|
Author
|
Topic: Does anything at all exhibit specified complexity?
|
Rex Kerr
Member
Member # 632
|
posted 10. March 2004 23:25
Human behavior produces human artifacts. The design filter is supposed to give information on how something is produced. So we are concerned with human behavior.
But we can also ask the question, how do we know when a human artifact is specified?
Also, you claim that "Where we don't know the relevant physical laws, Dembski's filter will force us to find out what we don't know."
How? If we know that we don't know the relevant physical laws, we can search for them perfectly well without the filter, as has been done in science for the past several centuries. If we don't know that whe don't know the relevant physical laws, Dembski's filter will return an answer of "design".
Finally, consider whether or not science can explain these natural phenomena well enough to predict the probability:
- Whether a protein of known structure will bind a small molecule of known structure.
- What the structure of a protein will be given that we only know the amino acid sequence for it.
- How much protein will be made in a particular cell given that we know the DNA sequence for the gene for that protein?
IP: Logged
|
|
kyle7
Member
Member # 191
|
posted 11. March 2004 02:16
Rex says the following:
quote:
Human behavior produces human artifacts. The design filter is supposed to give information on how something is produced. So we are concerned with human behavior.
You are distorting the purpose of the design filter for your own ends. We don't need to know anything about the designer. I say again. We don't need to know anything about the designer! The complexity-specification criterion detects design by establishing three things: contingency complexity, and specification. The designers behavior is not a factor in the design filter. Behavior of the designer is an extraneous concern. There is a key error in your statement. The design filter does not "tell how something is produced" but only tells if the artifact is designed by an intelligent designer. If we can show contingency, complexity and specification we can conclude the artifact in question is designed. Dembski says, "A design inference therefore does not avoid the problem of how a designing intelligence might have produced an object. It simply makes it a separate question" (Dembski, "No free Lunch" pg 112).
Rex also asks:
quote:
But we can also ask the question, how do we know when a human artifact is specified?
A better question is: "How do we know when an artifact is specified." "Specifications are the non-ad hoc patterns that can legitimately be used to eliminate chance and warrant a design inference" (Dembski "No Free Lunch" pg 12). Essentially, a specification is determined by using physical models, empirical data, and/or probabilistic models. For example, an ID researcher may want to determine if strange "writing" in the sand is due to natural phenomena or is due to an intelligent designer. Fluid models that include particles of sand could be run to develop a statistical data-base on sand formations. Also, empirical data could be generated by taking pictures of sand formations found in natural locations. Also, complexity measures could be developed showing the interconnectedness of the pattern. Taking all of this information together a probability could be calculated. I will post this now and respond to the rest of your points in another post.
[ 13. March 2004, 20:05: Message edited by: kyle7 ]
IP: Logged
|
|
Pim van Meurs
Member
Member # 541
|
posted 11. March 2004 12:05
Kyle7: You are distorting the purpose of the design filter for your own ends. We don't need to know anything about the designer. I say again. We don't need to know anything about the designer! The complexity-specification criterion detects design by establishing three things: contingency complexity, and specification.
I think in this case it is relevant to distinguish what the design filter tries to do which is attempting to infer design without any knowledge about the designer(s) and the reality that when applying the filter such knowledge seems to be essential for the filter to work practically.
i find this not too surprising since this is how science detects intelligent design all the time, using means, opportunity, motives.
In other words, I believe we disagree with Dembski's statement that 'who is the designer' is a separate question.
In that context I would like to ask the following question about specified complexity. If the law of conservation of CSI leads Dembski to conclude that algorithms (regularities and chance) can only displace CSI not generate it, why would the same not apply to intelligent designers? In other words, what makes intelligent designers to be 'above the law'? Has Dembski shown that intelligent designers can in fact create CSI and not merely displace it?
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 12. March 2004 13:53
Kyle, I think you're missing the point of scientific inquiry into a process. As I've stated before, I am uninterested in the stated purpose of the filter for this discussion. I am interested in what it will actually do when we throw various scenarios at it. Dembski may have created the filter for some purpose, but that doesn't mean the project was successful--for instance, we released South African cane toads into Australia to try to control accidentally imported cane beetles. Now there are lots of cane toads, and lots of cane beetles too, and both are big pests. Saying that someone pointing out this fact is "distorting the purpose of cane toad introduction for my own ends" really misses the point.
The reason I want to try using the filter to detect human design activity as if human design activity were natural is because I am testing the rhetoric that there is something special about the way we detect design. If there is something special about the way we detect design, then the filter should not be able to catch human design events.
Alternatively, if there isn't anything special about the way we detect (human) design, then the filter should be able to catch human design if we include what we know about human design processes as part of our chance processes and regularities. Indeed, we might expect that the filter would turn into a gravity-detector instead of a design detector if we add what we know about human design into our chance/regularity rules and remove everything we know about gravity.
I claim that the latter is the case. In fact, I claim that it cannot be anything *but* the case, because if we understand design well enough to match it to specifications, and understand humans well enough to know that they build things to specifications, it is a very short jump to coming up with probability estimates for humans creating certain designs. So if we can apply the normally-interpreted filter properly to artifacts which have been created by humans--which requires a specification--we should be able to construct an alternate filter that catches exactly those objects as "natural or human-created", but not "complex" (and thus not "complex and specified").
This would obviously raise the question of why we're using the filter we're using.
Also, Kyle, you seem to have completely ignored my (rhetorical) list of natural phenomena. The filter is supposed to be used on objects that involve those phenomena, not fluid models of sand.
IP: Logged
|
|
Claire
Member
Member # 725
|
posted 12. March 2004 21:05
Rex,
Are we wanting to measure design or understand design? This question by the way does not imply that I am a supporter of design.
Claire
[ 20. March 2004, 21:47: Message edited by: Claire ]
IP: Logged
|
|
kyle7
Member
Member # 191
|
posted 13. March 2004 12:25
Rex says:
quote:
Also, you claim that "where we don't know the relevant physical laws, Dembski's filter will force us to find out what we don't know.
How? If we know that we don't know the relevant physical laws, we can search for them perfectly well without the filter, as has been done in science for the past several centuries. If we don't know that whe don't know the relevant physical laws, Dembski's filter will return an answer of "design".
Let me explain what I meant. In the scientific investigation of design you will always have those who advocate design and those who don't. The one side will put forward evidence to show design, while the other side will attempt to refute the evidence. Dembski's filter is a useful tool in such discussions. It will force both sides to use science to justify their arguments. If the one side puts forward some artifact as evidence of design, then the other side will be forced to examine the side information used to establish the design inference. If all relevant physical phenomena are taken into account, then those who are against design will have to investigate new scientific laws or processes that may explain the "designed" object as natural. Science will then benefit from such discussions. I should point out here that I talk about "two" sides for the sake of simplicity. Actually, there are many sides in the design camp as the ISCID brainstorm threads exemplifies.
You ask whether science can explain some natural phenomena well enough to predict the probability. The list you give are the following:
1) Whether a protein of known structure will bind a small molecule of known structure.
2) What the structure of a protein will be given that we only know the amino acid sequence for it.
3) How much protein will be made in a particular cell given that we know the DNA sequence for the gene for that protein?
For the first phenomena, I think we could use molecular dynamics to determine a probability. The second phenomena can also be modeled using molecular mechanics methods. The calculation of a probability for the last example would be very difficult. I think you would need to know more information. Essentially, you would need to model the whole cell -- which is beyond our current capabilities.
But, it is ridiculous to use these examples to say that Dembski's filter is useless. We don't need to know every probability to use the filter. We can only expect the filter to be as good as the known scientific laws and theories and that is exactly what we want it to do. We don't want to throw in speculation and bias, for if we do, we will be outside the realm of science.
Also, I should state that there are different levels of analysis in using Dembski's filter. As with any analysis, it can range from the very simple to the very detailed. Currently the ID movement is increasing the level of analysis from the very simple to the more complex and detailed. Also, some are discussing the development of a computer program to automate the use of the filter. If those at the National Science Foundation would get off their high horses and put forward some money, then the utility of the method would be realized much sooner. But, this is anther subject so I will post this now.
[ 13. March 2004, 20:01: Message edited by: kyle7 ]
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 18. March 2004 01:53
I'm afraid I don't share your optimism about the utility of the EF to foster scientific debate, at least not among scientists. If it will encourage non-scientists to take a scientific approach, then I think that's a positive development--except that I think that the EF is an intellectually bankrupt (albeit in a somewhat subtle way) construct, as I've described above in this thread.
The main reason I don't share your optimism is that I've seen exactly zero careful treatments of specification applied to any specific problem. Nobody in the ID camp has done it, and nobody outside it has either; I have tried, but I can't figure out how to do it properly. It seems that this is exactly what the EF should encourage people to work on, and yet it's largely been ignored.
As to the examples, we can't actually do any of the three reliably. There are yearly tests of the first two, using molecular dynamics and mechanics methods (primarily), but unfortunately the result is that even the best algorithms only get sort of close, and then only on half to three quarters of the samples.
The reason I bring these up is that fundamentally, evolution involves changes in protein structure and regulation. If we cannot reliably predict the outcome of such changes (which can potentially be quite large), and we don't have experimental data that adequately addresses the question (and we don't), then of course our probability estimates are going to be off. And if our estimates are off and we apply the filter anyway, it will return an answer of "design", since the way Dembski set it up, that's what it does when you don't understand the system properly.
Most NSF funding goes towards fairly conventional project proposals. That doesn't mean that a fraction of the budget can't be used to support less conventional but related research--if things work out, then you put your positive results in your *next* grant application, and maybe it'll get funded. I see no particular reason why ID-inspired research should get any special treatment.
In any case, this is getting off topic. The discussion here is important, but it isn't directly relevant to whether there is a fundamental problem with Dembski's (artificial) separation between "intelligent" and "natural" causes when analyzing the probability of an event.
IP: Logged
|
|
Stephen Wright
Member
Member # 195
|
posted 18. March 2004 15:48
Rex wrote: quote:
“The main reason I don't share your optimism is that I've seen exactly zero careful treatments of specification applied to any specific problem. Nobody in the ID camp has done it, and nobody outside it has either.”
This is an unfair (to the continuity of your thinking) sound bite from your recent post. I just feel that I should defend the state of the art regarding how well dedicated folks develop specifications in modern technology. When I am at my work desk, I am surrounded with specifications. ISO documentation, process control manuals, lab reports, standard values, data sheets, MSDS sheets, Industry Standards and custom specs for particular special situations. Creating quality specs is an art and fundamental science. I think that the process of generating specs is a decidedly healthy part of our technological culture - so much so - it may be taken for granted.
Rex, Please help me again understand your viewpoint. I know that at the naïve level of specified complexity, to which I am referring, you agree that it is conceptually pretty straightforward. It is with the Dembskian use of specification in the EF, that you dispute. I am a very much oriented toward pragmatic understanding. Let me propose a concrete example. From today’s CNN website regarding the robots we have sent to Mars.
quote:
“Recognizable patterns
The identification of fossils is often difficult, explained Ron Greeley, Mars Exploration Rover team scientist from Arizona State University in Tempe, even by scientists observing them with the full spectrum of lab instruments.
"Remember that fossils are defined as the traces -- such as leaf imprints in rocks, or the remains, such as shells or bones -- of formerly living organisms. Typically, recognizable patterns are sought, such as bilateral symmetry," Greeley said. "Unfortunately, similar patterns often occur in rocks that result from non-biological processes, which make the identification more difficult."
On Mars, Greeley said, there is no reason to expect the same patterns as fossils that are seen on Earth. Nonetheless, patterns of some sort are being formulated by the astrobiological community -- so-called biomarkers -- in the on-going search for life elsewhere. Furthermore, while the Athena science gear onboard the two Mars rover are great assets, "analyzing patterns and other features remotely is not so easy," he added.”
By Leonard David
Those researchers will need specifications - as to what pattern characteristics exhibited by the natural materials we will find on Mars could lead us to infer life existed there at some prior date. I find the specified level of probability showing “design or life-pattern” proposed by W. Dembski very tight. Please explain how it would not be useful in addressing the issues these scientists face. Will it be misleading? Will it find “design” as life when it is not, if fact, warranted. Is it a right direction but not developed enough? How does it fail?
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 19. March 2004 02:19
Dembski has a specific quasi-mathematical definition of the term "specification", and I am using his term rather than the colloquial one.
A specification must "univocally" identify the pattern you observe, must lie within certain complexity bounds (or meet certain criteria relative to other possible specifications, if you use NFL instead of TDI), and must be independent of the event being studied (otherwise it is a "fabrication").
A handwaving determination of specification is easy. I want to see a rigorous one. How do you determine whether specifications are univocal or not? How do you judge the complexity of a specification? How can you demonstrate that the event is independent of the specification?
Presumably, Dembski put these conditions in for good reason. (I think there was good reason, as one can demonstrate that the method fails if any one of these criteria is omitted.) If they are there for good reason, they should be treated with due diligence when a determination of design is made.
If you review the published (and posted-online) treatments of specification, I think you will find that specification is never treated carefully, showing in detail that all the criteria are met.
As far as Mars goes, people there are not using Dembskian specifications, but rather standard induction. If life produces pattern X here, and we see an X-like pattern on Mars, then this is evidence that the same process is occurring in both places. (One can try to make life-is-designed arguments that look like this, but that is not what TDI is about.)
IP: Logged
|
|
Claire
Member
Member # 725
|
posted 29. March 2004 22:03
By my second edited suggestion, that is from my first question:
"This question by the way does not imply that I am a supporter of design"
could now not imply that I do not support design.
So I am about to say now, that I haven't yet suggested whether I am or I am not a supporter of design theory but the questions and statements are very directional and are "leading". This can be sometimes equated to how we use our sense about what appears to be logical reasoning in science, when it isn't because it is really judgmental reasoning. I am surprised that some people on this forum are supposed to be the former when in fact they often argue by the latter!(although I am still intrigued). The emotional content is very obvious. I think we can misread posts much like we can misread words that are supposed to describe aspects of our reality and their descriptions of complexity theory etc, as they are often leading one way only. We might then, "sometimes" benefit from looking the other way, even if our experiential judgment is very powerful.
Claire
[ 29. March 2004, 22:07: Message edited by: Claire ]
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 30. March 2004 19:01
Claire, I am interested in understanding the properties of Dembski's "Explanatory Filter". Does it detect design, or can we only show that it detects a lack of understanding?
If the tool is to be used, we should understand how it works. A yardstick, for instance, is not a good tool for detecting fever.
As to your points about logical vs. judgmental reasoning, I agree, I think, but I am not sure whether the comment is more applicable to this thread than the average thread on this board.
IP: Logged
|
|
Claire
Member
Member # 725
|
posted 31. March 2004 01:51
Rex,
I am interested in that theory too (The Filter) because months ago it reminded of an type of inference i was thinking about.
What I posted has very much to do with the thread I think. The type of inference via the filter is much like re training our logical soundness of about how we interpret information, in this case it was the words in a post.
I'll post more later this week about your thought.
Can an idea in a post be assumed to be designed? It's different, indirectly obvious. Think about this one.
Claire
[ 31. March 2004, 01:58: Message edited by: Claire ]
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 31. March 2004 12:07
I think the ideas in a post can be assumed or deduced to be designed, but I do not think it is possible for the EF to detect this.
I see your point about similarities between training to use the filter vs. training to use logically sound techniques for reasoning. This highlights why it is important that the filter be logically sound--it would be a waste of effort and misleading to train ourselves to use a technique that was fundamentally unsound or illogical!
IP: Logged
|
|
Claire
Member
Member # 725
|
posted 31. March 2004 22:15
Rex,
I said "re training" our logical soundness, not not be logical here. Interpretation is the fundamental part of the equiptment and vice versa. The logical inference for the EF could be tweeked, re trained, we copuld then interpret what we assume as a separate process of the EF as separate from the tweeking.
Claire
[ 31. March 2004, 22:28: Message edited by: Claire ]
IP: Logged
|
|
|
Printer-friendly view of this topic