|
Author
|
Topic: Can intelligence be described scientifically?
|
Cornelius G. Hunter
Member
Member # 81
|
posted 03. March 2003 15:17
Erik:
There's an old trick. Sit someone down and ask them to figure out your code. Roll dice or draw cards. With each event, the code produces a number, say from 1-5. The person looks at several rolls of the die, each time you telling him what the code produces. The person tries to figure out the code, but it is a trick, there is none. Each time you roll the die you sit back and fold your arms, hiding some fingers, making others visible. The number produced by each event is simply the number of visible fingers. The result is entirely contingent upon the whims of the trickster, there is no rationale involving the die. When you look at a circuit, you are assuming a rationale, repeatable process, not some trickster.
A fundamental precept in evolution, OTH, is that contingency dominates. It predicts essentially nothing. Whatever we find, it is the product of evolution. If there is a universal DNA code, then it was produced by evolution. If there are different codes, the too were produced by evolution. If fossils are gradual, that is evolution in action. If they appear planted there and then don't change, that is evolution. Everything is contingent so nothing is required.
While ID is not new, I would not say it is merely negative campaigning against the theory of evolution. You note Dembski's design inference method. Obviously, that is more than mere negative campaigning. You ask what it means to be "intelligently designed" in the context of his inference, as though he has not specified it. He has.
- Cornelius
IP: Logged
|
|
Cornelius G. Hunter
Member
Member # 81
|
posted 03. March 2003 15:26
RBH:
RBH writes: " This example is a "design inference" by functional analysis. Roughly, it says that If the function (not "purpose") of the nucleotides implies that they be readily accessed and read by other biochemical entities in the cell, and if there is relative selective advantage for easy availability for that, then it is likely that they're sticking out radially."
But why does the design inference by functional analysis require the evolutionary concept of selective advantage? For as you said:
RBH writes: "To interpret Pauling's reasoning as embodying something more than 'a structure suggested by functional analysis of the hypothesized operating context' is to load it down with connotations it can't bear."
Yes, I quite agree. So why do you load it down?
Then you say:
RBH writes: "Every evolutionist I know is fully aware that biological structures can be described in terms of little d "design." All of them that I know accept - indeed assert - that natural physical processes can produce that little d "design." "
Of course they do, this is the point that is in contention. I can just rightly say:
"Every IDer I know is fully aware that biological structures canNOT be described in terms of little d "design." All of them that I know accept - indeed assert - that natural physical processes canNOT produce that little d "design." "
-- Cornelius
IP: Logged
|
|
Erik
Member
Member # 160
|
posted 03. March 2003 15:41
In order to keep this thread on topic, I request that my fellow ID critics do not challenge Cornelius G. Hunter's claims about the (lack of) merits of evolutionary biology in this thread.
I will reply later.
IP: Logged
|
|
Erik
Member
Member # 160
|
posted 03. March 2003 17:48
Cornelius G. Hunter, you did not tell me precisely how it is circular to formulate scientific models for electrical circuits without a model of intelligence. Concerning the old trick you describe, I note that a scientific model does not need to provide a complete description (indeed, Ohm's law does not describe the detailed motion of every electron in a resistor). After a sufficiently large number of observations of rolled die (or drawn cards) together with the associated (de)coded data, one can estimate the correlation coefficient between the decoding procedure's input and it's output and one would find that they are uncorrelated. Furthermore, the more data one obtains, the more simple models can be ruled out (e.g. after a sufficiently large amount of observations one should in principle be able to say that your decoding procedure is not, e.g., a finite state-machine with less than 5 states). After a while, a sufficiently intelligent victim of the trick would conclude that the model that the input (dice/card outcome) and output (decoded data) are not only uncorrelated, but also have a mutual information* of zero. Such a model will describe the trickster well.**
Although I don't agree with your claims about the lack of merits of evolution, I will simply grant them for the sake of the discussion. They are irrelevant to the subject of this thread (i.e. "ID advocates' view of intelligence"). We can discuss them elsewhere, if you want me to answer; in this thread I will simply avoid them.
You claimed that Dembski has specified what it means to be "intelligently designed". Either we disagree about how vague such a specification can be and still be interesting or you have studied more of his work than I have (or both!). Can you explain exactly what you are referring to? Also indicate whether or not you think Dembski's specification of what it means to be "intelligently designed" constitutes a scientific model. If your claim is true, it will provide a good starting point for discussion and comparison with the views of other ID advocates.
Erik
* Mutual information is the correlation coefficient's more sophisticated big brother. The former measures the amount of statistical dependence, whereas the latter only measures the amount of linear statistical dependence.
** I strongly doubt that humans can be non-ergodic without the aid of an explicit rule to follow.
IP: Logged
|
|
Cornelius G. Hunter
Member
Member # 81
|
posted 03. March 2003 19:49
Erik:
Before going any further, I should say that I admit your statement that you don't need have a "model of intelligence" in order to do science is probably very common amongst scientists. Therefore, the burden in this discussion is on me. Here's my point. When we do our science we implicitly make certain metaphysical assumptions about the world, such as the existence of rational and consistent relationships, call them natural laws, uniformity and simplicity in the natural world, etc. When modern science had its beginnings, people like Francis Bacon had no problem placing science within a greater metaphysical worldview which supported these assumptions. But in the 400 years since, we have gone through a transition where a great many scientists, apparently including yourself, espouse the notion that science is somehow untethered and free from any metaphysical assumptions. When you say you don't need a "model of intelligence" to do science, you are denying the very basis for your scientific work.
Also, this view falsifies ID from the beginning. If we don't need a "model of intelligence" in order to do circuit work, we certainly aren't going to need one in biology. This, of course, does not imply that affirming the historical and fundamental metaphysics underlying science validates ID. ID critics need not deny uniformity and parsimony, they simply need to show that when it comes to biology, any appeal beyond naturalistic mechanisms is superfluous. This does not entail the more heroic claim that a "model of intelligence" is unnecessary in science, period. IOW, you are going farther than you need to in critiquing ID, and in doing so, it seems to me, have gone out on a limb.
So regarding this thread, what I'm trying to say in answer to your initial post (ie, "how do you guys model the "I") is that there is no special definition particular to ID. What distinguishes ID from evolution is the claim that naturalistic mechanisms are insufficient to explain the species and the origin of life. We need not get into deep metaphysical puzzles and questions about what the "I" is. I don't think we need to get into the details of Dembski's work to address your initial post.
-- Cornelius
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 04. March 2003 00:30
There are a couple of problems with not having a specific definition of I (intelligence).
First, suppose that naturalistic mechanisms are insufficient to explain the origin of species and life. Is absolutely everything either "naturalistic" or "intelligent"? For example, why can't it be "emotional"? Maybe some primitive cosmic angst is responsible for the appearance of design. Or maybe it is an accidental byproduct of sneezing. There are a lot of non-naturalistic explanations out there. I assume that the word "intelligent" was chosen for a reason. And therefore it is important to say what it means in this context (and why one would suspect intelligence instead of something else--this gets back to Dembski).
Second, if we look at examples of "science" that involve intelligence--psychology for instance--we see (and no offense to psychologists here; they have a tough problem) an astounding lack of progress compared to the non-intelligence-involving sciences like physics and molecular biology. Intelligence has frustrated our attempts at progress. If you can't offer anything beyond "well, the design was intelligent", the overwhelming response of the research community--even if they agreed!--ought to be to keep assuming that it wasn't for as long as there was even the tiniest shred of hope that the design wasn't intelligent. So far, we basically hit a dead end when it comes to intelligence--and that is with our own intelligence, which presumably is inferior to the intelligence that did the designing. Telling people that something is intelligent without enough additional details so that some experiments can be done is tantamount to telling people to give up, that this is something that inherently cannot be understood.
IP: Logged
|
|
Erik
Member
Member # 160
|
posted 05. March 2003 17:10
Let's first get very general. Science can be done without any metaphysical assumptions beyond the assumption that the external world exists, if it is done in the spirit of regression analysis and data compression. This is a much firmer and assumption-free framework than trying to base a philosophy of science on certain metaphysical commitments. Go to the library (or follow the links) and read the following articles (read Schulte's article if you only have the time or motivation to read one):
Solomonoff R. (1964) "A Formal Theory of Inductive Inference", Information and Control, 7 : 1-22
(Go here for free access to the preprint.)
Wolpert D. (1996) "The Lack of A Priori Distinctions Between Learning Algorithms", Neural Computation, 8 : 1341-1390
(Go here for a preprint. Note that the ordering of the pages is inverted in the PDF, i.e. the last page appears first.)
Schulte O. (1999) "Means-ends epistemology", British Journal for the Philosophy of Science, 50 : 1-31
(Non-subscribers can go here for a preprint.)
Hutter M. (2000) "Towards a Universal Theory of Artificial Intelligence based on Algorithmic Probability and Sequential Decision Theory", cs.AI/0012011
Some of these articles attempt to establish special inference methods. But that's not why I cite them. None of them provide a complete account of science (the actual data collecting as well as the refinement and invention of experimental techniques are not covered). Some of them are not even primarily intended as contributions to the philosophy of science. But I do not cite them in order to provide a complete account of what science is. I cite them because they exemplify and lay-out the general framework for scientific theorizing: The universe is a data-generating machine and the scientist observes a stream of data. The scientist's task is to, based on the data stream, provide a model for predicting the as yet unobserved part of the data stream. Any model will do, as long as it is accurate and general. No silly metaphysical concerns about "explanation", "causes", "mechanisms", etc. are involved. Just regression analysis and data compression. The data stream that scientists observe may well have more a priori structure than Solomonoff's semi-infinite bitstring, some subtle constraints may have been placed on the data-generating machine (aka the universe) in some of the articles, etc., but my point is the general framework, not the details.
Your reply begins with the claim that I must make metaphysical assumptions to do science. Even if true, it does not follow from that that I must make metaphysical assumptions about intelligence in order to do science. And it certainly doesn't follow that need a scientific model of intelligence in order to do science. Your reply did not contain an argument for the first of these claims, and you didn't even try to support the third claim.
Now let's get specific. Suppose that there are independent ways to recognize certain components called "resistors". I can take such resistors to the lab and observe how they behave in various situations. In particular, I can subject them to situations where a measureable quantity called the "voltage over the resistor" (U) has particular values, and I can observe which values of another measurable quantity called the "current through the resistor" (I) that tend to occur for given values of U. Suppose further than after extensive experimental work, I find it reasonable to advance the following hypothesis and model:
"For every resistor, there is a parameter R such that U = RI. The value of R may differ from resistor to resistor, but it does not depend explicitly on time."
After some additional time in the lab, I note that the model is accurate for certain ranges of U's and I's. Were it not for the fact that Georg Simon Ohm beat me to it, I would have advanced a good scientific model. And I would have done so without the need for a scientific theory of intelligence. Be specific about your disagreements if you still disagree!
You also wrote: "If we don't need a 'model of intelligence' in order to do circuit work, we certainly aren't going to need one in biology." That statement strikes me as completely absurd. First of all, there is no a priori between electrical circuits and biological phenomena. Second, the following is our currently best model of electrical circuits:
It is an historical fact that Maxwell's equations were obtained without a scientific model of intelligence. It is also a fact that Maxwell's equations constitute a scientific model. Hence the absurdity of your claim.
You also wrote: "This, of course, does not imply that affirming the historical and fundamental metaphysics underlying science validates ID. ID critics need not deny uniformity and parsimony, they simply need to show that when it comes to biology, any appeal beyond naturalistic mechanisms is superfluous." The words "naturalistic" and "mechanisms" are not defined within the general framework for science. They are metaphysical concepts, whose meaning I am not certain of, but I think most people use the words to refer to processes that are familiar from currently existing scientific models. What ID advocates need to show is that they have a scientific model which provides a (i) description that is applicable to a wider class of phenomena than any current model, and/or (ii) more accurate than any currently available model. What ID critics need to show is that ID advocates have failed to do so (an easy task, since ID advocates have not even tried to formulate a scientific model).
Furthermore, you wrote: "We need not get into deep metaphysical puzzles and questions about what the "I" is. I don't think we need to get into the details of Dembski's work to address your initial post." If Dembski's work contains what you claimed it does, it is very relevant to this thread (i.e. "ID advocates' view of intelligence"; can it be described scientifically? etc.). I insist that we get into the details, first the scientific aspects (if any), and then the metaphysical ones.
Erik
PS. Since you think "naturalistic mechanisms" are what distinguishes ID from evolution, I'd like to know precisely how you define the term. DS.
IP: Logged
|
|
Cornelius G. Hunter
Member
Member # 81
|
posted 05. March 2003 22:21
Erik:
I took a look at Solomonoff and Schulte. Thanks for the links. We're talking about the metaphysics behind science. I said some metaphysical assumptions are required to do science, such as uniformity and simplicity. You said:
Erik wrote: "Science can be done without any metaphysical assumptions beyond the assumption that the external world exists, if it is done in the spirit of regression analysis and data compression." and,
Erik wrote: "The scientist's task is to, based on the data stream, provide a model for predicting the as yet unobserved part of the data stream. Any model will do, as long as it is accurate and general. No silly metaphysical concerns about "explanation", "causes", "mechanisms", etc. are involved. Just regression analysis and data compression."
I don't see how Solomonoff and Schulte support your claim. Consider a few circuit examples:
1) One scientist concludes V = iR, the other concludes the observed data are strictly a function of time since the Big Bang, and a very complicated function at that (the function is not analytical, but is a table lookup -- his fit is perfect). Which one is scientifically right, and why?
2) One scientist concludes V = iR, the other concludes V = iR^1.00013 – 0.00000082. The latter has a better fit. Which one is scientifically right, and why?
Now what I'm getting at is this. If you say you cannot answer these questions then you're saying you can't do science, because these examples are basic to science. If you do answer them then you will have to introduce some assumptions beyond merely that the external world exists.
Do you agree? If so, then Maxwell is like the 1st scientist who concludes that V = iR. Maxwell's eqns are not derived merely from the data. He would have an a priori metaphysical view of how things should work. And, I think, it is this view that allows someone like Paley, when examining the eyeball, to conclude that it did not arise from known laws; or someone today, when looking at the bat's echolocation system or the DNA code, to say evolution does not account for this. This, of course, could all be wrong. If there were strong evidences for evolution I'd suspect it is wrong, but the scientific evidence for evolution is not strong.
--Cornelius
ps--I can respond to your other comments, but wanted to talk about this topic first.
pps--I define "naturalistic mechanisms" as the effects of natural laws we deduce in the natural sciences.
[ 06. March 2003, 00:18: Message edited by: Cornelius G. Hunter ]
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 06. March 2003 00:42
You also have to assume induction if you want to explain why we care about science. Induction is disprovable: in the past it has worked in many cases, but it need not have. However, although this shows that assuming induction is consistent, it doesn't "prove" induction.
But if science is just the study of what models have successfully predicted, then you don't even need induction. In the past, inductive models applied to distant-past events predicted near-past events.
With that in mind,
(1) The scientist who concluded V=iR at time T correctly predicted V at time T+t; the one who concluded that V=f(T) did not predict V at f(T+t). Therefore, the scientist who concluded V=iR had the better model.
(2) Scientists note that the world is simple. V=iR is simpler than V=iR^1.00013 - 0.00000082. This motivates an error analysis; in the past, if something was within error of a simple quantity, future measurements tended to converge on the simple quantity, so we might write V=iR as our law, and report V=iR^(1.00013+-0.00027) - (0.00000082+-0.00000129) as our data. Which is superior? That depends on what you're trying to model, and how much work you are willing to do. E=0.5*mv^2 is a good model for some cases, and E=m0*c^2/sqrt(1-v^2/c^2) in others. Is either scientifically right? They're both good models. The first is simpler, the second is more accurate.
Where in here do we need a model of intelligence?
[ 06. March 2003, 00:43: Message edited by: Rex Kerr ]
IP: Logged
|
|
Cornelius G. Hunter
Member
Member # 81
|
posted 06. March 2003 01:53
Rex:
In order to answer the questions you had to make some assumptions. On the first question, you say S1 had the right scientific answer. You're right, how did you know that? You note that Scientist #2 (S2) failed to predict whereas S1 correctly predicted V(T+t). Not quite. Actually, the situation is that while S2 failed to make a prediction, S1's prediction was wrong. You claim the prediction was correct because you think that S1's prediction would fall within some statistical measure of tolerance that you selected. You are right, it did. How did you know that the measurement would fall within your tolerance? I think you assumed the circuit behaves uniformly over time. This is more than merely assuming that the external world exists. S2 did not assume uniformity – he had to wait until the measurement was made before updating his model. What do you know that S2 doesn't know?
On the second question, you like S1's answer that V = iR because it is simpler and because future measurements tend to converge on that quantity. What made you think that future measurements tend to converge on that quantity? I wonder, because in this case you are wrong. We did a whole bunch of subsequent measurements and they did not converge on V = iR. I can only conclude that you would guess that future measurements would tend to converge on V = iR because you believe V = iR is a good model. I think you're right. Perhaps you have a good sense of simplicity. S2 did not understand this. You know something that S2 doesn't know.
You, as a scientist, have some very important information about how the world works which helps you tremendously in analyzing things. As important as these things are, you have no defense for them, at least no scientific defense for them. They are dogmatic beliefs which we all hold. They are a model of how the world works which we have no justification for. We rely on them, but then after achieving success we deny them. Will you permit me to say you have a model of the rationale of the world?
--Cornelius
[ 06. March 2003, 02:50: Message edited by: Cornelius G. Hunter ]
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 06. March 2003 02:55
I will permit you to say anything you want, but I reserve the right to disagree with it. And I most assuredly disagree with your last claim there.
I already said what I knew that S2 didn't know. I assume induction. This isn't necessary for science; post-hoc we can see that inductive theories are the ones which made the best predictions, and never say anything about the future. However, I assume induction, noting that this is consistent with the evidence so far. If you prefer, you can call it uniformity. The two are equivalent.
With S1's answer, I'm not saying anything about future measurements. If you read my previous sentences carefully, you'll notice that the tense is important. I'm presuming that in the past, measurements had tended to converge to V=iR to within experimental error. I pick V=iR out of the possible values because it's simple and consistent with the data. (Also, I probably know of other simple rules that work remarkably well, and I suspect that V=iR may be a consequence of one of those; if I accept the simple form of the other, I get V=iR essentially for free.)
So, as I said before, I have a model, and the model is that the external world exists (which inherently includes concepts like identity), and that induction works. I note that this is consistent--if I'm especially clever, I note that given Godel's incompleteness theorem, I'm likely to be unable to do better than note a lack of obvious inconsistency--and get on with the business of measuring stuff.
I don't need extensive metaphysics to do science. And this is good, because there are many possible metaphysics that include induction, and we have no good way to distinguish between them. For example: you could have a unifying intelligence that has a personality and that personality is the source of the uniformity; you could have fundamental Logical Forms including the Induction Principle; you could have a system generated with minimum information via local update rules and changing the rules would require more information; you could have every possible world existing, but only the one where induction works keeps us alive, so when we trace backwards from this instant it has been inductive the whole way; and so on.
Luckily, you don't need to deal with any of this mess until you start explicitly leaning on, say, the existence of every possible world, or the universal intelligence, or whatever.
IP: Logged
|
|
Cornelius G. Hunter
Member
Member # 81
|
posted 06. March 2003 03:55
Rex:
You wrote: "If you read my previous sentences carefully, you'll notice that the tense is important. I'm presuming that in the past, measurements had tended to converge to V=iR to within experimental error. I pick V=iR out of the possible values because it's simple and consistent with the data."
But no, the past measurement, in fact, did not converge to V = iR either. They converged to V = iR^1.00013 – 0.00000082 if you use that functional form. Of course, you can restrict your function to V being directly proportional to i, in which case you'll get some constant times R. In fact, we reran the batch experiment 2,000 times, and not once did it converge on V = iR. Even when we combine all those results from the 2,000 batches, it still did not converge on V = iR.
You are right that V = iR was within the experimental error, but so were an infinity of other models. Why do you pick out V = iR as your solution?
Then you said that your assumption of uniformity (or induction) is based on past experience and that it is not necessary for science. I'm not following. How does science work without the assumption of uniformity? S2 in the 1st experiment was unable to formulate a prediction because he did not assume uniformity. How can science predict without uniformity? Secondly, I'm not clear on how you justify your assumption of uniformity based on past experience. What does that have to do with the future?
--Cornelius
[ 06. March 2003, 04:12: Message edited by: Cornelius G. Hunter ]
IP: Logged
|
|
Erik
Member
Member # 160
|
posted 06. March 2003 10:52
Cornelius G. Hunter, thanks for the more specific disagreements. Your first scenario reads: "1) One scientist concludes V = iR, the other concludes the observed data are strictly a function of time since the Big Bang, and a very complicated function at that (the function is not analytical, but is a table lookup -- his fit is perfect). Which one is scientifically right, and why?" The first scientist's conclusion has the proper form for a scientific model. If it is sufficiently general and accurate it will be a good scientific model, because it will allow us to compress (maybe not losslessly) our tables of observed V and i values into single parameters R and it will allow us to interpolate (and extrapolate) between the measured values (i.e. make predictions). The second scientist's conclusion, no matter how right, is not a proper scientific model, because it doesn't achieve any compression at all. Its applicability is also severely limited to the data we already know. One requirement placed on scientific models is that they should relate observable phenomena to each other so that we can compress (not necessarily losslessly) and recover (at least some of) the data in our lists of observations.
Your second scenario reads: "2) One scientist concludes V = iR, the other concludes V = iR^1.00013 – 0.00000082. The latter has a better fit. Which one is scientifically right, and why?" A minor point is that we can simplify the second model to V = iR' - 0.00000082, where R' = R^1.00013. Since R is a free parameter, we might as well estimate R' instead at less cost and greater utility (R would have a horrible unit). So who, scientifically speaking, is right? Both*! As has already been noted by Rex Kerr, which model we should prefer depends on how accurate and general we need and want to be. A scientific model should describe as much of the data stream as accurately and simply as possible, but it need not be true in any deeper sense than being accurate. There is only one truth, but there can be many accurate models, of which some are better than others.
You defined "naturalistic mechanisms" as "the effects of natural laws we deduce in the natural sciences". There are problems with this. The natural laws we deduce in the natural sciences are mere descriptions of the universe. They are what we write on pieces of paper and they have no effects. Perhaps you meant that the phenomena described by our natural laws constitute "natural mechanisms"? If so, and assuming that "natural laws" can refer to any scientific model (as opposed to scientific models of a certain kind), then you have just defined ID out of science (since ID, by a claim in your previous post, would then be the claim that certain biological phenomena cannot be described scientifically).
Erik
* Assuming that both models are sufficiently accurate and general (even one happens to be accurate than the other).
IP: Logged
|
|
Cornelius G. Hunter
Member
Member # 81
|
posted 06. March 2003 12:18
Erik:
In your answers to the two questions, you make assumptions about the world such as uniformity (ie, "it will be a good scientific model, because it will allow us to compress … and it will allow us to interpolate and extrapolate") and simplicity (ie, "A scientific model should describe as much of the data stream as accurately and simply as possible").
Regarding your statement: "You defined naturalistic mechanisms as 'the effects of natural laws we deduce in the natural sciences'. There are problems with this. The natural laws we deduce in the natural sciences are mere descriptions of the universe. They are what we write on pieces of paper and they have no effects."
Sorry, I meant that naturalistic mechanisms are the effects of natural laws *which* we deduce in the natural sciences. Could there be naturalistic mechanisms arising from natural laws which we have *not* deduced? Of course. Could we be wrong about a natural law that we have deduced? Of course.
You then claim I have refuted ID, presumably because you believe scientific models assume nothing more than "the external world exists," but this is the point in contention.
--Cornelius
IP: Logged
|
|
Erik
Member
Member # 160
|
posted 06. March 2003 13:11
Cornelius G. Hunter, you read my post too quickly. I wrote:
"If it is sufficiently general and accurate it will be a good scientific model, because it will allow us to compress [...] "
Observe that my statement is conditioned on the requirement that the model is sufficiently general and accurate. If that condition is satisfied then it will be a good scientific model. That is hardly an assumption about the universe, and much less a metaphysical assumption. If anything it was an assumption about your example, not about the universe. The second statement you misread as an assumption about the universe was:
"A scientific model should describe as much of the data stream as accurately and simply as possible, but it need not be true in any deeper sense than being accurate."
You apparently think that this amounts to the assumption that the world is simple. It does not. I mean that simplicity is an end in itself, not that simplicity is necessarily a good predictor of accuracy. Generality, accuracy and simplicity are all ends in themselves, and we try to make our models as general, accurate, and simple as possible. To a very good approximation, that's what scientific theorizing is all about. All other things being equal, use the simpler model because that is... well, simpler. Although quite good non-metaphysical a priori arguments can be advanced that choosing the simplest model is the best inference method, it is not necessary for me to commit myself to a particular inference method for the purposes of this discussion.
Furthermore, I did not write that scientific models assume nothing more than the existence of the external world. I wrote that no metaphysical assumptions beyond that are needed in order to do science. Specific scientific models may well make additional assumptions (e.g. time translation invariance, low speeds compared to the speed of light, ...), but even those assumptions are not metaphysical assumptions. They are assumptions about the content of the data-stream, rather than about the inner workings of the real world.
I don't see how your new statement about "naturalistic mechanisms", "natural laws", etc. is different from the original.
Erik
PS. I doubt either of us will be able to convince the other about science and metaphysical assumptions, so it seems like a bad idea to postpone the discussion of the highly relevant stuff (e.g. Dembski's view of intelligence) until we've resolved this issue. DS.
[ 06. March 2003, 13:14: Message edited by: Erik ]
IP: Logged
|
|
|
Printer-friendly view of this topic