|
Author
|
Topic: Constraining the Search Space
|
Alix Nenuphar
Member
Member # 686
|
posted 11. April 2003 18:28
Since this is my first topic, I beg your indulgence for the preamble.
ID is an attempt to formalise the perfectly valid human intuition that the various events in the history of the universe are the result of conscious choice, rather than unconscious process. This intuition is obviously true: humans, crows, ants, etc. 'design' on a regular basis, and their designs are things with which we are familiar.
The more important question is whether this intuition can legitimately be extended into times and places in which no known designers are operative; the distant past, for example, or another planet. To that end, the ID movement has been looking for possible design events in those areas.
Since the universe is, as someone once remarked, "a rather large place, and has been for a rather long time," this search process requires some improvement in efficiency: it is impossible to examine every possible event in the history of the universe and query it for design. The current pool of possible design events is strikingly limited.
So far, most ID advocates have concentrated on identifying biological 'design events'; those which do not have currently well-established evolutionary pathways and/or which appear to be too improbable to be result of such pathways.
I suggest that a potentially more fruitful approach to the problem might be to begin with the designer: to hypothesize various categories of 'designers', 'motivations', and 'mechanisms'; this would enable us to identify potential classes of events to inspect. In addition, this approach might allow us to include non-biological events more easily.
Please keep in mind, I am aware of the general distaste for speculation about the nature of the designers; I am not advocating that we can deduce these things from our hypothetical examples, but simply use them to limit the search.
Perhaps a trivial example from archaeology might be benefit. When confronted with a set of 'apparently' shaped stones taken from a prehistoric site, the criteria used to select out those which might be tools include usage requirements based on the nature of the site; availability of the types of stone, etc. Scrapers, cutters, awls, etc. are all presumed to be needed. These represent classes of tools which might be found. A stone that appears worked, but does not appear to fit into any of the known tool categories is likely to be put aside as an 'undetermined' or unknown.
The point is, we've already hypothesized several things about the designers in this case: 'human' (or at least hominid); 'stone-age technology'; 'need for tools'.
Indeed, it might be suspected that it will be impossible for us to even recognise an 'artifact' created by a designer unless that particular designer shares to some degree our psychological, technological, and motivational constraints.
What kind of designers can we legitimately speculate about? Unconstrained designers are not a viable choice for this technique: no statements can be made about their psychology, technology, or motivation. This would appear to leave out spirits, disembodied essences, gods, etc, and allow inclusion of biological and mechanical designers.
Does this approach appear to have any validity?
(Caveat: this post represents a refinement of a concept begun on another discussion board. From those who might have already encountered it, I beg your patient indulgence.)
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 11. April 2003 20:41
I think this is a very valid methodology. (I do believe that there are political reasons that it will not get much interest here.) But this is exactly how archaeology, forensics, and other sciences that do already identify things that are associated with intelligent being proceed. Note that forensics, archaeology, and the like are often brought up as examples that ID is not a foreign subject to science, usually right before the presenter claims that ID is rejected by science. (These claims about tendencies in ID presentations are from my actual experience. They may not apply to this particular group and are intended only to be representative of how ID has been presented in ID seminars and meetings like DDD III.)
I think that such methods can be supported with sound mathematical and observational basis, such as with Bayes rule statistical methods. I would note that there are sections of Dembski's No Free Lunch that dispute the usefulness of idea, and someone should present something from NFL for discussion.
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 11. April 2003 21:17
I'm not certain that even unconstrained designers need necessarily be ruled out. Humans can create many more shapes out of stone than we actually observe, so it's not the capabilities of humans that make us produce artifacts of a certain type--rather it is our needs and/or aesthetics. It might be difficult to comprehend the needs of spirits or gods (maybe they don't have any needs!), but if the entity in question has enough psychological similarities to humans to be even vaguely comprehensible, we might be able to constrain their designs based on a supposed aesthetic sense of said entit(y/ies).
I'm not sure how far one could get by proposing such an entity, hypothesizing that certain objects were created by it, and then using that information about its aesthetic sense to predict future observations. (I'm even less sure that this would give different answers from a simple, "Gee, we observe this, so let's look for other stuff that's similar".) But it seems to me that it could be interesting if done carefully.
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 11. April 2003 22:27
Alix specified "various categories of 'designers', 'motivations', and 'mechanisms'". I'm not sure how any human could be totally unconstrained, for example by any sort of physical limitations on access to the events in question.
Forensics, for example, focuses on the evidence that relates to mechanism and access of the human, and to known patterns of human action, in part. Archaeology assumes, for example, that stones carved had to be carved by someone in physical locality to the stone in question, for the most part. These relate to "various categories of 'designers', 'motivations', and 'mechanisms'", without specifying the type of design that could have been accomplished, for example limiting the pattern in the stone per se in any way.
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 11. April 2003 22:46
Alix wrote quote:
The point is, we've already hypothesized several things about the designers in this case: 'human' (or at least hominid); 'stone-age technology'; 'need for tools'.
(Just goes to show you what hanging around a paleontologist does for you. I was paying attention, Alix! ![[Smile]](smile.gif) )
In fact, as you say, we know/hypothesize quite a bit about the potential designers. We often know (from fossils, geology, etc.) a good deal about the physical/ecological context, and hence the likely needs of the designers. That is, we can know (or infer) something about functions (and designers' purposes) independent of the assemblage of artifacts. We can assign the artifacts (as you say) to their various functions just becfause we know about those functions independent of our knowledge of the artifacts. (Note that this is not "specification" in the Dembskian sense!)
We can ourselves manufacture the tools and try them out - I remember sitting in a paleo course too many decades ago, watching fascinated as the lecturer sat cross-legged on the table in front of the hall lecturing whilst knapping a chunk of flint. It's not foreign technology to us. On my porch at this moment is a lovely canteloupe-sized chunk of Coshocton County Upper Mercer black flint waiting for me to get up the courage to whack it one.
Rex's remarks about aesthetics are very interesting. Last year in the Introduction to Multiple Designers Theory thread I suggested that people versed in art history and in attributing works to particular artists would be valuable members of a multi-disciplinary team devoted to discriminating among the products of multiple designers: quote:
B. Design Themes: As I have mentioned elsewhere, in the study of human-designed phenomena like works of art or literature, there are more-or-less well-developed research methods for assigning works to designers. Analysis of physical properties (e.g., characteristic brush stroke micro-patterns visible on a painting), statistical properties (e.g., distributions of vocabulary items or syntactic structures), and other properties of human-designed objects are routinely used to attribute an object to one or another creator. The same is true of the unembodied designers of MDT. It should in principle be possible to identify characteristic design properties and even different general design themes that differentiate one from another of the multiple designers. It is likely that the same methods that are used in attributing human-designed objects to one or another human creator could be adapted to attribute biological designs to one or another of the unembodied creators. That suggests the utility of multi-disciplinary research teams involving not only scientists but also those trained and experienced in discerning such things as individual esthetic themes and differing creative motifs among human artists. Their insights could form the basis for hypotheses that can then be tested scientifically.
It is not out of the question to incorporate those kinds of judgements into a Bayesian framework, as gedanken suggests, moving the enterprise away from a Fisherian hypothesis-testing approach to one more suited to discriminating among competing hypotheses. Or designers.
RBH
IP: Logged
|
|
Janitor@MIT
Member
Member # 125
|
posted 15. April 2003 13:58
The terms “need” and “purpose” have recurred here. Please define. On biological theory I’m fairly confident that I can subsume “purpose” under “need” and define “need” in purely biological (and the corresponding design-theoretic) terms.
Also there is a rigorous definition of “mechanism” in design science. Is there a corresponding definition in evolutionary biology? Dr. Dembski has asked if evolutionary theory has a “mechanism.” Does it? I suspect that we must be working with the same definition of “mechanism.” Are we?
And to continue in Rex Kerr's vein: Does “mechanism” (as rigorously defined) constrain (any or all) designers?
I suspect that the very definition of "design mechanism" we are working with, assuming that there is some commensurability here, is quite "user-dependent."
OTOH, I would certainly have problems understanding a "designer" who does things radically differnt from how I do them. Hmmm... I wonder how I might imagine such a designer?
Are there design "invariants?" Maybe we could work on that question? Are there "ways" that design must or can only be done?
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 15. April 2003 14:38
Apropos of Janitor's remarks on "need" and "purpose," I would note also that "purpose" and "function" are often conflated in these discussions.
Janitor suggested that we may be operating with several different meanings of "mechanism" in this (and other) discussions. I suspect he is correct. So many of the disputes and misunderstandings in this whole area derive from different (usually implicit) understandings of what terms mean. Here was my try at "mechanism" in the now-closed "Intelligent design is not a mechanistic theory" thread:
quote:
I understand a "theory" in science to have several characteristics and functions. Among them (very sketchily outlined) are:
1. Organize phenomena into classes, where the classes are connected by relationships described in general laws and law-like principles that are corroborated by a body of empirical research.
2. Provide an explanation of phenomena, where "explanation" means to assign the particular phenomenon to be explained to the appropriate class from among those mentioned above and invoke the law-like relationships - causal, correlational, exclusionary, etc. - among the classes to account for the particular phenomenon. That is, explanation is subsumption of particular phenomena under larger classes to which the various relations specified in general laws apply.
3. Generate predictions of new observations of phenomena. A simple-minded example is if a newly found phenomenon a is tentatively assigned to Class A, and a general law-like principle is 'If A, then B,' go look for b, a member of Class B, in the specified association with a.
4. Be as consistent as possible with the laws, principles, and generalizations in relevant neighboring domains of scientific inquiry. E.g. if one is hypothesizing about neural information-processing activity, one is not permitted to invoke FTL information transmission rates.
"Mechanisms" enter in several of the points above, but I mostly understand describing a "mechanism" to be describing the physical (matter and energy) transformations that connect the classes referred to in numbers 1-3 above. That is, the notion of "mechanism" is intimately bound up with the relations between classes. It goes beyond correlational relations, where the occurrences of phenomena in A are more-or-less regularly associated with occurrences of phenomena in B, to a description of transformational relations which describe how members of A are transformed into members of B by a series of physical transformations/operators, where again "physical" means embodied in matter and energy.
(As noted, there are other potential relations between A and B; here I'm focusing just on "mechanism." Also, while I've written of "A" transformed to "B" as though one class generates [i.e., is transformed into] another [under certain theory-specified conditions], there are often (always?) many-one and/or one-many relations among classes.)
My editing of the previous post is in brackets. Notice that what I have attempted to define is what is meant by describing a mechanism. After all is said, done, and/or calculated, theories are descriptions of one sort or another. On my definition above, the answer to Dembski's question is an emphatic "yes," evolutionary theory has a number of mechanisms.
I would be interested to see what Janitor says is "a rigorous definition of 'mechanism' in design science."
RBH
[ 15. April 2003, 14:42: Message edited by: RBH ]
IP: Logged
|
|
Janitor@MIT
Member
Member # 125
|
posted 16. April 2003 15:20
“To iterate is human. To recurse is divine.”
Usually, the rigorous definition of “mechanism” is said to have been provided by Church and Turing (independently): Partial recursive function = mechanical. I always wondered what that “partial” is all about. Apparently it supplies the ambiguity necessary to make this definition applicable to just about anything one can imagine. The hackers tell me (quite confidently) that that every recursive function has an (I/O) equivalent non-recursive implementation. Does that mean that every “mechanism” has its “non-mechanical” equivalent? But what they are talking about strikes me as even more mechanical than mechanical, if ya know what I mean? "Non-recursive" for them seems to mean unnecessarily, elaborately, and inefficiently “mechanical.” (Whereas a some recursive functions have a certain simple elegance to them—in my aesthetic opinion. But as everyone knows, programming is to aesthetics what beans is to cuisine.)
The “user dependency” enters in a number of different ways. The “mechanism” doesn’t work (except for pure mathematicians) w/o (user-provided?) “plug-ins.” It has to be supplied with “resources,” including but not limited to information and space-time. Where does the information come from? From some other recursive function(s)? Another ontogenetic loop?
It’s natural to see these resources, and in particular space and time, as constraints or costs (again user-defined). Usually the definition stipulates to finitude (test to halt). But notice that this is not intrinsic to the recursion as written on the page. This stipulation really is because we are drastically finite creatures ourselves. We require (“need”) our solutions (approximations) in P-time (~ experimental time ~ lifetime). We want our procedures to halt in real-time/anytime.
(I noticed in Schneider’s, now famous, ev program, e.g., that either he intervened to halt the program or it “consumed” all resources and halted.)
Is evolution “mechanical,” by this (C-T) definition? Hmmm…
OTOH I can say, with some confidence, that it is exactly the introduction of these “user-dependencies” (resource boundings) that makes design a nearly, if not exactly or exclusively, mechanical procedure… by this definition.
But all that’s BS. Uh, I mean “brainstorming.”
As to the “psychological” investigation of designers it’s all very interesting in that purely speculative sort of way, but I think I just reduced all such “psychological” terms to computational terms. Per the demands of, not computer theory, but biological theory! A designer is just an anytime evolutionary algorithm. A “machine!” But… but… machines do have “purposes,” even if they are not their own…
LOL
IP: Logged
|
|
|
Printer-friendly view of this topic