|
Author
|
Topic: No False Positives and the Lust for Certainty
|
Frances
Member
Member # 169
|
posted 07. November 2002 13:48
Paul asks the following excellent questions
Do I agree or disagree with Wilkins and Elsberry's statement
quote:
...why are we ever forced to a rarefied design conclusion? Surely we can content ourselves with regularities, chance, and "don't know" explanations. (p. 721)
I believe their conclusion is not dissimilar from mine. When we have eliminated all known chance hypotheses why should we infer Intelligent Design and not "we don't know"? What makes ID a better explanation than 'we don't know'? Absent any independent evidence Such overreaching inferences as a rarefied design inference carry a heavy metaphysical burden, and the onus is on the proponents of such an a priori assumption to justify it.
Source
and
quote:
Also, using your own words, please explain the criteria that distinguish ordinary and rarefied design inferences
"Independent evidence" allow us for inductive generalizations.
quote:
Independent evidence and explanatory power need to work in tandem and for one to outpace the other typically leads to difficulties.
W. Dembski NFL p. 91
Ordinary design as found in for instance archaeology or criminology provides us with indirect and independent evidence allowing us to make non ad-hoc generalizations.
Rarefied design "based on inference from ignorance", or what is left over when our chance hypotheses fail or when we have no such hypotheses yet and we have no additional independent data.
IP: Logged
|
|
Paul A. Nelson
Member
Member # 26
|
posted 07. November 2002 15:11
Again, to Frances:
Thought experiment: Let's suppose that life was designed and brought into existence by a non-human intelligence. That's what actually happened, and we want science to be able to discover (or infer) what actually happened. So, using biological evidence, we infer that life was intelligently designed.
Now, Wilkins and Elsberry would call this a "rarefied" design inference. According to their analysis, in science "no rarefied design inference is needed" (p. 722). Therefore science cannot infer certain possible states of affairs, despite the fact that (under our thought experiment), those states of affairs are not only possible, but actually the case (i.e., true).
I'd say there is something profoundly wrong with the Wilkins/Elsberry analysis. The thought experiment stipulates that an empirically possible conclusion is true (i.e., life was intelligently designed). The Wilkins/Elsberry analysis says, however, that this true conclusion cannot be known or inferred.
In your understanding, when could one legitimately infer so-called rarefied design?
About the ordinary vs rarefied distinction. Is the search for extraterrestrial intelligence (SETI) employing an ordinary or rarefied inference?
[ 08. November 2002, 06:37: Message edited by: Paul A. Nelson ]
IP: Logged
|
|
Terence A-H Tan
Member
Member # 234
|
posted 07. November 2002 15:36
This is written in keeping with the original thread. Regarding Axe's paper ... i'll read it in my other other spare time.
Continuing my 2 cents worth of spare time mullings,
What strikes me about the false positive issue is the importance of understanding the paradoxical dual nature of fasifiability and confirmatory in any particular scientific research.
Drawing from insights into how my profession thinks in medicine, physicians are challenged to constantly form theories/hypothesis for each patient encounter as this is a pre-requistie to solving their problems. In the search for overall accurate answers reflecting the true nature of the black box, we not only engage in searches for specific archaeological signs of mechanisms that have ceased to exist but also for signs of phenomena that continue to exist. Now if we were to draw the analogy of Normality being theory X which we are trying to prove, than any signs of normality would be considered confirmatory while any signs of disease falsifies theory X. That doesn't mean that falsification is not useful or redundant in the research. In the way we approach patient problems, we physicians not only hunt for confirmatory signs to prove theory X but also for signs of no falsification or no disease signs to prove theory X. This is where the paradoxical nature of falsifiable signs is unveiled. When one says there is no falsifiable sign, then that sign becomes confirmatory to theory X, the truth being in the paradox. Hence confirmatory research conducted to prove evolution is the same as falsifiable research conducted to prove ID and when the outcome is not falsifiable, than it is the same as confirmatory result for ID. Thus specified complexity with its EF can be justified as a confirmatory scientific philosophical point for ID and is especially useful for this unique scientific informational category in which the design is finished and all prior work or mechanisms used is history. Can we attempt to try and reconstruct or understand the original mechanism used to create the design as confirmatory research ? Yes we can, by creativie chemistry, as we are all aware of designer oligos being synthesized everyday. The question as Dembski puts it is not in the reliability of the theory but in the object to which it is applied.
If we can find multiple specified complexities within an object, this would further strengthen the case for ID. A simple eg. from human intelligent intervention in biology: A clone of human cancer cell line expressing green fluorescent protein to which the latter's spontaneous presence in a place where it does not belong can be safely regarded as an impossibility. Thus we can say clearly that the presence of GFP in the human cell line indicates intelligent design.
A further philosophical point to mull on will be the unidirectional nature of signs to which no false positive can be attributed. This, I wish I have more time to mull on. Another day ...
IP: Logged
|
|
Janitor@MIT
Member
Member # 125
|
posted 08. November 2002 00:28
Thoughtexperiment:
With Dr. X’s prescription and results in hand, you set up and prepare to repeat his experiment. Perfectly routine experiment. There is nothing new to be discovered here. The experiment has been repeated many times, and you know what to expect. But then the results all come out wrong! They’re nothing like what you expected! They are not even close to Dr. X’s results!
What is the first thing you think?
IP: Logged
|
|
Terence A-H Tan
Member
Member # 234
|
posted 08. November 2002 10:13
" thoughtexperiment: You are a scientist conducting an experiment, perfectly routine experiment. All you are doing is refining a measurement, adding n+1 digit to a physical constant that is already known with high precision to the nth digit. But then your experiment returns gibberish, totally unexpected results. “These numbers are all wrong!” What is the first thing you think? "
I for one am adverse to using riddles and mystery names on a discussion table. They generate too many confusion without any clarification. This is because each thought experimenter has his/her own unique thought box to process the experiment. The output answers for the thought experimenter is straightforwards to himself/herself. To convert it to riddles only add to the confusion.
But since the strategy has been used, the only strategy left is to answer a riddle with a riddle.
Thoughtexperiment:
End of the world nuclear scenario that originated from someone pressing the wrong button because of a moment's faulty thought process. Nothing much left in the ashes. Scattered bits of clues here and there. A remnant of youngsters somehow survives with no clues as to what happened. A whole new generation of humans manage to propagate. Among which was a highly intelligent janitor who happened to be cleaning some of the ashes and stumbled upon a laboratory. He/She explored and discovered some flasks containing cell lines expressing GFP. What is the first thing he/she would think ?
Thoughtexperiment: Two scientists performing identical experiments with identical results. One scientist perusing the results shouts, “Eureka! Thank you National Science Foundation!” The other scientist peruses the same results and shouts, “Hallelujah! Praise Jehovah!” Which one is not a scientist?
IP: Logged
|
|
Janitor@MIT
Member
Member # 125
|
posted 08. November 2002 11:14
I apologize, Terence A-H Tan. It wasn’t a “strategy” or a “riddle.” I’ll answer the question myself. There could be many things wrong with this experiment, and many possibilites migth suggest themselves, but the first thing I would think is that there is something wrong with its design—such as I did not follow Dr. X’s instructions correctly? So the first thing I would do is recheck the set up.
But what exactly has gone into my thinking here?
Against some background of expectation I have received an unexpected result. The very first thing I do is consider a “design explanation” for the result. In the process I have immediately, and one might say even unconsciously, eliminated either “chance” or “necessity” as an explanation. Simultaneously, I have associated in my mind a possible design explanation with an improbable or unexpected result or occurrence.
I’ve only read “The Design Inference.” I’ll be quite honest, I thought it was interesting, but otherwise unremarkable. (With all due respect to Dr. Dembski, I don’t mean to diminish the effort and the insights in the least.) The reason I found it to be so is because of what Dembski said right at the beginning of the book, and which seems to have been lost in all the debate over the formalism he develops: All people think this way. Including all natural scientists.
Any experiment that is really worth performing is not going to have a simple and obvious result. These are the kind of experiments that contribute most to our understanding. It is not even a question of positives or negatives, false positives, or false negatives! Every experiment simultaneously tests three things: a method, a theory, and a scientist. By far the most problematic and generally unreliable isolable component in every experiment is the scientist. It is critical to account for the scientists’ expectations.
All the thoughtexperiment demonstrates is that these expectations quite naturally and in a perfectly logical and unforced way include design inferences and explanations.
Dr. Dembski’s point is well taken, at least by myself. None of the arguments of his critics has even addressed what I have repeatedly observed with my own eyes!
(The “Eureka!” or “Hallejuah” question occurred in a completely different context. And, taking the part of my own critic, it wasn’t even very relevant there. It really is has nothing to do with this question.)
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 08. November 2002 12:26
Paul,
An interesting thought experiment. How would one using biological evidence infer that life was intelligently designed? Based on what evidence was this inference made?
Before we can refer to this as rarefied design we need to know the details of the inference.
You seem to be worried that it may be possible that we cannot infer design when design actually happened or that we infer ignorance when in fact it was design. But the same applies if life arose not through intelligent design but through a processes yet beyond our reach of understanding. We would be infering ID when there was no such thing.
Dembski refers to the your example as false negatives and the second example as false positives. Dembski is clear that the design inference may have false negatives, that is, it does not infer ID when ID did happen.
In this case it is our ignorance of details that has left us unable to infer ID or natural pathways. Such is life...
Wilkins and Elsberry approach shows correctly how we should infer ignorance in both cases. That we 'know' the correct answer is of no relevance to these hypothetical examples.
Rarefied design is a tricky concept which does not seem to be different from ignorance of natural or chance pathways. As such the collapsing of the filter may be defensible.
Is SETI ordinary or rarefied. I would say its ordinary in the sense that SETI is not claiming that it will identify all SETI signals but only a small subset based on certain assumptions. Thus there will be inevitably false negatives and the pulsars showed how false positives could easily sneak in...
SETI is looking for narrow band signals similar to our radio signals which are less likely to have natural equivalents. It makes the assumptions that Intelligent Life out there may be using the same means of communication as we are.
As such inductive generalizations seem possible with the understanding that we are limiting the ID search to a subset of ID.
IP: Logged
|
|
Paul A. Nelson
Member
Member # 26
|
posted 09. November 2002 10:21
Frances wrote:
quote:
An interesting thought experiment. How would one using biological evidence infer that life was intelligently designed? Based on what evidence was this inference made?
It doesn’t matter. The thought experiment tests whether the Wilkins & Elsberry analysis allows us to infer so-called “rarefied” design. We stipulate that rarefied design is the case (true), and then, working backwards, ask if we could discover that truth using the Wilkins & Elsberry approach. We can’t -- in principle (see below). Therefore something is profoundly wrong with the Wilkins & Elsberry analysis.
What’s wrong isn’t hard to find. The reason the biological evidence doesn’t matter is Wilkins & Elsberry pointedly subordinate the evidence to “knowledge of designers” (p. 718). Since "rarefied design" is defined as any inference to a non-human intelligence for which we lack direct observational knowledge, and since Wilkins & Elsberry make such knowledge necessary for any legitimate design inference, one cannot use their method to discover causation by an unobserved non-human intelligence. Simple as that.
In fact, using the Wilkins & Elsberry filter, it is impossible to get beyond “we don’t know,” no matter how compelling the biological evidence. Cells could come stamped with bar codes (so to speak), and it wouldn’t make the least bit of difference. The biological evidence is strictly irrelevant.
That’s why Wilkins & Elsberry write:
quote:
...why are we ever forced to a rarefied design conclusion? Surely we can content ourselves with regularities, chance, and “don’t know” explanations. (p. 721)
and
quote:
...no rarefied design inference is needed. (p. 722)
To use a sporting metaphor, they’ve sewn a net that lets the fish of rarefied design escape every time.
For a skeptic who just doesn't want to catch the fish of design, that's an acceptable outcome. For any curious and open-minded scientist, however, who allows that design might have happened -- and wants to give the evidence half a chance -- the Wilkins & Elsberry approach is going to be an utter disaster.
You write:
quote:
Wilkins and Elsberry approach shows correctly how we should infer ignorance in both cases.
Nonsense. Their approach is designed –- pun intended –- to always favor “we don’t know,” no matter what the biological evidence indicates. Indeed, their heavy reliance on David Hume, who didn’t care how much damage he did to science as long as he could avoid design inferences, is telling. The next time I see Wesley Elsberry (he’s a friend), I’m going to ask him under what circumstances he would infer rarefied design. If the answer is something like “Only when the designer himself knocks on my office door,” I’ll know the game is up. [Wesley or John, if you’re reading this, please jump in at any time.]
Incidentally, using the Dembski filter, it is always possible to discover new natural regularities. Unlike the Wilkins & Elsberry filter, there is no in-principle exclusion of certain empirical possibilities.
About SETI and the ordinary/rarefied distinction. You wrote:
quote:
Is SETI ordinary or rarefied. I would say its ordinary in the sense that SETI is not claiming that it will identify all SETI signals but only a small subset based on certain assumptions.
OK. Suppose we assume that a non-human intelligence intended us to discover its actions by making biomolecules such as RNA impossible to originate, naturally, under plausible early Earth conditions. Would that be ordinary or rarefied?
[ 09. November 2002, 10:28: Message edited by: Paul A. Nelson ]
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 09. November 2002 14:21
Paul,
I seem to understand why you consider Wilkins/Eslberry's filter to be 'insufficient' because it fails to detect something which you believe must exist. Indeed the EF as formulated by Dembski seems to place ID in a privileged position. What Elsberry and Wilkins have shown so eloquently is that while the filter may work for ordinary design, it is not helpful in detecting rarefied design.
There is nothing inherently wrong with the filter, in fact I would argue that it captures how we infer design more accurately than the EF proposed by Dembski. Sure there will be instances in which the filter cannot determine ID, or chance and will have to reserve judgement until more data becomes available.
Let's look at the following scenarios:
1. Some intelligent designer(s) created life using purely law and chance and in fact have made all efforts to ensure that we cannot distinguish their actions from natural law. Lets call this 'front loading' or as Murray calls it 'stacking the deck'.
2. Some intelligent designer(s) created life and intentionally left indicators that life was designed. In order for this to be detected we need to however eliminate all chance hypotheses.
3. Life is fully naturally and we can explain its occurrence through natural law although our knowledge at any given time will be limited.
4. Life is fully naturally but gives the impression that there are obstacles which it cannot overcome. However this is because there is a mechanism which remains hidden from our knowledge.
Here we have four scenarios for the filter to deal with.
The EF filter will likely rule for ID in case 2 and 4 and may infer design, in cases 1 and 3 depending on the state of our knowledge.
So we may have some false positives and false negatives. So how are we now going to distinguish between case 2 and 4? If one presumes that intelligent design exists although one has no independent evidence then one may want to have ID be the privileged hypothesis(es) but if one is a skeptic one may want to realize that we cannot really distinguish between our ignorance and such rarefied intelligent design. So will one give preference to a particular hypothesis in such case or will one assume 'we don't know yet' and work on additional ways to determine which of the various hypotheses is better.
Why does Paul consider that rarefied design cannot be infered through elimination a problem? After all Dembski does not seem to consider false negatives to be a real problem for his filter. In fact there may always be cases in which we can not determine ignorance or 'ID' or 'chance' and thus rather than giving a single hypothesis a privileged status, we will just have to raise our hands and claim 'we just don't know'.
Paul objects that Wilkins and Elsberry's filter favors 'we don't know' but one may raise a similar objection to the EF which favors 'ID' hypotheses. Rather than doing damage to science, they save science from embarassements of false positives or junping to conclusions not warranted by the evidence.
Natural regularities can be detected independently of the EF proposed by Dembski I would argue. And Wilkins and Elsberry do not exclude rarefied design, they merely accept that rarefied design and ignorance are two concepts which cannot be resolved using the EF approach. For the same reason that Dembski's filter cannot resolve the problem of false positives due to our ignorance.
Dembski's filter gives in such cases the benefit of the doubt to ID when Elsberry/Wilkins give the benefit of the doubt to 'we don't know'.
About Paul's final question, non human inteligence has hidden its existence behind a naturalistic unlikelihood. Is this rarefied or ordinary design he asks.
Well, I would argue that it depends on our knowledge. Did these aliens inform us in any indirect or direct manner or is the only way to detect it through eliminating our ignorance? If it is the former then we may be able to propose ID hypotheses which can be tested side by side with chance hypotheses. But if we have only chance hypotheses to be eliminated, how do we know that there is a real ID hiding behind the curtain or merely a process which we may perhaps never be privy to?
What if the ID did all its magic before the Planck constant? Such front loading would make the ID indistinguishable for all practical purposes from chance. Would one argue that the filter's inability to infer design makes the filter unusable? Or are such false negatives unavoidable in any probabilistic eliminative approach?
May I also suggest that Paul refrains from suggestions that Elsberry and Wilkins are 'skeptics' who do not want to catch design and opposes them to 'curious and open minded scientists'. Such ad hominems are unwarranted.
From Wilkins/Elsberry's paper
quote:
However, it has one glaring problem - it blocks any inferences of design, and that is too much. There are well attested cases of design in the world: we humans do things by design all the time. So an explanatory filter had better not exclude design altogether. How can it be included here? When is a design inference legitimate
Does this sound like scientists who do not want to capture design?
The authors also address another issues about evolutionary algorithms being unable to generate CSI. In short, when examples of known ID is being tested Dembski seems to use the probability of chance to determine complexity but when looking at for instance evolutionary algorithms, Dembski uses the probability of the algorithm to determine complexity. Since the probability of the algorithm is likely to be close to 1 it is refered to as a probability amplifier.
But wait a minute, is an intelligent designer also not a probability amplifier? After all what is the probability of an ID hypothesis? I would argue close to one but that would mean that the complexity for ID is actually quite low.
So if one uses uniform chance to calculate complexity but then both evolutionary algorithms as well as intelligent designers can generate complexity. Alternatively both ID and evolutionary algorithms fail to generate any CSI.
So unless we allow for a distinction of rarefied versus ordinary design we have a real problem here in that the EF cannot separate evolutionary algorithms from ID.
But what if we know more about the intelligent design? What if we can build empirical knowledge? Would that be helpful in deciding ID or chance? I would argue that this is a requirement.
Of course this will require us to separate design into two classes, empirically detectable, inductive generalizable ordinary design versus a form which provides us with no generalizations aka rarefied design.
At least this approach will allow us to separate ID from for instance evolutionary algorithms.
[ 09. November 2002, 18:29: Message edited by: Frances ]
IP: Logged
|
|
Terence A-H Tan
Member
Member # 234
|
posted 10. November 2002 15:07
Likewise, I must apologize to Janitor for any misunderstandings I may have generated and any misquoted riddles.
I myself am new to this design debate and as such am in the reading and learning phase at present ... during my spare time again of course ![[Smile]](smile.gif) ... I chip in occasionally some of my thoughts where I think perhaps they may be useful on the discussion table ... or not really depends on who the thought processor is ... but I must admit my presuppositional bias stems from my faith in the intangible concepts of christianity … but then again whoever says he/she has no presuppositional bias cannot be held true ….
I'm going to chip in another 2 cents worth of mullings for now … this set of mullings are adapted and applied from Lester S.King’s, Medical Thinking, A Historical Preface where he writes in a different context about proximal causes. But the same concepts maybe useful in this particular ID context.
This would concern the unidirectional nature of signs. If we were to define an ideal sign in all its pure form, given all things are known, as unidirectional then it must point to one thing and one thing alone. Hence if a sign seems to point to more than one different location, that means we have either not used the most proximal location next to the sign yet or this proximal location is still unknown to us. If this proximal location has already been defined, then the proximal location either becomes a dead-end or a list of causes.
Simple eg. again taken from my profession:
We frequently request for a blood test ESR (erythrocyte sedimentation rate) for various reasons. Now causes of a raised ESR can range from infections to inflammation to autoimmunity etc. Similarly, there are lots of causes for a low ESR. One then wonders, how can ESR be a useful sign or even a sign at all if it points to so many different things. But if one were to scale the multi-directionality back to the most proximal location to which it points to (which is most often its very own definition) then we can appreciate that ESR is a sign that unidirectionally points to the rate of settling of erythrocytes in anti-coagulated blood. This proximal location becomes a dead end as we cannot further extrapolate from this. (let’s call this type of sign A)
We then have to invoke raised ESR or lowered ESR as new individual signs and the most proximal unidirectional location they point to according to the books are enhanced erythrocyte aggregation and decreased erythrocyte aggregation respectively. Further using enhanced erythrocyte aggregation as a sign, it unidirectionally points to increased in levels of serum proteins according to the books again. (let’s call this type of sign B). Increase in levels of serum proteins can then be broken down into a list of causes, but per se is not a sign.
What’s the difference between sign A and sign B ? In sign A, the most proximal location of ESR is its very own definition and hence no false positive or negatives can be attributed to it. Sign B however is different, as far as we know, raised ESR means increased erythrocyte aggregation causing a faster sedimentation rate. But who’s to say that some other mechanism, which we do not know yet, might not be causing the increase sedimentation rate as well. And who’s to say enhanced erythrocyte aggregation is always due to increase serum protein levels. Maybe the red blood cells have lost their repelling –ve charges by some non-protein mechanism and clump together. Sign B thus has not reduced itself to the most proximal location yet. Hence for a sign to be the perfect ideal sign without assumptions, it must eventually define itself because of our limitation in knowledge. This is not because of the sign itself, but because of our limitation in knowledge. But if we were to invoke that this sign is a statement of fact until proven otherwise, by nature of the definition of sign itself, it is purely unidirectional. Perhaps this is what Dembski is alluding to. That if the evolutionary camp has been using this same strategy for so long and still continues to use it, then by default on fair grounds, IDers can use the same strategy too.
But for all practical purposes, assumptions are a necessary part of life. We attribute more faith to highly probable and logical assumptions. Also because of the nature of signs and our limited knowledge, it is hard for a good theory to be founded on just one sign alone. The building of a good and accurate hypothesis to solve patient problems require the synergistic combinations of different signs. This I believe is true for any theory with incomplete knowledge.
IP: Logged
|
|
Noel Rude
Member
Member # 516
|
posted 10. November 2002 19:01
For what it's worth ... historical linguistics presents a bit of a Popperian enigma: Linguists have no way of establishing that any two or more languages are NOT related ... they can only establish that certain languages ARE related. One cannot propose a hypothesis of relationship that can be refuted, one can only propose a hypothesis of relationship that can be confirmed. It is always possible, for example, that English and Chinese are related at some remote point ... just as it is possible that they are not. But we can never confirm the latter, only the former.
[ 10. November 2002, 19:32: Message edited by: Noel Rude ]
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 10. November 2002 20:23
Linguistics is somewhat of a hobby of mine and I find the analogy of evolution within linguistics quite fascinating. Terms such as mutation, selection, common descent seem to have found their ways into certain areas of linguistics and there may be good reason for using this analogy.
Santa Fe institute
Family trees for languages
I am not sure if the statement that linguistics cannot disprove certain relationships. It might be unfeasable to disprove certain hypotheses due to lack of historical data but that does not mean that they are not in principle refutable. This may be an interesting topic for a separate thread.
As far as the assertion about no false positives, I would encourage Noel to give some examples. Grimm's law seems to have been a important advance in linguistics in that it introduced a methodolgy into historical linguistics.
How this relates to no false positives is somewhat unclear to me though.
[ 10. November 2002, 23:09: Message edited by: Frances ]
IP: Logged
|
|
Noel Rude
Member
Member # 516
|
posted 10. November 2002 21:20
Yes, for the record, I think it worth emphasizing that historical linguistics allows for no false positives. Because when change is introduced into a language system (sound, grammar) it tends to spread through the whole system, this allows us to make an inference of genetic relationship. For example, "Grimm's Law" (among other such "laws") establishes the relationship between Germanic and the rest of Indo-European: where Greek, Latin, Slavic, Indic ... have p, t, k, etc., Germanic has spirantized versions of these sounds, e.g. pod- vs. fish, tri- vs. three, kent- vs. hundred ... Given a sufficient number of such correspondences (sufficient complexity) to make the inference, there are no false positives.
IP: Logged
|
|
Paul A. Nelson
Member
Member # 26
|
posted 11. November 2002 10:58
Frances:
Under no circumstances does the Elsberry & Wilkins filter allow one to infer so-called rarefied design. As they put it:
quote:
...no rarefied design inference is needed...(p. 722)
This conclusion has nothing to do with (as you put it) “something which you [Paul] believe must exist.” It’s a logical consequence of the Wilkins & Elsberry analysis. If life was designed by an unobserved intelligence, we could never discover that using their approach.
You write, of Wilkins & Elsberry:
quote:
Rather than doing damage to science, they save science from embarassements of false positives or junping to conclusions not warranted by the evidence.
Again, this is nonsense. The “embarrassment” they wish to avoid is the problem of induction. But the problem of induction is unavoidable, because it is logically inescapable in any empirical investigation. As I said several posts earlier in this thread, science cannot get around the possibility of error. Design inferences are fallible. Yet Wilkins & Elsberry ask design theorists to do the logically impossible, i.e., to rule out all chance hypotheses –- both the known and possibly relevant (which Dembski's filter requires) as well as the unknown (which cannot be done). Since the logically impossible cannot be done, it’s no surprise that “we don’t know” always comes out ahead in the Wilkins & Elsberry analysis.
You write:
quote:
Wilkins and Elsberry do not exclude rarefied design.
Read their paper again. “Surely we can content ourselves with regularities, chance, and ‘don’t know’ explanations” (p. 721). What’s on the menu today? Only regularities, chance, and “we don’t know.” But isn’t the Wilkins & Elsberry cafeteria also offering rarefied design, since that is also an empirical possibility? Nope. Not today -– not ever.
Given that Wilkins and Elsberry haven’t jumped into this thread yet, I’m going to send a direct invitation to both to respond here. Incidentally, you quote them as allowing inferences to design:
quote:
There are well attested cases of design in the world: we humans do things by design all the time.
This is not what they call "rarefied" design, which is what's at issue here. Wilkins & Elsberry permit design inferences only where one has independent evidence of the designer. Their filter cannot detect rarefied design, because its logical structure (as they explicitly intend) always favors ignorance over any inference of rarefied design.
Frances wrote:
quote:
About Paul's final question, non human inteligence has hidden its existence behind a naturalistic unlikelihood. Is this rarefied or ordinary design he asks. Well, I would argue that it depends on our knowledge....how do we know that there is a real ID hiding behind the curtain or merely a process which we may perhaps never be privy to?
Do you see that this same worry applies to SETI? There might be an unknown natural process that produces narrow-band radio transmissions. Yet the SETI researchers don’t worry about that much, because there might be an unknown natural process producing any effect we ascribe to intelligent causes. This is the problem of induction that Wilkins & Elsberry want solved. They have a lust for certainty that cannot be satisfied, in principle.
I’ll ask Wilkins & Elsberry about how their approach handles SETI.
Frances wrote:
quote:
May I also suggest that Paul refrains from suggestions that Elsberry and Wilkins are 'skeptics' who do not want to catch design and opposes them to 'curious and open minded scientists'. Such ad hominems are unwarranted.
My apologies. I withdraw the comment, and rephrase it as follows:
The Wilkins & Elsberry design filter makes it logically impossible to detect what they call “rarefied design,” i.e., design by an unobserved intelligence. Anyone who wants to learn the truth about the world, and thinks that such design might be possible, should reject the Wilkins & Elsberry approach. Dembski’s explanatory filter, by contrast, allows one to consider the evidence for any natural cause.
P.S. I just sent an e-mail invitation to Wesley and John to join this thread.
[ 11. November 2002, 11:42: Message edited by: Paul A. Nelson ]
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 11. November 2002 12:17
Dear Paul,
I think that it is important to realize that what Wilkins and Elsberry have done is not eliminate rarefied design but shown that rarefied design and 'we don't' know' are indistinguishable. If life were designed by an unobserved intelligence, Dembski's filter would also not be able to distinguish between 'we don't know' and 'intelligent design' but since Dembski gives a privileged position to 'ID', ID would be infered. But this means that the ID filter will be open to many false positives in instances of 'we don't know' and 'no intelligent design'. Thus any ID inference is immediately suspect to be merely a 'ID of the gaps' argument.
That's what Wilkins and Elsberry have shown. It may be a disappointment to those who believe that there is unobserved intelligent design in nature but for all practical purposes it cannot be distinguished from 'we don't know' not matter which filter one uses.
Induction may be inescapable in some instance but in an eliminative filter with rarefied design it is an unwarranted leap of faith. After all how can one differentiate between 'we don't know' and rarefied design?
Design inferences may be fallible but they are in case of Dembski also privileged. That is the default assumption is design not we don't know, even if we have no way to independently test the ID hypothesis. This makes Dembski's ID inference hopelessly susceptible to false positives.
May I also suggest that when you quote me, you quote me correctly?
quote:
Wilkins and Elsberry do not exclude rarefied design.
which should be compared to my actual quote
quote:
And Wilkins and Elsberry do not exclude rarefied design, they merely accept that rarefied design and ignorance are two concepts which cannot be resolved using the EF approach.
I think that your final conclusion is correct, a design inference filter that gives a privileged status to ID based on the presumptions that ID (in its rarefied form) must exist, cannot be scientifically defended.
If one wants to stack the deck in one's favor then by all means use an inference approach which will give plenty of potentially false positives and no way to distinguish between false positives and intelligent design.
This 'ID of the gaps' has failed to work in the past, why should we believe it will work better now?
quote:
So instead of design being the penultimate default hypothesis in the decision tree, rarefied design becomes, at best, a tenuous conclusion to draw. There is an in-principle difference between rarefied and ordinary design inferences, based on the background knowledge available about ordinary, but not rarefied, design agencies. Rarefied design inferences tell us nothing that can be inductively generalized. Consequently, analogies between artifacts of ordinary design, which are the result of causal regularities of (known) designers, and the "artifacts" of rarefied design do not hold (as Philo noted in Hume's Dialogues Concerning Natural Religion, Book V).7 Indeed, we might even conclude that the specified small probability of rarefied design is itself an artifact of our prior expectations. As our background knowledge changes and grows (due to the "irrational" inferences of people like Actual Charles), so too do the specifications, and sp/SP can become HP or IP. Why is there a rarefied design option in the filter at all? Dembski has not dealt with such Humean objections. His a priori expectation is that events of specified small probability (relative to whichever specification) do not happen by themselves through chance or regularity, and hence require some other "explanation". But if this is merely a statement about our expectations, and we already require a "don't know" or "don't know yet" option in our filter, why are we ever forced to a rarefied design conclusion? Surely we can content ourselves with regularities, chance and "don't know" explanations. Such overreaching inferences as a rarefied design inference carry a heavy metaphysical burden, and the onus is on the proponents of such an a priori assumption to justify it. Otherwise, "don't know" is adequate to the empirical task where background information is equivocal. This leads us to a truncated filter of the form of the following flow diagram:
Source
Wesley also seems to have a Powerpoint presentation online.
May I also point out that Wilkins and Elsberry are not the only ones making such arguments. Sober uses a different approach to show that in absence of a (positive) hypothesis of ID, we cannot really conclude anything.
First of all Fitelson et al notice that:
quote:
Notice that the filter is eliminativist, with the Design hypothesis occupying a special position.
Source
and then comment on the following
quote:
Dembski’s parsimony ordering, even if correct, makes it puzzling why the Filter treats the likelihood of the Chance hypothesis as relevant, but ignores the likelihoods of Regularity and Design.
As Fitelson et al point out however regularity would not be able to explain CSI since acceptance of the hypothesis that E was due to regularity means that E was highly probable given the conditions. But the same would apply to intelligent design. Acceptance of ID would mean similarly that ID cannot explain CSI since the probability of E under the hypothesis of ID would be close to 1 and thus the complexity would be low.
Fitelson et al come to a similar conclusion
quote:
When prediction is probabilistic, a theory cannot be accepted or rejected just by seeing what it predicts (Royall 1997, ch. 3). The best you can do is compare theories with each other.
To test evolutionary theory against the hypothesis of intelligent design, you must know what both hypotheses predict about observables (Fitelson and Sober 1998, Sober 1999b). The searchlight therefore must be focused on the design hypothesis itself. What does it predict? If defenders of the design hypothesis want their theory to be scientific, they need to do the scientific work of formulating and testing the predictions that creationism makes (Kitcher 1984, Pennock 1999). Dembski’s Explanatory Filter encourages creationists to think that this responsibility can be evaded. However, the fact of the matter is that the responsibility must be faced.
Both Sober and others have expanded on these ideas by showing that modus tollens arguments have no probabilistic equivalent. This makes it inevitable that one has to propose competing hypotheses. But competing hypotheses require that one can formulate in an non ad-hoc manner generalizations. Rarefied design however does not provide us with means to do so.
[ 11. November 2002, 13:12: Message edited by: Frances ]
IP: Logged
|
|
|
Printer-friendly view of this topic