posted 07. September 2003 10:28                

code:

---

  Initial  | Actual | Reliable 
  Infer ID | Cause  | “no false positive”?
  ---------|--------|---------------------
1 Not ID   | Not ID |  Yes
2 Not ID   | Yes ID |  Yes   *
3 Yes ID   | Not ID |  NO    **
4 Yes ID   | Yes ID |  Yes   +
 

---

* Type ‘2’ was a “false negative”.

** We can see that only type ‘3’ is of a failure to produce “no false positives”.

+ Type 4 is interesting in being differentiated from type 3 when we find we are in almost universal agreement on actual cause. What allows us to find universal agreement? (This must be deferred to “Part 2”.)

What is of greatest interest is cases that demonstrate, or get close to type 3, an actual false positive identification of design. So no such case can actually be one in which the “actual” (universally agreed) cause is classifiable as “intelligent design.” So of course the Easter Island figures, for example, are either type 2 or 4, and thus are not directly demonstrative of our issue – but do they lead us to cases that will demonstrate the issue?

The most interesting distinctions are between types 1 and 3, and between types 3 and 4. By using examples from type 1 or type 4, we can see if we are close to a type 3 example. If we find that type 3 examples can be constructed, then we show the reliability problem for the explanatory filter, which may lead us to methods to improve the filter’s pattern recognition procedure steps.

Here are some pictures of stone “figures”, first two are almost universally considered to be natural process occurrence, the third are the Easter Island figures in question:



Two classes of issue, first is distinction between types 1 and 3: (The second type, 3 and 4 distinction, must be delayed to “Part 2” due to length):

Are the first two figures cases of type 1 or 3? (Do these figures falsely appear to be “designed” because they match the conditions for the explanatory filter?)

To answer these questions I have asked for where we would find the “specification” of the Easter Island figures. I want to see if the “specification” would not match the first two figures as well.

Now the detail I want to concentrate on is exactly what constitutes a “specification” that is given independently of the event.

Now of course we could give a specification for the Easter Island figures: “massive stone figures with heads and body representation”. But we have a problem here – “representation” is a statement of motive, an intention to “represent”. So let’s change to “massive stone figures that have an appearance that we recognize a comparison to portions of the human head, and to portions of the human body, assembled in proper relation so as to be recognized as a whole as such.” Now where do we find that specification given independently of the event of the Easter Island figures? Does that exist, or was that specification constructed specifically for what is seen? We have a minor, but probably not insurmountable problem.

Here is a partial specification: “A portion of a stone figure in three dimensions that we would recognize as the eye area of the face.” Could we agree that is found somewhere independently of the stone figures in question? We might find some language description somewhere of this, and this is in my opinion universally agreeable that this partial “specification” is given somewhere independent of the event. We could continue with other portions of specifications, for the various aspects of the head shape.

The problem we have here is that those specifications are given as separate aspects, disjoint rather than as an assembled single “specification”. Does that still qualify as “specification” for the event, if the specification consists of a conjunction of partial specifications given separately that must be combined? Let’s accept that for the moment and see where it leads us.

So we take as the “specification” of the Easter Island figures, given independently of the event, as the conjunction of descriptions of the parts of the figure as recognizable as parts of the human form, existing in stone. And we choose a particular assembled list of those independently given partial specifications as the entire specification.

Now let’s apply to the Easter Island figures. Surely the probability that each aspect would be found in stone independently is of fairly low probability, and the probability by any natural process. Let’s not argue about the difficulty of such calculation for the moment, as that can only further degrades the reliability of the EF. I would agree that a multiplicative assembly of those partial specifications would describe partial events that make the total assembly of the Easter Island figures of lower than 10^-150 probability. So we have met the conditions of the EF for those figures by using an assembly of independently given sub-events or sub-patterns recognized.

Now apply the same procedure to the first two stone figures pictured above. In the first figure we must include aspects like shape of the rim of a hat, relation of such a figure to a nose figure, etc. And in the second figure we see the “cubist” representation of the eye and nose – just as examples.

What we see is that by combining the separately given sub-specifications for the first two figures, we can most likely come up with a sufficiently long list of those sub-specifications that the probability of those occurring in stone at one location, multiplied together, are also lower than 10^-150 probability.

The key here is that we are relying on finding sub-specifications for portions of the figures as occurring disparately, rather than finding such a specification given in whole but independently of the event. We most certainly cannot look at the figure and describe it and thus count that as given independently of the event!

So – do we have a problem? Do specifications for the first two figures exist, wherein the probabilities for meeting the sub-parts in natural processes can be multiplied together and form less than the universal probability bound? Thus the first two figures may provide examples of type 3 failures of the explanatory filter. Dr. Nelson has pointed us toward such a failure in his example – only the first example suggested.

NOW can we improve the explanatory filter in differentiating these, and do so with characteristics of the hypothetical agents of construction? Would “motive” be useful to consider?

What I have learned from this example is that the type 1-3 distinction relates more closely to figuring out just what could be considered a “specification”, and that type 3-4 distinction will have a greater degree in which “motive” is potentially useful. So we leave that for part two of this post. The key here is to examine what constitutes a “specification, given independently of the event”, and the type 1-3 distinction shows problems that need to be addressed. Without refining what it means to give a “specification” we have a clear failure of type 3 in the first two figures.

I repeat my question, can someone please show where a “specification” of the Easter Island figures exists independent of the event?

[ 07. September 2003, 12:38: Message edited by: gedanken ]

posted 07. September 2003 15:41                    

quote:

---

To answer these questions I have asked for where we would find the “specification” of the Easter Island figures.

---

quote:

---

I want to see if the “specification” would not match the first two figures as well.

---

posted 07. September 2003 16:13                   

quote:

---

In the case of New Hampshire's "Old Man of the Mountain," for instance (alas the formation collapsed recently), the human-like features of the pattern disappear at nearly all angles other than the one shown in the photograph above.

---

posted 07. September 2003 18:51                

quote:

---

The specification is the anatomical form of Homo sapiens.

---

But I would be quite surprised if we are to take as “specifications” any existing object, having been observed. Then any form of commonality, though of low probability, which is found again in another part of the universe would then be declared as “design”. There is no connection required in this notion of specification to any known “intelligent” description of the object, only duplication of form? (Surely you are not taking as premise that the “human anatomical form” is itself “designed” as the argument that it forms a “specification”, are you? – that would be circular reasoning on part of ID conceptualization.)

In addition I would argue most strongly that the Easter Island figures are not accurate representation of human form.

Here is a highly accurate bronze representation of a human form, next to the Easter Island forms:


The Easter Island forms are not directly observable human forms, they are representational in the sense that the lines, edges, and shapes are approximations to our form with specifically sharp edges corresponding to features of the human form that we are particularly adept at recognizing. Now I have previously warned about using “representation” as a descriptor for identifying “design”, since itself it implies design and would be a tautological statement that “representation” implied design. So we need a different word.

What we have is that the edges and various aspects of the Easter Island forms are those which we would easily recognize as a human form – very different from actually matching closely the human form.

The aspect of the image only appearing at certain angles is basically irrelevant. A flat relief human-representational sculpture is not visibly “designed” to appear as human image from angles other than nearly face on so that it is readily visible. That aspect speaks much more to the issue of probability than it does of specification, as the specification is only met in a narrow set of views rather than the specification being met in a wide angles of viewing.

But the use of “anatomical form of the Homo sapiens face” is quite an adequate and similarly produced specification, if that is to be considered as sufficient for the old man or other face images. It is not the matter of “specification” that is in question, it is the matter of probability of meeting the specific specification as given that is relevant to the explanatory filter. I hardly see a way to deny that the first two face images can have an equally acceptable specification.

Yet I must emphasize most strongly that this appearance of meeting the lines that we recognize as a figure that is the specification, not the portion of the human form itself that is the specification. The Easter Island figures are considerably abstracted and not crisply coincident with actual human anatomy! That is why we can easily classify the Easter Island figures – because we recognize that these meet the conditions of representation (advisedly used) that are produced in various cultures of human beings.

We can, for example, effectively make a comparative argument for the representation of the human form as cause for the Easter Island figures. (In this paragraph I am not per se rejecting eliminative methods, that will be dealt with in other paragraphs, and I am using “representation” advisedly in my meaning as human designed). The comparison would be drawn between possible natural geological, water, etc., processes that could produce the figures, and the possibility of human construction according to patterns of representation that we observe regularly for humans. We compare human action to be “representational” to these other non-intelligent natural processes. So I argue that a comparative method most certainly provides a strong argument for our accepting the “ID” classification as “true” or universally agreed for the Easter Island figures. We are not in disagreement that the Easter Island figures are either type 2 or 4 case in above table, universal agreement is “Actual cause – Yes ID” category. The question at hand for other paragraphs is whether the eliminative method of the explanatory filter adequately distinguishes the various stone figures.

So I don’t agree that the “specification” of the Easter Island forms has been adequately described independently of the event.

[Added in edit ] ===================

I am adding this to clarify what I have written, and to deal with issues that have some subtlety. The eliminatory ID method is not terribly unreliable, just unreliable under certain types of circumstances and more reliable in others. (I assert that it is therefore not inherently “reliable” as is claimed by ID enthusiasts.)

But I am not saying that an accurate “representation” of a human form cannot be made into a valid “specification”. I’ll illustrate by assuming a variant case. Assume that the stone figures had in fact been highly accurate reproductions of human anatomy, of representational quality like the Rodan sculpture, but in stone material. (Many such ancient sculptures are found, just perhaps not as distant in knowledge of their origins. But let’s assume that the case is of the same age and location as the Easter Island figures.) In such a case the explanatory filter would probably be quite reliable (even if for the wrong reason).

Now the specification could not be that the form was a “representation” per se, because that would indeed be circular reasoning. Rather the specification would be that it “matches” an accurate representational form. Here we do not inherently claim it is a representation, rather that some specification has been given independently that forms a narrow tolerance to lifelike representations of human form.

It just happens that Dr. Nelson has chosen an example that makes my demonstration easy – there was no assurance that an ID enthusiast would choose such a case.

The key is that the Easter Island figures degree of abstraction needs to be accounted for in the specification. As we allow our “specification” to account for greater degree of abstraction we have various difficulties. Of course as the degree of variance in forms (e.g. a “tolerance” concept”) increases, a larger number of random patterns could fit that “specification”. In other words as the specification per se is loosened to allow for more forms to match, then a greater probability occurs that a randomly generated form could match the specification.

We cannot just state that the specification is something that matches representational forms with some arbitrary amount of abstraction – we have a problem of finding that particular degree of abstraction given as a specification independent of the event.

To be very specific the degree of abstraction cannot just be arbitrary, or the randomly generated stone face figures are perfectly valid cases under that particular specification. So let’s analyze what happens in that case:

In that case the specification given has sufficiently high probability of a random occurrence that the universal probability bound is exceeded. In this case, no “design” is inferred in the Easter Island figures directly. Now be very careful, I am not saying that the configuration found in the Easter Island case was therefore a configuration that is likely by random processes – far from it. What I am saying is that the configuration that has been specified (“a match to representations of human form with arbitrary degree of abstraction”) does not capture the specifics of the Easter Island figures sufficiently to be of low probability. It allows figures like the random geological figures to fall in the specification.

So the specification needs to be tightened up from that one that is too general, but yet needs to not be tightened up so much that it does not capture the Easter Island figures. This is where the problem occurs!

The problem is that it is not just abstraction of human form that is apparent in the Easter Island figures. There are a great many details there that have to do with details of possible human construction techniques, things that human seem historically to do (as in “representational forms”), details of how the form was on the land in a place not likely for geological processes to generate the same event.

In other words the event was indeed clearly very unlikely by a non-intelligent process! The problem is that to capture what is known about the unlikely-ness – to capture the “specification” – it is difficult to do so while remaining “independent of the event”.

What we wind up with is a case in which comparative methods succeed, but eliminative methods fail. But in this case the failure may not be one of producing a “false positive” directly – rather the eliminative method may simply not allow for a specification to be given that is independent of the event and captures the event.

What we get is a willingness to want to capture that event anyway to make it a positive result example of eliminative ID method. So the ID advocate may start making use of lack of clarity in what “specification” means, so as to appear to produce a specification that can be argued to be independent of the event, and also to be of low probability by known natural processes. This failure is a different one than I had been contemplating, but shows the unreliability of the explanatory filter in practice by showing the subjectivity in forming the “specification”.

And what is notable in this example is that the comparative method (as opposed to the eliminative method) can make use of aspects like “motive”, in the abstract sense that humans are known to have motives toward making representational form, and constructing them. Thus a comparison does not need to be made to a “specification” which must be independent of the event in question. The processes can be quite dependent on the hypothesized event because details of the event are considered (as opposed to eliminative approaches that must ignore all such details). So the comparative method allows scenarios (abstracted) to be constructed in which the Easter Island figures are intentionally constructed by humans using primitive tools and construction techniques – even though those methods, intentions, and techniques are considered in a fairly abstracted way. The comparative analysis still gives much higher posterior probability to the human construction of the Easter Island figures.

[ 07. September 2003, 23:42: Message edited by: gedanken ]

posted 07. September 2003 19:13                    

quote:

---

The Easter Island figures are considerably abstracted and not crisply coincident with actual human anatomy!

---

posted 07. September 2003 21:09                    

quote:

---

That seems to me to be a dangerous position for ID.

---

posted 07. September 2003 22:44                   

quote:

---

Actually, this is no more dangerous than saying that we can *know* anything. What is knowledge? Justified true belief? What is justification? Warrant. So what is this warrant that allows us to call our beliefs justified, or "knowledge"?

The real world is ambiguous. Holding the design inference to higher standards than "knowledge" is not realistic. It doesn't take our epistemic situation seriously.

---

posted 07. September 2003 23:18                

quote:

---

Many chinese characters are heavily abstracted. Consequently, I'd have a hard time telling the difference between a legitimate chinese character and a meaningless, fake character unless I'd studied chinese writing for a long time.

---

posted 08. September 2003 01:02                   

quote:

---

What this means, of course, is that Dembski's design inferene will not be particularly useful either in initial recognition or identification of design

---

quote:

---

So typically, patterns that are likely candidates for design are first identified as such by some unspecified ("mysterious") means, then with the pattern in hand S picks out side information identified (by unspecified means) as releavant to the particular pattern, then sees whether the pattern in question is among the various patterns that could have been constructed from that side information.

---