|
Author
|
Topic: The Other Flagellum II
|
Nel
Member
Member # 614
|
posted 23. September 2003 20:50
Ged writes:
quote:
But if you read that thread, you will see that I fully accept that one can “infer design”. I accept fully that one can infer design of the Easter Island figures—and also of “Happy Birthday Ged!”. (If you had read the thread, you would realize that. Why not actually read it?)
I've read it Ged. But you seem to be making the claim that the EF is unreliable. The only thing, in my mind, that would render the EF completely unreliable is a sufficient number of cases of false positives and a good indication that the causal history of these false positives was completely non-telic. Thats where I stand. I'm being brutally honest with you. Sometimes I think your threads don't apply to me so I get bored. I'm no master of the EF, and have restricted my design inferences to pure Biology, but I'm looking to change that. But thats the way I think.
Ged writes:
quote:
I too am interested in how one could actually differentiate a human or even other-worldly construction. My point in that thread is that the strict conditions of the EF were not being met—not that the cases could not be differentiated. So the EF fails to be reliable in various ways. One of the failures is ID enthusiasts claiming that the conditions of the EF have been met, when in fact they have not.
In my opinion the two conditions for the EF is specificity and complexity. So far, your example of clouds and several geological processes that kinda form a human-like figure may be specified, but not complex. In that sense, if one condition is not met, the EF catches it as non-intelligently designed.
Ged writes:
quote:
The same problem occurs in this thread. The “frequency” measurement has not been justified as being statistically significant with regard to anything that anyone is trying to infer. Working out the actual probability theory will show unexpected results if you don’t understand it.
I showed how one can do a co-option frequency both mathematically and with respect to the actual Biology. So I will have to disagree with you on this one.
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 23. September 2003 21:07
At risk that entire sequence will be deleted:
Right in part, you understand in part—but then jump to the wrong conclusion about what I have said.
Once again there are two different cases. On type of case of claiming of inferring design when the EF conditions were actually not met. (That is probably the case here with regard to any relevance of cooption frequencies.)
The other case was quite different, showing false positive. (I’ll not discuss further here.)
Easter Island figures were figures in which we are all agreed we can “infer design”. But the conditions of the EF were not met! It is the EF that is not providing a “false positive” it is providing a “false negative”. I’m happy you are comfortable with that. Don’t claim that the EF provided a positive result without justifying it, possibly with numbers. (Do the same here with numbers here with regard to what you are inferring.)
[Edit for clarity:
Alonso correctly identified that matching of the human figure was of high probability, and thus did not meet the conditions of the EF—precisely because they did not meet the condition of “complexity”, wherein complexity is equated to low probability. (Why a different word, rather than simply say it was not of low probability?) In any case now we have established that the “specification” of the human form can be met with high probability. Thus that specification cannot be used as a claim of inferring design with with the EF, whether the EF is reliable or not. To do so is to make a false claim of a positive result. Now of course we can recognize that the Easter Island figures were done by humans, but the way we do that is not to simply recognize that they are of a specification of a human form—having just agreed that specification has high probability of being met. Rather we need a more detailed specification, and when we add more details we are no longer “independent of the event”. However if we make two comparative hypotheses, we can reject one over the other. And two comparative hypotheses have no requirement that the specification used be “independent of the event”. We can use all the specification that we might derive from a detailed analysis of our hypothesis of human construction, for example, even including issues of “motive”.
End edit]
The problem here is not actually delineating what is being “inferred”. Alonso talks about ability to do cooption frequency. While I have doubts that one can do that, once again if you read back in this thread you will see that I agreed one could do that in principle. My point is then to follow up mathematically with an inference. Just presenting a frequency does not infer anything—even if it is very low. Something has to relate to that probabilistically, and has to be shown to be statistically significant. Read my loaded die example above. Frequencies were measured, but were irrelevant.
[ 23. September 2003, 22:30: Message edited by: gedanken ]
IP: Logged
|
|
Nel
Member
Member # 614
|
posted 23. September 2003 21:17
Ged writes:
quote:
Easter Island figures were figures in which we are all agreed we can “infer design”. But the conditions of the EF were not met! It is the EF that is not providing a “false positive” it is providing a “false negative”. I’m happy you are comfortable with that. Don’t claim that the EF provided a positive result without justifying it, possibly with numbers. (Do the same here with numbers here with regard to what you are inferring.
I think the EF would provide a positive result with the easter island statues. This is because it is both specified and complex. Remember that Dembski's math simply reflects how we infer design from everyday life. If you want the actual numbers, you will have to wait until I'm sufficiently versed in doing the math. However, I would probably never do it for the statues, probably motors, or molecular motors (synthetic ones and biological ones).
Ged writes:
quote:
The problem here is not actually delineating what is being “inferred”. Alonso talks about ability to do cooption frequency. While I have doubts that one can do that, once again if you read back in this thread you will see that I agreed one could do that in principle. My point is then to follow up mathematically with an inference. Just presenting a frequency does not infer anything—even if it is very low. Something has to relate to that probabilistically, and has to be shown to be statistically significant.
Are you saying that one has to show that the co-option event itself is more probable then a non-co-option event? In otherwords, in order to calculate the frequency we must first establish co-option probabilistically? If so, actually delineating the co-option event with a probability calculation is probably not needed, as co-option events can occur whether of high probability or low probability.
Ged writes:
quote:
In any case now we have established that the “specification” of the human form can be met with high probability. Thus that specification cannot be used as a claim of inferring design with with the EF, whether the EF is reliable or not. To do so is to make a false claim of a positive result. Now of course we can recognize that the Easter Island figures were done by humans, but the way we do that is not to simply recognize that they are of a specification of a human form—having just agreed that specification has high probability of being met. Rather we need a more detailed specification, and when we add more details we are no longer “independent of the event”.
First, it is no suprise that specifications can arise with high probability. Since, high probability events and low probability events are distinct from specifications. Consider for example randomly banging on your keyboard and accidently writing the word "the". Nonetheless, it is the marriage of independant patterns and low probability events which causes us to infer design. There is no need to add more information the independant pattern, such as with the easter island statues, because the independant pattern itself is embedded within a complex state such that inference to design is trivial (which is why people don't think that the human faces on clouds they think they see are designed). Complexity breaks the ambiguity. Thus, it is completely possible to calculate the probability of an event arising by chance, and independantly identifying the event.
[ 24. September 2003, 00:10: Message edited by: Nelson-Alonso ]
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 24. September 2003 10:27
Alonso, I have quoted your last point in my thread Does intelligence imply “motive”? and answer it there.
The problem in your last point is that your method allows one to toss a coin 500 times and call that particular result “specified”. It will also be highly complex pattern (e.g. P<10^-150). Thus that would be inferred as “design”. One has to create a “specification” and then check the probability of that particular specification, not simply the probability of what you see in the event.
With respect to this thread, I have asked repeatedly if any kind of specification can be given for any kind of “inference”. If so, what is that specification?
IP: Logged
|
|
Nel
Member
Member # 614
|
posted 24. September 2003 20:53
Ged writes:
quote:
The problem in your last point is that your method allows one to toss a coin 500 times and call that particular result “specified”.
Thats called a fabrication, not a specification.
Ged writes:
quote:
With respect to this thread, I have asked repeatedly if any kind of specification can be given for any kind of “inference”. If so, what is that specification?
I've answered it a couple of times actually. The pattern I'm looking for is if the gene itself performs differently in the descendant then it did in the ancestor.
[ 24. September 2003, 20:55: Message edited by: Nelson-Alonso ]
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 25. September 2003 00:07
Right (on first point)!
The difference between a "specification" and a "fabrication" is whether it is independent of the event. If the written pattern is dependent on the event, it is not a "specification" in strict Dembski language, it is a "fabrication". And that was my point. (And Dr. Dembski himself sometimes lapses into colloquial use of ‘specification’ and then invokes independence requirement—but technically correct would be to use “pattern” which is classified as “specification” upon demonstration of independence of the event.)
quote:
The pattern I'm looking for is if the gene itself performs differently in the descendant then it did in the ancestor.
Taken as a "specification" I agree this is a fair "specification". And if so, truly taken as a general search criterion (not further limited to some specific gene for example) then the rejection region is very large and highly probable. That is because in general there are a great many cases of genes that perform differently in an ancestor. Even observed in the lab for bacteria, if I am not mistaken. If so, that makes the probability of the rejection region 1.0.
Alonso, if you actually meant to restrict the rejection region's specification to a specific gene or some other limitation, please write that into your specification.
So I ask again:
What is the specification?
Alonso, the reason I am focusing on this is that there is a logical fallacy that seems common among ID enthusiasts. That fallacy is to use one meaning of a term in one place in an argument, then use another meaning of that same term in a different place in the argument. (Like A implies B, B’ implies C, “Therefore” A implies C.) I just wanted to be sure this is not happening in this case.
Because to be a useful “specification” for drawing an explanatory filter inference (according to ID methodology), the pattern must be demonstrably “independent of the event”. And to be “complex” it must not be common, rather it must have low probability. I see the possibility that you could invoke one understanding of the pattern to indicate independence, and another to invoke complexity (e.g. low probability). Now you wound not want to do a logical fallacy like that—so I am asking for a single clear and unambiguous declaration of the pattern so we can determine if it is both a “specification” and “complex.” To be unambiguous, it must also be complete, by that I mean that you must not have understood limitations that are only understood when you are examining complexity, but not understood when examining the appearance of being “independent of the event”.
[ 25. September 2003, 10:19: Message edited by: gedanken ]
IP: Logged
|
|
Nel
Member
Member # 614
|
posted 25. September 2003 15:06
Ged writes:
quote:
Taken as a "specification" I agree this is a fair "specification". And if so, truly taken as a general search criterion (not further limited to some specific gene for example) then the rejection region is very large and highly probable. That is because in general there are a great many cases of genes that perform differently in an ancestor. Even observed in the lab for bacteria, if I am not mistaken. If so, that makes the probability of the rejection region 1.0.
There is no such thing as a probability of 1. when we are talking about historical events where we must infer such things. There is no such thing as "observed" cases of co-option unless someone is sitting there watching it happen. I don't know why you simply reword your arguments (although it does make replying to you trivial).
Furthermore, it is a non-sequitor to say that because we see examples of co-option that the "probability of the rejection region" is 1. It could be a very complex (improbable) event and still occur in which case we would make a design inference.
Ged writes:
quote:
Alonso, if you actually meant to restrict the rejection region's specification to a specific gene or some other limitation, please write that into your specification.
Well I gave an example using the Brachyury gene. But I'm not writing "specifications" or using Dembski's math to calculate co-option frequencies. Dembski's math is used to infer design in the first place, and we could use that to detect it in such specific events. I've developed a "prototype" formula for which to calculate a co-option frequency wrt convergence (just brainstorming this)
quote:
gcf = N/Y
Where N is the number of times a gene was co-opted in a lineage that is Y years old.
Ged writes:
quote:
Alonso, the reason I am focusing on this is that there is a logical fallacy that seems common among ID enthusiasts. That fallacy is to use one meaning of a term in one place in an argument, then use another meaning of that same term in a different place in the argument. (Like A implies B, B’ implies C, “Therefore” A implies C.) I just wanted to be sure this is not happening in this case.
Give me an example.
Ged writes:
quote:
Because to be a useful “specification” for drawing an explanatory filter inference (according to ID methodology), the pattern must be demonstrably “independent of the event”. And to be “complex” it must not be common, rather it must have low probability.
This is a non-sequitor. Just because something is common doesn't mean it is of high probability.
Ged writes:
quote:
I see the possibility that you could invoke one understanding of the pattern to indicate independence, and another to invoke complexity (e.g. low probability). Now you wound not want to do a logical fallacy like that—so I am asking for a single clear and unambiguous declaration of the pattern so we can determine if it is both a “specification” and “complex.”
This is unnecessary to determine a co-option frequency. For example, I can calculate the gcf for lens crystallins and fish antifreeze proteins. Such a thing would be trivial since not only does it fit the specification for "co-option" but it is a simple example. I wouldn't say it is "complex".
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 25. September 2003 15:13
Thanks Alonso,
You have confirmed my suspicion that you do not have an "Explanatory Filter" inference described. (Because if you do, you would be willing to put the "specification" as used in the EF.)
That was all I was interested in.
You could explain the mathematics behind the inference in what you are doing, if you feel it has any importance to the reader. Without that it has the same problem as Mike's position: There is a vague claim being made, but no math to back it up. Once again probability equations can be surprising, and the math needs to be written out in detail.
IP: Logged
|
|
Nel
Member
Member # 614
|
posted 25. September 2003 15:19
Ged writes:
quote:
You could explain the mathematics behind the inference in what you are doing,
I already did.
quote:
gcf = N/Y
Where N is the number of times a gene was co-opted in a lineage that is Y years old. Hope this helps.
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 25. September 2003 15:43
Take a look at Dembski's books.
A "design inference" of any sort would be in terms of P(H|conditions) for some hypothesis H to be verified and conditions discussed. Or at least in the eliminative "inference" in terms of some P(R|H) wherein R is shown of low probability under some constraint. (In which I dispute the reliability.)
I don't see probability even discussed. No math of an "inference" has been drawn here.
BUT hey, if ID enthusiasts want to insist that
gcf = N/Y
is the mathematics of an inference, then I am happy for you. ![[Smile]](smile.gif)
[ 25. September 2003, 16:11: Message edited by: gedanken ]
IP: Logged
|
|
Nel
Member
Member # 614
|
posted 25. September 2003 17:20
Ged,
What does your post have to do with what I wrote? Why would I look at Dembski's books when I am calculating a co-option frequency? The co-option frequency that I discussed would show which genes are or are not constrained to be co-optable. I can use Dembksi's math to do a design inference of the specific events, but that has nothing to do with anything anyone wrote in this thread. And for certain, one can make a design inference without Dembski's math. Why would you mention "Dembski's books"? What is the relevance of your posts?
[ 25. September 2003, 17:32: Message edited by: Nelson-Alonso ]
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 25. September 2003 21:09
quote:
quote:
You could explain the mathematics behind the inference in what you are doing,
I already did.
I assumed since you said "I already did" to a question about an "inference" that you intended to produce an "inference".
Since you have not produce an "inference" with any mathematics, I pointed that out.
Let me get a little more specific. An inference means making a decision, not just giving some number or measurement. A measurement is not an inference. To do an inference mathematically, one must give a probability, one way or another. Now one can do logical inferences that avoid math, but I asked about a "mathematical" inference, and you responded on that subject. Even a "logical" inference is one of probability, just not given numerically--distinguishing highly likely (near 1.0) and not likely (near 0.0). Again, a mathematical "inference" is in terms of probability.
And if you are not using the "explanatory filter" as your means of "inference", then you need to show what the basis of the inference is. I don't think an inference, logical, or probability, has been supported whatsoever. In fact the details of the inference logic have not even been discussed.
quote:
And for certain, one can make a design inference without Dembski's math.
What "inference" of ID does not specify "low probability"? (Once again I see no probability.)
If any "inference" logic was discussed, one could consider probability mathematics to determine if it was "logical".
Alonso, you could at least tell us what is being "inferred". Maybe we could work backward from that. (And if you tell me "design" then I need speak no more unless you provide a "design" inference method used by ID enthusiasts that does not include probability, or supply details of your own.)
[ 25. September 2003, 21:35: Message edited by: gedanken ]
IP: Logged
|
|
Nel
Member
Member # 614
|
posted 26. September 2003 10:48
Ged writes:
quote:
I assumed since you said "I already did" to a question about an "inference" that you intended to produce an "inference".
Since you have not produce an "inference" with any mathematics, I pointed that out.
I showed an inference that I obtained with biological data, and I showed how one can calculate a co-option frequency with a mathematical formula. Unless you have a reply to my brachyury gene example, I'm not quite sure what the relevance of your posts in this thread is.
Ged writes:
quote:
Let me get a little more specific. An inference means making a decision, not just giving some number or measurement.
No an inference is usually defined as:
quote:
The act of reasoning from factual knowledge or evidence.
With the brachyury gene example, and by measuring co-option frequencies, I am "reasoning from factual knowledge or evidence".
IP: Logged
|
|
|
Printer-friendly view of this topic