible cases that match the definition.)
In my tentative test “MP-2-Test-2” we would identify A,B,C as an “IC Core”. And by extension of the method, we presumably would also identify (A,B,D), (A,C,D), and (B,C,D) as additional “IC Cores”. So is this still a case of IC according to MP-2?
Now let me give another case that is similar to the one just above, but with more parts:
Specified function occurs when A,B,C,D,E,F are present.
(Some intermediate cases, monotonically reducing to the following.)
Specified function occurs when A,B,C are present.
Specified function occurs when A,B,F are present.
Specified function occurs when A,E,F are present.
Specified function occurs when D,E,F are present.
(Now note that this case is monotonic -- that is all cases that have more than the “cores” listed also have the function occurring.)
Now this case is somewhat similar to the last, except starting with more total parts. Primarily it has 4 listed “cores”, just the same as the previous case that was declared to be IC.
Are (A,B,C), (A,B,F), (A,E,F), and (D,E,F) each “irreducible cores” of (A,B,C,D,E,F)?
Note that tentative test “MP-2-Test2” above identifies each such case as an “irreducible core” and thus A,B,C,D,E,F is “IC” because it has irreducible cores.
I’ll deal with the more complex case, and also “EP” later, after we have resolved this. The above is partially studying the issue I tried to capture in the hard-to-understand diagram.
Now suffice it to say that the “evolutionary definition” simply followed one potential pathway -- then declared “IC”. I suspect that more than one pathway needs to be followed -- is that not true? And if it is true, then the “test” description does not suggest that, and needs to be amended. It needs to say where one goes on to other potential pathways. Alonso’s reply shows the tendency to refer to some large body of discussion involving many topics. This is precisely why I want the test procedure to be compressed down to a single (possibly lengthy) description. At least it should be outlined in terms of its complexity and issues that cannot be resolved in a reasonable length post should be mentioned as left for later. Without any example test procedure, I don’t know how to find even one case of EP-1. I’ll examine your later post in more detail to see if a test can be constructed from it that I can understand.
[ 19. March 2003, 03:10: Message edited by: gedanken ]
IP: Logged
|
|
Daniel Edington
Member
Member # 421
|
posted 22. March 2003 16:52
quote:
Behe's Original Definition:
MP-1
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)
So what we want is a effective and conclusive means of testing the general theory of evolution, right? Toward that end I don’t believe Irreducible Complexity as it is currently defined (that is the definition as presented by Behe) represents such a test. Even Dembski’s “enhanced” definition did not fix the problems I see with IC. In order to correct this deficiency I propose discarding the current definition in favor of something more effective.
So why is it that I disagree with the current definition of Irreducible Complexity? Because it occurs to me that simply removing a component and seeing if the system loses its function is not necessarily the same as showing that evolutionary intermediates cannot exist. Thus such a demonstration tells us nothing about how such a system may or may not have evolved. However, I think that with a little work we could probably develop a conclusive test for the general theory of evolution based on the concept of IC.
I believe that this would involve a radical redefinition of the concept of IC. Although, I am not exactly sure what the final form of such a redefinition would be. But, while reading a review of Behe’s book “Darwin’s Black Box” I started thinking about it seriously.
“Does Irreducible Complexity refute neo-Darwinism?”
by Gert Korthof
http://home.wxs.nl/~gkorthof/korthof8.htm
As an initial first stab at such a definition, I would say it might be something along the lines of an IC system is one for which viable evolutionary intermediate forms, in going from some ancestor organism without said system to the modern organism with said system, can be shown not to exist. I realize this definition is rather clumsy, but it is only a preliminary working definition and is subject to change.
What we would then need is a test for is to see if any such evolutionary intermediates could or could not exist for suspected IC systems. We would essentially want to be able to press the evolutionary UNDO button and see what we get. Which brought to mind an article I recently encountered in the journal Nature. In which scientists claim to have done just this.
Nature, 06 March, 2003, Vol 422 No 6927, pp68-72
A test for IC would then possibly be something like this: By first proposing mechanisms for suspected IC systems (which has already been done in many cases) and then (the hard part) attempt to determine what the intermediate forms for that system might have been. Then genetically engineer organisms with those intermediate forms in place and see if they work. Not being a biologist I expect there are some technical problems with this idea, but in principle it should work as a test of evolution.
Dan
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 23. March 2003 00:51
Dan, that's an interesting idea, but it poses some problems. Specifically, design is an extremely powerful hypothesis; unless carefully constrained, it can explain almost anything. For instance, on the way into this building, I walked across a parking lot that had on the order of five hundred fallen leaves on it. Now, I could hypothesize that the position and size of these leaves was carefully chosen by a group of students who were skipping out on their finals. However, the parsimonious explanation is that the leaves fell from trees, were pushed around by cars and feet and bicycle tires, and eventually ended up where they are now.
It would not be good enough as a test to see if someone could place leaves in that pattern; I suspect that with a little attention to traffic patterns, someone could do it quite nicely.
So while I support the idea of tests for IC theories, I am not sure how they'd avoid the pitfall illustrated above. It could quite likely be done--but one would have to pay careful attention, I think.
Added in edit: what one would have to pay attention to is that the tests weren't just tests of our design skill instead of any test of what might be going on (evolution or design by someone other than us or whatever).
[ 23. March 2003, 00:54: Message edited by: Rex Kerr ]
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 23. March 2003 02:16
Alonso, back on page 2 you said:
quote:
I'll call my definition MP-3. But I digress.
Since this was not next to any definition, I can’t tell what definition it is referring to. I guess we can hold the designation reserved for your definition, wherever it is, but I ask that we stick to Micah’s specific format for designating definitions, and Yersinia’s suggestion of putting the definition in a quote form to identify it. So if you want a label to be accepted in this thread, it must be clearly attached to its definition. That is the whole point of this thread -- to make clear and unambiguous definitions and tests that do not require bouncing around in multiple threads to find the primary information. (If required, one can link to specific external documentation, but this must be identified by specific page bounds and properly designated references.)
This is the only way that we can have clarity in what we are discussing -- as my belief is that part of the difficulty in discussing IC is that various sides claim that there is obfuscation going on when definitions are not clear.
quote:
A knockout, as I think you are using it, is when a gene is removed from the system causing either preservation of function (a redundant gene product) or loss of function. I've never heard of a knockout restoring function. Of course you could not be referring to knock-outs done in the laboratory. In short, if you could give a short summary of the "knock-out" table I will be able to respond. Nonetheless, that even one test case successfully delinates and IC system (the first and second one) thus far, 3 test cases show a confirmation of MP-2. Furthermore, you keep distinguishing MP-1 from MP-2 as if MP-1 states that removing any component from any component from an IC system regardless of redundancy will destroy function. A charitable reading of Behe shows this is not the case, such as when he did not mention Hageman as the part of the "IC core" of the clotting cascade. Examples of a singleton part functioning has been shown to you as Hemoglobin, Dawkins's membranes, and PPi synthase.
I may have been using inappropriate terminology. By “knockout” I was simply referring to removal of a part. Definitions of IC are based on “remove a part”, and that was what I meant. That will lead, of course, to the issue of distinguishing just what a part is, whether the part is the genetic code such as a gene, or is the part a particular protein, or is the part the complex structure. And for a complex structure, where does one draw the boundary? These will be important issues, which are distinctions that are added to the issues already identified. I am wanting to very specifically identify things in this order: First a specific definition -- referred to by its designation, and that definition presented here in clear and unambiguous terms. Second construction of a logical level test for that definition. The logical level test does not have to capture all cases that apply to the particular definition -- so there may be many logical-level tests for a given definition of IC. Then such logical level tests could be refined to actual laboratory procedures (where by “laboratory” I don’t necessarily mean that one has to do it in a lab, but that one organizes a procedure of identifying actual cases by direct observation of nature). Here I am specifically interested in examining the properties of the logical-level test for two aspects: 1) Is it actually reducible to a laboratory procedure -- eg. has it escaped the difficulty of being an “argument from ignorance”. And 2) does it have any particular cases of interest -- e.g. does it contradict a case that was already accepted as having a different outcome, and why.
quote:
One can falsify whether a particular evolutionary pathway is IC by showing that each step is functional and therefore selectable. In this definition of irreducible complexity we can define it as some ID critics erroneously interpreted MP-1.
quote:
A single system composed of several well-matched, interacting parts that contribute to any function of the system, wherein the removal of any one of the parts causes the system to be completely non-functional.
This is a structural definition that can be tested without knowing that Darwinian theory exists.
I don’t understand what you mean “as some ID critics erroneously interpreted MP-1.” Micah Sparacio presented and labeled MP-1. He also presented and labeled MP-2, and made clear that they were separate definitions.
Let me see if I understand, by constructing one test for MP-1:
quote:
(tentative) MP-1-Test-1
To test whether this is an IC system using MP-1 I remove either A, B, C, or D from the system. If the function F* is lost then the system is irreducibly complex according to MP-1. However there must not be any singleton subset of the original components A,B,C or D alone which preserves function F*, and in that case the system is not IC due to lack of complexity.
If you want to argue their equivalence, then do so by constructing tests for each and demonstrate that the same logical-level test applies to both. (In other words list MP-1-Test-x and MP-2-Test-y as the same test, yet applied to different definitions MP-1 and MP-2, and we can examine if Test-x/y appears to test for the properties listed in MP-1 and MP-2.)
In previous quote you said “Furthermore, you keep distinguishing MP-1 from MP-2 as if MP-1 states that removing any component from an IC system regardless of redundancy will destroy function.” (Edited redundant phrase.) Now Micah presented MP-2 as Dembski’s definition, from (No Free Lunch, 285). (Note that while short, this is a complete and fully identifying attribution.) I was at Darwin Democracy and Design II conference, June 2001, where I personally heard Dembski discuss that there were problems with Behe’s definition, and that No Free Lunch would deal with those issues.
My specific notes from that are:
quote:
One thing in NFL is it supposedly "tightens" up Behe's arguments: Irreducible complexity is not sufficient. Something called "minimal complexity" will be the new argument.
This was several months before NFL came out, and indeed NFL contains the definition MP-2 as presented and discusses the “IC Core” concept in great detail. And it was after NFL that the terminology “IC Core” came into common usage -- whether by Behe or Dembski or others. I am interpreting “minimal complexity” argument that Dembski discussed as the identification of the IC core. You may wish to equate MP-1 and MP-2 definitions, but I am trusting Dembski that there was a need for a distinction. (What are the logical implication is another thing, my point was not to trust Dembski’s arguments unexamined, but that if even ID promoter Dembski thought there was a problem needing solution that there must be something needing scrutiny.)
Above “A charitable reading of Behe shows this is not the case, such as when he did not mention Hageman as the part of the ‘IC core’ of the clotting cascade.” This gets to the issue of where redundant parts are allowed, and where some parts that are not “redundant” are required. Can you or someone construct a test that applies to MP-2, for example, that gives a “correct” answer for this case. I shall make an attempt here, but someone else tell me if this is correct:
quote:
(tentative) MP-2-Test-3:
Start with system X with parts A,B,C,D that have function F*.
To test whether this is an IC system using MP-2 I remove either A, B, C, or D from the system. If the function F* is lost then the system X is irreducibly complex according to MP-2.
If we remove D from the system, and function F* is preserved, then we are left with system A,B,C where D was a redundant part. (Later one must try all other possible cases of redundant parts after done examining this case.)
With D removed, then remove either A, B, or C from the system. If function F* is lost then we have an irreducibly complex “core” of A,B,C according to MP-2. System X is “IC” because it has irreducible core A,B,C.
However if we then remove A from the system, and if function F* is preserved then we have system B,C where A and D were redundant. If we then remove B and if function F* is preserved then we have the system C and therefore it is not irreducibly complex according to definition MP-2. Any reduction to a singleton part preserving function F* means that the system is not “IC” due to lack of complexity of the singleton “core.”
Now D has been shown to be redundant, but for example test for combination A,B, D by removing C from original system X. If the function F* is preserved, but function F* is lost upon removal of any of A,B,D, then we have alternate redundant parts C or D in subsystem (A,B,C) or (A,B,D). However each of (A,B,C) and (A,B,D) are irreducible cores of System X, and therefore System X is “IC”.
As mentioned above, all combinations must be tested, and the sequences above are examples among all logical combinations and all such combinations need to be tested. Exits are: NOT IC if a singleton core is found to preserve function F*, due to lack of complexity. But with no singleton cores any number of irreducible cores of subsets may exist and each is labeled an “irreducible core” of system X. If removal of any combination of parts results in function F* being lost, then the system is IC, so long as there are no singleton cores that preserve function F*. Furthermore the example methods are generalizeable to arbitrary number of parts and not limited to 4 parts.
Will someone let me know if this is an accurate test for MP-2? Does it miss any important aspects? Does it capture too many cases? (I mean this without getting into the issues of identification of what the “parts” are, like issues of what the boundaries are -- let’s assume that there are clear identifications for the moment.)
It looks to me that MP-1 and MP-2 are distinct because MP-2 allows there to be multiple IC Cores. There is no requirement, as in MP-1, that there be a single element of the entire set that when removed causes function to cease.
What I don’t understand it this:
Take case A,B,C,D,E,G, producing function F*. Reduce to A,B,C and function F* is preserved. Reduce to D,E,G, and function F* is preserved. Each subset (A,B,C) or (D,E,G) is an IC core and can’t be reduced and still preserve F*. (Furthermore there are no singleton sets that produce function F*.) Is this IC? Can someone demonstrate the equivalent of any labeled test (tentative or not) that shows that this is NOT IC, if this is not considered IC by MP-2? And if this is a contradicting case, can someone then write a test that captures MP-2?
[ 23. March 2003, 03:10: Message edited by: gedanken ]
IP: Logged
|
|
Daniel Edington
Member
Member # 421
|
posted 23. March 2003 06:39
Rex Kerr,
This I believe is the beauty of my new definition. It is not designed as a test of design. It is designed as a test of naturalistic evolution. In fact it is designed as the test that Charles Darwin himself proposed.
quote:
If it could be demonstrated that any complex organ existed, which could not possibly have been formed by numerous, successive, slight modifications, my theory would absolutely break down – Charles Darwin
Obviously if any biological system were found for which evolutionary intermediates could not exist, then Evolution, as a purely naturalistic process, would break down. And while design would not be confirmed by this test, I think that even the most skeptical scientists would have to consider it at that time.
Dan
[ 23. March 2003, 06:41: Message edited by: Daniel Edington ]
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 23. March 2003 13:04
Dan: Obviously if any biological system were found for which evolutionary intermediates could not exist, then Evolution, as a purely naturalistic process, would break down. And while design would not be confirmed by this test, I think that even the most skeptical scientists would have to consider it at that time.
What I see to be the major problem with this is that unless ID can propose pathways, and predictions and mechanisms, it may be considered but should be quickly rejected as well, especially when scientific hypotheses are to be considered. At most one can claim that for some instances it has been shown that Darwinian mechanisms seem to be insufficient. But that seems to be no reason to conclude or even consider intelligent design.
Furthermore evolutionary pathways which proceed through co-option for instance may be much harder to reverse engineer.
But IC is not just used to assert the inability of Darwinian pathways but also the likelihood of intelligent design. The latter one seems to be 'begging the question' imho.
Unless ID can propose such pathways, mechanisms etc, ID's contributions to science seem to be limited to formulating rejection scenarios for naturalistic pathways. If ID wants to become a replacement for methodological naturalism it will have to do more than this. The question is: Can it?
Given Dembski's latests assertions it seems that such expectations of ID may be unrealistic. Others like Murray have argued much of the same from a slightly different perspective.
Another question to ponder: If we can reverse engineer the pathway we believe was naturalistic, have we not found that ID is required to increase CSI? Or would it show that CSI can arise through natural processes?
[ 23. March 2003, 13:08: Message edited by: Frances ]
IP: Logged
|
|
Daniel Edington
Member
Member # 421
|
posted 23. March 2003 18:41
quote:
Dan: Obviously if any biological system were found for which evolutionary intermediates could not exist, then Evolution, as a purely naturalistic process, would break down. And while design would not be confirmed by this test, I think that even the most skeptical scientists would have to consider it at that time.
What I see to be the major problem with this is that unless ID can propose pathways, and predictions and mechanisms, it may be considered but should be quickly rejected as well, especially when scientific hypotheses are to be considered.
I have no clue what your point is? However, whatever it is I don’t think it is my concern.
quote:
At most one can claim that for some instances it has been shown that Darwinian mechanisms seem to be insufficient. But that seems to be no reason to conclude or even consider intelligent design.
I don't think you understand the significance of what negative results for the above test would be. If biological system were found for which evolutionary intermediates could not exist, then Evolution as a purely naturalistic process, would be impossible for that system. While such a result would not indicate design, it would most certainly open the door to the possibility.
quote:
Furthermore evolutionary pathways which proceed through co-option for instance may be much harder to reverse engineer.
It goes without saying that some systems would be harder to reverse engineer that others.
quote:
But IC is not just used to assert the inability of Darwinian pathways but also the likelihood of intelligent design. The latter one seems to be 'begging the question' imho.
I think a that most ID thinkers have gotten ahead of themselves with respect to the significance of IC. The place to start is by testing current naturalistics theories. I think IC can be used as a definitive test of the concept of naturalistic evolution. A negative test on that account would, as I have already stated, open the door to the possibility of design. However, it would not say anything about the likelihood of design.
quote:
Unless ID can propose such pathways, mechanisms etc, ID's contributions to science seem to be limited to formulating rejection scenarios for naturalistic pathways. If ID wants to become a replacement for methodological naturalism it will have to do more than this. The question is: Can it?
Given Dembski's latests assertions it seems that such expectations of ID may be unrealistic. Others like Murray have argued much of the same from a slightly different perspective.
I am not in a position to say what ID wants or what it doesn’t. However, as I see it ID is currently limited to formulating rejection scenarios for naturalistic pathways and that seems to be the way they like it. So your question may be irrelevant.
quote:
Another question to ponder: If we can reverse engineer the pathway we believe was naturalistic, have we not found that ID is required to increase CSI? Or would it show that CSI can arise through natural processes?
I am not sure that it would show it, but I think that it would strongly suggest it. However, I am not exactly sure.
[ 23. March 2003, 18:46: Message edited by: Daniel Edington ]
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 23. March 2003 21:28
Dan: I don't think you understand the significance of what negative results for the above test would be. If biological system were found for which evolutionary intermediates could not exist, then Evolution as a purely naturalistic process, would be impossible for that system. While such a result would not indicate design, it would most certainly open the door to the possibility.
Why is it that the possibility for design would depend on the feasibility of evolutionary pathways? What about design through evolution for instance? Furthermore, why do you suggest that design would be non-naturalistic? In fact although design is proposed to be by some to be a replacement to methodological naturalism, it seems so far unable to differentiate itself from it. Michael Murray in "Natural providence (or design trouble)" shows how ID fails in this aspect.
My concern with ID is as you mention that it does seem to limit itself to formulation rejection scenarios. But that again makes ID not much different from methodological naturalism either.
[ 23. March 2003, 21:30: Message edited by: Frances ]
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 23. March 2003 22:40
quote:
My concern with ID is as you mention that it does seem to limit itself to formulation rejection scenarios. But that again makes ID not much different from methodological naturalism either.
I really don’t want this thread to become derailed, but this one issue seems relevant.
One needs to mind Rex Kerr’s point:
quote:
Specifically, design is an extremely powerful hypothesis; unless carefully constrained, it can explain almost anything.
The problem, of course, is that to do so would really get off topic. It could become relevant if a test or definition were posed here that explicitly dealt in a relationship of design to IC. But as it is we are just probing the logic of IC itself, and its relationship to evolutionary scenarios. I really prefer here to stick to the details of whether tests can be constructed for IC definitions.
Now very closely related to the issue of whether the tests for IC (as part of ID concepts) are “much different from methodological naturalism.” I think there is something to this, in that what I’m about to note about Daniel’s “test” is that it is only rejecting and not confirming of a particular definition of IC itself. We could get into issues of the degree to which science is strictly producing relational models which simply are not violated, and the degree to which science is producing relational models that are predictive in a useful way. I think that evolutionary theory appears to have a wide range of outcomes within any predictive framework that are consistent with the relationships of evolutionary theory.
I think that by examining the definitions of IC and tests for IC that we might be able to evaluate whether IC concepts share any such useful predictivity. But to do so I think we first have to find out if IC concepts even obey basic concepts of logic, and relevance to observation of nature.
So, do you want me to mark Daniel’s definition and “test” with numbers? I’ll do so in part, even though Daniel is not an ID proponent and thus his definition may not carry much weight in the ID movement. I just question whether I should be discriminating, or should label anything anyone wants to propose with a number, to be evaluated later by whomever decides to do so.
This may be the best policy. However I have tried to make a distinction between a “test” for IC as a laboratory procedure (or logical procedure from which a laboratory procedure could be derived by refinement), which additionally has the property of terminating in a distinct answer in a finite amount of time. Procedures that can only falsify an IC definition, but which don’t confirm an IC definition as relevant to a case are not “tests” but rather what I was calling “falsifications”. (These might be labeled “falsificaiton tests,” but I wanted a one word designator so I was using “falsification”.) This gets us to the issue of MN being a matter of falsificaiton, and not confirming a theory. I don’t think this gets in the way of my purpose in making the distinction, because all we are asking for is whether a given case is known to fit the definition of IC or not. If we have a procedure that terminates in a reasonable and finite time (with reasonably derived laboratory procedures) then we have a test. If we only have a procedure that falsifies the definition of IC as applying, then we have a “falsification”. And making the distinction in labeling is important in my opinion.
So here is Daniel Edington’s definition, numbered:
quote:
CS-3:
As an initial first stab at such a definition, I would say it might be something along the lines of an IC system is one for which viable evolutionary intermediate forms, in going from some ancestor organism without said system to the modern organism with said system, can be shown not to exist. I realize this definition is rather clumsy, but it is only a preliminary working definition and is subject to change.
(Clearly this definition deals in “construction sequence” issues, according to Micah’s categorization scheme.)
But for the “test”, I claim this is only a falsificaiton, and will label it thus. You can argue with me on that and I will change the label if you ask, but then I would claim that the “test” does not obey the character of a test as defined in this thread and thus is not valid as such.
quote:
CS-3-Falsification-1:
A test for IC would then possibly be something like this: By first proposing mechanisms for suspected IC systems (which has already been done in many cases) and then (the hard part) attempt to determine what the intermediate forms for that system might have been. Then genetically engineer organisms with those intermediate forms in place and see if they work. Not being a biologist I expect there are some technical problems with this idea, but in principle it should work as a test of evolution.
(Sorry about including commentary, and authors should revise their definitions as they wish.)
I include this as a “falsification” because I don’t think that the “hard part” of the test is one that can terminate in finite time of work. There are arbitrary number of pathways that could be taken. Now there may be ways to limit the “length” for the evolutionary pathway, but I see no way to constrain all possible cases. I understand that DNA is “digital” and thus has finite possibilities, for example, but the number of combinations which could be gone through is sufficiently large that no actual genetic engineering task could physically go thorough all the possibilities within a reasonably bounded string length of the genetic code within the lifetime of many generations of scientists. Thus I label this a “falsificaiton” and not a terminating “test”.
Now this labeling of the test is quite important in my view, when considering Rex’s issue that ID “…is an extremely powerful hypothesis; unless carefully constrained, it can explain almost anything.” We have yet another argument to be made to get from IC to “design”. And the test is only a falsification of IC, not a positive indication for this particular definition. Compare this to evolutionary theory, and definitions of types of “descent with modification” processes. Here we have a sufficiently narrow (though still surprisingly wide) range of potential outcomes for any case that we can verify that predictions are actually being followed. This is an important distinction -- the tests for evolution are based on predictive scenarios.
The usefulness of definition; of IC of CS-3 depends on an ability to test, and “CS-3-Falsification-1” seems to suffer from being in essence an argument in the form of an “argument from ignorance”. We only find that we have not falsified CS-3 when we lack a plausible evolutionary scenario, and if there were such a scenario which we were “ignorant” of the result of the “falsificaiton” test would change simply by making a theoretical discovery. Most outocmes in science, even though they are in essence subject to the falsificaiton issue, are not strictly written in terms of an argument from ignorance, and the predictive value of the theory behind the definitions and tests make them more substantial in their usefulness.
[ 23. March 2003, 22:42: Message edited by: gedanken ]
IP: Logged
|
|
Daniel Edington
Member
Member # 421
|
posted 24. March 2003 07:41
quote:
Now this labeling of the test is quite important in my view, when considering Rex’s issue that ID “…is an extremely powerful hypothesis; unless carefully constrained, it can explain almost anything.” We have yet another argument to be made to get from IC to “design”. And the test is only a falsification of IC, not a positive indication for this particular definition. Compare this to evolutionary theory, and definitions of types of “descent with modification” processes. Here we have a sufficiently narrow (though still surprisingly wide) range of potential outcomes for any case that we can verify that predictions are actually being followed. This is an important distinction -- the tests for evolution are based on predictive scenarios.
I think the type of scenario that you would call a “test” is one that you will never find. However, I have no problems with you labelling my "test" as a falsification. Of course if you could tell me how ID could possibly be constrained, I think I could give you a test for it.
quote:
The usefulness of definition; of IC of CS-3 depends on an ability to test, and “CS-3-Falsification-1” seems to suffer from being in essence an argument in the form of an “argument from ignorance”. We only find that we have not falsified CS-3 when we lack a plausible evolutionary scenario, and if there were such a scenario which we were “ignorant” of the result of the “falsification” test would change simply by making a theoretical discovery. Most outcomes in science, even though they are in essence subject to the falsification issue, are not strictly written in terms of an argument from ignorance, and the predictive value of the theory behind the definitions and tests make them more substantial in their usefulness.
I think all previous definitions of IC suffered from the same problem and my definition comes the closest to being testable. While it is true that evolution is highly sensitive to the initial conditions and the boundary conditions that arise over time, I don’t believe that there are in fact an arbitrary number of plausible scenarios. What we are looking for here would not be just any scenario with any intermediates. We would need to show that the proposed intermediate would have to be part of any plausible evolutionary scenario (an “all roads lead to Rome” sort of thing.) It all sort of depends on what system one is working with, but It should only take one. If ID is in fact correct then such a system would in fact have to exist.
At any rate it is all more or less academic, considering as you said I am “not an ID proponent” and thus I expect this definition to be rejected outright.
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 24. March 2003 15:20
quote:
I think the type of scenario that you would call a “test” is one that you will never find.
I’ve stated repeatedly that I am not qualified to produce these definitions for IC, that must be left for those who think they have useful definitions. What I am trying to do is to examine any such definition that is presented, and to find out the logical consequences as stated. This may point the way for those who wish to make such definitions to find better ones. Or it might point the way to show that they are not useful. I don’t know the outcome. And specifically on the test, I think that the way you posed the definition CS-3 is one where you are probably right. That’s why I am asking the ID community to provide the best definitions that they can come up with, and the best tests that they can come up with, so we can examine their logical consequences to see if they need improvement.
Without presentation of clear definitions, and without understanding their consequences, there will be a continuation of charges and counter charges that definitions are not clear, and that implications of those definitions are not being examined. Here I want to examine them in a somewhat abstract manner.
quote:
I think all previous definitions of IC suffered from the same problem and my definition comes the closest to being testable.
I completely disagree here. Now we have not yet even examined the issue of how crisply the concept of “part” has been defined, and if there are problems when the “part” being examined can have slight function changes. At present I am trying to examine logical consistency issues without even getting into that issue yet.
But doing so (and assuming a crisp concept of “part” as viable) I want to examine the consequences of those definitions. I’m not convinced that those definitions provide any problem for evolution just because they are matched, but with that proviso I think that they can be matched with effective tests. Go back and review the tests for MP-1 and MP-2 (including “tentative” tests) and tell me which of them does not terminate in an up or down answer, presuming one can objectively follow the issue of what a “part” is. I think they all terminate in a finite time, and the number of combinations that have to be examined are relatively small, for example for a 4 part system there are only 32,768 combinations of what happens when parts are removed. And the number of combinations is relatively small if monotonicity is assumed (e.g. there are no “inhibitive” aspects of parts such that removal allows function in a combination). And to perform the test for a 4 part system, there are only 16 combinations that need to be tested for function F*, in the tests provided. This is doable in reasonable time.
What is not clear is what are the implications of such a definition being met. But I don’t pretend to know the implications, that is why I am providing example cases within given tests and asking if those are supposed to be IC or not according to definitions given. At this point I am not sure that ID “experts” have described what I am calling “logical tests” (from which laboratory tests could be organized by extension to observation of actual nature) in sufficient detail to even see if the definitions given make sense.
I would have thought that the experts on ID would have had a greater interest in clearing up the clarity issues. I have provided tentative “tests” above, and it seems to me that someone could tell me if they accurately reflect the intended definitions.
[ 24. March 2003, 15:25: Message edited by: gedanken ]
IP: Logged
|
|
Daniel Edington
Member
Member # 421
|
posted 25. March 2003 15:39
quote:
I’ve stated repeatedly that I am not qualified to produce these definitions for IC, that must be left for those who think they have useful definitions. What I am trying to do is to examine any such definition that is presented, and to find out the logical consequences as stated.
Don’t short change yourself. If you are qualified to examine such definitions then you are also qualified to produce them.
quote:
for example for a 4 part system there are only 32,768 combinations of what happens when parts are removed. And the number of combinations is relatively small if monotonicity is assumed (e.g. there are no “inhibitive” aspects of parts such that removal allows function in a combination). And to perform the test for a 4 part system, there are only 16 combinations that need to be tested for function F*, in the tests provided. This is doable in reasonable time.
The problem here is that the answer to a question such as what happens when you remove a part from a system can only be answered in light of a specific system. There are no universal rules for this sort of thing. Unfortunately Biology is not like electrical or mechanical engineering and thus these sorts of questions must be framed in the context of specific systems. Still I am not clear what it all has to do with evolution.
Take your hypothetical four part system X and then lets say that it meets the criteria of IC based on MP2. Start with system X with parts A,B,C,D that have function F*.
Now say we test whether this is an IC system using MP-2 and we remove either A, B, C, or D from the system and the function F* is lost then the system X is then said to be irreducibly complex according to MP-2. The question left is, despite having show it was IC by definition MP2, have we also show that system X could not have evolved? I don’t believe we have. Definition MP1 and MP2 assume the only pathway open for the evolution of these systems is serial direct Darwinian evolution. That is that they had to evolve by the sequential addition of components.
A + B -> AB + C -> ABC + D -> ABCD
This however is not the case, and I believe only my definition takes the other possible routes into account. Although both MP1 and MP2 would work as a limiting case, in situations where it can be shown that the system must have evolved via a serial direct pathway.
The Journal of Theoretical Biology, 203: 111-116, 2000
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 25. March 2003 17:33
quote:
Don’t short change yourself. If you are qualified to examine such definitions then you are also qualified to produce them.
I’m not a biologist, and I’m quite sure that readers will note that I have been extremely critical of ID. As such having me produce definitions, and then examine them for consequences could be viewed as choosing that which will support some point I might make. I would never come up with ID theory like that myself -- and in that sense I really don’t think I will then be able to find a “working” definition of IC.
This is why I am asking those who have been supporting various ID positions to come up with the definitions. I just want to examine them for their consequences, and to understand how to test for them. For example if there are two different types of definitions -- one that has problems for evolution but can’t be tested, and the other that can be tested but doesn’t produce problems for evolution -- that would be significant. Conflating the two would be an exercise in obfuscation. But I was hoping that someone who thinks this is not the case would present some clarification of the definitions and tests for them, so I could understand how IC might actually pose a problem for evolution.
So far it doesn’t seem that there is much interest in clarification -- it might in fact be the case that such clarification is counterproductive to the aims of ID advocates. I would think that clarification would be important to ID advocates, so that controversy can be resolved rather than increased. But they’ll have to choose for themselves, we can simply watch and read for ourselves.
quote:
… The question left is, despite having show it was IC by definition MP2, have we also show that system X could not have evolved? I don’t believe we have. Definition MP1 and MP2 assume the only pathway open for the evolution of these systems is serial direct Darwinian evolution. That is that they had to evolve by the sequential addition of components.
Patience…
I can’t even find out if MP-2 precludes redundant systems. If it doesn’t even preclude redundant systems, then it doesn’t even pose difficulty with “direct” pathways. Below I shall re-ask my basic question, hoping that someone can answer it based on Behe’s or Dembski’s literature. Clearly MP-1 precludes redundant systems, but then poses little difficulty for evolutionary pathways that are direct in terms of being selectable at each point, yet that involve steps like you mentioned that are not strictly constructive addition of components while leaving the others alone. I think that is why Dembski posed MP-2, and if you read my notes from my listening to his personal statements on this you will see that Dembski felt that such adjustments were needed.
As to “I believe only my definition takes the other possible routes into account” you might be right. However CS-3 seems far too close to the intended consequence to be demonstrated, rather than the observable evidence to produce such a conclusion. You can’t just say we have a problem for evolution when we observe that we have a problem for evolution. I know you didn’t word it quite that way, but I don’t think that any ID advocate is going to accept CS-3 precisely because it can’t be demonstrated and can only be implied by an “argument from ignorance”.
---
So here is my question again. I ask it to help me to understand if the definition clearly shows what is being implied by arguments presented on the subject:
quote:
Dembski's Enhanced Definition:
MP-2
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)
Now tests have been given for MP-2, such as
quote:
MP-2-Test-1:
Start with system X with parts A,B,C,D that have function F*.
To test whether this is an IC system using MP-2 I remove either A, B, C, or D from the system. If the function F* is lost then the system is irreducibly complex according to MP-2. If we remove D from the system, and function F* is preserved, then we are left with system A,B,C where D was a redundant part. We then remove A from the system, and if function F* is preserved then we have system B,C where A and D were redundant. We then remove B and if function F* is preserved then we have the system C and therefore it is not irreducibly complex according to definition MP-2.
And my
quote:
(tentative) MP-2-Test-3:
Start with system X with parts A,B,C,D that have function F*.
To test whether this is an IC system using MP-2 I remove either A, B, C, or D from the system. If the function F* is lost then the system X is irreducibly complex according to MP-2.
If we remove D from the system, and function F* is preserved, then we are left with system A,B,C where D was a redundant part. (Later one must try all other possible cases of redundant parts after done examining this case.)
With D removed, then remove either A, B, or C from the system. If function F* is lost then we have an irreducibly complex “core” of A,B,C according to MP-2. System X is “IC” because it has irreducible core A,B,C.
However if we then remove A from the system, and if function F* is preserved then we have system B,C where A and D were redundant. If we then remove B and if function F* is preserved then we have the system C and therefore it is not irreducibly complex according to definition MP-2. Any reduction to a singleton part preserving function F* means that the system is not “IC” due to lack of complexity of the singleton “core.”
Now D has been shown to be redundant, but for example test for combination A,B, D by removing C from original system X. If the function F* is preserved, but function F* is lost upon removal of any of A,B,D, then we have alternate redundant parts C or D in subsystem (A,B,C) or (A,B,D). However each of (A,B,C) and (A,B,D) are irreducible cores of System X, and therefore System X is “IC”.
As mentioned above, all combinations must be tested, and the sequences above are examples among all logical combinations and all such combinations need to be tested. Exits are: NOT IC if a singleton core is found to preserve function F*, due to lack of complexity. But with no singleton cores any number of irreducible cores of subsets may exist and each is labeled an “irreducible core” of system X. If removal of any combination of parts results in function F* being lost, then the system is IC, so long as there are no singleton cores that preserve function F*. Furthermore the example methods are generalizeable to arbitrary number of parts and not limited to 4 parts.
The problems are that Mp-2-Test-1 don’t cover complex cases of 4 parts. And my tentative Mp-2-Test-3 doesn’t look for a non-reduncant core, it simply trims down until it finds cores and so long as none of them are singular it declares “IC”.
So I ask again is this IC:
System X consisting of ABCDEG produced function F*.
Subsystems ABC produce F*. Also DEG produces F*. Is X “IC”? Are ABC and DEG “irreducible cores”?
The problem is that the tests MP-2-Test-1 and MP-2-Test-3, simply extended to 6 parts, would give positive answers to my question.
There could be a different meaning more in line with a “core” concept, that is that there is a requirement that a fixed subset of all the “IC cores” must have at least two members. (Perhaps this is what is to be called the “IC core” rather than the individual subsets that produce function F*.)
If that is the case, then we need to write yet another test, as no test procedure has yet been written that identifies that condition.
--
Once this is resolved, then I would intend to pose cases where there are different functions, say H*, J* for different subsets of the parts. And then we could add into the discussion the issue of parts not being crisply identified. This might show if there are any logical requirements that evolutionary pathways not be present -- I suspect highly that this will not be the case but I need to work with the actual definitions and clear statements of how they are to be tested for.
Now I’ll tell you what I suspect:
I suspect that MP-2 will be shown to need modification, because selectable pathway cases will be demonstrated (at least in logical terms) that stay within the definitions and tests for that condition that are of finite length. What I am interested in is whether tests that start to capture other aspects (like Daniel is referring to) will not start to change from being finite to being open-ended and non-terminating. If the tests need to be modified to get at what is intended in IC, and such modifications wind up making them in the form of an “argument from ignorance,” then we would find that there are not any suitable tests that actually terminate and demonstrate difficulty for evolutionary processes.
But if someone can pose a test that actually stands up to logical analysis, and which terminates, that will be very interesting.
[ 25. March 2003, 21:19: Message edited by: gedanken ]
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 26. March 2003 00:29
I'm not sure if there are actual examples of this (DNA replication, perhaps), but I think Dembski's definition of MP-2 classifies the following obviously evolvable scenario as a case of IC:
Z binds somewhere in a cluster and catalyzes a reaction.
Z is duplicated to Z and Y.
Z loses the ability to bind to anything but Y.
Y loses the ability to catalyze a reaction.
Z's ability to catalyze the reaction is diminished through drift.
Y is duplicated to Y and X.
X accidentally helps Z's catalysis activity a bit.
Z loses the ability to catalyze anything without X.
(And so on.)
At this point we started with a one-component enzyme and ended up with a three-component system. Yet, if you remove any one component, the system fails, and thus meets the MP-2 criterion.
I don't think people who doubt macroevolution would even be disturbed by the successive degradation scenario given here.
IP: Logged
|
|
|
Printer-friendly view of this topic