|
Author
|
Topic: Critical Questions
|
Geoff
Member
Member # 865
|
posted 05. August 2003 00:30
Critical questions.
I am trying to understand the defining concepts of irreducible complexity and intelligent design. I have read articles and a book on the subjects but am having trouble with a couple of ideas. The problems I’ve encountered lead me to five critical questions.
Let me start with irreducible complexity. As I understand it, an irreducibly complex system is, quite simply, one that cannot function when any of its components is taken away. Irreducibly complex systems are apparently identifiable in biological organisms but the common layman’s example seems to be a mousetrap, which obviously will not function when one of the pieces of the trap is removed. The implication is that the trap was created as a whole unit by an intelligent designer. The extension to biological systems is made with the basic inference being that an irreducibly complex system cannot have evolved according to neo-Darwinist theory, which assumes that gradual changes building up over long periods of time are responsible for the complex systems we see today. Contrary to this viewpoint, irreducibly complex systems must have been designed and constructed all at once because there is no obvious precursor system that could have produced the observed system through Darwinistic means. We assume that irreducibly complex systems could not have evolved from a precursor system because we cannot see or imagine a precursor system that could have functioned and survived long enough to evolve. At the same time, Darwinists assume that these same irreducibly complex systems must have evolved through transitional forms that have not been found, because it is difficult to imagine that an intelligent designer would create them from scratch. So both sides are making an assumption, based on something they have not found, because of a difficulty in imagining alternate possibilities. This leads me to my first critical question.
1) Until more evidence is found, what reasoning or logic dictates that one position is more sound than the other?
Perhaps this information will never be found. If that’s the case, then it may suggest more appropriate directions for efforts to encourage people to consider probing questions about evolution.
My second and third critical questions stem from an analogy to computer programming. Computer programs are coded information that a processing unit uses to perform a specified task using the tools it has available. There are very strict rules that the instructions must follow based on the abilities of the tools and the possible ways in which the tools can be used together. At work I often write programs on my computer and because of the strict rules they must follow, most of these programs seem to be irreducibly complex. If you take away any piece, even one line of code, they will not function at all. However, these programs also evolve in a sense because I am never able to write the full and final version all at once on the first try. I have to write small versions that perform a subset of the tasks I want or that carry out the tasks at a reduced level. Then I use what I’ve learned to make the next program, in some cases by modifying the existing one, but in many cases, because of the particular application developer, by starting from scratch to create new functionality. So it seems that each version of my program represents an organism or a system that is irreducibly complex but that evolved, albeit by a guiding intelligence, from an ancestor. There are then two critical questions that this analogy raises.
2) Is irreducible complexity really often the result of the strictness of the rules that dictate the organism’s function as opposed to the complexity, delicate balances, and interrelatedness of the organism’s parts?
3) Does irreducible complexity completely exclude all evolutionary ideas or does it in fact coexist with the slow gradual changes and precursor systems that evolution assumes?
Next let me ask about Intelligent Design, which says something about the ability to infer design in an event from the details that can be known about the event. From what I’ve read, a design inference about an event can be made if the event is “contingent” (it didn’t have to happen), it is “complex” (chances are it is not just random) and it is “specified”. From what I have read, this last criterion means that the event can be mapped onto a pattern that is not generated in an ad-hoc manner to fit the observations. However, what this seems to mean is that in order to determine whether an event is specified, you have to be able to construct a pattern describing the event that is independent of the event itself and that is also based on some aspect of intelligence – a set of meaningful information. But this seems to imply that you have to be able to figure out what the event means in order to know that it is truly specified. My fourth critical question follows from this observation – it seems that the design inference could be circular.
4) If you have to decode the event in order to determine that it was specified, why do you need the design inference criteria at all?
It seems that the real intent of Intelligent Design, the basis of its appeal, is the possibility that it will allow us to infer design in an event without being able to decode the event. In fact, in order to have a hope of competing with evolution as a “scientific” hypothesis, it must attempt to do this. This brings me to another analogy, in this case cryptology. Suppose for a moment that the NSA receives a letter in the mail that consists of a twenty pages of characters in a small font. There are no spaces or capitals in the characters to delineate words or sentence breaks but twenty pages should provide a large enough sample set to decode any message that might be there, if a message is there at all. However, that is the crux of the matter since there is a chance that the characters are randomly generated. Should the agency take the chance and begin spending computer time on the twenty page character string, or is there another test they can perform beforehand that will tell them whether an intelligent designer generated the string? The same thing applies to the concept of inferring design in an event and this leads me to the last critical question.
5) If you cannot decode an event, is there another test that can tell you something about the probability that the event was designed?
I hope these questions are constructive and that there are some simple answers that I have not yet encountered in my reading. I am also extremely interested in what people familiar with these subjects think. Thanks.
Geoff
IP: Logged
|
|
Christopher M. Langan
Member
Member # 264
|
posted 05. August 2003 11:55
Regarding the debate between proponents of Darwinism and ID, Geoff writes: "Until more evidence is found, what reasoning or logic dictates that one position is more sound than the other? Perhaps this information will never be found. If that’s the case, then it may suggest more appropriate directions for efforts to encourage people to consider probing questions about evolution."
Since Geoff has mentioned logic, I'll provide a reminder that there is only one reasonably well-defined logic-based (as opposed to exclusively "biology-based", "information theory-based" or "physics-based") theory of ID and/or evolution. Called the CTMU (short for Cognitive-Theoretic Model of the Universe), it uses the tautological basis of logic to make inferences about the natural world. (Browse recent threads and back issues of PCID for more information.) This theory logically relates function to structure and easily accommodates the oft-noted tautological dimension of Darwinian evolution with respect to first-order Markovian laws of nature, obviating the necessity to specifically "decode" evolutionary events (although this remains a viable possibility in certain cases).
[ 05. August 2003, 12:06: Message edited by: Christopher M. Langan ]
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 05. August 2003 12:51
I'll take a shot at addressing (if not answering!) one of Geoff's questions. He wrote quote:
Computer programs are coded information that a processing unit uses to perform a specified task using the tools it has available. There are very strict rules that the instructions must follow based on the abilities of the tools and the possible ways in which the tools can be used together.
[description of human programming excised]
So it seems that each version of my program represents an organism or a system that is irreducibly complex but that evolved, albeit by a guiding intelligence, from an ancestor.
and then he asks quote:
3) Does irreducible complexity completely exclude all evolutionary ideas or does it in fact coexist with the slow gradual changes and precursor systems that evolution assumes?
There is some considerable evidence that the notion of "irreducible complexity" does not completely (or even necessarily partly) "exclude all evolutionary ideas." To stay with the computer programming analogy, there is an increasingly well-developed area of research in genetic programming in which the process of programming is given over to evolutionary systems. Those genetic programming platforms can evolve what are "irreducibly complex" systems, operationally defined in terms of the knockout criterion for determining ICness, where the programmer does not "guide" the process except to define the desired functionality.
Once again, the Lenski, et al., study is pertinent. The 'organisms' in that simulation are assembly language programs: sequences of instructions. Leaving aside the various arguments about biological verisimilitude and numbers of intermediates, it is indisputable that programs capable of performing complex logic functions evolved in a context where the processes invoked by evolutionary theory were operating (replication with heritable variation; competition for limited resources; selection on phenotypic function; cooption of simpler functions). The programs that evolved met the operational definition of irreducible complexity. So irreducible complexity cannot "exclude all evolutionary ideas."
I'll take mild issue with the phraseology "slow gradual changes." Surely an evolutionary process must take place on a fitness landscape where selectable incremental change occurs. "Gradual" must refer to the local topography of fitness landscapes - evolution occurs in selectable phenotypic increments produced by genotypic change in one or several dimensions on one or another fitness landscape. But the several fitness landscapes available in a given evolutionary context have different topographies. A single selectable step on one landscape may be a large leap on another. For example, if there are just two variation-producing operators, point mutation and single-point crossover recombination, in an evolutionary system, then there are two quite different fitness landscapes available to the system. For example, given two 'organisms' with binary gene strings:
Organism A: 11111111
Organism B: 00000000
the two variation-producing operators induce two different landscapes. The 'nearest neighbors' of the two organism on the point mutation landscape are quote:
Organism A Organisms B
11111110 00000001
11111101 00000010
11111011 00000100
11110111 00001000
11101111 00010000
11011111 00100000
10111111 01000000
01111111 10000000
On the recombination landscape, though, where A and B recombine by crossover, the nearest neighbors of A are: quote:
01111111 11111110
00111111 11111100
00011111 11111000
00001111 11110000
00000111 11100000
00000011 11000000
00000001 10000000
Notice that only two of the 'nearest neighbors' of Organism A on the point mutation landscape are also its nearest neighbors on the recombination landscape. "Gradual change" must take into account variation of a population on both landscapes. The implication is that there is more 'room' for evolution, even in a constrained simulation, than is apparent at a glance. Recombination is one of the ways in which evolutionary processes can coopt simpler functions to construct organisms performing more complex functions, though cooption can occur in asexually reproducing organisms as well.
All of the 'nearest neighbors' on all of the fitness landscapes are accessible to evolutionary processes. I am aware (thanks in part to Rex Kerr) that one can construct a manifold of the several fitness landscapes. But (pace Rex) I find it useful (at least pedagogically) to think in terms of several landscapes induced by several operators.
It seems to me that a fair amount of the incredulity over what evolutionary processes can accomplish in unguided (by intelligence) selective contexts derives from a failure to appreciate the complexity (in the ordinary language sense) of the topopgraphies of high-dimensioned natural fitness landscapes and the consequent availablity of multiple different pathways on those landscapes. We humans tend to think in terms of two- or three-dimensional landscapes and can easily envision unjumpable cliffs and unreachable plateaus on them. But it is more difficult to envision landscapes with continuous slopes and ridges connecting fitness peaks in high-dimensioned spaces. It is surely the case that there are evolutionary cul de sacs in high-dimensioned spaces, but they are much rarer than one might expect from intuitions based on two- and three-dimensioned analogies.
RBH
[ 05. August 2003, 13:07: Message edited by: RBH ]
IP: Logged
|
|
Christopher M. Langan
Member
Member # 264
|
posted 05. August 2003 21:56
In acknowledging that "there is some considerable evidence that the notion of irreducible complexity does not completely (or even necessarily partly) exclude all evolutionary ideas," RBH invokes certain results in the field of genetic programming. Although I agree with RBH's basic assertion regarding the compatibility of IC and evolution, and even that computational simulations may support it in certain computational contexts, genetic programmers currently lack any coherent mapping from computational to naturalistic contexts. Thus, they are severely limited in the extent to which they can present any computational simulation as "evidence" of anything in the natural world (and I'm talking about a deeper level of correspondence than that discussed in recent criticisms of Lenski's work).
To realize the evidentiary value of such a simulation, one would need at least a minimal computational model of nature ... e.g. the CAMU (Cellular Automaton Model of the Universe), which attempts to map some part of nature into a metaphysical cellular automaton. If, on the other hand, one prefers a self-contained language-based computational model, as is suggested by any likening of organisms to assembly language programs, then one is driven toward SCSPL and thus toward the CTMU. The CTMU follows from the fact that any comprehensive description of nature must account for the generation of computational syntax along with mere syntactic content, and must thus portray nature as telic and protocomputational rather than merely computational in essence. Nature has the capacity to self-generate and regenerate, whereas computational automata and even computation theory do not. This militates against the notion that nature is fully analogous to standard computation theory.
In other words, the reason that the Cellular Automaton Model and other so-called "computational models of reality" fail as models of evolution is that they cannot account for the evolution of computation itself, the array or machine or circuit in which computation occurs, or even the relationship between the array and its contents. This leads naturally to the topic of manifolds. A cellular array is a discrete manifold in which neighboring cells are separated and connected by boundaries. In language theory, the concept of a cellular array corresponds to structural syntax, but the correspondence breaks down where the grammatical or generative part of syntax requires more of a setting than a mere discrete manifold can provide. Similar objections apply to any manifold in which discrete fitness landscapes are regarded as "points".
Thus, when RBH concludes that "a fair amount of the incredulity over what evolutionary processes can accomplish in unguided (by intelligence) selective contexts derives from a failure to appreciate the complexity (in the ordinary language sense) of the topographies of high-dimensioned natural fitness landscapes and the consequent availablity of multiple different pathways on those landscapes," it is important to remember that there is more to a model of nature than any mere fitness landscape or manifold thereof can provide (regardless of dimensionality). Obviously, the properties of such landscapes can ultimately explain nothing without an explanation for the existence of fitness landscapes themselves.
IP: Logged
|
|
Argon
Member
Member # 276
|
posted 05. August 2003 23:04
Well, the properties of the landscapes can often explain the immediate activities of objects within them. So I wouldn't say they explain "nothing". They just don't explain "everything".
Anyway, everyone knows it's turtles all the way down.
[ 05. August 2003, 23:05: Message edited by: Argon ]
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 06. August 2003 02:21
Langan correctly observes that the results of the various lines of genetic programming research do not instantiate CAMU. I'm not sure that's a problem, though, because the goals of GA and GP research, to the extent that they have to do with simulating nature rather than developing applications, are much more modest. Rather than attempting to model the universe as a whole at one go, they're directed at simulating more-or-less well delimited portions of the universe. (As an aside, I'm not entranced with cellular automata as generic models of a whole lot. Wolfram apparently believes they're the cat's pj's, but I don't, at least not in the sense of 'simulation model' I described in the Strachan thread.)
Moreover, to some extent the genetic programming efforts and many of the ALife simulations are not aimed at directly modeling (isomorphically representing) specific parts of the (biological) universe, but are rather aimed at testing at a slightly more abstract level the capabilities and limitations of the processes hypothesized to account for portions of the universe, in the present case the biological evolution part. (Of course, since the simulations are themselves part of the universe, they constitute completely veridical models of themselves!) Clearly, though, it's important to clarify the correspondences (at multiple levels) between simulations, computational and otherwise, and nature. I concur with Langan's remarks about the inadequacies of purely local models - CAs being a prime example - as models of syntactic structures in which embeddings, for example, are possible. Chomsky made that point at length in his 1959 review of Skinner's Verbal Behavior. It's one reason I don't like CAs as models of much of anything having to do with biology, because life processes involve actions at a distance, where causal influences span both space and time. And evolutionary processes operating at a local level can generate time- and space-spanning structures.
I have to confess that as I read Langan's posting the phrase Argon used ('turtles all the way down') was actually running through my mind too. Argon is correct: the GAs and GPs and CAs don't explain everything, but they do help explain some very interesting things. That they don't account for the origin of 'natural' fitness landscapes themselves is not an immediate problem for me nor, I think, for most working scientists in the area. The properties of those landscapes are of interest, but not necessarily their origin. One can chase turtles if that is one's inclination. But until and unless I see some fairly specific mappings that describe the correspondences leading from a general theory of everything to the operation of a GA on my machines, I'm comfortable not worrying overmuch about the cosmic level of analysis. To criticize a theory or research approach because it doesn't account for everything including the proverbial kitchen sink is not real helpful. That nature and computational models are "fully analogous" is not a claim that I or anyone I'm aware of has made. (Well, except Wolfram, I guess. I haven't yet fought my way all the way through his book. And I may not do so: human lifetimes are finite.)
RBH
IP: Logged
|
|
Christopher M. Langan
Member
Member # 264
|
posted 06. August 2003 03:09
Argon responds "Well, the properties of the landscapes can often explain the immediate activities of objects within them. So I wouldn't say they explain nothing. They just don't explain everything."
Good point, and I only have one point to make in return. What I wrote was "Obviously, the properties of such landscapes can ultimately explain nothing without an explanation for the existence of fitness landscapes themselves." I meant that an "ultimate explanation" differs from an "explanation" in that it must cite a model of causation and an ultimate cause therein, or to apply Argon's colorful phrasing, "the bottom turtle". In the present context, this comes down to the wherewithal of fitness landscapes themselves.
RBH adds "But until and unless I see some fairly specific mappings that describe the correspondences leading from a general theory of everything to the operation of a GA on my machines, I'm comfortable not worrying overmuch about the cosmic level of analysis." The cosmic level of causation is the ultimate level of causation. So while I agree that "to criticize a theory or research approach because it doesn't account for everything including the proverbial kitchen sink is not real helpful," this doesn't apply to theories that claim to offer ultimate explanations of (e.g.) life. While one could argue that Darwinism doesn't explicitly claim to offer an ultimate explanation of life, I get the distinct impression that many of its proponents think that no deeper explanation (e.g. ID) could possibly be required. It was for them that my criticism was intended.
While fitness landscapes fail to qualify as ultimate explanations, I readily concur that they can be useful at intermediate stages of explanation. I'd merely add that if they have explained more than the smaller turtles atop them, it is only because they have stood on the shoulders of giant turtles beneath.
IP: Logged
|
|
Argon
Member
Member # 276
|
posted 06. August 2003 10:28
Yes, the proximate explanations provided by fitness landscapes and experimental work are useful. In fact, the IC->ID hypothesis of Behe is specifically formulated at this level. Likewise Dembski's filters & stuff. These approaches are predicated on the ability to know the historical pathways (fitness landscapes) available to a system. If that mapping is suspect then the support for the hypotheses could be indeterminable.
[ 06. August 2003, 10:30: Message edited by: Argon ]
IP: Logged
|
|
Christopher M. Langan
Member
Member # 264
|
posted 06. August 2003 11:42
Argon agrees that "the proximate explanations provided by fitness landscapes and experimental work are useful. In fact, the IC->ID hypothesis of Behe is specifically formulated at this level. Likewise Dembski's filters & stuff. These approaches are predicated on the ability to know the historical pathways (fitness landscapes) available to a system. If that mapping is suspect then the support for the hypotheses could be indeterminable."
Quite so. However, the mapping is what it is. And what it is is the CTMU, which implies ID even without attention to specific fitness landscapes (in fact, the CTMU is the only picture in which specific computational fitness landscapes can be coherently mapped to nature). I merely meant to point out that the richness of higher-dimensional fitness landscapes cannot be taken as "evidence" in favor of anti-ID hypotheses. Because the CTMU implies a form of ID, anti-ID hypotheses are anti-CTMU hypotheses. So on a rational (logical, pre-empirical) level, any attempt to use fitness landscapes as evidence for such a hypothesis would undermine the theory-to-universe mapping on which it critically depends.
IP: Logged
|
|
Argon
Member
Member # 276
|
posted 06. August 2003 12:48
Then Dembski and Behe have got some trouble with the CTMU because they are basing their hypotheses about ID on proximate causes. That is, they will accept proximate explanations for the evolution of IC systems (if they exist) as evidence against ID. In the past, Dembski has specifically argued against theistic evolution and deistic evolution concepts as being "no friend" of intelligent design.
So what is being discussed by Behe et al. at these levels is not whether ID is necessary for the fundamental underpinnings of the universe. Instead, their argument is about the manner by which intelligence is manifested in local situations such as with the emergence of some types of biological features. It's about the nature and method of intelligent "involvement", not whether intelligence is necessary at a lower level to create a world in which evolution is possible. Basically, their IDea of ID is not exactly your IDea of ID. They are "proximate IDists".
As it stands now the CTMU does not differentiate between potential, proximate mechanisms for the emergence of say, factor VIII in the mammalian clotting system. However, empirical and modeling methods potentially can, with a reasonable degree of confidence.
[ 06. August 2003, 12:50: Message edited by: Argon ]
IP: Logged
|
|
Christopher M. Langan
Member
Member # 264
|
posted 06. August 2003 13:36
Argon states that "Dembski and Behe have got some trouble with the CTMU because they are basing their hypotheses about ID on proximate causes." Even if this were so - and I don't agree that it is - it would be no big deal. After all, the proximate causes to which Argon refers reside directly on the route between the CTMU theory-universe mapping and nature. Argon continues: "That is, they will accept proximate explanations for the evolution of IC systems (if they exist) as evidence against ID." In that case - and I must question Argon's ability to speak authoritatively on what Professors Behe and Dembski would actually do if someone (impossibly) produced "empirical evidence" against ID - then perhaps we'd have a minor disagreement. However, we'd remain in substantial agreement regarding ID itself, as opposed to anybody's idea of what it takes to confirm or disconfirm it. Regarding Argon's assertion that "Dembski has specifically argued against theistic evolution and deistic evolution concepts as being 'no friend' of intelligent design," I suspect (assuming the accuracy of Argon's statement) that the reasons may have something to do with the extent to which these hypotheses require front-loading. The CTMU does not require that ID be exclusively front-loaded.
Argon goes on: "So what is being discussed by Behe et al. at these levels is not whether ID is necessary for the fundamental underpinnings of the universe. Instead, their argument is about the manner by which intelligence is manifested in local situations such as with the emergence of some types of biological features ... Basically, their IDea of ID is not exactly your IDea of ID. They are 'proximate IDists'." Again, I doubt that anyone is in a position to speak definitively for individual ID theorists but the theorists themselves. In any case, there is no reason to believe that empirical ID research programs need be incompatible with CTMU-level mathematical reasoning. Such programs require a supporting model of causality, and that's non-negotiable.
Argon asserts that "As it stands now the CTMU does not differentiate between potential, proximate mechanisms for the emergence of say, factor VIII in the mammalian clotting system." If Argon is insinuating that the CTMU is incapable of differentiating between front-loading and proximate loading (of design), this is flatly incorrect. On the other hand, when Argon asserts that "empirical and modeling methods potentially can [distinguish between these alternatives] with a reasonable degree of confidence," they certainly cannot do so without sound rational underpinnings. To my knowledge, Behe and Dembski have not ruled out further exploration of these underpinnings. In fact, I'm quite sure that they have not.
IP: Logged
|
|
Argon
Member
Member # 276
|
posted 06. August 2003 19:12
Chris writes
quote:
Argon continues: "That is, they will accept proximate explanations for the evolution of IC systems (if they exist) as evidence against ID." In that case - and I must question Argon's ability to speak authoritatively on what Professors Behe and Dembski would actually do if someone (impossibly) produced "empirical evidence" against ID - then perhaps we'd have a minor disagreement.
I'm speaking of Behe's and Dembski's writings. For example, Dembski's "no friend" comment comes from a paper available at ARN. Likewise, one hopes Behe's "Darwin's Black Box" and later presentations are helpful at discerning his opinion (I heard Behe's lecture at the Interpreting evolution conference at Haverford College just over two years ago).
It's not that one could produce empirical evidence against ID (I agree with Chris here and I should have chosen my words better in original post - mea culpa). It's that a particular method of design, such as one that bridges IC "gaps" by introducing new, functional systems, can be rendered unnecessary (from a proximate perspective). The criteria for ICness is formulated in proximate terms and that's the level at which it is evaluated.
quote:
In any case, there is no reason to believe that empirical ID research programs need be incompatible with CTMU-level mathematical reasoning.
I agree. I'm not suggesting that empirical ID research programs are incompatible with CTMU. For instance, if it is found that IC systems can naturally arise within organisms, one could not say that an ID was not involved because, as Chris has mentioned in this thread, the intelligence could be acting at a root levels (e.g. by setting the original conditions of the universe).
quote:
Argon asserts that "As it stands now the CTMU does not differentiate between potential, proximate mechanisms for the emergence of say, factor VIII in the mammalian clotting system." If Argon is insinuating that the CTMU is incapable of differentiating between front-loading and proximate loading (of design), this is flatly incorrect.
Actually, I didn't mean to insinuate anything about the CTMU and certainly not what Chris might tentatively suggest. I wrote that CTMU does not differentiate between proximate mechanisms, not that it is incapable of distinguishing between proximate and distal causes. It could help if I try to put is another way. IC, as a potentially, experimentally accessible hypothesis, tackles proximate causes. ID doesn't depend on whether IC is inaccessible to evolution or not. Whether a clotting protein finds its way in an organism because it was put there by scientists (intelligent designers of a fashion) or via evolutionary steps are both -- and I hope on not going too far out on a limb here -- compatible with the CTMU.
quote:
On the other hand, when Argon asserts that "empirical and modeling methods potentially can [distinguish between these alternatives] with a reasonable degree of confidence," they certainly cannot do so without sound rational underpinnings.
I agree that sound, rational underpinnings are essential for validating models and methods.
[ 06. August 2003, 19:13: Message edited by: Argon ]
IP: Logged
|
|
Christopher M. Langan
Member
Member # 264
|
posted 07. August 2003 00:27
Good post. A few comments may be in order. If a new functional system were introduced that could deterministically "bridge IC gaps", it could still be interpreted as an instance of front-loaded ID without precluding ad hoc ID. Causal explanations for IC phenomena can be formulated on different levels; the most important is the most basic and general, by which all other levels are explained. Argon seems to acknowledge both of these facts when he writes that "intelligence could be acting at a root level...by setting the original conditions of the universe". And precisely because CTMU reasoning begins on the root level, its conceptual apparatus is capable not only of distinguishing among proximate mechanisms, but of providing them with a fundamental model of causality. Without such a model, what does or does not constitute a "causal explanation" is moot.
IP: Logged
|
|
Argon
Member
Member # 276
|
posted 07. August 2003 08:32
I generally agree with Chris' comments. I'll add that although CTMU may be to provide a fundamental model of causality, the actual connections to specific biological phenomena may not be readily determinable or predictable from the root theory. That's not because there isn't a real connection, only that, in practice, as in weather forecasting, such modeling could be computationally impossible for us to perform. So there will have to be a lot of "fill-in" work that needs to be performed before the connection can be linked.
[ 07. August 2003, 13:02: Message edited by: Argon ]
IP: Logged
|
|
Christopher M. Langan
Member
Member # 264
|
posted 07. August 2003 13:41
Argon comments that "although CTMU may be to provide a fundamental model of causality, the actual connections to specific biological phenomena may not be readily determinable or predictable from the root theory." Although I appreciate the use of "may", whereby Argon seems to acknowledge the speculative character of his statement, it pays to be careful when making statements of this kind...especially to the authors of theories to which they refer. The author of a theory or model can usually be expected to know somewhat more than a layman about the potential for specificity, if only due to his sheer familiarity with the material. In particular, while it may not be superficially apparent how a global model of causality might (computably) generate specific predictions, its ability to do cannot be ruled out on the basis of a general introduction. Because the CTMU is in a unique position with respect to the derivation of logical truths about the natural world, a little caution may be appropriate when it comes to setting limits on its implicative power.
IP: Logged
|
|
|
Printer-friendly view of this topic