Member # 594
posted 30. January 2003 11:48
quote:Stone circles would possibly be a good example. And with the people doing the test unaware of any of the physical explanations that are now available (e.g. instructed to explicitly ignore the recently gained knowledge). In the ID method, there needs to be an explanation of why the EF did not produce a positive result. Stone circles appear to have the same level of complexity as the most primitive of crop circles (not that of the more pictorial and advanced crop circles, mind you). So this case run today would be more of a gedanken experiment.
quote:I think this is an excellent idea. Can you give an example of the sort of pattern you have in mind? Thanks.
The tests must include the case in which the ID advocates are willing to acknowledge no ID is active, and yet there is to those running the test no known knowledge of how a physical cause could exist for the event.
Another interesting case are the rings of Saturn. (As an aside specific to the rings of Saturn, they are not predicted by physical law as it is currently understood now. See NASA link.) And of course RBH has already brought up Oklo reactor.
Now what is very important here is these explanations of why they might not be picked as “designed” is that there are likely to be explanations found for these structures/events. Remember that scientists think that they have found explanations for the bacterial flagellum, or at least that such explanations would be likely to exist.
The EF depends on “sweeping the field clean” of explanations. But there is no way to prove that there is no as yet unknown explanation -- so this is a subjective determination of whether such an explanation is likely. But the ID advocates won’t acknowledge that if there is a probability of greater than the UPB (10^-150) of an explanation being found, then the known probability of the event is not less than 10^-150 (or other UPB).
(Actually this is that the product of the probability that an explanation will be found, times the probability that the “found” explanation would produce the structure found, is less than 10^-150, summed over all such pairs. Since most “explanations” would result in a probability of much greater than 10^-50, for example we are only need a 1 in 10^100 probability of such an explanation being found to show that there is a greater than 10^-150 probability of a physical cause for the event even if we don’t yet know that explanation -- and thus the EF never produces a positive result when faithfully applied! Since it never produces a positive result, except when the positive answer is already known, then it is 100% reliable, and useless as a test!)
The point of finding why these examples were declared as not passing the EF, even though no explanation is known to the experimenter, is that it will in my opinion be regularly shown that the reason is that the experimenter has an opinion or assumption that an explanation can be found and thus declares that the pathway of low probability is not found by the EF -- even though the actual physical cause has not yet been found.
The absolute “set” compliment aspect of the explanatory filter makes it actually logically sound from a certain standpoint. By that, I mean that the EF is sound in the sense that its definitions preclude its ever failing to be correct if complete knowledge of all mechanisms are known -- e.g. far beyond any reasonable state of science into omniscience. The problems with the EF are that it is not sound in a practical world because it is supersensitive to the lack of knowledge of mechanisms and the outcome is easily reversed upon a gaining of a small bit of new knowledge. Therefore it is not “reliable” in the real world, and it’s apparent reliability stems from its being a tautology.
Now we must compare the EF (in its nature as tautological) to scientific theories that also have an element of being tautological. In instances we find that the descriptive language of a theory in in terms of the theory, and some describe this condition as making scientific theories tautological. What is important is the nature of these theories fitting to the evidence repeatedly, and providing a degree of predictivity. That there is a degree of circular reasoning does not make a scientific theory invalid -- in fact it is almost a requirement, and that is why science is not “proven” but rather a theory can only be “disproven” by significant counterexamples. What is important is to distinguish a theory from a test. A theory is a construction which in itself makes predictions (even if they are not hard predictions like object A will move to position B within close tolerance, but rather are of the form object B will be related to object A with certain constraining relationships). A test makes no such predictions, rather can be used with a theory as a means of experimental verification of an aspect of a theory because the test is found to be in concordance with the theory. Furthermore many tests don’t have validation within a particular theory, rather they are validated independently by observation of “reliability” in the real world.
Since the EF is a method of test, not a theory of how the world works, we don’t judge it by the same standards as a theory of how the world works. We judge it by standards of how well the test works in the real world as a test of the conditions it is designed to test, just as we would judge a test for finding a particular chemical, for example. If the test is found to be supersensitive to extraneous information (like something we didn’t know previously) then we would throw out the test. (And this is precisely what we find, by the way.)
Now the EF does seem to work in some circumstances. This is in part because of its tautological nature. But it is also due to its logical form. When used in circumstances in which there is high prior probability of a designer producing the effect, then it tends to be reliable. (The reasons for this are embedded in Bayesean theory -- the mathematics works out when we have as a given that there is relatively high prior probability of a “designer” producing the observed effect.) But of course that the prior probability of a designer producing the effect is reasonably high is a knowledge of the “designer” pathway of action. Since the EF explicitly eschews any such knowledge, the EF can’t be reliable. It can be fixed simply by adding the condition to the EF that the prior probability of a designer acting to produce the effect is relatively high -- but then of course we simply have an inference to the best explanation and not an “eliminative” method.
So any test of the EF that will show how it fails will by necessity be a case in which there is extremely low prior probability of a “designer” producing the observed effect. (And by the way, all the examples given by ID advocates to show the test being “reliable” happen to be cases in which the reader would agree that the prior probability of a designer acting to produce the result is relatively high probability.)
Due to the tautological nature of the EF, it can’t be falsified. This is in the sense that if we found a failure in the real world ID advocates would actually declare the method to be a success, because we found the previously unknown mechanism that made the event have a higher probability of causing the effect than the UPB (e.g. 10^-150).
These reasons are why the EF is so supersensitive to lack of knowledge, and to assumptions that must in some circumstances be philosophical and religious. We note that religion and philosophy is not something that is supposed to be free from requirements of consistency with observation -- just that religion and philosophy has additional aspects that go beyond scientific assumptions. For this reason, the EF is unreliable because it depends on philosophical assumptions over which there are a great deal of disagreement in society. Namely the EF applied to biology, for example, depends on assumptions that an unembodied designer acts regularly in the world.
Since no experimental verification of the EF “test” can be done because of its tautological nature, we can only hope to show that the EF fails as a reliable test when based on real-world assumptions. There are problems with my original proposal: "The tests must include the case in which the ID advocates are willing to acknowledge no ID is active, and yet there is to those running the test no known knowledge of how a physical cause could exist for the event.” The problem of course is that all these cases (like stone circles, rings of Saturn, Oklo reactor) are cases in which the ID advocates would also say that there is good reason to expect a physical explanation to be found, if it has not yet been found. Thus the EF should take the pathway of there being a higher than UPB probability of the events being caused by natural processes.
What is left out by the ID advocates is that if this reasoning causes the stone circles, rings of Saturn, and Oklo reactor to be declared as not meeting the UPB and thus not a positive outcome of the EF, then we must apply the same standard to the bacterial flagellum, for example. It must also be declared to not be “designed” -- not because the explanation for the flagellum is complete, but because there is expectation in the scientific community that a possible pathway will be discovered (if not already done so). And note that since the EF is a matter of “sweeping the field clean”, the field for the flagellum has clearly not been “swept clean” of explanations.
In the case of the bacterial flagellum, together with the larger set of so-called “irreducibly complex” structures in biological forms, we have the ID advocates assuming for the most part that an unembodied designer has a high prior probability of acting. This assumption makes the ID cause a reasonable conclusion (based on assumptions). But those who find that science has a good history of coming up with physical explanations for physical structures and events find a high prior probability that an explanation will be found for the development pathway of the flagellum. I think that RBH and others have already identified this problem of the method depending too much on subjective aspect -- and thus the EF as a “test” is clearly too subjective to be considered scientifically reliable.