|
Author
|
Topic: Evolving Inventions
|
gedanken
Member
Member # 594
|
posted 04. March 2003 13:07
I wanted to also thank John for re-engaging in the discussion while being extremely busy. I’m trying to understand that position as well as clarify my own thinking. I’ll do my best to make my self both clear, and to suggest how things could be clarified in the TRIZ presentation so I could understand better.
The thread Erik Larson: Ray Kurzweil's Impossible Vision which discusses Dennett’s claims and the “frame problem” has a number of relevant discussions. I feel there are four major problem areas in TRIZ (and in many cases they are similar to or relate to problems with Dennett’s argument):
Issue 1: Over estimation of ability of the mind (i.e. inventiveness)
In my opinion many readers here and people who view problems as does Dennett are making a mistake of overestimating the capabilities of the mind. Very specifically in TRIZ paper Bracht identifies TRIZ as a method of going beyond “trial and error”:
quote:
Psychological inertia is the key weakness of trial and error because it limits the sorts of solutions the problem solver can discover; he can only find solutions that may be reached by variation of concepts he is already familiar with. As we shall see later, this has far-reaching implications for the types of problems that are solvable via trial and error and for the inventive process itself.
In studying the process of invention, Altschuller focused a great deal of attention upon the sorts of problems that need solving. He found that there are two types: inventive problems and routine problems. Inventive problems always involve some sort of technical contradiction that must be the resolved. A contradiction, in logic, refers to a statement that cannot be true, written as “A & ~A.” A technical contradiction embodies the same sort of logic.4 In an inventive problem, making one aspect of an object better will cause another aspect to get worse, thus, there will be a need for A and ~A. Altschuller points to the icebreaker ship for an example of a technical contradiction in an inventive problem.5
(Note that I shall refer to aspects of this quote in other contexts to follow as well.)
Now we could model the mind as requiring something beyond physical reality for it to work -- for example if the “mind” is some sort of antennae for spiritual or beyond-physical processes that do our thinking for us. Problems with this interpretation are the lack of evidence, and the strong evidence for the affect of physical aspects of the health of our physical brains in varying aspects of the thought process. (I’m making no claim about “spirit”, quite the opposite -- rather my point is that there is observable evidence of functionality of the brain to do what we observe in “thought”.)
So unless we are to depend on such assertions of beyond-physical aspects of the thinking process, then we must at least examine the information flow processes involved in inventiveness.
Here is an example that we clearly would consider “inventiveness”. What was Alexander Graham Bell doing when he and his workers accidentally invented the telephone? Answer: In one aspect, he was studying how sound could be translated to visual appearance. See (one) History of the Telephone:
quote:
Bell's father was a teacher of the deaf, and he imparted this knowledge to his only surviving son--Bell's brothers died of tuberculosis. Bell was the only telegraph inventor who understood how much we communicate via speech. He tried to build devices that would allow the deaf to see the pattern of speech sounds they were making and compare them with a kind of template. One of these devices was built around an actual human ear, which gave Bell an intimate understanding of how sounds are turned into waves by the bones of the middle ear.
These experiments led Bell to propose that the way to create a 'speaking telegraph' was to 'follow the analogy of nature'. Inventors often adopt this strategy: if nature has solved a problem, why not imitate nature? Bell, with help from his assistant Watson, built a kind of electromechanical ear in 1875, in which one spoke into a tube that ended in a membrane like the eardrum; a piece of metal was glued to the membrane and attached with a hinge to an electromagnet, which amplified the vibrations much like the bones of the middle ear [3]. This first telephone came about as a result of an accident. Watson had been trying to transmit telegraph signals; when one of the metal reeds got stuck, he plucked it in an effort to break it free. Bell came rushing into the room to find out what Watson was doing, because he had heard a strong signal in the other room from the receiving reed, strong enough so that he knew he could use this simple device to transmit speech. Like many other inventors, Bell took advantage of accidents and mistakes that would have been ignored by someone else.
What we see in this example of “inventiveness” is the effect of serendipity and of prior information that prepared the inventor to have that “click” moment. We must note that the TRIZ argument does not suggest that the “click” moment could not be a computable phenomenon -- as matching constraints and “goal” patterns is not the issue for the “frame problem” for example. The issue was the pathway to reaching the particular “inventive” state for consideration in the first place, not its recognition once arrived at. Very specifically refer to the Kurzweil thread mentioned in my introduction, and one sees that the objections in the “frame problem” were not about recognizing a goal state, but were limitations of search procedures suggested for the AI systems.
In some respects, the serendipitous input to inventiveness can be in “inheritance” of information (rather than direct structural genetic inheritance). In this regard there may not be a “contradiction” to overcome in the inventor’s mind, given his/her specific informational history, yet there is such a “contradiction” apparent to those who observe from the outside. There is a relationship of the inheritance of “information” in invention by minds, and inheritance of structure in potential invention by descent with modification. (And once again examine the Bell example of this “inheritance” of information.)
For the TRIZ argument to be plausible to many of us, it must accommodate the observed limitations on intelligence. The history of invention is replete with notions of the connected-ness to ideas that were current. (See Connections series and books by James Burke.) So unless “inventiveness” is to be associated only with extreme absence of any connection to ideas of the time or available to the inventor, or to be associated only with something akin to “divine intervention”, it should acknowledge the connection to available information in inventiveness. As we shall see in the conclusion, one of the difficulties I see is a lack of objectivity of application criterion of “inventiveness”.
(Once again I am not claiming that there are no examples that someone would wish to consider “divine intervention” or in which there is no understood way that the contradiction could have been resolved with recourse to information available to the inventor at the time. What I am trying to get at is the common occurrence of invention that appears to be within the scope of the TRIZ description, and consider information available to the inventor in those common cases being examined in that light in the large part.)
Issue 2: Underestimation of facility of random change in evolution
In a prior post, I gave an example of a potential pathway for a GA to “invent” the ice breaker, yet do so with only small change rather than leaps of logic. Does this mean that the ice breaker is now not an “inventive” solution after all? Or does it mean that the GA can design the ice breaker with only types of problem solving that are found in nature in evolutionary systems? That we do not yet understand the natural pathways that we model in the GA does not imply that natural local systems do not produce the results that we see as consistent in the vast grid of evidence for evolution.
While John may wish to question how the aspect of dividing up the hull into portions for consideration of changes could arise, I have shown various ways that this is a common aspect of both GA and evolutionary processes. The Force et al. papers I was discussing with Alonso are example of division of control for parts of an organism. The book Genetic Programming III: Darwinian Invention and Problem Solving by Koza et al. (1999 Morgan Kaufmann) is filled with examples of the inherent dividing up of control for changing of parts of a structure, including duplicating parts of structure. Duplication of genetic material leading to separated control is slightly different from actual duplication of structural elements themselves (such as in a program or FPGA logic structure), but the notion of dividing and refining parts separately is not at all unexpected.
John still seems to consider my example in a prior post of the stepwise evolution of the “Hull” to be an immediate addition of a “hole”. (John’s comments immediately started referring to “hole”). But the hull inherently contains a hole in the middle in which the bowels of the ship are held, and my example of narrowing of a section is simply a choice by the evolutionary process to change shape of a part of the hull structure rather than the overall cross-section. [Note here that my description is of an “I” shaped hull, not one exactly as John Bracht describes in the TRIZ paper, as I had slightly misunderstood the original example.]
In part I see a failure to recognize the length of circuity of evolutionary processes. By posing as a “contradiction” to be solved, one may fail to note that a long circuitous route has no step-by-step contradiction to resolve, though a direct pathway has such a “contradiction”. On this route the advantages may be much more slight than the jump in advantage by direct resolution of the contradiction. (And here I concentrate on how the slight jumps can be by relative or constrained random change of state, with selection by comparing advantages either in actual nature or by mental modeling to compare usefulness.) This relates to issue 3 and also to issue1 wherein the inventive solution may also have been found by a pathway of information that has no contradiction to resolve with the inventor, just with those who view the invention process from the outside with a compressed history.
This issue goes hand-in-hand with the next issue 3 identified, where the issue of small change giving a pathway to a solution is observed to have a degree of better fit rather than a complete resolution of the “contradiction”. A solution does not have to be a complete exact fit. I gave an example of small change designs that gradually evolved the ice breaker, ignoring the logical contradiction. The random “descent with modification” change can probe these pathways, and should not be underestimated.
To be meaningful to many of us, the TRIZ argument must divest itself of the apparent “argument from ignorance” form, and go beyond simply arguing that we don’t yet understand how natural systems could produce the “inventive solution” to the apparent contradiction.
Issue 3: Posing as strict logical problem, especially as “crisp” logic rather than approximate reasoning.
“Fuzzy Logic” and other methods related to neural networks allow partial solutions to be considered as though they were relevant by analogic reasoning in the human, thus opening up “search” hypervolumes that were not initially present. Things are considered that would normally be rejected in strict search because of initial inconsistencies found in a strictly “logical” framework, and thus the partial fit is recognized and allows for knowledge of correct focus for actual “search” later. (This is highly related to issues of the “frame problem” described as a limitation of AI.)
In Bart Kosko’s Neural Networks and Fuzzy Systems: A Dynamical Systems Approach to Machine Intelligence (1992, Prentice Hall)) he explains:
quote:
Fuzzy theory holds that all things are matters of degree. It mechanizes much of our “folk psychology.” Fuzzy theory also reduces black-white logic and mathematics to special limiting cases of gray relationships. Along the way it violates black-white “laws of logic,” in particular the law of escluded middle either A or not-A, and yet resolves the paradoxes or antinomies [Kline, 1980] that these laws generate. …
Refer back to the quote under Issue 1, to see the basis of TRIZ on the “contradiction” issue. And as noted in Issue 2, the random change works hand in hand with the small improvement. Fuzzy logic, as view for recognition of partial solution of the contradiction, allows for consideration of partial resolution of the apparent contradiction in small steps.
For example the “paradox” or contradiction of the mid-section of hull that is essentially a hole connecting the upper and lower hull sections is really just a modified hull itself. The “hole” is really just the same body cavity that always existed as part of the hull when it contained many “guts”, but the “guts” of the ship were gradually shifted up or down in my scenario.
The slight improvement is sorted out in the evolutionary process. This overcomes the “frame problem” in which there is no settling on a focus that is relevant, because the actual contradictions are not as stark and hard as has been suggested in Bracht’s TRIZ paper. The missing pathway may actually not be missing at all, just not yet imagined in our own human “frame problem” in which our own mental search for the actual pathway was not successful because we lacked information that nature already had.
New discoveries that relate to approximate reasoning techniques will allow AI systems to advance beyond the limitations of theorem proving systems that have computability limitations. These suggest how humans get beyond such limitations on computability. The TRIZ argument depends in part on the “frame problem” argument technique, and both must cast off those limitations. Solutions to the so-called “frame problem” of AI (which are used as arguments that AI systems will never become inventive) are relevant as well to understanding the TRIZ issue.
Issue 4: Posing as “search” and not recognition of “fit”.
This goes hand in hand with #3, in that the analogic nature of the problem solving is not recognized in a strict logical search. But this is also true as to inventiveness in terms as though the inventor was “searching” for a solution to the particular problem that was actually solved in the “invention”. (See Bell example from before, wherein Bell recognized the importance of a result he was not “searching” for. Remember that recognition of “fit” or satisfaction of constraints, even partial or fuzzy, is not the limitation on AI search.)
Very important is our recognition (as I repeat) that the problem in the AI “frame problem” is not a failure of the mechanistic AI system to recognize a solution once it was found by the process. The argument of the “frame problem” is that the AI search will not arrive at that solution because it spends to much time on focus that is not relevant and thus never gets around to examining the actual solution. This is claimed as an essential problem of AI “search”.
RBH has repeatedly emphasized this issue. GA systems must not be “greedy” so as to go for a quick casting off of somewhat unfit states, or else they will not produce interesting results. Modeling the GA as a “search” for a particular solution can create a model in which premature “selection” occurs and thus reduce the chance of something interesting actually happening.
And my long presentation on duplicate genes by Force, Lynch, et al. are about the very “preservation” of those states so that they may appear later in new circumstances, in the field of actual biological evolution.
Furthermore, for example, there is no reason for evolutionary processes to “search” for a solution like the ice breaker. Just if there happened to be a reason for an organism to have a niche in such an environment, then there might be a locally occurring fitness that drives toward that type of solution -- but evolution should not be modeled as is some of invention as a search for a particular type of solution or solution of a particular problem. As in the Bell example, the inventive solution was not found to the problem being “searched” for, but to a different problem serendipitously recognized. And once again, I emphasize that the “frame problem” which forms the basis of part of the TRIZ argument does not claim that recognition of value of a solution is the difficult aspect for physical systems (as in selection according to fitness to an environment), rather it is the pathway to arrive within a reachable and likely jump to the state in question.
One possible and very fair recognition of the difference between “inventive” solution and non-inventive solution is that the aspects must be assembled in a mental simulation before physical testing. RBH posted some research findings that suggest that “partial individuals” might be of interest in GA search procedures fir AI -- and this is because the subsystem being generated by constrained random GA trials could be varied independently in the AI search. This is not, of course, available to actual evolutionary processes of descent with modification, which must make whole individuals.
Mental processes (themselves modeled as some form of GA based process) would not be limited to such creation of “partial individuals”. So the intelligent agent could (using GA-like procedures) possibly be identified as making a connection to solve a contradiction by construction of “partial individuals” in a mental model (with mental simulation), wherein evolution must construct a complete individual.
But as in the Bell invention scenario, one must not over-emphasize the “search” aspect such that pathways that occur because of unrelated constrained random events and environmental change events can very often lead to serendipitous results. These could appear “inventive” and occur regularly and with very high frequency, yet never because of “search” for solution to a specific problem that is the contradiction envisioned in TRIZ. So TRIZ can provide limitations on true “search” that are not limitations of evolutionary behavior.
For many of us to take the argument seriously, it must take account of the serendipitous result that occurs because evolution is not search. And the argument must not get caught up in other aspects of the search metaphor that are not features of real evolutionary systems.
Moving the goalposts? -- NO, goalposts set on the fractal attractor.
I disagree that there has been significant “moving the goalposts” as has been in part suggested by some, and I will elaborate on what I mean by that.
An example of a fractal is a coastline, that stays very complex and jagged no matter how closely one looks. A “fractal attractor” is a dynamic pulling aspect of a nonlinear system that keeps the system moving in a chaotic manner, such that a fractal pathway is generated. Issues of supersensitivity to change are often associated with fractal attractors.
(I note that ISCID’s “complex systems” include “topics such as artificial life, cellular automata, chaos, criticality, evolutionary computation, fractals, parallel computation, self-organization, and others.”)
What I see are arguments that have a form of very complex conundrum or paradox with the only resolution in beyond-physical causation. (And I bring up the “beyond-physical” because TRIZ is applicable to the long history of evolution and not just a single point. As such a physical incursion of an intelligence to effect a solution to the contradiction would be likely to leave evidence beyond the change itself, at least if changes were occurring repeatedly over billions of years. For example what embodied or physical agent is available over billions of years to repeatedly effect such changes in our biological history? The lack of such independent evidence means that any such physical incursion has the effect of a beyond-physical action, even if it is not actually so, being totally beyond any experience with physical intelligent agents that could have made such effects.)
This supersensitivity is observed on one side, if one sees an observed sequence of evolution exhibiting a characteristic (e.g. “irreducibly complex” or “inventive”), that might be considered impossible without certain externally applied causal factors (yet factors for which we have seen no additional evidence whatsoever). On the other hand a slight change in the meaning, definition, or interpretation of terms would mean that the observed sequence did not have the original characteristic (e.g. it was not “IC” after all, or was not “inventive” after all).
Rather than moving the goalposts, the discussion has been constructed logically to fall within that boundary in which these slight sifts in definition make the issue supersensitive to flipping to one of the forms or the other. But then the entire issue may have been motivated by factors that make sitting on the fractal edge beneficial -- thus providing an “attractor” (in the fractal sense) to the supersensitive edge. If there are no obvious cases of “IC” or “inventiveness” according to a definition that is given, then there would be a loss of interest in the condition being described because there would be no challenge to evolution. And on the other side if there were obvious understandable or observable processes that created “IC” or “inventiveness” in biological evolutionary processes that were understood, then having the definition shade on that side would likewise provide no challenge to evolution. So psychological forces come into play so that the definition needs to be shifted slightly toward the fractal or “supersensitive” edge again.
We note the above “issues” 1 through 4. Vagueness in the dimension discussed in each issue allows the various issues to interact, and definitions in TRIZ to appear to conflict in the midpoint or ambiguity between aspects identified in these issues. As noted in the discussion of the four issues, there are strong interactions between the issues, and definitions provided give rise to ambiguity as to which case applies as differentiated by pairwise consideration of these issues.
These “fractal” property definitions may be occurring because of a desire to find problems in evolution that cannot be solved by understanding of physical reality in a scientific manner -- and by that I mean in a manner that exhibits understandable relationships like the relationships of physics in which the processes can be well understood with only local causation. So the goalposts are not being moved -- rather they are inherently placed in the supersensitive fractal space between possible interpretations so that they must shift -- unless an answer is acknowledged that some non-physical external causation is involved.
Now there is no necessity for us to confine ourselves to causation that is not beyond the physical. (ISCID is devoted to allowing such extension of consideration.) But to be scientific, we must be able to make distinctions based on observation.
We have not had a history in science in which there have been vast areas of observable processes that could not be understood. (I’m not now referring to historical sciences in which evidence must inherently be missing in significant cases, but to observation of physical processes that can be endlessly probed.) So unless we are to conclude that such “beyond physical” causation must additionally only be present when it is inherently lost in missing evidence, we must be dealing with causation that can be either analyzed in increasing detail, or one should give sufficient evidence that we should accept beyond physical causation as a regular feature of nature. We have seen a prime example in the issue of the nature of intelligence discussed above, which can be analyzed by present-day observational techniques and is not limited historical observation with inherently missing data points.
To improve on the acceptance of the TRIZ (or IC concept for that matter) I see an important aspect needing to be demonstrated: The relationships need to be made more clear so that readers don’t think that the arguments are being carefully placed on the fractal edge as described. If the description is being intentionally placed on the fractal edge, then arguments will inherently spin out with no solution because every attempt to resolve the issue with observable techniques will cause a shift in the participants to bring the issues back to a state of dynamic change. (I suggest this has happened with Behe’s shifting sequence definitions of “IC”.) Thus inherently nothing can be resolved for the participants involved. Resolving this appearance will help in the willingness of scientists to examine the issues.
[ 05. March 2003, 00:49: Message edited by: gedanken ]
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 04. March 2003 13:09
PS, at this point I shall only post brief responses to direct questions or issues of my posts. Work is calling.
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 04. March 2003 18:31
I found the post very illuminating, gedankin, but I take issue with two aspects of it.
First, evolution is perfectly well modeled as a search; the issue is not that it is not-a-search, because evolution and genetic algorithms are encompassed by the optimization-theory definition of search. The issue, IMO, is that people think of it like searching for one's keys, and then get all confused about how a random process could find their keys as well as they could. This intuition of everyday human search fails badly when the number of trials is immense, time periods are long, and the space to be searched is radically different than anything we are normally explicitly aware of. (E.g. there are extremely many "good solutions", whereas with keys, you either have found them or not.)
I don't disagree with your points, just the phrase "evolution is not search". I suspect that you may not disagree with my points either.
Also, I think the fractal attractor is an interesting metaphor, but needn't have any relevance to the ID debate. If your conclusions are hypersensitive to the exact form of your definitions, it just means that you can move your goalposts a long way without changing very much. I don't see how designing in supersensitivity has any redeeming qualities. Maybe the problem just doesn't admit any other type of description, but that means we'd have to be lucky to recognize a solution if we stepped on it.
For example, in the discussion of TRIZ, I see no evidence that the discussion has been designed to be supersensitive. We have some basic terms involved, as in any discussion. We haven't any solid definition of, say, "inventive" as it applies to biology. This isn't an issue of supersensitivity to definition so much as insufficiently precise definitions to begin with. It'd be just as hard to do hard science involving "fish", if you didn't know whether "fish" meant "had scales" or "lives underwater" or "chordate" or "is sold at Rubio's" (a Mexican food chain notable for their fish tacos). That's a problem when you're trying to use sloppy everyday concepts to do hard science--and that's a problem when words like "intelligence" and "inventiveness" and so on are used without careful definition and examples relevant to the field at issue.
And anyway, I don't see that IC/ID/TRIZ/whatever has to do any fancy fractal balancing to make important points. Find an invariant--irreducible complexity, for instance; show that biological specimens clearly have that invariant; show that natural processes clearly cannot produce that invariant; show that intelligent agents can. Done. Behe tried to do this, apparently, but it took him a few iterations to get a definition of IC that did what he wanted it to do; but by then, the clarity of the examples was mostly lost.
It would be fun to see a truly "fractal" discussion. But I just don't see evidence of it in this case--not any more than when it comes to any other discussion.
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 04. March 2003 22:00
I should probably say something here since I'm one who argues strongly that the search metaphor is misleading and deceptive when applied to biological evolution. I do so for several reasons.
Except in archaic form, or in a form that implicitly endorses some sort of 'great chain of being' conception, evolutionary theory does not itself require or even invite the search metaphor. It identifies some variables, some evolutionary operators, and some constraints, and purely through the automatic action of those variables and operators within the contraints, the evolutionary process sometimes results in populations occupying local optima in complex high-dimensioned fitness spaces. The occupation of the local optima is a by-product, a side effect of the operation of the system, not its goal, function, or purpose. In Rex's analogy, biological evolution isn't even looking for its keys!
Using the search metaphor invites all sorts of confusions, among them the notion that biological evolution has a goal, that we can look at the current state and say something about the system having sought it, and that purpose is inherent in, or imposed upon, the evolutionary process. None of those are the case.
Further, the search metaphor invites using what we know or hypothesize about search technologies (and optimization theory) as they are used in human applications, for example GAs, as somehow being direct models of biological evolution. They are not. Those kinds of human applications embody goals, designer-imposed contraints, and designer-imposed operations. All of those are red herrings in thinking about biological evolution, phantom rabbits that have been chased down many a garden path in everything from NFL to MESA.
Modeling evolution as a search process is a slightly different issue. Provided they are carefully designed, evolutionary search models can help illuminate some contentious issues, or at least help one frame hypotheses in a rigorous manner. For example, good AL programs can allow one to begin to look at questions about the role of nonlinear local interactions among 'dumb' agents in producing population-level behaviors of interest. There are a number of examples out there that are helpful in that respect.
I really think the connotations of "search" are more misleading than helpful, much as I think Dawkins' use of "selfish" in the title of his book is more misleading than helpful. It carries way more connotational weight than is justified, and leads to all sorts of unfortunate and unhelpful confusions.
RBH
[ 04. March 2003, 22:02: Message edited by: RBH ]
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 04. March 2003 23:00
Thanks, Rex,
On search, I think we are pretty much in agreement (e.g. RBH, Rex, myself). There are shadings of the terminology that can be misused. For example (using the metaphor of the ice breaker) there would not be a “goal” of finding a working ice breaker in real evolution. Rather if there was an advantage to be had, and a mutation went towards that advantage of cutting through ice, then a reproductive niche could be found.
RBH pretty much covered the subject in his reply.
But let me re-emphasize the issue of the “greedy search”. The GA that gets greedy will fail sometimes because it does not take large enough leaps to get out of local maxima. What Force et al. shows is that the genes (control elements) can develop for a while without getting “greedy” in real evolution. Researchers have found the need for precisely that in GA simulation, if I am not mistaken.
On fractal edge:
quote:
For example, in the discussion of TRIZ, I see no evidence that the discussion has been designed to be supersensitive.
But then look at John’s reply to me some posts back:
quote:
This is an interesting example because it illustrates my thesis so nicely. The flaw in this example is that it incorporates inventiveness in the re-working of the "genetic" regulatory system such that different regulatory networks result. Gedankin calls them "extra design segments" but as far as I can tell these are separate regulatory modules controlling the shape of separate parts of the hull. This is precisely the sort of inventiveness I'm talking about--it's a re-engineering of the hypervolume such that new inventions are possible. However, I'm not at all sure that Gedankin is suggesting a hull with a hole through it; it almost sounds like an I-or-T-shaped cross-sectional shape to me. At any rate, it is surely possible for a ship with an "inventive" hull to exist but to be morphologically identical to a "normal" hull, if the diameter of the hole is set to zero. However, my argument is that getting to the inventive hull in the first place (where one has the requisite "genes" for hole size, position, connectivity, etc) is impossible without intelligent input. By assuming the inventive change occurs without intelligence, Gedankin has simply assumed the point in question and made a nice, logical circle.
First to John,
I was mistaken in my initial reading and indeed have been describing an “I” shape (not with a hole between sides, but the only reference to “hole” as the trunk holding the top and bottom of the “I” with a hole passing between decks. I think this is also “inventive” so the issue should remain similar. But then the “hole” as John described is more dramatic of a topological change. In some respects it differs no more than a more extensive morphological change, but I think my 1 deck to 3 levels is sufficiently novel.
However continuing my point, I think there are difficulties with the “hole” pattern with thin “sides” that cut the ice. I don’t think that I could describe an evolutionary approach because I don’t think it would work. (Are there any real tested icebreakers that work that way?) Since to my estimation it would not work (because it could not turn in the ice) I think my idea is better. As such I could not give an evolutionary pathway because I personally think that the side supported version would be selected out -- and as such I don’t see how I can construct a reasonable scenario.
Now back to Rex,
This points out the very point I was making about being on an “edge” between possible interpretations. (And a “fractal” edge because the personalities make small changes in how they describe the meaning -- strictly ones of interpretation and not reasonably called “moving the goal posts”.) Here we have a question of just what is “inventive”. Is my simpler “I” form construction “inventive”? It has a possible pathway. So if it has a pathway, can it be declared as not inventive? And if it is “inventive” could it actually develop at all -- meaning are there any examples in actual biology that actually are inventive? (Is going from a land animal to the whale just “sifting through alternatives”?)
What I mean by supersensitive is that the decision (is it “inventive” or is it “not inventive”) is very sensitive to interpretation precisely because it is situated in a difficult to interpret region of vagueness. If things that are easily understandable by microevolution were “inventive” then we would have no argument. And if we are talking about a degree of evolution that is sufficiently complex as to not have readily available explanation -- then we don’t have readily available explanation
quote:
And anyway, I don't see that IC/ID/TRIZ/whatever has to do any fancy fractal balancing to make important points. Find an invariant--irreducible complexity, for instance; show that biological specimens clearly have that invariant; show that natural processes clearly cannot produce that invariant; show that intelligent agents can. Done. Behe tried to do this, apparently, but it took him a few iterations to get a definition of IC that did what he wanted it to do; but by then, the clarity of the examples was mostly lost.
Behe’s shifting definition is precisely the point. It suffered from the same problem. The original definition simultaneously (due to lack of clarity) had some people claiming that there were no examples of IC in biology, and others were claiming that there were well explained examples of IC. The slight shift of viewpoint (within the lack of clarity) pushes people to accept one view or the other or to be undecided in the middle. A slight shift in interpretation makes nothing IC. A slight shift in interpretation the other way and IC is found commonly enough to be explained and therefore supporters claim it wasn’t IC in the first place.
My claim is that all the definitions of IC are carefully positioned to be in that fuzzy inconclusive region. But it has become less fractal as the latest definition approached a tautology: (Paraphrasing) It’s IC and therefore hard to produce by evolution if it is improbable to be produced by evolution.
(Note it’s still right on that edge. The slope or clarity of the transition has increased, become more sharp. Nothing can be declared as present in evolution, and IC, simultaneously. Either it wasn’t IC, or it wasn’t evolution. Of course that was Behe’s point. But now find the missing information in the ‘argument from ignorance’ form and even this version of IC changes its declaration -- as soon as the evolutionary pathway is found then it is no longer IC because it is now probable in evolution! The only stabilization is because that fact is so blatantly obvious that nobody argues it any more with the latest definition. It’s still a dynamic loop, but possibly not “fractal” because it is so easily understood that nobody gets caught in the logic loop.)
But while the problems were not so obvious, the unclear definitions allowed for easy shifting past the demarcations. As long as an explanation of how the structure came about was not forthcoming, it could be a good example of IC. But once explained, the interpretation could be shifted to a claim that it was not IC after all. That lack of clarity -- just where beneficial to the IC argument -- gives supersensitivity of whether or not it is IC to slight shifts of interpretation. It is the question of the shift wherein the output (IC or not) is supersensitive to a shift in interpretation of the definition.
(Yes this allows the moving of the “goalposts” in reality -- but with plausible deniability. That’s my opinion. But it was so easy for someone who has the idea that they were on to something to cling to a particular model that I don’t think they were really “moving” the goalposts because they were consistent all along, even if changing the claim from “IC” to something else. The lack of clarity allowed this without the claimants actually doing anything that they would think was a subterfuge. The fault lies in allowing something with such lack of clarity to be used in claims as though they were objective.)
To bring this back to “evolving inventions”, I think this is precisely the case with “inventiveness” here. If changes are made in the definition of “inventiveness” they will not turn out to be such that they either show lots of known cases of evolution are “inventive”, or that there are no cases of “inventive” in actual biological forms (except of course really genetic engineered). Now if there really are clear cases of unembodied designers making inventive jumps in real organisms, then there might be clear cases that can be presented that will show this property -- both clearly inventive, and clearly impossible for an evolutionary pathway. But if there are not actual cases of unembodied designers making inventive jumps, will the definitions be shifted so as to not be in that fractal region between being demonstrated and not having plausible cases observed? (It becomes “fractal” due to the lack of frequency of cases of clear inventive jumps occurring without any possible evolutionary scenarios -- while the participants are still looking for those cases. There is no desire to put it on the fractal edge -- it just happens due to the nature of the search for an unembodied designer that leaves no evidence except that which can be searched for with an “argument from ignorance” form of argument.)
(Sorry, that wasn't short, was it.)
[ 05. March 2003, 00:22: Message edited by: gedanken ]
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 05. March 2003 02:13
Hi John,
I appreciate your encouragement. And while I believe that I and others have made a pretty convincing case, in the interest of fairness I would like you to help the discussion move forward by providing for a more rigorous definition of the terms used by you. After all as you suggest, a rigorous argument is required and while I believe I have provided such based on your own publication as well as based on the publications of others I do entertain the possibility that you our differences are due to problems with lack of a rigorous analysis and definition on your part. So before I spend more effort explaining the world of GA's I want to make sure that we are still on the same track. I am looking forward to a more rigorous definition that may help us resolve some of our differences. So far however I would argue that both your assertions as well as the various works on inventive design indicate that GA's which can vary their state space/hyperspace can indeed be considered candidates for creative/inventive solution.
I think we can all benefit from you defining more rigorously the terms used by you.
Examples I would encourage you to address include:
Hypervolume
Creative/inventive
As I have said before the many evidences seem to undermine your basic assertions about GA's namely that they cannot increase their hypervolume/parameter space.
Let's first resolve this.
In Christ
F
IP: Logged
|
|
Nel
Member
Member # 614
|
posted 05. March 2003 12:19
Ged:
Behe’s shifting definition is precisely the point. It suffered from the same problem. The original definition simultaneously (due to lack of clarity) had some people claiming that there were no examples of IC in biology, and others were claiming that there were well explained examples of IC. The slight shift of viewpoint (within the lack of clarity) pushes people to accept one view or the other or to be undecided in the middle. A slight shift in interpretation makes nothing IC. A slight shift in interpretation the other way and IC is found commonly enough to be explained and therefore supporters claim it wasn’t IC in the first place.
Nelson:
I really don't understand the logic here, and it is most certainly a perversion of the definition of IC. What I think you mean by "explained" is that an evolutionary story can be told about it. However, that does not show that it is not IC. In fact, if you are telling a story and the story is exclusively always a co-option story (as it is with the T3SS) then you are definitely dealing with an IC system. The concept of IC is quite simple, when you have a collection of parts that contribute to the basic function, there exists a core set of parts which you cannot remove or you lose that function.
Where Behe gives the definition for IC, on the next page, he realizes that different parts could have different functions when he allows for circuitous routes. Nonetheless, Behe's definition is clear even in the most basic form, you know a system is IC when you can't reduce it's basic function to one part. If you can't do that, you have found an IC system.
There was no shifting of this definition, in my opinion. The definition was simply put in more precise language.
Ged:
While John may wish to question how the aspect of dividing up the hull into portions for consideration of changes could arise, I have shown various ways that this is a common aspect of both GA and evolutionary processes.
Nelson:
Of course what is important is how a gradual process can invent something new, and not how a chance mutation can partition gene (or a hull). In fact, no biological example has yet been given as of yet (or that I have thus far seen).
Ged:
The Force et al. papers I was discussing with Alonso are example of division of control for parts of an organism.
Nelson:
Actually what the Force paper shows is one thought to be common consequence of gene duplication, which resulted in a partitioning of function. This changes the classical result where the original function was retained, and the other gene aquired a new function. I know this seems like I'm not disagreeing with what you are saying, but there is an important assumption that you have in the way you are describing the situation, in that it will inevitably lead to a new invention (or IC system or whatever). By explaining what the raw data shows, this can be seen as nothing more than question begging. Since all that is left is the hope of beneficial revision of parts and their subsequent step by step evolution to a new IC system. Unfortunately this has not been demonstrated.
Now, there is an important twist to this story. Ged states:
quote:
However since the duplicate in the DDC model only partitions control of function or morphology, without changing it, there is no disadvantage. This means that the duplicated gene will be more likely to be carried to future generations than in the classical model! (And furthermore that is confirmed by observation!)
However, the reason why Zebrafish would keep the partioned genes is because it needs the two functions. This would more likely point to functional constraint. Whereas in the classical model, the gene was free to mutate all it liked without any harmful effects, finally arriving at a new function.
It is also not true that there is no "disadvantage". I am not going to argue that it is impossible for gene duplications, or any other form of mutation, to give rise to inventions or IC systems. I am saying that it is quite unlikely, and that such inventions are either the result of direct intelligent intervention or the result of programmed "smart" systems, as with Shapiro's natural genetic engineering hypothesis. We see inventions happening even today as with the 2,4-DNT example that I think is a nice candidate for natural genetic engineering.
I am also not saying that gene duplications cannot give rise to new functions. Just that there are some rare (and simple) examples and that this is also unlikely, not only because of the caveats that John Bracth mentioned, but also because of the reasons I will add below.
My interest in starting a new thread concerning gene duplication is that there may be a hint of teleology when all is said in done. And this is partly inspired by reading a quote from one of the authors of Gedankin's references. But I think I'll just leave it as a reply in this thread for now until I get more info on these mechanisms or things like them.
Speaking of Lungfish, Ohno stated:
quote:
"By establishing such a system [NA:rerring to gene duplication] the organism effectively forfeited an opportunity for further evolution. In a manner of speaking, the genome became frozen, while containing enormous genetic redundancy. It is clear that in doing so such a lineage reached an evolutional dead end."
Ohno S., "Evolution by Gene Duplication", 1970.
Gene duplication in Drosophila seems to have also caused either slightly reduced eyes or severely reduced eyes. Clearly there is no only a reduction in function, as seen with Zebrafish, but a reduction in structure.
I may be wrong here, but it began to dawn on me that this is as much a problem for ID as it is for Darwinian evolution. Since gene duplication doesn't seem to be a mechanism used by the organism to evolve, how would a designer handle the problem of gene duplication, especially if the consequences can lead to an evolutionary dead-end. I then tried to find a paper that gave me some idea of the rate of gene duplication.
One of two things can happen when a gene is duplicated. The classic scenario can be described as follows:
quote:
New genes, in addition to the old ones, must be made. It is thought that this occurs as a two-step process. First, a gene (or group of genes) is accidentally duplicated so that a chromosome now carries an extra copy of it. Because only one good copy is needed to produce the original protein, the extra copy is free to accumulate mutations without harming the organism. After a time, enough changes may have accumulated in the duplicate to give it a new function."
Lewontin R.C., "Human Diversity", 1995
As I cited in earlier in this thread, a couple of papers by Force et. al. show that the majority of gene duplications do not end up with new functions. In fact, there are several models of gene duplications. The paper argues that the most common form of gene duplication results in partition of function.
A review by Lynch shows that the rate of gene duplication is not only exceedingly high but it seems to result very commonly in gene silencing.
quote:
The most common fate of duplicate genes appears to be the simple silencing of one member of the pair. The average time before silencing of one duplicate gene pair member is ~4 million years in animals
Evolution and Gene Duplication Michael Lynch,Science (2002)
They go on to propose that the silencing of these genes may play a significant role in the origin of new species. I think that is very interesting from a teleological perspective, but this has not been tested.
This paper (Lynch, M. and Conery, J.S., 2000 (Nov. 10)). The evolutionary fate and consequences of duplicate genes. Science, v.290, p.1151-1155, shows kind of the same thing that gene duplication, more often than not, become quickly silenced ,what they refer to as "nonfunctionalization".
In this review (Epigenetics: Regulation Through Repression Wolffe et. al. Science 1999 October 15; 286: 481-486.) when these researchers introduced copies of genes into an organism, what they got was gene silencing.
Another paper, (Otto SP, Yong P. Adv Genet 2002;46:451-83 The evolution of gene duplicates), shows the same thing but offers this insight:
quote:
"In summary, gene and genome duplication events are commonplace. There is currently no good evidence that organisms have evolved mechanisms to
increase the chance that their offspring carry gene or genome duplicates, although mobile elements within organisms clearly have
evolved mechanisms to promote their own replication. In contrast,evidence from several species suggests that a variety of mechanisms
have evolved to eliminate or silence gene duplicates soon after they arise. Thus, there is every reason to believe that the appearance of
duplicated segments within chromosomes represents nothing more than a series of historical accidents."
It appears therefore, that complex systems are used to ensure that gene duplication does not occur to often. There are complex mechanisms to promote the duplication of certain genes but not all genes, and to silence most but not all gene duplications. This can be interpreted through a teleological filter.
In short though, any story about gene duplication must also take into account these issues.
[ 05. March 2003, 15:23: Message edited by: Nelson_Alonso ]
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 06. March 2003 00:49
Alonso suggests that Behe’s definition is clear after all. So why does Dembski have to “salvage” it on page 280 of NFL? Here is a discussion of that, I don’t have NFL at hand at the moment.
What I mean by “changing” is that interpretations by different people are different. Just look through the discussions on IC, say on ARN, and you see different people posing systems all the time that meet the requirements, then someone else showing an interpretation that claims the structure was not IC after all. If the definition was clear and objective, it would not be so difficult. Science is about repeatability.
quote:
What I think you mean by "explained" is that an evolutionary story can be told about it. However, that does not show that it is not IC.
But the common anti-evolution claim of a “story” is that it is not highly accepted consensus that the particular pathway is the one and only valid explanation, rather that it is one of many possible pathways. But the claim of IC is that it cannot have a developmental pathway with sufficient probability to be reasonable. (And if we are to apply the EF as meaningful, it must have a possibility of such a pathway occurring of less than 10^-150 -- hardly such a case when a possible pathway can easily be given.)
One problem with the “irreducible core” is exactly the same problem as with the TRIZ argument -- to get us to the subject of this thread. It shows multiple co-option pathways as likely. (Indeed co-option is the story of evolution, to suggest a problem with all examples being of co-option is just plain silly.) What the “core” shows is the likelihood of loss of complexity at some point so as to remove unused aspects that are not part of the current function during the intermediate transitions. In the TRIZ argument it suggests that the structure was reachable and thus not “inventive”, even though in the contracted form there is an apparent “contradiction” in the claimed IC structure.
You can always argue that one essence of the definition of IC, for example, was preserved over the variations given by Behe and Dembski. But during those variations, there were also changes in how many cases it actually applied to as others interpreted those definitions, and also changes in how probable were occurrences of ways that were understood to get around the apparent difficulty. So in this we see not only shifting of interpretation, but shifting of raw definition (even if certain essence were maintained) and this shifting is part of my point of being on an “edge”.
And in the second place Behe himself has made a series of different wordings: His initial definition did not speak of “core”. Those differences are differences, even if they speak to similar issues. For example in “Reply to My Critics,” Biology and Philosophy 16: 685–709, 2001, Behe said:
quote:
However, commentary by Robert Pennock and others has made me realize that there is a weakness in that view of irreducible complexity. The current definition puts the focus on removing a part from an already-functioning system. Thus, seeking a counterexample to irreducible complexity, in Tower of Babel Pennock writes about a part in a sophisticated chronometer, whose origin is simply assumed, which breaks to give a system that he posits can nonetheless work in a simpler watch in a less demanding environment.5 The difficult task facing Darwinian evolution, however, would not be to remove parts from sophisticated pre-existing systems; it would be to bring together components to make a new system in the first place. Thus there is an asymmetry between my current definition of irreducible complexity and the task facing natural selection. I hope to repair this defect in future work.
But what I really mean by “shifting” is that there is a lack of clarity, and thus the reader (or multiple readers among themselves) can come up with different interpretations. By “shifting” I mean choosing among different interpretations, selecting one then another interpretation. And doing this and observing the application of the (shifted interpreted) definition gives a result of “IC” or “non-IC” (and possibly an analog unclear in between).
--
quote:
Nelson:
Actually what the Force paper shows is one thought to be common consequence of gene duplication, which resulted in a partitioning of function. This changes the classical result where the original function was retained, and the other gene aquired a new function. I know this seems like I'm not disagreeing with what you are saying, but there is an important assumption that you have in the way you are describing the situation, in that it will inevitably lead to a new invention (or IC system or whatever). By explaining what the raw data shows, this can be seen as nothing more than question begging. Since all that is left is the hope of beneficial revision of parts and their subsequent step by step evolution to a new IC system. Unfortunately this has not been demonstrated.
Now, there is an important twist to this story. Ged states:
quote:
However since the duplicate in the DDC model only partitions control of function or morphology, without changing it, there is no disadvantage. This means that the duplicated gene will be more likely to be carried to future generations than in the classical model! (And furthermore that is confirmed by observation!)
However, the reason why Zebrafish would keep the partioned genes is because it needs the two functions. This would more likely point to functional constraint. Whereas in the classical model, the gene was free to mutate all it liked without any harmful effects, finally arriving at a new function.
I have an introductory question about why Alonso chooses to mischaracterize my argument by saying “… but there is an important assumption that you have in the way you are describing the situation, in that it will inevitably lead to a new invention…”. I highlighted “inevitably” because I never said this always happens, rather I was disagreeing with Alonso’s implication that it happens with such low probability as to be negligible. There is a significant difference, and I am surprised that Alonso worded my argument in this way. It is certainly easier to disagree with a different argument than the one presented.
No, Alonso, you don’t seem to be disagreeing with what I was saying in many regards. But the Force et al. paper was not researching whether those genes eventually took on new modifications or participated in further new functionality. It was researching the shorter and mid-term preservation of duplicated genes, as I read it.
Now in the classical model a given duplication would be less likely to produce a fit individual in the first place! That’s because the duplicate, to be retained, had to furthermore take on a beneficial change of function, not just a neutral change of function association (e.g. partition), in order to be retained in the first levels of reproduction. But with the DDC model, there is simply partition.
And once again, you don’t seem to be disagreeing on Zebrafish needing the two functions. That is of course why the duplicated gene would be retained in future generations, and also becoming active in controlling structure, as opposed to becoming noise or so-called “junk DNA” or pseudogene as I understand the article. Because of this retention into the future, there is a basis for change, and this was pointed out by the writers of the article itself:
quote:
On the other hand, by expanding the time period for which genes are exposed to selection, the preservation of duplicates by the DDC process facilitates subsequent opportunity for the evolution of new functions.
The mutation of the gene to produce new function is not discussed in the paper, which was focusing on “preservation”. Preservation is the same gene, not a mutation thereof.
Now I’m not a biologist, and I could have easily made a mistake in my interpretation above. But note that Alonso didn’t answer my question: Are you disagreeing with common descent?
Let’s assume that we do have designers (say unembodied ones that can do anything). But they choose to leave an apparent pattern of descent with modification for reasons unknown, so as to be consistent with observation. So in this to follow I’m not arguing with ID processes.
So we have the changed structures caused by the changed genetic patterns that we see in these descendents of common ancestry, and they indeed have different structures corresponding to changes in the genes.
So by whatever process (unembodied ID or otherwise) the genes were changed by a process equivalent to duplication, addition, and modification. And these changes resulted in the changed organisms that we observe in that descent with modification tree.
So now dropping the assumption of an ID, I don’t see that we just have only a “hope of beneficial revision of parts and their subsequent step by step evolution to …” (whatever). We actually observe such descent with modification. What Force et al. provides is a description of one mechanism by which such descent with modification operations occur, and observation and modeling of the frequency of those types of changes.
Or Alonso, are you disagreeing with common descent? If so, the Force et al. paper is irrelevant, as its evidence is based on common descent based arguments, and if so the premises of that paper are not meaningful to you so how can the conclusions be accepted for use in your argument?
All I’m asking for is clarity in argument and definitions.
[Edits for clarity, and an additional point about my language, since at this point no responses have been given.]
[ 06. March 2003, 13:23: Message edited by: gedanken ]
IP: Logged
|
|
Nel
Member
Member # 614
|
posted 06. March 2003 15:19
Ged:
Alonso suggests that Behe’s definition is clear after all. So why does Dembski have to “salvage” it on page 280 of NFL? Here is a discussion of that, I don’t have NFL at hand at the moment.
Nelson:
I don't see this as Dembski "salvaging" the definition, since, as I showed, it is quite self-explanatory. What Dembski did was put it in more precise language in order to make the definition technically sound and where you can make no mistake with regard to the consequence of the definition.
Ged:
What I mean by “changing” is that interpretations by different people are different. Just look through the discussions on IC, say on ARN, and you see different people posing systems all the time that meet the requirements, then someone else showing an interpretation that claims the structure was not IC after all. If the definition was clear and objective, it would not be so difficult. Science is about repeatability.
Nelson:
People can come up with different interpretations all the time, in fact, usually it's the ID proponent who proposes an IC system and, no surprise, an ID opponent who disagrees that the system is IC. If you look at ARN under the thread "List of IC systems" (which I need to get back to at some point, haven't had time to post much at ARN do to work) you will see just the opposite. The only attempt that was made to remove a part from one of the IC systems I proposed was with the OriC. When an attempt was made to remove a part from one of my systems (an ID opponent referencing of a paper on OrIC) it came up short. In fact, the paper was completely irrelevant. From then on, stories about the evolutionary pathway was all that was left. No further attempt was made to reduce the basic function to one part (or one subfunction).
Nelson:
What I think you mean by "explained" is that an evolutionary story can be told about it. However, that does not show that it is not IC.
Ged:
But the common anti-evolution claim of a “story” is that it is not highly accepted consensus that the particular pathway is the one and only valid explanation, rather that it is one of many possible pathways.
Nelson:
I'm going to forgo addressing your use of the fallacious label "anti-evolution claim". Nonetheless, it is simply not true that the common retort to a "story" about a particular IC systems origin is that it is not widely accepted. Rather, the common rebuttal is that not only is the story undetailed, and imaginary, but it contains fatal errors. Such as the fact that the export machine to bacterial flagella is really not a gradualistic pathway, it rather invokes 10+ co-option events all due to chance as natural selection was pruning the system for the function that already exist. Another example is that the system in question, the type III secretory system they point to, post-dates the bacterial flagellum.
As if that weren't enough they also ignore the fact that the various subsystems within machines like bacterial flagellum are also IC.
So the statement that the common retort is that it is not the most widely accepted is simply false . For more information take a look at:
http://www.arn.org/docs2/news/wd_still_spinning.htm
quote:
The scientific literature shows a complete absence of concrete, causally detailed proposals for how coevolution and co-option might actually produce irreducibly complex biochemical systems In place of such proposals, Darwinists simply observe that because subsystems of irreducibly complex systems might be functional, any such functions could be selected by natural selection. Accordingly, selection can work on those parts and thereby form irreducibly complex systems. All of this is highly speculative, and accounts for cell biologist Franklin Harold's (2001, 205) frank admission: "There are presently no detailed Darwinian accounts of the evolution of any biochemical or cellular system, only a variety of wishful speculations."
and
quote:
But the TTSS, as Mike Gene (see citation at end) notes, is restricted "to animal and plant pathogens." Accordingly, the TTSS could only have been around since the rise of metazoans. Gene continues: "In fact, the function of the system depends on intimate contact with these multicellular organisms. This all indicates this system arose after plants and animals appeared. In fact, the type III genes of plant pathogens are more similar to their own flagellar genes than the type III genes of animal pathogens. This has led some to propose that the type III system arose in plant pathogens and then spread to animal pathogens by horizontal transfer.... When we look at the type III system its genes are commonly clustered and found on large virulence plasmids. When they are in the chromosome, their GC content is typically lower than the GC content of the surrounding genome. In other words, there is good reason to invoke horizontal transfer to explain type III distribution. In contrast, flagellar genes are usually split into three or more operons, they are not found on plasmids, and their GC content is the same as the surrounding genome. There is no evidence that the flagellum has been spread about by horizontal transfer."
Ged:
But the claim of IC is that it cannot have a developmental pathway with sufficient probability to be reasonable. (And if we are to apply the EF as meaningful, it must have a possibility of such a pathway occurring of less than 10^-150 -- hardly such a case when a possible pathway can easily be given.)
Nelson:
Showing the basic errors with evolutionary stories such as the bacterial flagellum plus showing just how many co-option events are needed to occur just to get some kind of motor with a filament (and how irrelevant this is especially when taking brownian motion into account). All of this contributes to the notion that "it cannot have a developmental pathway with sufficient probability to be reasonable."
Ged:
One problem with the “irreducible core” is exactly the same problem as with the TRIZ argument -- to get us to the subject of this thread. It shows multiple co-option pathways as likely. (Indeed co-option is the story of evolution, to suggest a problem with all examples being of co-option is just plain silly.)
Nelson:
I bolded the 2 parts I think is blatantly false. First, as I show above, neither TRIZ nor IC show that mutliple co-option pathways are likely. In fact, they show the opposite. Now, I'll admit that I'm not an expert on TRIZ or even IC, but I do know biochemical systems and I am fairly familiar with the concept of IC (having argued it off and on for 3 years). No doubt you will either retract this statement or say you meant something else. But as far as I can interpret what you write here, it is completely false. Second, co-option pathways are not the story of evolution for the very reason that they admit too many random chance factors of mutli-part systems. Orr himself attested to this when he wrote you might as well hope your air conditioner starts working with your engine. And yes, this is exactly what the T3SS story is attempting to show. Here is Mike Gene:
quote:
But we've seen that the Ur-IC state [NA: ancestral state] of the bacterial flagellum likely entailed around 20 gene products. As a result, a figure outlining the gradual cooption of parts should contain at least 20 components (A-E, G-U), 38 pre-flagellar functional states (F1-F38), and 19 cooption events. In other words, to propose a series of IC systems that gradually increase in complexity, step-by-step, (a 3-part system becomes a 4-part system becomes a 5-part system, etc.), 38 non-flagellar functions are involved. One can attempt to cut down on the 38 functions by arbitrarily eliminating some (due to the ad hoc load of positing 38 unknown functions), but to do so, one must dip from simultaneous cooption (since we are constrained to explain the origin of these 20 functional sequences) and thus weaken the whole hypothesis [1].
The footnote makes the point very clearly.
quote:
1. Recall that the advantage to gradual cooption is its simplicity, where only one protein has to match another rather than a group of proteins matching another group.
http://www.idthink.net/biot/flag5/index.html
What I regard as the main story of evolution, what I think is the most probable pathway an evolutionary story can take, is what David Ussery and Thornhill discussed in this paper (Thornhill, R.H., Ussery, D.W. 2000. "A classification of possible routes of Darwinian evolution." J. Theor. Bio. 203: 111-116). What is interesting is that direct paths are completely eliminated by irreducible complexity. What I mean by a direct path was nicely demonstrated by Richard Dawkins in climbing mount improbable:
quote:
"The point is that ninety-one membranes are more effective in stopping photons than ninety, ninety are more effective than eighty-nine, and so on back to one membrane, which is more effective than zero. This is the kind of thing I mean when I say there is a smooth gradient up Mount Improbable. We would be dealing with an abrupt precipice if, say, any number of membranes above forty-five was very effective while any number below forty-five was totally ineffective. Neither common sense nor the evidence leads us to suspect any such hidden discontinuities."
But in an IC system you don't have this direct pathway for the simple reason that if the 45 membranes were required to stop photons, 44 membranes wouldn't be less effective in stopping photons, it would be completely ineffective in stopping photons. That is an IC system and that is an abrupt precipice. That is the invention that it so hard to get in TRIZ. That is what makes both TRIZ and IC such powerful indicators of intelligent design.
So whats the solution that Ussery and Thornhill give? Adoption from a different function and elimination of funciton redundancy. We cannot rule out that the T3SS , an IC system, evolved. Indeed, it is the consensus now that this is what likely occured. But, it evolved from a more complex precursor, itself IC. It is obvious that such a scenario is useless to Darwinian evolution and useful to ID as I think that most IDers would agree that the original state of an IC system we see today was most likely more complex and not less. This isn't exactly what is meant by elimination of functional redunandacy but even so, a more complex ancestor is being assumed by this pathway, which itself , did not do away with an irreducible core. As Mike states:
quote:
While the originally proposed 6-part system does not need all six parts, it still contains an essential IC core. That is, this system still requires a four-part interaction involving A, B/E, C/G, and D to elicit F; IC is simply embedded in the redundant complexity.
http://www.idthink.net/back/ic/
So the only option left , really, is adoption from a different function (co-option). An unlikely, and almost always undetailed, pathway to an IC system. So, when you see a Darwinist tell you a story about how bacteria got it's flagella, or how the cell got it's F-ATP synthase, or how the cell got it's recombination machine, or how the cell got it's ribosome, or whatever, you will always here a co-option story and that in and of itself tells you that the system you are dealing with is IC.
Ged:
What the “core” shows is the likelihood of loss of complexity at some point so as to remove unused aspects that are not part of the current function during the intermediate transitions. In the TRIZ argument it suggests that the structure was reachable and thus not “inventive”, even though in the contracted form there is an apparent “contradiction” in the claimed IC structure.
Nelson:
This is not true either. What the core shows is quite simply that you cannot reduce the current function
to one part (or one subfunction), because the current function needs these interactions and subfunctions to work. So you cannot reduce motility of the bacterial flagellum to flgE. What the TRIZ argument shows is that when you have an export machine, the gradualistic scenario needs to "invent" a filament, it needs to "invent" a motor, it needs to "invent" the way the flagellum is built in the first place (since that is IC). One of these inventions , even if it can get a less efficient filament had to ovecome this technical contradiction:
quote:
. Bacteria are constantly being buffeted by water molecules and thus live in a "Brownian storm." The simple fact is that because bacteria are so small, they swim through a Brownian storm. Brownian motion will knock bacteria off course after 3-4 seconds. [4] And this highlights a serious problem with the EFM hypothesis. The flagellum is a highly sophisticated machine. Even if one believes it evolved, what we study today is the product of billions of years of evolutionary modification. Yet even this high sophisticated/highly evolved system barely overcomes the Brownian storm. Thus, just how advantageous would some proto-wiggle really be? Imagine a boat in the ocean during a tropical storm. Would a propeller that spun once every second really be any better than no propeller?
This is a technical contradiction not unlike that which was experienced with the ice breaker ship example. But in this case we can see just how much the trial and error process being invoked by the main Darwinian story about the evolution of bacterial flagella is simply incapable of overcomming it.
Ged:
You can always argue that one essence of the definition of IC, for example, was preserved over the variations given by Behe and Dembski.
Nelson:
No need, since the entire definition of IC was preserved in it's entirety.
Ged:
But during those variations, there were also changes in how many cases it actually applied to as others interpreted those definitions,
Nelson:
There was no change in how many cases it actually applied to. At least not by the biggies of ID.
Ged:
and also changes in how probable were occurrences of ways that were understood to get around the apparent difficulty.
Nelson:
There was no change in the assessment of the probability of any IC system by Darwinian pathways. At least not by biggies of ID.
Ged:
So in this we see not only shifting of interpretation, but shifting of raw definition (even if certain essence were maintained) and this shifting is part of my point of being on an “edge”.
Nelson:
You're going to have to specifically address the shifting of "raw" definition. Since this was not done, I can only suspect that there was actually no change in the raw definition of IC.
Ged:
And in the second place Behe himself has made a series of different wordings: His initial definition did not speak of “core”.
Nelson:
But it was quite obvious, for those who read the book, that he was speaking of an irreducible core. For example, take the blood clotting cascade.
quote:
"the blood-clotting system fits the defintion of irreducible complexity. That is, it is a single system composed of several interacting parts that contribute to the basic function, and where the removal of any one of the parts causes the system effectively to cease functioning. The function of the blood clotting system is to form a solid barrer at the right time and place that is able to stop blood flow out of an injured vessel. The components of the system are fibrinogen, prothrombin, Stuart factor, and proaccelerin. "
But there are other components of the system than this. Why didn't Behe include Hageman, for example? Blood clots in vivo without Hageman. But it cannot without prothrombin. Or without Stuart factor, or without proaccelerin. Or without fibrinogen. It is obvious that if you see a system where the function,
in this case, The function of the blood clotting system is to form a solid barrer at the right time and place that is able to stop blood flow out of an injured vessel. , cannot be reduced to one protein or one subfunction of that protein, then you are dealing with an irreducibly complex system.
Ged:
Those differences are differences, even if they speak to similar issues. For example in “Reply to My Critics,” Biology and Philosophy 16: 685–709, 2001, Behe said:
quote:
However, commentary by Robert Pennock and others has made me realize that there is a weakness in that view of irreducible complexity. The current definition puts the focus on removing a part from an already-functioning system. Thus, seeking a counterexample to irreducible complexity, in Tower of Babel Pennock writes about a part in a sophisticated chronometer, whose origin is simply assumed, which breaks to give a system that he posits can nonetheless work in a simpler watch in a less demanding environment.5 The difficult task facing Darwinian evolution, however, would not be to remove parts from sophisticated pre-existing systems; it would be to bring together components to make a new system in the first place. Thus there is an asymmetry between my current definition of irreducible complexity and the task facing natural selection. I hope to repair this defect in future work.
Nelson:
What Behe is referring to here, is that the relationship between IC systems and natural selection is not apparent in the definition. In other words, when an ID opponent takes an IC mousetrap and uses the base of the trap as a doorstop, he folds his arms, says IC has been debunked, and sits back down with a big smile on his face. Meanwhile, the audience is still left with the question "uhh, but how did it evolve?". Notice the second definition that Behe gives in order to accentuate the problem for natural selection:
quote:
"While thinking of Keith Robison’s scenario [NA: About the evolution of a blood clotting cascade], I was struck that irreducible complexity could be better formulated in evolutionary terms by focusing on a proposed pathway, and on whether each step that would be necessary to build a certain system using that pathway was selected or unselected."
Furthermore, the definition adds to the IC definition in that we can now count the degree of irreducible complexity, and even perhaps give a selection co-effecient, perhaps even a rate of mutation, as in to say a system y is more irreducibly complex then system x in an evolutionary sense. This is why he goes on to say:
quote:
"The focus is off of the “parts” (whose number may stay the same even while the nature of the parts is changing) and re-directed toward “steps.” "
I think this is a lot more helpful to those Darwinists trying to classify a system as IC and then showing natural pathways. I have heard some of the most ridiculous things being IC, from atoms to the European economy. I have even seen people give me stories that were completely unfalsifiable and ever changing because of the fact that it is "undetailed". This is why Behe states:
quote:
"Envisioning IC in terms of selected or unselected steps thus puts the focus on the process of trying to build the system. A big advantage, I think, is that it encourages people to pay attention to details; hopefully it would encourage really detailed scenarios by proponents of Darwinism (ones that might be checked experimentally) and discourage just-so stories that leap over many steps without comment. So with those thoughts in mind, I offer the following tentative “evolutionary” definition of irreducible complexity: 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."
<http://www.discovery.org/viewDB/index.php3?program=CRSC%> 20Responses&command=view&id=442
Ged:
But what I really mean by “shifting” is that there is a lack of clarity, and thus the reader (or multiple readers among themselves) can come up with different interpretations. By “shifting” I mean choosing among different interpretations, selecting one then another interpretation. And doing this and observing the application of the (shifted interpreted) definition gives a result of “IC” or “non-IC” (and possibly an analog unclear in between).
Nelson:
Perhaps you can give me an example of an IC system and show that it lacks clarity. Perhaps together we can come to see how useful the concept is to Biology, even if we don't use biological examples.
Back to Gene Duplication
Nelson:
However, the reason why Zebrafish would keep the partioned genes is because it needs the two functions. This would more likely point to functional constraint. Whereas in the classical model, the gene was free to mutate all it liked without any harmful effects, finally arriving at a new function.
Ged:
I have an introductory question about why Alonso chooses to mischaracterize my argument by saying “… but there is an important assumption that you have in the way you are describing the situation, in that it will inevitably lead to a new invention…”. I highlighted “inevitably” because I never said this always happens, rather I was disagreeing with Alonso’s implication that it happens with such low probability as to be negligible. There is a significant difference, and I am surprised that Alonso worded my argument in this way. It is certainly easier to disagree with a different argument than the one presented.
Nelson:
No this argument is just as easy to disagree with. Simply replace the word "inevitably" with
"that it happens with high probability". What would happen with high probability in this case is that since the two funcitons are already fixed by natural selection, you most likely won't get much more. Add to that the difficulty in getting a re-wiring of the system, such as, how exactly when a gene is involved in a synthesis pathway, is removed from that synthesis pathway, and start to build tiself around a completely new reaction. Then there is the problems I outline above (and some new ones in this post) where there exists complex mechanisms to silence a gene duplicate. That what we see in the laboratory is less structure. As even Ohno stated:
quote:
However, a far more likely fate [NA: of gene duplication] is degeneracy due to the loss of promoter sequence, premature chain terminations, frameshifts etc.
Ged:
No, Alonso, you don’t seem to be disagreeing with what I was saying in many regards.
Nelson:
Actually there is significant, and fundamental disagreement between us. You seem to think that you are somehow (which is very odd to me) in a better situation with the DDC model than you were with the classical model. You aren't.
Ged:
But the Force et al. paper was not researching whether those genes eventually took on new modifications or participated in further new functionality. It was researching the shorter and mid-term preservation of duplicated genes, as I read it.
Nelson:
Exactly, this supports my point. This preservation of duplicated genes, as I stated, happens because the functions become fixed in the organisms, because it absolutely needs these two genes to be expressed in both the hindbrain and the bud. This points to functional constraint, not a significantly different invention.
Ged:
Now in the classical model a given duplication would be less likely to produce a fit individual in the first place! That’s because the duplicate, to be retained, had to furthermore take on a beneficial change of function, not just a neutral change of function association (e.g. partition), in order to be retained in the first levels of reproduction. But with the DDC model, there is simply partition.
Nelson:
Wrong. There is no need, in the classical model, for an immediate beneficial change. As your own cite states:
quote:
The points I wish to make are: 1) Natural selection is an extremely conservative force. So long as a particualr function is assigned to a single gene locus in the genome, natural selection only permits trivial mutations of that locus to accompany evolution.
If we see only a partitioning of gene function, we see that (1) now applies. The DDC model embraces (1) and therefore "permits trivial mutations of that locus to accompany evolution.
Now to the classical model:
quote:
Only a redundant copy of a gene can escape from natural selection and while being ignored by natural selection can accumulate meaningful mutation to emerge as a new gene locus with a new function. Thus, evolution has been heavily dependant on upon the mechanism of gene duplication.
Ouch. Heavily dependant on a mechanism that occurs rarely, and, in Ged's own words, is disadvantageous.
But I digress. So this mechanism allows the redundant copy to escape the clutches of natural selection since it is not needed by the organism, in contrast to the DDC model, where the conservative force (natural selection) holds the two functions in place for an indefinte period of time, where now only the hope of a new function lingers in the horizon.
So, in the classical model, all we needed, for the organism not to have any harmful effect, was the duplicate gene having the original function, and the redundant gene undergoing mutation until it arrived to a new function, where natural selection would finally conserve the new function. This is quite different from the DDC model where the two functions are immediately conserved, and must break from this conservation in order to aquire new functions.
In the next paragraph Ged simply repeats what I say above, without addressing my concern over functional constraint, in fact, what Ged stated next supported my point, so I deleted it.
But on to another point, Ged then asks:
Ged:
Are you disagreeing with common descent?
Nelson:
I do not disagree with common descent and the reason why I didn't answer the question is that it is irrelevant and still is. Whether gradualistic (RM&NS) evolution is the mechanism behind common descent or not is the question. Common descent does not force the assumption that the descent occured because of simple, gradualistic changes. In fact, with the evidence I add in my previous post, and above, that complex mechanisms exist that nix mechanisms like gene duplication in the first place, and that these mechanisms may themselves faciliate speciation, points to intelligent design, not a blind watchmaker.
But my main point here is that the gradualistic story is wanting. For example, what we see , in addition to gene silencing and loss of structure, is harmful effects as well, even the duplication of a single chromosome causes lethality of a human embryo early in development. All of this is relevant to intelligent design and RM&NS.
[ 06. March 2003, 16:55: Message edited by: Nelson_Alonso ]
IP: Logged
|
|
Nel
Member
Member # 614
|
posted 06. March 2003 15:51
I hope John doesn't mind me taking up one of the replies to his post.
John wrote:
Contrary to your assertions, I haven't changed terms or definitions. I never intended a hypervolume to refer to the number of genes or number of basepairs in a genome, and I've made that abundantly clear from the very beginning (go back and read the ARN thread where I deal with th | | |
