|
Author
|
Topic: An Argument: No Increase in CSI After the Origination of Life
|
Tom Stalnaker
Member
Member # 114
|
posted 07. March 2002 05:33
On this post, I am going to present an argument that there has been no increase in complex specified information (CSI) in any life-forms subsequent to the first single-celled organisms. In other words, Dembski's explanatory filter does not indicate any additional design beyond that which occurred prior to and during abiogenesis. As an aside, please note that my purpose is to do my best to argue against Dembski's arguments regarding intelligent design so that we can learn more about how his arguments work.
I am hoping that someone can show me where this counter-argument to Dembski's arguments goes wrong.
As an example, I will examine the appearance of humans, and argue that humans contain no more CSI than bacteria, and thus the design inference does not indicate that any intelligent intervention has occurred after the appearance of bacteria. (I am assuming that common descent is true, and therefore, if you believe that human life arose independently of other life forms, this argument does not apply.)
The argument starts as follows: if bacteria merely reproduced and proliferated up until the present, and the descendents always remained similar to their progenitors, we would not regard that as an increase in CSI. No intelligent design is needed to account for this event because, once life has become established, its contiuance, unchanged, is a matter of regularity. Therefore, my preliminary point is that in as much as the human life form is an elaborated version of bacteria (!), it reflects no additional intelligent design.
However, it seems that the assumption of the ID argument is that humans show an increase in CSI, relative to bacteria, that must be accounted for; in other words, the design inference indicates that an intelligent designer must have added CSI in order to create humans. My response is that any increase in complexity manifested in humans is not specified . In Dembski's explanatory filter, an event must be both complex and specified in order not to be attributable to chance. To know if an event is specified in Dembski's sense, we should ask "Given an event, would we be able to formulate a pattern describing it if we had no knowledge which event occurred?" (Dembski, _The Design Inference_, p 15) So, the correct question is: If we had no knowledge of which life forms appeared subsequent to bacteria, would we be able to formulate a pattern describing the human life form (or any other)? The only specified pattern describing humans (that I can think of) is that they are able to propagate genes. However, this specified pattern is identical to the one already existing in bacteria billions of years ago, and the maintainance of a specified pattern does not indicate an increase in CSI. Thus, insofar as humans show an increase in complexity over bacteria, it is not specified, and insofar as humans reflect a specified pattern, it is the same as that in bacteria. Therefore, the explanatory filter fails to rule out chance (or, more likely, chance and regularity working together) for the evolution of humans from bacteria.
[ 07 March 2002, 05:35: Message edited by: Tom Stalnaker ]
IP: Logged
|
|
Tom Stalnaker
Member
Member # 114
|
posted 07. March 2002 05:38
Another way to look at this argument is as follows: Dembski uses the example of an evolutionary algorithm seeking a target to show that such an algorithm cannot find a specified target any better than a blind search, unless the algorithm is somehow provided with additional information by an intelligent designer. This argument appears to be unassailable. However, we should look closer at what real Darwinian evolution is supposed to be doing (according to Darwinists). Once life has been established, Darwinian evolution is not seeking a target. The only "target" (better called the "starting point") for Darwinian evolution is to have replicating genetic information (i.e. life). Once that starting point has been reached, Darwinian evolution has no further targets, and thus needs no further information. It need only maintain its position on the starting target, and that can be done perfectly well with a blind search (assuming the target does not move extremely quickly). Any subsequent formation of more complicated life forms will not be due to the seeking of a pre-specified target, but rather will be due to random elaborations of the already formed CSI. Thus, evolutionary algorithms that are seeking a target, such as those set up by programmers, are not analogous to Darwinian evolution, which has no target. When people set up an evolutionary algorithm, they often want it to solve some pre-specified problem for them. Dembski's point is that these algorithms can only solve problems insofar as they have been designed specifically to solve a certain kind of problem. Darwinian evolution is often portrayed as if it solves problems, such as how to overcome certain environmental obstacles; however, any problems that it solves are not pre-specified -- i.e. it only solves problems in an ad hoc sense. Thus it does not generate, nor does it need to generate, complex specified information.
[ 07 March 2002, 05:42: Message edited by: Tom Stalnaker ]
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 07. March 2002 12:24
quote:
Originally posted by Tom Stalnaker:
Another way to look at this argument is as follows: Dembski uses the example of an evolutionary algorithm seeking a target to show that such an algorithm cannot find a specified target any better than a blind search, unless the algorithm is somehow provided with additional information by an intelligent designer. This argument appears to be unassailable.
Dembski's argument is based on the NFL theorem but recently it has been shown that NFL theorems do not apply under certain circumstances and it has been argued that under most common circumstances NFL theorem does not apply. In fact this make sense given that genetic algorithms do seem to work quite well (Tom Schneider, evolution of biological information)
quote:
Abstract
In a recent paper [3] it was shown that No Free Lunch results [5] hold for any subset F of the set of all possible functions from a finite set X to a finite set Y if F is closed under permutation of X. In this article, we prove that the number of those subsets can be neglected compared to the overall number of possible subsets. Further, we present some arguments
why problem classes relevant in practice are not likely to be closed under permutation.
and the conclusion is that
quote:
Conclusion
Based on the results in [3], we have shown that the statement "I'm only interested in a subset F of all possible functions, so the NFL theorems do not apply" is true with a probability close to one (if F is chosen uniformly and Y and X have
reasonable cardinalities). Further, the statements "In my application domain, functions with maximum number of local minima are not realistic" and "For some components, the objective functions under consideration will not have the maximal possible steepness" lead to scenarios where NFL does not hold.
To me it seems that the claim that "... such an algorithm cannot find a specified target any better than a blind search" is hardly unassailable.
For instance in the Ev case Dembski argued that the CSI was introduced by the special way ties were handled.
quote:
William A. Dembski claims that the ev program does not demonstrate an information increase because of the SPECIAL RULE. His claim is:
For instance, in the crucial paragraph from his article that I quoted above, Schneider remarks parenthetically: "To preserve diversity [of organisms], no replacement takes place if [the number of mistakes is] equal." Schneider's Pascal source code reveals why: "SPECIAL RULE: if the bugs have the same number of mistakes, reproduction (by replacement) does not take place. This ensures that the quicksort algorithm does not affect who takes over the population. [1988 October 26] Without this, the population quickly is taken over and evolution is extremely slow!" (9) Schneider is here fine-tuning his evolutionary algorithm to obtain the results he wants. All such fine-tuning amounts to investigator interference smuggling in complex specified information.
Assume that Dembski had been correct, what is the relevance of this? What if nature had a similar rule as used by the Ev program. What would prohibit nature for employing a similar rule and thus generating CSI?
I believe that the argument "CSI is smuggled in" will fail to rebut the observations that algorithms do seem to be able to generate CSI. Certainly it has not been established that algorithms could not generate CSI now that the NFL theorem seems to have been found possibly inapplicable?
[ 07 March 2002, 12:37: Message edited by: Frances ]
IP: Logged
|
|
Janitor@MIT
Member
Member # 125
|
posted 07. March 2002 14:34
In the movie Sequence Logos, Machine/Channel Capacity, Maxwell's Demon, and Molecular Computers: a Review of the Theory of Molecular Machines, Tom Schneider appears as the reverent novice in the dramatic scene where Maxwell’s Demon is exorcised (Parental discretion advised.). The much anticipated sequel (Evolution of Biological Information) takes a totally unexpected plot twist as we see the formerly cherubic cleric attempting to resurrect the very monster whose demise he had celebrated in the previous movie!
See the movie! And say your prayers!
IP: Logged
|
|
Tom Stalnaker
Member
Member # 114
|
posted 07. March 2002 18:27
Actually Frances, my offhand statement about Dembski's argument _appearing_ unassailable was just that ... offhand. But now I will defend his argument. In Schneider's example, I believe the CSI is "smuggled in" with the way that the "organisms" are selected. That fitness function is specifically designed by Schneider so that organisms are guided towards the goal that he has set for them. Unintelligent nature, on the other hand, does not have the ability to choose the fitness function that would direct its organisms towards a particular target. Schneider even points out that as soon as one of his organisms gets one step closer to the target (i.e. makes one less error), it very quickly takes over the population. So therefore its not surprising that his organisms very quickly progress to making no errors -- they are being guided by a fitness function that was intelligently designed. Dembski clearly acknowledges that evolutionary algorithms are very good at finding solutions for particular classes of problems once a good fitness function for that class of problems has been designed. However, that doesn't mean that such algorithms are truly intelligent, or that they independently generate CSI.
My argument against Dembski's argument takes another approach, which is to point out that life forms do not really have any CSI except for the CSI, present in all life, of having the ability to propagate genes.
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 07. March 2002 22:41
quote:
Originally posted by Tom Stalnaker:
Actually Frances, my offhand statement about Dembski's argument _appearing_ unassailable was just that ... offhand. But now I will defend his argument. In Schneider's example, I believe the CSI is "smuggled in" with the way that the "organisms" are selected.
I find that interesting but when Tom tested Dembski's claim it was found that it did not make a difference. In fact there is no real 'fitness function', the simulation uses a simple reproduction/mutation/selection criterium just like found in nature. If the argument is that the fitness function in nature is what generates CSI then I have no objection, in fact I believe that this is what actually happens.
quote:
That fitness function is specifically designed by Schneider so that organisms are guided towards the goal that he has set for them.
Could you perhaps tell us what that goal is? I have looked at Schneider's code and not found evidence of a goal per se. But I understand fully that this may be due to my own oversight and would like to have others take a look at well.
quote:
Unintelligent nature, on the other hand, does not have the ability to choose the fitness function that would direct its organisms towards a particular target.
I do not think that Tom selected a fitness function different from what unintelligent nature could do.
quote:
Schneider even points out that as soon as one of his organisms gets one step closer to the target (i.e. makes one less error), it very quickly takes over the population. So therefore its not surprising that his organisms very quickly progress to making no errors -- they are being guided by a fitness function that was intelligently designed. Dembski clearly acknowledges that evolutionary algorithms are very good at finding solutions for particular classes of problems once a good fitness function for that class of problems has been designed.
Again, what fitness function has been designed here?
quote:
However, that doesn't mean that such algorithms are truly intelligent, or that they independently generate CSI.
So how does one show that an algorithm cannot generate CSI. Dembski suggests that CSI is smuggled in but so far there has not been evidence presented of such smuggling in of CSI.
quote:
My argument against Dembski's argument takes another approach, which is to point out that life forms do not really have any CSI except for the CSI, present in all life, of having the ability to propagate genes.
An interesting argument indeed and I do not want to distract from your argument here.
In fact Schneider shows that there is no effect in case of no selection and that it seems clear that it is the selection step that leads to the genome increasing its Shannon entropy.
In fact it seems to me that nature is determining what mutations are relevant and what mutations are not.
IP: Logged
|
|
Bryan Cross
Member
Member # 51
|
posted 08. March 2002 10:38
CSI is not had for free if the CSI is frontloaded into nature. It just pushes the question of the origin of the CSI back one step.
quote:
“If we had no knowledge of which life forms appeared subsequent to bacteria, would we be able to formulate a pattern describing the human life form (or any other)?”
Who is the “we” in this sentence?
quote:
“Once that starting point has been reached, Darwinian evolution has no further targets, and thus needs no further information.”
Whether Darwinian evolution ‘needs’ any more information at that point or not does not establish that humans have no more CSI than bacteria.
quote:
“Darwinian evolution is often portrayed as if it solves problems, such as how to overcome certain environmental obstacles; however, any problems that it solves are not pre-specified -- i.e. it only solves problems in an ad hoc sense. Thus it does not generate, nor does it need to generate, complex specified information.”
The conclusion is a non sequitur. Just because Darwinian evolution only solves problems in an ad hoc sense does not entail that it cannot generate further CSI.
- Bryan
IP: Logged
|
|
Alex Wild
Member
Member # 165
|
posted 08. March 2002 12:22
quote:
In Schneider's example, I believe the CSI is "smuggled in" with the way that the "organisms" are selected. That fitness function is specifically designed by Schneider so that organisms are guided towards the goal that he has set for them. Unintelligent nature, on the other hand, does not have the ability to choose the fitness function that would direct its organisms towards a particular target.
Sure, but the notion that organisms have particular targets is a metaphysical assumption that is not empirically testable. Evolutionary algorithms invariably place lineages in higher optima than the immediate surroundings in a fitness landscape. The question becomes, was a particular optimum a pre-ordained target, or just an optimum that the algorithm found and climbed?
How do we know, for example, that modern porcupines were a target? What if there is a higher optimum, say, winged porcupines, that the porcupine lineage was intended to have reached but didn't? Is there any empirical way to test this idea?
IP: Logged
|
|
Tom Stalnaker
Member
Member # 114
|
posted 09. March 2002 01:01
Hi guys. Bryan: I agree that if CSI is front-loaded into nature, you don't get it for free. But I think that one should be able to determine whether it is in fact front-loaded, or whether it is being added on a regular basis. My argument is that there is no need to postulate CSI being added after the origin of life. If you want to argue that CSI is in fact added after the origin of life, then you need to come up with an affirmative answer to the question that I asked, quoted below.
I asked “If we had no knowledge of which life forms appeared subsequent to bacteria, would we be able to formulate a pattern describing the human life form (or any other)?” , and you said "Who is the “we” in this sentence?"
"We" is anyone who is trying to determine whether humans exhibit CSI above and beyond bacteria. Feel free to substitute the pronoun that makes sense to you: ("you" or "one", or if you are posing the question to yourself: "I").
Alex:
I agree with you completely! There is no evidence that organisms are evolving towards particular targets. That is why I think that Dembski's argument about evolutionary algorithms might be a little misleading. That argument is based on the idea that such an algorithm cannot find a particular target without an infusion of CSI. According to Darwinists, Darwinian evolution is not seeking particular targets, and therefore I don't think that Darwinian evolution needs an infusion of CSI.
IP: Logged
|
|
Bryan Cross
Member
Member # 51
|
posted 13. March 2002 11:12
Tom,
quote:
My argument is that there is no need to postulate CSI being added after the origin of life. If you want to argue that CSI is in fact added after the origin of life, then you need to come up with an affirmative answer to the question that I asked, quoted below.
I asked “If we had no knowledge of which life forms appeared subsequent to bacteria, would we be able to formulate a pattern describing the human life form (or any other)?” , and you said "Who is the “we” in this sentence?"
"We" is anyone who is trying to determine whether humans exhibit CSI above and beyond bacteria. Feel free to substitute the pronoun that makes sense to you: ("you" or "one", or if you are posing the question to yourself: "I").
You are not just arguing that no CSI is added after the origin of life; anyone who holds to frontloading can agree with that. Nor is that the claim I am contesting. You are also arguing that humans have no more CSI than bacteria. This latter claim is the stronger claim, and I don’t think your argument substantiates it. You attempt to prove this stronger claim by asking whether if “we” had no knowledge of which life forms appeared subsequent to bacteria, “we” could formulate a pattern describing the human life form. If “we” cannot answer ‘yes’ to this question, then you conclude that humans have no more specificity than bacteria.
In Dembski’s formulation of the criterion for specificity, (which you quote from p. 15 of DI), the “we” refers to humans. It refers not just to any one member of the human race, but collectively to all humans having a sufficient degree of intelligence, speech capacity and understanding. (For example, just because a human baby cannot formulate a list of consecutive prime numbers does not mean that such a list is not specified.) All such sufficiently intelligent humans necessarily have at least some degree of self-understanding. Therefore, it is impossible for this set of humans to lack completely knowledge of their own existence and humanity. Therefore, the protasis of the conditional in your question can never be met. Therefore your question is an ill-formed question. Hence, your claim that humans have no more CSI than bacteria is not established by your argument.
- Bryan
IP: Logged
|
|
Tom Stalnaker
Member
Member # 114
|
posted 15. March 2002 17:17
Bryan,
"We" does not necessarily refer to humans. It is being used as a non-specific pronoun, in the sense of this: "If someone had no knowledge of which life forms appeared subsequent to bacteria, would that someone be able to formulate a pattern describing the human life form (or any other)?" The "someone" is abstract, in the sense that it is not any individual or group in particular, but rather means, in general, is it informationally possible to formulate a pattern describing the human life form, given the information state described by the conditional. The intelligence needed to evaluate this question is independent of any particular manisfestation. In other words, it could be an alien, a computer, or whatever. By the way, you can use pronouns in abstract counterfactual conditionals like that example, regardless of the physical possibility of a particular person or group of persons being able to fulfill the conditional. For instance, you could ask "If someone were viewing the earth from the sun, what would the earth look like?" Yes, it might be hard to answer, but no, that is not an illformed question.
You might be thinking that it is logically impossible that a human could be unaware of which life-forms appeared subsequent to bacteria, and that a logically impossible conditional is not meaningful. However, that is not the case here. It is even logically possible that a human could be unaware that he is a human, or that he is a life form. Any state that it is logically possible for one human to be in, is also logically possible for all humans to be in.
Anyway, all of that aside, you seem to be saying that it is impossible to answer (or even ask) the question that Dembski poses as part of the generic explanatory filter, in this particular case. That would suggest that it is impossible to apply Dembski's explanatory filter to humans. Thus we can't know whether humans were intelligently designed or not. Is that your position? If so, I think your biggest opponent on that point would be Dembski.
Finally, I think I see what you mean about me making a weaker claim and a stronger claim, with the weaker claim being equivalent to arguing for front-loading of CSI, and the stronger claim being that humans have no more CSI than bacteria. I was confused about this point when I wrote the original post, and so thanks for clearing that up a little. I actually meant to stick to the stronger claim, and I am still honestly interested in someone showing me which bit of complexity in humans is "specified", in Dembski's sense. In DI, Dembski says that all creationists and evolutionists agree that life is specified, and I would agree that life, per se, is specified. But is any _particular_ life form specified? If not, then I still hold that, aside from the origin of life, there is no need to postulate any intelligent design that was either front-loaded or added bit by bit. In other words, my argument is that the Intelligent Design argument applies exclusively to the origin of life.
[ 16 March 2002, 07:41: Message edited by: Tom Stalnaker ]
IP: Logged
|
|
nobody
Member
Member # 145
|
posted 19. March 2002 01:36
Tom: "In other words, my argument is that the Intelligent Design argument applies exclusively to the origin of life."
You do?
Are you claiming that the information content of bacteria DNA is equivalent to the information content of human DNA?
IP: Logged
|
|
Bryan Cross
Member
Member # 51
|
posted 19. March 2002 22:25
Tom,
quote:
You might be thinking that it is logically impossible that a human could be unaware of which life-forms appeared subsequent to bacteria, and that a logically impossible conditional is not meaningful. However, that is not the case here. It is even logically possible that a human could be unaware that he is a human, or that he is a life form. Any state that it is logically possible for one human to be in, is also logically possible for all humans to be in.
I don’t disagree with you about the logical possibility that all humans could be unaware that they are human. My claim is that any being capable of asking the question you are asking is necessarily self-aware. That is, any being capable of asking your question will necessarily know that it is a being capable of asking questions. Hence, that being will necessarily be consciously aware of at least one property or capacity describing him or herself, i.e. rationality. Thus any *human* asking and understanding your question will necessarily be aware of his or her own rationality, and aware of the nature of bacteria. In other words, any human capable of asking your question will necessarily have awareness of at least one property describing him or herself, and not describing bacteria. So any human capable of asking your question will necessarily know that if he or she appeared subsequent to bacteria, then there is a least one property describing his or her own life form, i.e. rationality.
Dembski’s explanatory filter does not rule our false negatives, only (he claims) false positives. So it does not matter that it is logically possible that all humans could be unaware of their humanity. That would not show that humans are not designed. If even one person can formulate the pattern independently of the event, then (according to Dembski’s explanatory filter) this indicates intelligent design.
quote:
Anyway, all of that aside, you seem to be saying that it is impossible to answer (or even ask) the question that Dembski poses as part of the generic explanatory filter, in this particular case. That would suggest that it is impossible to apply Dembski's explanatory filter to humans. Thus we can't know whether humans were intelligently designed or not. Is that your position? If so, I think your biggest opponent on that point would be Dembski.
I don’t know what Dembski would say about my comments above. As I said above, Dembski would be the first to admit that his explanatory filter does not eliminate false negatives. So even if we human could not formulate a pattern describing the human life form, that does not indicate that humans have no more CSI than bacteria. But if I am right, then there are certain patterns or properties that rational beings are necessarily capable of formulating. If that is true, then according to Dembski’s explanatory filter, those patterns either always indicate design or they simply cannot be tested by Dembski’s explanatory filter because they are not detachable (but in the opposite direction, i.e. from within, not from without).
Just to be clear, I’m clearly not saying that the map of the human genome is something we can necessarily formulate if we didn’t know what life forms appeared after bacteria. I have been talking about rational and linguistic capacities. But what about all the information in human beings? Your argument does not show that human beings have no more CSI than do bacteria. At best, your argument would only show that Dembski’s explanatory filter cannot tell us whether human beings have any more CSI than do bacteria.
Moreover, you have raised the bar unnecessarily high for specificity. Your criterion for specificity requires that the inquirer be capable of formulating the pattern of *what would have occurred*. Dembski’s criteria for specificity allow that the inquirer be capable of formulating the pattern of what is present. Hence, for Dembski, the inquirer does not have to know which event is going to occur (or would have occurred) in order for the event to be designed or specified. In other words, Dembski's explanatory filter does not only apply to events; it also applies to information. You have described in such a way that it only applies to events. But Dembski applies it also to objects, without prior knowledge of the events of those objects' formation. Hence for Dembski we can examine, say, the organs and systems of the human body and determine whether they are specified by determining whether the function they perform corresponds to any functional pattern of which we can independently formulate. Then if these organs and systems are complex, and if they are not had by bacteria, then the human would have more CSI than do bacteria.
- Bryan
[ 19 March 2002, 22:27: Message edited by: Bryan Cross ]
IP: Logged
|
|
|
Printer-friendly view of this topic