|
Author
|
Topic: Design Inference Game II
|
RBH
Member
Member # 380
|
posted 19. January 2004 20:42
general_re, quoted by FRDiamond, said quote:
And so in order to ensure that the filter itself is the thing being tested - and not simply one's knowledge of the sorts of things that humans make - requiring the application of the filter seems reasonable enough.
Since the Filter requires that one calculate a probability of assembly (that's what "complexity" means, remember), and in order to calculate that probability one must know what the "parts" of the object are, we need (a) an analysis of the object into parts, and (b) some numbers.
Cre8ionist said quote:
1. A reference class of possible events (Any letter of the alphabet could appear at any position);
2. A pattern that restricts the reference class of possible events (here the English language);
3. The precise event has occurred ( here the complex specified English text).
Therefore, it triggers a design inference.
But there's no probabilty calculated there, and hence no reference to complexity in Dembski's sense, namely improbability. Humans are notoriously bad at 'horse-back' estimation of probability, so Cre8ionist's putative design inference does not (yet) follow from his analysis.
Further, there's no justification offered for Cre8ionist's selection of the two reference classes he identifies (letters of the alphabet and the English language). It might be the case that the appropriate reference sets are an alphabet of just 8 symbols: {i l n o s t u w} and a language with just two permitted combinations of those symbols: {wilson solutions}.
Similarly, in #2, in order to calculate an improbability (= complexity) we need to know what the parts are, and I have not the faintest notion what they might be. I could make some guesses, but they would be just that: guesses. Hence I can't apply the Filter to it.
To put it plainly, in order to apply the Explanatory Filter in the way TDI requires, or in order to determine Specified Complexity in the way NFL requires, we must have numbers for the (im)probabilities, not horse-back guesses. If all we have are the latter, then we have only vaporware of the high-temperature variety. ![[Smile]](smile.gif)
RBH
IP: Logged
|
|
Pim van Meurs
Member
Member # 541
|
posted 19. January 2004 22:54
Creationist: Therefore, the object in picture two could be designed, but without some accurate way of measuring it's complexity etc... I don't see the EF as rendering a conclusive result.
So what would be needed to measure its complexity? What makes 2 different from 1?
As far as my question about your design inference, I believe that Dembski's design inference has elimination of regularity as the first step. Your posting in which you explain your rationale for infering design for 1 includes
1. A reference class of possible events
2. A pattern that restricts the reference class of possible events
3. The precise event has occured
Therefor it triggers a design inference.
What makes you change your mind when it comes to 2?
IP: Logged
|
|
Evan
Member
Member # 164
|
posted 19. January 2004 23:22
I concur with RBH, and I think his and other’s remarks take me back to some questions I asked of general_re, via Diamond, early in this thread.
Just how much “background information” are we assumed to have? General_re says
general_re, quoted by FRDiamond, said,
quote:
Perhaps I should clarify what I mean by historical knowledge and historical context - I simply mean one's background knowledge of that *particular* object. The game was designed as an exploration of the explanatory filter - if I post a picture of, say, a frying pan, one could certainly look at it and say "ah, ha - that's a frying pan, frying pans are known to be made by intelligent agents - namely, humans - and therefore this object is designed." All well and good, but it doesn't make use of Dembski's tools in arriving at that conclusion, as undoubtedly correct as it is. So, although it sounds as though it might prove to be difficult as a practical matter, I think it is necessary to set aside one's background knowledge of such objects, and treat each one as a new and novel thing, with which one has no first- or secondhand experience or knowledge of.
But ask of general_re (are you still with us, Diamond), how do we decide how much “first- or second hand knowledge” do we have? Do we know about written language? Do we know how many letters of the alphabet they usually have? Without knowing about existing designers (humans - how can we think about this without thinking about designers), how are we to assign any probabilities - which are, as RBH said, absolutely necessary in order to apply Dembski’s filter.
So I ask of general_re, or any other players here, the same question I asked earlier - let’s have some explication of how he (or Diamond) defines and structures the game so that the EF can be properly applied; i.e., with probabilities associated with the events that might have led to the existence of the object in question.
In particular, describe
1) What criteria separate useable background knowledge from unusable background knowledge?, and,
2) What background knowledge about known and unknown designers are allowed, and why or why not?
In respect to this issue of the designers, general_re says,
quote:
Dembski says that while one *may* take the conclusions of the filter and the qualities of the object, and *then* infer from that some or all of the qualities of the putative designer, one is not required to know anything about the designer *in advance* in order to successfully/productively apply the filter. Given that, that's the other reason that certain sorts of historical knowledge were ruled out of bounds - you're not supposed to *need* to a priori know anything about the designer to infer design.
It is difficult to see how it could be any other way, actually - even if we agree that certain biological structures are designed (by non-human agents), discussions of the qualities of the designer or designers are A) speculative, and; B) likely to be rather tendentious. If you deduce that an object was designed based on what you *think* you know about the designer, your deduction can only be as good as what you *actually* know about the designer.
Exactly - and that is why all design inferences are “speculative” and “tendentious” (a very good and apt word, by the way). All design inferences are based on assumptions about the putative designers, taken, in my opinion, from a combination of our experience with human beings and our idealized personification of natural processes. If you genuinely exclude all speculations about the nature of the designers, then you can infer design about anything.
There is a vicious circle that I discussed in my original response to Diamond. Design is not a property of the qualities of something, it is a property of its history. Of course, the central ID argument is that certain qualities cannot come into existence without the involvement of an intelligent agent, but this is the central point under contention - this is by no means a proven proposition.
Furthermore, the EF is (as PvM has regularly mentioned) is an eliminative filter: if we can show that an the probability of natural processes being the cause of the object is extraordinarily low, then we can conclude designed. But to do this, we either have to:
1) eliminate all possible natural causes, or
2) know the actual history of the thing, be able to assign probabilities to the events of that history, and calculate the resultant probability of all the events of that history.
But no one does this. Instead, people make estimates based on their background knowledge of what known designers do, and jump to speculative conclusions. Sure, they throw in a little probability in (there are 26 letters in the alphabet) and make some conclusions about specificity (which is again nothing more than comparisons with objects made by a known type of designer), but this is not really an application of the EF.
So, again hoping that Diamond comes back, I offer these critiques in response to this game.
[ 19. January 2004, 23:30: Message edited by: Evan ]
IP: Logged
|
|
FRDiamond
Member
Member # 1072
|
posted 20. January 2004 11:59
Evan, I'm still here, and I appreciate your valuable critique. You have raised some very valid points. Let me address one of them here.
I think that in science, no knowledge of any kind need be off limits. What branch of science limits knowledge of things?
However, for the purposes of this game, at least in my understanding, it simply would not suffice as a valid application of the EF to just offer a direct historical account of how any particular object came into being.
What this limitation means to me in practice is that for example in picture #1 we do know about written languages such as the English language, font types, codes, the chemical properties of the markings on the surface of the "moisture absorbing" material, the chemical properties of the "moisture absorbing" material of the sphere itself, the properties of objects inflated with air, and so on.
So then we could postulate if we wanted to, that at least one possible independent matching pattern in Picture #1 is that some of the markings and lines bear a striking resemblance to the conventions of a preexisting code or convention used by intelligent agents for communication, known as the English language. But the problem, as RBH has elucidated, and to which you have alluded, is in the numbers. For example, what are the probabilites of those markings on the surface of the sphere being simply the result of the natural operation of the laws physics and chemistry alone, and how are such probabilites actually calculated?
Since I do not know the first thing about probability statistics, that's where expert advice on the subject would definitly come in handy.
Cordially,
IP: Logged
|
|
David Garrett
Member
Member # 759
|
posted 20. January 2004 12:29
It seems like this game is a good way to stimulate discussion, but I can't help but be critical of a couple of key points about what is missing - function and material composition.
First, function is such a critical aspect of the ID argument that it is really difficult to judge by shape alone. The first item is a very basic shape yet it does possess writing. Even if the writing were in Korean (and none of us spoke Korean) we would still recognize it as a form of language by recognizing the craftsmanship of the writing. But, by its shape alone, we would be unable to know unless we knew the function of the item. If we knew that its function was to be bounced then that would generate a good discussion just on whether or not nature would produce a spherical object that "needs" to bounce. The second item bears no markings or writings except for what appears to be striations or lines between spokes and the spokes themselves. Without knowing the function of this item it is very difficult to say more.
Secondly, without knowing the material composition of the items it is hard to know. If we knew the basketball was made of rubber then we would infer ID even without the writing. The second item looks to be crystalline but I can't say what exactly it's made of.
Just some thoughts,
Dave
IP: Logged
|
|
|
|
David Garrett
Member
Member # 759
|
posted 20. January 2004 15:26
quote:
Dave, it seems to me that the notion of function implies a definite end or purpose that the object in question serves or a particular kind of work it is intended to perform, which in turn implies design. Notions of function (or even dysfunction, for that matter) are notions that very difficult (imho) to justify, or even comprehend as emmenating from non-teleological, random sources.
Maybe the word "funtion" is a bit leading. What I meant to suggest might be better termed "natural state" or "natural behavior". How the object interacts with the world. Having a little bit of a medical background, I tend to think in terms of Anatomy and Physiology. We must not only consider the Anatomy of the object but also its Physiology. And while we might not be dealing with living organisms in this game, it is still worth considering how an inanimate object appears or behaves in nature.
Dave
[ 20. January 2004, 15:27: Message edited by: David Garrett ]
IP: Logged
|
|
Evan
Member
Member # 164
|
posted 20. January 2004 16:31
But the concept of IC (which is the property of an object that implies that it might be designed) is defined in terms of function - therefore function is a relevant concept.
Furthermore, we have what we might call proximate vs. ultimate functions. "Ultimate" functions in the sense of purpose would necessitate knowing something about the motivations of the designer, but "proximate" functions just have to do with how one part of something interacts with something else. Function is this second sense is central to the idea of IC, and therefore is relevant to the design inference.
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 20. January 2004 18:06
But the original suggestion was to use the design inference (Dembski's work) for this game, which does not refer to function.
IP: Logged
|
|
Evan
Member
Member # 164
|
posted 20. January 2004 18:57
I see your point. However, my understanding of Dembski's reasoning goes like this:
1) to establish design, we have to eliminate chance and law as causes
2) to do this directly, we have to establish that the history of an object's coming into its current form is very highly improbable, and do to that we have to know its history.
3) However, if we don't know its history, we can use a certain property of the system, that of irreducible complexity, to infer that no matter what its history was, natural causes could not have brought it into its current form with the aid of intelligence.
4) But ICness is defined in terms of function - not function in terms of what the object taken in its entirety does, but in terms of the functions of its parts - but those of course are related to the function of the object as a whole.
Therefore, in the absence of information about history, information about how the object functions is relevant.
This may be faulty reasoning, but it is a summary of the point I was trying to make.
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 20. January 2004 22:52
David Garrett wrote quote:
Even if the writing were in Korean (and none of us spoke Korean) we would still recognize it as a form of language by recognizing the craftsmanship of the writing.
Well, now, that's not always the case. Consider the Voynich manuscript. It's been around since the 17th century, and people still don't know if it in fact bears a language. While it's catalogued in the Museum of Hoaxes, there are still those who believe it will be someday be deciphered. But at this stage, we don't even know if it's a form of language in spite of the craftsmanship of the marks.
I'm not at all convinced that the English language (and its alphabet) can be justifiably used as a reference set for #1. In the emergency services we try to use a very stereotyped and limited vocabulary in order to reduce the probability of miscommunication under stressful and noisy conditions. The probability of a given utterance in that context is considerably different from the probability of occurrence of that same utterance in a context without the limitations we attempt to impose. Hence context of the utterance is critical to calculating the probability of occurrence of a given sequence of words. In the case of #1, the word "Wilson" is almost certainly a fixed sign, not a linguistic symbol. If we looked at lots and lots of such objects, the number of different signs occupying that 'space' on the sphere would be very limited, and hence seeing any one of them is a relatively high probability event. We don't have to know that "Wilson" is a manufacturer of basketballs to gather the data necessary to make that probability calculation. All the words in the English language is not an appropriate reference set.
It follows that inferring "design" from a single object is not possible. If there's just one object, there's no probability to be calculated without subjectively choosing a reference set for the denominator of the calculation. Principles that govern the choice of that reference set are not given by the Filter or its analogs and descendants. Let me choose the specification (as Dembski does by not providing a principled method) and I'll deliver whatever is desired by way of design inferences. (A way to test my assertion is the following: Suppose you and I disagree on the appropriate reference set for calculating a probability. Is there a principled way of settling our disagreement?)
FRDiamond wrote quote:
Dave, it seems to me that the notion of function implies a definite end or purpose that the object in question serves or a particular kind of work it is intended to perform, which in turn implies design. Notions of function (or even dysfunction, for that matter) are notions that very difficult (imho) to justify, or even comprehend as emmenating from non-teleological, random sources.
"Function" doesn't imply "design" with anything approaching necessity. To attribute a "function" to something is a statement about a human's analysis of a system. The "function" of the bottom of a mountain is to hold up the top: that is neither intentionality nor design. The function of a riverbed is to channel the flow of water that carves the riverbed; again, function doesn't entail intention nor design. One of the things the ozone layer does is filter UV; is that even a function? Function language does not automatically translate to (intelligent) design language; "function" does not translate as "intention." Note that "intended" sort of quitely eased into FRDiamond's "... particular kind of work it is intended to perform, ...". That conflation is the source of a whole lot of confusion. Biologists use function language a whole lot because it comports with the analysis of systems into interacting parts, whose parts severally contribute to the operation of the system as a whole. But that function language doesn't entail "intention" language. Again, conflating the two contributes only confusion. In order to make the leap from "function" to "intended function," one must have independent evidence of the intention and/or the intender.
Finally, Evan wrote quote:
Therefore, in the absence of information about history, information about how the object functions is relevant.
The inference of design in the absence of information about the designer - its intentions, methods, and capabilities - is an inference of "No History." Hence all that's left to look at is function.
RBH
[ 20. January 2004, 23:03: Message edited by: RBH ]
IP: Logged
|
|
Cre8ionist
Member
Member # 140
|
posted 20. January 2004 23:17
I understand your point RBH about the calculation not being done. Since we have the necessary formula for doing the calculation wrt the probability of any given English phrase provided by Dembski ala Dawkins (NFL pg 181), it's really a mere formality. And I probably should have put "calculation provided upon request."
Disregarding case and considering only Roman letters and spaces , take the length of the specified sequence, in this case, under close inspection, I can make out the words Wilson, + Solution, + Moisture + Absorbing + Technology + 4 spaces, again, using the Dembski/Dawkins formula, the length of the identifiable sequence is 45. The number of possibilities at each location is 27. Therefore, you have I believe, and I'm no mathematician, 1 chance in 27 raised to the 45th power of arriving at said text.
As to whether we're dealing with a smaller alphabet or not doesn't really seem to make sense here. Let's suppose for instance our picture contained the same exact text, but the text wasn't on a basketball, let's say the picture was of a Scrabble board and the words were written with the tiles. Could you then admit a design inference based upon probability? Or suppose you found the words scrawled in the sand on some deserted beach, could you rightly conclude that someone with a knowledge of English had been there before you? If so, picture selection has a lot to do with the direction of this thread, it's possibly muddling the point. I don't think it should. If you don't make a design inference in those cases, why not? In any case, surely we all knew picture 1 was a designed object, I wonder what methods Pim and the rest of you used and if you could try to formalize your methods? I maintain that Dembski's method is right, in the presence of SC design can be inferred......Cre8
[ 20. January 2004, 23:21: Message edited by: Cre8ionist ]
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 21. January 2004 01:03
Why choose letters instead of "pen-strokes" or instead of whole words or instead of ink molecules?
With a scrabble board, we understand the probability of each letter appearing, because whe know where the letters came from--the specific set containing 9 a, 2 b, 2 c, 4 d, 12 e, 2 f, etc.. (Although we may have to account for the probability that letter tiles weren't actually present in those numbers. For example, if we see the word "fuzzy", neither "design" nor "chance" are good explanations given the availability of letters, as there is only one z!)
But with a basketball, if we don't understand the origin of the thing, we can't say what process produces the letters, and if the words somehow recombined from other sporting goods, it wouldn't be very improbable to find those words on a basketball.
If we do understand the origin of the thing and use that in our analysis, we don't need a filter--we already know the answer!
[ 21. January 2004, 01:06: Message edited by: Rex Kerr ]
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 21. January 2004 01:11
The design inference for the basketball actually fails on Cre8's probability argument: 27^45 is not less than the Universal Probability Bound of 1 in 10^150, and thus the Explanatory Filter on Dembski's rules cannot infer design. The Filter yields the inference that the basketball occurred by chance; for the Filter it's a false negative given the calculation.
Cre8ionist wrote quote:
As to whether we're dealing with a smaller alphabet or not doesn't really seem to make sense here. Let's suppose for instance our picture contained the same exact text, but the text wasn't on a basketball, let's say the picture was of a Scrabble board and the words were written with the tiles. Could you then admit a design inference based upon probability? Or suppose you found the words scrawled in the sand on some deserted beach, could you rightly conclude that someone with a knowledge of English had been there before you? If so, picture selection has a lot to do with the direction of this thread, it's possibly muddling the point. I don't think it should. If you don't make a design inference in those cases, why not? In any case, surely we all knew picture 1 was a designed object, I wonder what methods Pim and the rest of you used and if you could try to formalize your methods? I maintain that Dembski's method is right, in the presence of SC design can be inferred.
I don't "admit" a design inference based on probability in any of those cases. I could infer that someone who knows English had been there before me and wrote the letters, but not because of some probabilistic argument but on the basis of knowing what English is and who typically produce English sentences, humans. In other words, unlike the cases in which Dembski really wants to apply the Filter, I know the typical causal history of English sentences, and don't have to mess around with numbers in a straiend analogy.
With respect to the basketball, again to classify it as designed I need make no probabilistic arguments. I know (having looked them up on the Web) the design specifications and method of manufacture of that sort of object, and classify it as designed because I know them: humans make them according to known specifications and for known purposes (not merely "functions").
So #1 is due to chance according to the Filter, and due to design if one knows the design specs and method of manufacture. How about #2? Who's got some numbers there?
RBH
[ 21. January 2004, 01:13: Message edited by: RBH ]
IP: Logged
|
|
Pim van Meurs
Member
Member # 541
|
posted 21. January 2004 01:25
cre8: In any case, surely we all knew picture 1 was a designed object, I wonder what methods Pim and the rest of you used and if you could try to formalize your methods? I maintain that Dembski's method is right, in the presence of SC design can be inferred......
We are using Dembski's method which seems to bear some similarity to your approach but seems to include regularity and chance as possible explanations and in addition seems to use a 10^-150 bound for any design inference.
Maybe Dembski could provide us with some guidance here?
[ 21. January 2004, 01:26: Message edited by: Pim van Meurs ]
IP: Logged
|
|
|
Printer-friendly view of this topic