|
Author
|
Topic: Information creation and transcendence
|
warren_bergerson
Member
Member # 262
|
posted 22. January 2003 08:59
Rex,
Quote: An individual organism does not respond to its environment using an evolutionary algorithm. The closest it gets is responding by reading off answers from its genetic code--but those are stored, and no novel information is created by reading off what is already there. So when you say
Whether you intended to or not, you have identified one or more of the central issues being discussed here. You are ‘ASSUMING’ that 1)some evolutionary process involving random variation and natural selection can generate information, and 2)there is no within lifetime process or mechanisms with a similar capacity to generate information.
We know/accept from analysis with evolutionary algorithms that a repetitive process involving variation, selection, and storage or preservation can find an adaptive solution from a set of possible solutions. As generally accepted and as I have formally defined it, these adaptive or teleological solutions represent information(where the volume of information can be quantified in terms of complexity or teleological complexity). Similarly, the force, intelligence or information processing required to generate/extract teleological information can be measured or quantified in terms of the ‘expected number of iterations or cycles needed to find a teleological solution using a ‘random variation/efficient selection’ process. [As far as I am aware, all mathematically precise definitions of biological information and biological information processing are variations of the definitions I have proposed. Gedanken certainly has not offered an alternative definition. ]
Your assertion that there is no within lifetime generation of information is based on the assumption or assertion that there is no type of selection other than Natural Selection. This assumption or assertion is clearly not valid. There is a lot of evidence for forms of selection which do not result in death or failure to reproduce. If there are forms of selection other than Darwinian Natural Selection and if the other components of teleological information processing exist, then biological systems have the capacity for generate information. [If you reject the possibility of within lifetime information processing and within lifetime information generation, then you end up claiming the Mona Lisa and Shakespearean plays must be coded in the genes.]
If you accept that biological systems have the capacity for within lifetime information processing/generation, then you ask the question what is the relative magnitude of the two types of information generating capacities. We know that Darwinian process involve a maximum of one cycle per lifetime (in fact the figure in much smaller). Within lifetime information processing capacity as I have demonstrated is far in excess of 10^1000 cycles per lifetime.
When viewed in terms of biological information processing, Darwinian theories are based on the claim/assumption/assertion that biological designs are generated by the 1 Darwinian information processing cycle per life and that the other 10^1000 plus information generating cycles play no role in the process. When viewed from the perspective of biological information (as defined) Darwinian theories seem nonsensical.
The discussion of front loading was in part an attempt to explain, or begin explaining the processes and mechanisms by which make the huge within lifetime information generating capacity possible. The claim is not the a multi-cellular organism performs 10^1000 processing cycles, but rather that the organism performs processing equivalent to 10^1000 cycles. There are in fact many relatively simple processes that can find complex adaptive solutions with relatively little actual 'processing'. If a logic machine capable of performing information processing is programmed or ‘front loaded’ with the appropriate search routine, the machine can find very complex, very improbable solutions very quickly.
Gedanken,
The fundamental flaw in your entire argument is that you have not provided a mathematically precise definitions of information and information processing. Your arguments bounce around making assertions that can not be evaluated because you can not be pinned down on what you mean. You started this discussion with the assertion that ID was in trouble because it failed to provide a precise and workable definition of information and information processing. It appears that your entire argument is flawed because of this same problem.
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 22. January 2003 12:40
Warren, on your previous post:
You say that there are “within lifetime generation of information”. I propose that Warren start or work within an existing thread to demonstrate how this conforms with actual observation of nature.
Irving said quote:
quote:
So Irving, I’m still not clear, are you saying that “conservation of information” only applies to some systems, or is it not a characteristic you accept?
It's not a concept I can accept or reject, since it's pretty meaningless to me without a firm definition of just what information is supposedly being conserved.
I agree completely with Irving here. Some, for example, claim that Dembski’s definition of “information” is equivalent to negative scaled entropy of physics (the “information” measure that has been associated with physics). If it is so defined, then it is indeed known to be “conserved” in the sense of the second law of thermodynamics, as it is equivalent. (Second law says entropy will stay the same or increase in an isolated system for example, so as negative “information” would stay the same or decrease.)
However Dembski claims that his is not simply a restatement of the second law. So my challenge is for someone to explain a definition of “information” that is one in which there is an observation of “conservation of information”. I will then be interested in the consequences of that definition. If no one can provide a definition of information or an argument for “conservation”, then I must assume that no consistent definition exists, just as Warren suggests I am saying.
Warren said:
quote:
Gedanken,
The fundamental flaw in your entire argument is that you have not provided a mathematically precise definitions of information and information processing. Your arguments bounce around making assertions that can not be evaluated because you can not be pinned down on what you mean. You started this discussion with the assertion that ID was in trouble because it failed to provide a precise and workable definition of information and information processing. It appears that your entire argument is flawed because of this same problem.
Warren, thank you for this, as it always helps to know where my writing might lack in clarity.
But it is not my place to provide the definitions used by a writer who is making a claim of ID. What I am doing is examining the properties of particular definitions. I ask: Does this definition have this property? Then we look at what the implications are if it does have that property. Then we look at the implications if it does not have that property. (I may have only covered some particular cases, and may go on to clarify some of those cases individually.) But my point will not be to take a particular definition, but rather to work with all definitions that fall within a particular class. (Surely the concept of working with all of an aspect that falls in a class is not foreign to ID concepts -- as this is a concept regularly used by Dembski.)
This is the reason that I have been asking people to tell what the properties of the definitions they use. Of course they may have misunderstood my question, or intentionally avoided it.
So people can read my argument and ignore it if they choose. But the property of implication is a well defined logic and mathematical concept. If we speak of a class of all definitions that have a particular property -- and there are implications from that property, then all definitions that have that property will have those implications. The only argument is to show that the generalization is wrong, or to show that the definition you wish to use does not have that property.
As I am out of time, I will return later with specifics, this time concentrating more on single issues rather than broad strokes.
In the mean time, the opening post talks about various possibilities. And Warren is the only person who has attempted to provide or point to a particular definition of “information”. (And I have problems with Warren’s concept based on lack of consistency with current day observation -- but this thread is not the proper place to argue that as it is the subject of other threads.) So I would like to see others either point to or define what they mean by “information”, and do so in terms that relate to the opening post. And likewise from Warren, I would like to know how his definition of information relates to the following questions:
Using your definition of information:
Does it have a “conservation of information” principle?
If so, explain it, and explain how one determines the amount of information in a current system, and how one justifies the “conservation” argument in the face of examples of apparent “information” being generated by a given system. (For example Dembski argues that “displacement” of the source of information allows for “conservation of information” in a computer algorithm, so that all the “information” that is output by that program was in fact already contained within that program. This is very closely related to the questions of the opening post, because the issue of “transcendence” and whether it is required is wholly dependent on whether we have a natural explanation understanding of what is proposed.)
Explain how “regress” affects the definition you are using. This means that “information” has a source. And that source can in turn have further prior sources. Does your definition deal in time series progression or flow of information? If it does not, then it cannot have a “conservation” principle of any sort, and thus is not subject of this particular thread. If it does, then I am interested in the rules or observations you have on that “regress” issue -- how does your understanding operate as information flows through levels of systems? (This will relate to the opening topic of “transcendence”, as “transcendence” is a concept that relates to generation of new information outside of natural physical principles, so first I want to understand the relationships of your particular definitions of “information” as how they relate under natural physical constraints.)
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 22. January 2003 17:24
Warren,
I think maybe I am finally starting to understand your point. I had (apparently wrongly) assumed that you were concerned about how much information was in the genome, and that an evolutionary algorithm would have difficulty accounting for it all.
But--correct me if I'm wrong--I think your actual point was supposed to be that biological organisms process a vast amount of non-hereditary information, and that it doesn't make sense for an evolutionary algorithm, with such limited processing capability, to be able to come up with a biological system that has enormous processing capability.
We might doubt that low-information-processing-capacity systems could give rise to high-I.P.C. systems based on a thought experiment. Can a slime-mold build a human? Can a human build a nearly omniscient god-like intelligence? Well, no. But this isn't a good analogy.
There are multiple indications that systems that have low-information designs can have higher-information outputs. For example, I mentioned earlier where I had a computer that had a design that encompassed no more than about 100MB of information, but had processed 10TB of information in its lifetime. Also, if we look at living organisms, the majority of information seems to be in the genetic code--1GB or so for humans, without intricate compression--and yet we process more visual information than that each minute.
Is your point that this disparity in genomic information content to environmental information processing capability requires "front loading"? If so, I would claim that the front loading occurred at the level of the fundamental physical laws of the universe. I'm not sure if it is possible to say anything meaningful in terms of design there, because both design and non-design processes have to be described in such untestable and imprecise terms. I doubt that I will be able to say anything meaningful either way, so I won't try.
If you take the fundamental physical laws as a given (and more or less constant), then I don't see any particular problem remaining.
Do you see instances of front-loading that occur after the physical laws of the universe have been set?
IP: Logged
|
|
Mark Elkington
Member
Member # 120
|
posted 23. January 2003 07:03
Gedanken,
quote:
My principle aim was to follow the various options -- essentially following the breakdown (or "flow chart") and analyze each option. And I'm sorry if you eventually wind up thinking I hijacked your thread to discuss a different issue, namely consistency of the ID positions given any particular answer to the breakdown.
Not at all, that was my aim in enumerating the options.
quote:
Now I think we completely agree "…human choices and behaviour must be nondeterministic in some way." (Though I'm not sure we would agree on the implications of your larger statement.) Specifically, I don't believe that Gould's "Non-overlapping magisteria" principle is entirely accurate. Gould points out that the "magisteria" of science and religion are "interdigitating in wondrously complex ways along their joint border." I think that the border is not a succinct and crisp line. I think that science can inform more on religion that than Gould (or even Pope John Paul) would like. And I also have come to understand that there can be influences of religion (including complete disagreement with) on the studies of scientists.
I can't help but suspect NOMA of being an attempt to marginalise theology. For me the border between science and religion, if there is one, is a circle, i.e. science lies wholly within a metaphysical framework. Of course, for much of the time science is done without the direct influence of this framework (e.g. atheists get the same lab results as Buddhists).
quote:
It is my firm belief in the principle of attempting to root out most philosophical biases of science beyond what we might call a "naïve realism" position, so as to gain the maximum amount of information from direct observation of nature. Science (or what is to be considered science) is for me something that must extend beyond any particular religious view, and that which is accepted as science must be accessible to those with varying religious viewpoints. So when one may take a combination of an initial philosophical/religious position, and combine it effectively with scientific-like observation to derive very powerful philosophical or religious implications (which must then depend on both the evidence from nature and the religious predicates), we must then be careful to not call these "science". My disagreement with Gould's NOMA is that I think that Gould and some scientists fail to recognize the extent to which some religious claims can depend on observational techniques as from or related to science and are highly logically developed -- and thus this boundary is not so crisp as was suggested. But let me then emphasize that any "scientific" position must then show itself to be based independently of any such religious assumptions.
As we approach the limits of the reach of science (be it cosmology, consciousness or particle physics), the questions seem to become increasingly philosophical/religious. Perhaps what's needed is a recognition by science of how close it is to this boundary in each instance. Equally, religion claiming truth should have no fear of scientific freedom.
quote:
Returning to Quantum Mechanics and its implications:
Modern physics has concluded that randomness is at the heart of reality, that so far there is no essential identification of "determinism" as one found in the older classical physics. I agree with Mark that an essence of "freewill" can be found in this. And I agree with Dembski that within that randomness, there can be a dimension of the expression of "transcendence".
However, within classical physics the lack of determinism is practical only (i.e. rapid divergence of nonlinear systems for small variations in initial conditions - chaos theory.) With what little understanding I have of QM, while it appears to provide a mechanism for real nondeterminism, the question of causal agency apart from randomness remains.
From memory, in NFL Dembski was I think suggesting that the QM-level intervention of a designer in biological evolution would be only historically detectable. I got the impression it was a back door mechanism (arguably of maximal elegance, unlike a crude moving of the particles), but of minimal laboratory observability. Interesting idea, though open to the charge of god-of-the-gaps.
I'm enjoying your perspectives here. One more thought, on the question of displacement and regress. Postmodernism has been criticised as almost laughably inconsistent for making the absolute claim that there can be no such claims. Many argue that a similar flaw is at the heart of naturalistic evolution. I think ID has intuitively seized on that, and is now struggling to break through and frame a strong hunches and half-theories in the form a testable widely and acceptable theory with working scientific utility.
Mark
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 23. January 2003 09:43
Gedanken,
Quote: In the mean time, the opening post talks about various possibilities. And Warren is the only person who has attempted to provide or point to a particular definition of "information". (And I have problems with Warren’s concept based on lack of consistency with current day observation -- but this thread is not the proper place to argue that as it is the subject of other threads.)
By current day observations, I assume you mean current day modeling of information. Current practices in biology misinterprets descriptive transformations such as ‘‘DNA’ changes to ‘living organism’’ as causal relationships. The suggestion or misinterpretation is that because such transformations are observed to recur they are causal relationships. This is in turn misinterpreted to mean that the transformation does not require generating additional information.
If this transformation is modeled (or if we model the portions of the transformation that we understand) using ‘teleological information’ mathematics, then we quickly discover that what might at first be described as an information neutral transformation does in fact require vast amount of information processing or information generation.
Teleological information and teleological information processing is, or at least is based on the concepts of information and information processing used in computers and logic machines. I believe you will find lots of ‘consistency with current day observations’ if you look at computer modeling and simulations of complex processes. .
I am making the claim that teleological information mathematics combined with the ‘dynamic and teleological’ interpretation of scientific determinism makes it both possible and practical to model, simulate, and reverse engineer the operations of biological systems. I have not yet found any ‘simple’ way of demonstrating or testing this claim. At least one of the problems in testing the claim is convincing biologists that their descriptive, non-causal transformation models are not rigorous scientific models and that the models are not adequate to construct predictive theories.
Quote: Using your definition of information: Does it have a "conservation of information" principle?
As I understand it, the conservation of information principle is a mathematical principle or property of closed systems. If you treat, view, or model a biological system as a closed system then such a system can not create information. Teleological information mathematics treats or models biological systems as open systems which interact with an external environment. In such an interactive system, teleological information is generated (the teleological or adaptive options are found) by interactive processes such as selection (not just Darwin’s Natural Selection) and information is destroyed when the external environment changes and the adaptive or teleological property of a causal relationship is lost.
You can demonstrate the creation, destruction, and conservation of information as defined in an abstract mathematical universe involving an ‘abstract interactive teleological information processing system’ and ‘an abstract external environment’. I can show you a logic machine program or ‘pseudo program’ that not only creates teleological information, but that also evolves and generates creative teleological solutions.
Quote: Explain how "regress" affects the definition you are using. This means that "information" has a source. And that source can in turn have further prior sources.
This is, I believe, actually two issues. First, it is the intuitively difficult mathematical concept of a system operating on itself, redesigning itself, and redesigning its own information processing/design capabilities. There are mathematical techniques for addressing this type of regress.
The second issue is the philosophical issue of absolute truth or absolute knowledge. To paraphrase, if science is the search for absolute truth and if absolute truth exists then there must exist ultimate source of information. My answer to this is that life, teleological information processing, and science are not about finding absolute truth but about finding from a limited currently available set of options the forms of truth/teleological information which are most likely to increase the chances of survival.
Quote: Does your definition deal in time series progression or flow of information? If it does not, then it cannot have a "conservation" principle of any sort, and thus is not subject of this particular thread. If it does, then I am interested in the rules or observations you have on that "regress" issue --
I am not entirely sure what you mean by the terminology you are using. As I defined it, ‘information’ is a point in time concept measured in terms of teleological complexity and ‘information processing’ is the set of complex operations responsible for transforming information. Information processing would be denoted or represented by logic machine programs. Information processing would be the operation of logic machines using such programs.
There are no theoretical problems with observing and measuring point in time information. ‘Observation’ of information processing, like the observation of all causal relationships is indirect. The technique of observing teleological information processing is the engineering or reverse engineering standard. In simplified terms, a model or simulation is valid or acceptable if the model can match or exceed the performance standard of the phenomena being observed.
As I have mentioned elsewhere, if you use the dynamic and teleological interpretation of scientific determinism, then the reverse engineering standard is the scientific hypothesis testing standard.
Quote: This will relate to the opening topic of "transcendence", as "transcendence" is a concept that relates to generation of new information outside of natural physical principles, so first I want to understand the relationships of your particular definitions of "information" as how they relate under natural physical constraints.)
ID, as I understand or interpret it, is the claim or observation that biological design is the product of some ‘intelligent designer’ or ‘intelligent design process’ other than those described or explained by traditional science. A not insignificant portion of ID advocates recognize that ‘intelligent design processes’ may have natural physical basis. I would define ‘transcendence’ to mean information or designs generated by causal processes or natural laws other than the permanent and universal laws recognized by traditional science.
I think it is important to distinguish between ID as metaphysics and ID as science. The topic introduced by Mark involved what I would consider metaphysics. When you introduce the idea that ID is flawed by the lack of a precise definition of information, then you are addressing not only ID as a science, but also biology as a science. It is, IMO, inappropriate to find fault with ID as a science based on criteria that can not be satisfied by conventional science.
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 23. January 2003 12:05
Rex,
Quote: I think maybe I am finally starting to understand your point. I had (apparently wrongly) assumed that you were concerned about how much information was in the genome, and that an evolutionary algorithm would have difficulty accounting for it all.
Relevant to this thread, I am saying that the definitions of information and information processing used in evolutionary biology are flawed and/or inadequate. I would even go so far as to suggest that evolutionary biology lacks a mathematically precise definition of information processing.
Evolutionary biology in one form or another claims that the genome contains the information needed to produce the living organism. Evolutionary biology suggests that the transformation from DNA to living organism does not require the generation of information. Biologists can not demonstrate that such an information neutral transformation is possible. I claim, and can easily demonstrate that such a transformation requires generating vast amounts of information(as I define information and information processing).
Evolutionary claims that the ‘information’ in the genome can be transformed or evolved based on a one cycle per lifetime process. Because the definitions used in biology are mathematically imprecise, it is not possible to either demonstrate the validity of this claim or to test and disprove the validity of the claim. Using my definitions of information and information processing it can be demonstrated that modeling or simulating known changes in the genome requires the equivalent of very large numbers of within lifetime processing cycles.
Quote: Also, if we look at living organisms, the majority of information seems to be in the genetic code--1GB or so for humans, without intricate compression--and yet we process more visual information than that each minute.
You have offered several good arguments supporting my claim that the volume of information associated with a life form is far greater than the information in the genome. You also sight examples of systems expanding information processing capacity. The definitions I proposed simply make it possible to quantify the amount of information at different points in time and to quantify the amount of force/intelligence/processing needed to produce different volumes of point in time information. Applying the quantification rules I provide, it is easily shown that biological systems involve far more information and far more information processing than suggested by neo-Darwinian theories.
It is somewhat interesting/amusing how easily people can move between the precise and complex concepts of information and information processing used in discussing computers, and the simplistic, metaphysical concepts of information used in evolutionary biology. People are willing to accept on the one hand that information processing in our primitive computers is highly complex, but information processing in biological systems must be extremely simplistic. Not only are most scientists/academics unable to see the inconsistencies in these positions, but they will argue strongly that no such inconsistency exists. Interesting, at least IMO.
Quote: If so, I would claim that the front loading occurred at the level of the fundamental physical laws of the universe.
This is an interesting position since it clearly contradicts what we know from the study of man made logic machines. Think of the genome as representing initialization values and front loading as the starting point computer program. I am suggesting that both the starting values and the starting point program(front loading) can be modified or can evolve in order to change the organism which will develop. [ If development can not be modeled as a simple transformation, then it would seem the only logical alternative is to view or model a biological system as a self programming logic machine. ] Once you start analyzing/modeling biological systems as self programming logic machines, it seems fairly obvious that both the starting values and the starting point programs can and must evolve.
Quote: If you take the fundamental physical laws as a given (and more or less constant), then I don't see any particular problem remaining. Do you see instances of front-loading that occur after the physical laws of the universe have been set?
I think it is often easy to forget that ‘fundamental physical laws as more or less constant’ is an assumption developed to make science ‘practical’, not a statement of reality. It can be demonstrated that causal relationships can take, or can be expressed in many logial/mathematical forms including dynamic and teleological (this is based on mathematical transformations of the relationship ‘A always causes B’.) The issue is not whether it is always necessary to express the laws of nature as permanent and universal, but whether scientific analysis can be productive and practical expressing the laws of nature in some alternative logical/mathematical form.
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 23. January 2003 12:34
Mark,
Thanks for your comments.
I think that your point about systems with supersensitivity and chaos responses being taken as “randomness” is extremely important. Even Dembski recognized this in a particular way -- in that he calls “randomness” things that include probability derived from “law”. However Dembski makes mistakes in claiming an absolute rather than relative probability -- as probability must always be measured with regard to a descriptive starting state. All events, due to the “supersensitivity” issue would be of probability less than 10^-150 if calculated from a starting point of the Big Bang -- without regard to QM. The important point is that “randomness” of classical models is simply a lack of knowledge of the details. (e.g. a die roll could be analyzed by the physics of inelastic collisions from the point of leaving the thrower’s hand. What makes it “random” is that the 6 outcomes are equally likely given supersensitive variations of the pathway leaving the thrower’s hand when averaged over those initial states -- not that the thrower could not in principle determine the outcome with sufficient control and knowledge.)
quote:
From memory, in NFL Dembski was I think suggesting that the QM-level intervention of a designer in biological evolution would be only historically detectable. I got the impression it was a back door mechanism (arguably of maximal elegance, unlike a crude moving of the particles), but of minimal laboratory observability. Interesting idea, though open to the charge of god-of-the-gaps.
Yessss indeed! This is Dembski’s “back door” attempt to save his concept from the difficulties of regress. The biggest problem is of course that it implies “descent with modification” is entirely correct. It is entirely incompatible with the explanatory filter and irreducible complexity arguments, which must demonstrate that such subtle information gains are insufficient. But this is where there is conflict! If humans are not to be considered exempt from requirement of consistent relationships in nature, then the information issue of human thought must be considered. If thought (by humans) can produce “design,” and the quantum mechanics connection of nature to ID must take a length of time of historical proportions so as to make it unmeasurable -- then we have the difficulty of explaining human thought! In this case it must be related to the computer algorithm in information and regress terms (e.g. “creativity” issue is equally applicable.).
Warren,
If modern science is missing out on a “teleological” perspective in its theories of how biological systems operate, then how about the misinterpretation by ID of how evolutionary systems are equally “teleological”?
Once again I think that RBH gave an important perspective in “Front Loading”, and Rex Kerr’s questions need to be answered somewhere. A new perspective on how systems operate in today’s world will easily be amenable to verification by observation. Are the consequences of your concept all consistent with direct evidence from observation of nature (in today’s world)? Any new concept needs to be checked in a great deal of detail for such consistency. I suggest proposing or working toward ways to make those checks, however difficult they may be (in a thread appropriate to the detailed subject). There is no value in a concept that cannot be verified against observation of the subject it discusses, even if it must go beyond “science”. Even religion has a degree of verifiability by experience, though that might be very different from “scientific” verification.
quote:
As I understand it, the conservation of information principle is a mathematical principle or property of closed systems.
Bingo!
But this is not entirely true -- and that is the issue I am calling “regress”. Because information flow (using for example an “entropy” definition) can be studied. The “information” definition as negative entropy is analyzed both from a “flow” perspective and from a “closed system” perspective. It obeys a form of “conservation of information” principle in the sense that no new information is created but is rather transformed, and is based on aspects of “energy”. Indeed the “information as a form of energy” perspective may be lacking in utility for understanding biological systems. But the point is that there are no violations of the “conservation” principle of the second law of thermodynamics (and note that it is not a normal “conservation” principle as in conservation of energy, as it states that negative entropy will stay the same or decrease within an isolated system boundary.)
In the paragraph that follows you made my point about information processing system of evolution. Since the population is not “isolated” form the environment, it can potentially pick up information from the environment -- and this principle would be true in absolutely any “conservation of information” principle, it does not depend on how it is defined as it is inherent in the “conservation” notion. If a principle is a “conservation” principle, it obeys this property -- I don’t need to have a precise definition of the principle to make that observation; as all “conservation” principles have that aspect precisely as part of their definition.
quote:
You can demonstrate the creation, destruction, and conservation of information as defined in an abstract mathematical universe involving an ‘abstract interactive teleological information processing system’ and ‘an abstract external environment’. I can show you a logic machine program or ‘pseudo program’ that not only creates teleological information, but that also evolves and generates creative teleological solutions.
Dembski admits somewhere that the “intelligent designer” could be the processes of evolution themselves. (Thus admitting precisely what is stated here -- a complex system can have the potential to do a tremendous amount of “information processing”.) This points to other failures of definition -- if we are to understand that an external designer is implied by an observation, then if the proposition is that there must have been either an external or an internal design process we have not made any progress in making that distinction. (In other words the “explanatory filter” in that regard has not provided any clarification.)
quote:
This is, I believe, actually two issues. First, it is the intuitively difficult mathematical concept of a system operating on itself, redesigning itself, and redesigning its own information processing/design capabilities. There are mathematical techniques for addressing this type of regress.
But remember the “open system” issue above!
quote:
The second issue is the philosophical issue of absolute truth or absolute knowledge. To paraphrase, if science is the search for absolute truth and if absolute truth exists then there must exist ultimate source of information. My answer to this is that life, teleological information processing, and science are not about finding absolute truth but about finding from a limited currently available set of options the forms of truth/teleological information which are most likely to increase the chances of survival.
Science has provided many ways of increasing human survival. (Medical knowledge, knowledge of effects on environment, etc.) Therefore the scientific method is very important. This does not deny that other ways knowing and understanding are also important, just affirms the importance of that particular way.
Here I choose to examine propositions made in ID from a philosophical standpoint. Regress may have a mathematical definition, but the propositions of ID have a metaphysical aspect that an “unembodied designer” is suggested for part of what are called ID “explanations”. So I am examining the ID proposals from this philosophical standpoint. We still have a requirement of obeying basic rules of logic in rational thought -- ID is not exempt from the basics of rational thought.
quote:
Quote: Does your definition deal in time series progression or flow of information? If it does not, then it cannot have a "conservation" principle of any sort, and thus is not subject of this particular thread. If it does, then I am interested in the rules or observations you have on that "regress" issue --
I am not entirely sure what you mean by the terminology you are using. As I defined it, ‘information’ is a point in time concept measured in terms of teleological complexity and ‘information processing’ is the set of complex operations responsible for transforming information. Information processing would be denoted or represented by logic machine programs. Information processing would be the operation of logic machines using such programs.
There is a lot to discuss in this and what you said following this.
First of all we can use logic machine processing as a model for certain physical activity, but it is an incomplete model. It does not model all aspects of the physical systems and thus is both abstract and incomplete in its relationships to physical nature. Western reductionism uses the technique of abstraction to find simplifying principles -- and this is where such modeling can be important. But when the model does not fit reality, it can often be explained as due to the incompleteness of the model. (Consider a simple classical model of planetary motion -- when it does not fit accurately, we consider missed elements. Then finally we had to include relativity. Etc. etc. etc. These are not indications that the original abstraction was not extremely useful -- it is simply an example of the application of the “art” of science.)
quote:
As I have mentioned elsewhere, if you use the dynamic and teleological interpretation of scientific determinism, then the reverse engineering standard is the scientific hypothesis testing standard.
I’m lost. I’ll probably regret quoting this. “Scientific determinism”? …
quote:
… A not insignificant portion of ID advocates recognize that ‘intelligent design processes’ may have natural physical basis. …
quote:
I think it is important to distinguish between ID as metaphysics and ID as science. …
But ID proponents often demand that “unembodied designers” be accepted as part of the “flow chart” of possible “explanations” for the apparent “deisgn”. This puts a metaphysical perspective on the entire ID presentation that it cannot escape from, even by recognizing the “or” of “physical or metaphysical”. When a claim is made that an observation implies “design” somewhere in the regress, and that must have come somewhere either in the “physical” or the “metaphysical”, and furthermore the method does not make for a way to observationally distinguish which case is operant -- then we have a problem with observability and confirmation that the method is “reliable”.
The method can never be verified by observation, because each failure of the method could have been a “metaphysical” origin case and poses no falsification. Yet whenever this is challenged by critics, the ID writers point to how there are “physical” pathways of design and “thus” the method is subject to falsification. The indecision on this appears often to be due to a desire to make the concept appear to be viable from a political perspective, rather than a willingness to come to grips with essential logical problems.
quote:
… When you introduce the idea that ID is flawed by the lack of a precise definition of information, then you are addressing not only ID as a science, but also biology as a science. It is, IMO, inappropriate to find fault with ID as a science based on criteria that can not be satisfied by conventional science.
Biology and evolutionary theory do not depend on a single precise definition of “information”. If you have a particular definition that you want to apply, and can demonstrate observationaly that some understanding is developed, then you may be contributing to science. But you have to demonstrate a specific inconsistency of the scientific theory in terms of its logic, or in terms of conflict with observation of physical nature, in order to have a valid concern about past statement of science. (These by the way occur regularly in scientific journals.)
I’m not sure where there is a failure of “conventional science” to be consistent with observation or internal logic of its claims.
For example, descent with modification can be viewed loosely as an “information” problem. The “natural selection” is a process of taking information from the environment and affecting the population. I don’t see why a more precise definition is needed to make that observation. If one is offered, and it is found useful, then it may be valuable -- but where is there a conflict demonstrated in the statements of science? You will have to be specific.
My complaint is not that ID lacks a precise definition of “information”. ID in fact has many so-called “precise” definitions of types information. These definitions are the basis of ID claims -- thus making the definition key to the claim being made. My problem is that these definitions are changed from point to point in the argument -- and that the implications of a given definition and portion of the argument offered are not examined for their implications.
It is my proposal that ID writers should examine their writing for such inconsistencies and distinguish various types of “information” with distinct names. (Also terms like “complexity”, etc.) This way the reader can know when the definition has changed in working through the argument, rather than giving a false appearance of a logical argument when none has been offered.
IP: Logged
|
|
Mark Elkington
Member
Member # 120
|
posted 24. January 2003 07:38
Gedanken,
quote:
Yessss indeed! This is Dembski’s “back door” attempt to save his concept from the difficulties of regress. The biggest problem is of course that it implies “descent with modification” is entirely correct. It is entirely incompatible with the explanatory filter and irreducible complexity arguments, which must demonstrate that such subtle information gains are insufficient.
Very good point. Using only QM back door insertion of information to design/create irreducibly complex structures, the designer would need to induce a large minimal set of mutations in a single generation...Hopeful monsters ID-style! Might be worth raising this in a new thread (I must recheck the relevant NFL chapter on this too).
quote:
But this is where there is conflict! If humans are not to be considered exempt from requirement of consistent relationships in nature, then the information issue of human thought must be considered. If thought (by humans) can produce “design,” and the quantum mechanics connection of nature to ID must take a length of time of historical proportions so as to make it unmeasurable -- then we have the difficulty of explaining human thought! In this case it must be related to the computer algorithm in information and regress terms (e.g. “creativity” issue is equally applicable.).
Yes... Where have we got to in this thread?
Mark
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 24. January 2003 09:47
Warren said:
quote:
Once you start analyzing/modeling biological systems as self programming logic machines, it seems fairly obvious that both the starting values and the starting point programs can and must evolve.
If by "starting point program" you mean "existing organism", then I agree. The genome of an animal is not sufficient to reproduce that animal--you need a specific physiology (a pre-existing animal) as well.
In terms of evolution, you can think of life as a fitness evaluation function, f(G,P) where G is the genome of the organism and P is the physiology of its parent. (The dependence on the environment is implicit.)
Now we have a question: how much information is in G, and how much in P? We can gain some insight into this question by looking at the relative frequency of lethal mutations to sterile and maternal effect lethal mutations.
A lethal mutation in a genome G' causes f(G',P)=0, because the organism dies (typically before being born/hatching/whatever). A maternal effect lethal often has f(G',P) = f(G,P), i.e., no first-generation cost in fitness. However, when the organism grows up into a parent P', then its progeny all die: f(G',P') = f(G,P') = 0. There are also classes of sterility mutations that have the same effect except that the progeny are not produced rather than dying.
It turns out that both lethal and sterile/maternal effect lethal mutations are known, confirming your claim that both the starting values (genome) and starting point program (composition of the parent) are important. I don't have exact numbers off the top of my head (or easily accessible), but it turns out that the ratio of lethal to sterile/M.E. lethal mutations is very large. If there were vastly more information in the configuration of an organism than in its genome, you would expect the ratio to be reversed: since organisms are built from genes, mutations would be likely to mess up the organism's configuration, rendering it unable to produce progeny.
Furthermore, there are few known instances of non-genetic heritability; again, if there were a vast excess of information in the organism's configuration apart from what is in the genome, you would expect to see the transmission of that information from generation to generation show up in non-genetic heritable traits. (There are a very few of these, but they are quite rare.)
These two observations is obviously far from conclusive, as it may just be that organism configuration has a different fitness landscape than genome identity. But I can't think of a better line of evidence to assess the contribution of each component.
quote:
It is easily shown that biological systems involve far more information and far more information processing than suggested by neo-Darwinian theories.
I have already given a calculation showing that the information in the human genome is consistent with the amount of information gain per generation that you suggested (1 bit/55 generations or somesuch). Based on the relative number of lethal mutants to sterility mutants, I would suggest that adding in the information saved in organismal configuration is relatively small compared to what is in the genome. (Especially since the genes specify the proteins out of which the configuration will be/was built!)
There is a somewhat separate issue of organisms being able to process vastly more information in a lifetime than exists within their genome. Note that this does not involve the "creation" of heritable information. Rather, the organism is simply filtering or transforming the information in external input into behaviors and/or developmental regulation. There is no problem here unless you can show that the input from the environment is insufficiently information rich to account for the information in organismal behavior and development. I don't think you were claiming that, so I will not address it further.
Finally, as I've said before, note that there is nothing inherently contradictory about a simple architecture having vast processing capability. For example, my polarized sunglasses seem extremely simple, but they perform a vast photon-sorting function based on the polarization of the incident photons.
IP: Logged
|
|
Irving
Member
Member # 535
|
posted 24. January 2003 10:39
Gedanken,
quote:
The important point is that “randomness” of classical models is simply a lack of knowledge of the details. (e.g. a die roll could be analyzed by the physics of inelastic collisions from the point of leaving the thrower’s hand. What makes it “random” is that the 6 outcomes are equally likely given supersensitive variations of the pathway leaving the thrower’s hand when averaged over those initial states -- not that the thrower could not in principle determine the outcome with sufficient control and knowledge.)
Aren't "supersensitive variations" just another way of phrasing "lack of knowledge?" Isn't then, everything determined by the initial state of the Big Bang? If human thought is governed by the same principles, then isn't this a pre-destination versus free-will issue? Now why does that sound familiar...
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 24. January 2003 11:25
Irving,
quote:
Aren't "supersensitive variations" just another way of phrasing "lack of knowledge?" Isn't then, everything determined by the initial state of the Big Bang? If human thought is governed by the same principles, then isn't this a pre-destination versus free-will issue? Now why does that sound familiar...
That’s why quantum mechanics’ essential randomness (that physics cannot peer beyond) is still important. In the “randomness” vs. determinism issue, at the quantum level (usually microscopic but not necessarily), physics can only make a prediction of a distribution of events. Combine this with supersensitivity -- e.g. a macroscopic event is dependent on a quantum variation -- then we have a clear case that macroscopic behavior is not deterministic within the statistical lack of knowledge of quantum events. (That’s assuming that we somehow magically knew the entire quantum state of the universe or system at a particular point in time, for a particular reference frame.)
What this means is that there is a degree of determinism predicted by physics, but as time progresses the known pathway diverges to a greater extent.
Now here is an example where quantum events could be easily understood to interact with something that we could probably do a fairly accurate job of predicting classically. We have a well-characterized surface, and a virtually perfect and well-characterized pair of dice. In addition there is a Geiger counter on the table surface. If there is a pop, it will mechanically deflect the surface of the table for a period of time to come. Let’s assume that the dice are thrown hard with some sort of precise measurement of their trajectory after leaving the player’s hand. Classically we might be able to do pretty well at predicting the bounce pattern if the table had no latent vibrations. But we don’t know the state of the latent vibrations of the table -- because quantum events effecting the Geiger counter cannot be predicted (even though we can know the number of counts per second statistically.) Supersensitivity of the bouncing of the die to table vibration state can easily affect the outcome, making our prediction ineffective. Thus quantum mechanical randomness in our thought experiment affects the macrostate of the “pseudorandom” throw of the dice, making it reasonably random.
This was just a thought experiment, and the number of quantum events that can effect various macroscopic motions is great. Effectively, our lack of knowledge of the specific outcomes of quantum events makes the prediction of long-term physical process progress impossible, even in principle. But this does not mean that short term prediction of macroscopic processes is difficult. For example understanding of the short to mid-term electrophysical processes of neural systems can undoubtedly be done with fairly good predictivity.
But a most important point is that we are dealing with “in principle” quantum variations that are quite predicable within a statistical distribution. If there was an effect of some external influence on the quantum states that had an aggregate effect, this could “in principle” be measured, and would constitute a violation of the prediction of the distribution of quantum mechanics. So if there is some sort of “quantum antennae” in parts of the brain for some sort of external influence -- this can in principle be measured as a violation of the predictions of quantum mechanics!
But remember that quantum mechanics is still only giving a distribution, so that individual events are not predictable, only ensembles of events at the quantum level.
So we ask “Is thought deterministic?” The answer depends on what you mean. If you mean “is thought completely predictable?” Then surely quantum events affect supersensitive outcomes and make complete predictability of thought processes impossible, even in principle. However if you mean “is thought understandable in principle from a physical level (if it is a result of physical brain processes)?” Then the answer is that thought could be understandable in principle. Being understood does not imply deterministic or complete prediction of outcomes. (Evolution theory is like this -- it constrains outcomes but does not entirely determine them.)
This reminds me of a Catholic philosopher who taught a philosophy class I took. He made a point was that if we were not free to choose our actions in a rational and predictable manner that we would not have “freedom” of will. Predictability (within limits) of behavior does not imply a loss of free will -- and this is not a matter only of having a complete neural map of the brain processes, it includes cases like simply knowing how humans react and use logic and self-interest and predicting from those characteristics.
[ 24. January 2003, 11:31: Message edited by: gedanken ]
IP: Logged
|
|
Irving
Member
Member # 535
|
posted 24. January 2003 20:57
quote:
That’s why quantum mechanics’ essential randomness (that physics cannot peer beyond) is still important.
Hmmm...just because we don't know, doesn't mean the particle doesn't. But if it only matters what we know, then all that we know is only what matters.
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 25. January 2003 09:53
Rex,
I compliment you on at least attempting to demonstrate the validity of your position. The analysis, however, is fatally flawed by the fact that it is based on at least three assumptions that are known to be false. First, you assume the environmental conditions or the fitness evaluation function as you call it, is stable over the life of the organisms. It is easily demonstrated that the actual environmental conditions under which an organism exists change rapidly and frequently and that in many instances if the organisms fails to adapt to changing conditions it will die and fail to reproduce. Second your analysis is based on the assumption that both genetic and evolutionary change are gradual and steady. The evidence clearly shows this is not valid. Finally, your analysis assumes that there is an identifiable functional relationship or mapping from genotype to phenotype and from phenotype to genotype. No such mappings are known or have been identified. The analysis of developmental processes clearly shows that development requires the generation of large volumes of teleological information. If development increases the volume of information, then the existence of a genotype to phenotype functional mapping is a mathematical impossibility.
In the business world it is considered misleading to use assumptions which are known to be incorrect and which are known to distort analytical results. I have always found it curious that that standard does not appear to be applied in evolutionary biology.
You raise the question of between generation heritability which is, of course, central to the difference between the ‘low volume of information Darwinian models’ and my ‘high volume of information models’. I think we would agree that the information involved in between generation heritability in most multi-cellular organisms is contained in the single cell formed when male and female sex cells combine.
In terms of information processing, this cell contains 1)stored genetic material, 2)a physical processing environment(hardware) and 3)a processing logic or program (software). Standard Darwinian genetic theory suggests that all the information needed to operate a complex organism is contained in the genetic material. Based on our existing knowledge of information processing in computers, it is easily demonstrated that the functionality observed in a complex organism such as a human or a mouse could not be defined by 30,000 relatively simple pieces of information.
What is the alternative? My suggestion is that the vast majority of information needed to produce a living multi-cellular organisms is generated within the organizations lifetime. We know that during its lifetime a multi-cellular organisms creates very large volumes of new information processing hardware by creating new cells. We also know from observing cells that the processing logic or software can be and is modified because we know that organisms create many different types of cells that interact with the environment in many different ways.
In order to demonstrate that my approach is feasible, we need to demonstrate that 1)it is mathematically possible for simple single cell information processing systems to evolve/develop into complex multi-cellular processing systems in a single generation(not a difficult mathematical task) and 2)demonstrate that given a limited amount of stored genetic information, the developmental/evolutionary process can be steered so that similar organisms are generated generation after generation(identifying all the steering mechanisms will be difficult, but I am certain that we are aware of some such processes).
Viewed from the high information perspective, there are numerous complex processing mechanisms that could account for rapid and complex ‘evolutionary changes’. If you begin looking for such mechanisms from the high information perspective then the mechanisms should be easy to identify. The significance of the mechanisms would be lost when viewed from the low information perspective.
The issue in this thread is ‘definitions of information and information processing’. While some are concerned with the metaphysical aspects of information, I am concerned with the mathematical and scientific implications. Starting with mathematics, we can define low information systems which express conventional Darwinian and genetic concepts. Using relatively standard concepts from computer science, it is also possible to define high information, rapid processing systems that both generate large volumes of teleological information and which also have the capacity to rapidly increase their information processing capacities.
Given both high information and low information mathematical systems, the question from a scientific perspective is which type of mathematical system is more effective and useful in modeling biological systems. It is easily demonstrated that models based on low information mathematics are not adequate(this require misleading and unrealistic assumptions in order to model or simulate evolutionary changes).
I claim, although I have not yet demonstrated it, that high information mathematical systems, based on the teleological definitions of information and information processing can be used to model, simulate and develop predictive theories of life forms.
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 25. January 2003 10:49
Warren,
I’m sure that Rex will give a much more complete and relevant answer, but I couldn’t let some of these statements stand unchallenged.
quote:
In the business world it is considered misleading to use assumptions which are known to be incorrect and which are known to distort analytical results. I have always found it curious that that standard does not appear to be applied in evolutionary biology.
Warren apparently wishes to follow Jonathan Wells’ approach in the book Icons of Evolution. For some of the claims essentially follow that book. Most important is the backhand charge of “fraud” that Wells levels, by making a comparison to the business world, and then using strawman arguments and innuendo to convince the reader that biologists are guilty of such fraud. Notable also is following Wells approach of claiming that genetic material does not determine for the largest part the form of the organism.
quote:
The analysis, however, is fatally flawed by the fact that it is based on at least three assumptions that are known to be false. First, you assume the environmental conditions or the fitness evaluation function as you call it, is stable over the life of the organisms. It is easily demonstrated that the actual environmental conditions under which an organism exists change rapidly and frequently and that in many instances if the organisms fails to adapt to changing conditions it will die and fail to reproduce.
Clearly all biologists recognize that the conditions of the environment change from second to second. A carnivorous predatory animal notices another animal moving about in its immediate environment. Clearly this is changing from second to second. A chase ensues. As the animals run, they use their vision to guide their motion -- and feedback from their visual systems helps adapt their motor activity to the actual “landscape” of the land itself during the chase (by way of the nervous system).
So the environmental conditions (fitness) are changing second by second in the sense that the prey may escape or the predator may make a catch based on feedback loops that each uses to adapt to the immediate environmental “landscape”.
No biologist has failed to observe this, nor failed to describe it as important aspect of fitness. But there are averages of these conditions over the life of the organism, and mathematical aggregation by averaging is a perfectly valid scientific methodology. Since “fitness” is a function that is measured over populations, the fact of individual adaptation by learning and feedback capacity of the organisms in no way invalidates the well defined technique of mathematical averaging.
quote:
Second your analysis is based on the assumption that both genetic and evolutionary change are gradual and steady. The evidence clearly shows this is not valid.
I don’t see the words “gradual” or “steady” in Rex’s statement. Biologists have long recognized that evolution is not entirely gradual nor steady -- and concepts like “punctuated equilibrium” have been introduced as part of the biologists presentation of that fact on the larger scale (e.g. variations in evolutionary rates over large numbers of years as compared to even larger time periods.)
But then Warren apparently wants to disagree with mathematical averaging techniques and the law of large numbers.
quote:
Finally, your analysis assumes that there is an identifiable functional relationship or mapping from genotype to phenotype and from phenotype to genotype. No such mappings are known or have been identified. The analysis of developmental processes clearly shows that development requires the generation of large volumes of teleological information. If development increases the volume of information, then the existence of a genotype to phenotype functional mapping is a mathematical impossibility.
Ah, yes, Wells’ claim that genetic information does not determine the biological form. Warren, the lack of a complete mapping does not negate that the physical characteristics of an organism are almost completely determined by genetic information. Just thee first link in Google for “genotype phenotype” here. So Warren, are you denying the validity of the entire study of genetics (from Mendel on)?
An organism has an accident right after birth, and looses part of a limb. Clearly the grown organism will most likely be missing that limb. This is not part of the genetic information of the organism, rather it was a clear environmental influence on the development of the organism. So what? It does apparently make a genotype to phenotype functional mapping a mathematical impossibility in some strict mathematical (philosophical) sense, doesn’t it?
The problem here is trying to enforce some sort of strict absolute conformance of a rigid mathematical model onto a changing world. The biological models cannot be held to such a rigid standard, they are approximations of the real world. The real world has so much complexity that no generalization is going to be a complete 100% fit. But the biological evolutionary models fit a vast grid of evidence within very reasonable experimental tolerances.
[ 25. January 2003, 11:11: Message edited by: gedanken ]
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 25. January 2003 12:02
warren wrote quote:
The analysis, however, is fatally flawed by the fact that it is based on at least three assumptions that are known to be false. First, you assume the environmental conditions or the fitness evaluation function as you call it, is stable over the life of the organisms. It is easily demonstrated that the actual environmental conditions under which an organism exists change rapidly and frequently and that in many instances if the organisms fails to adapt to changing conditions it will die and fail to reproduce.
He is here committing the same error that I remarked on in another thread (that I cannot locate at the moment), the error of identifying the individual organism as the appropriate unit of analysis when evolutionary biology (and population genetics) takes the population as its unit of analysis. Evolutionary theory is a theory about change in populations, not (directly, at least - see below) about the moment-to-moment behavioral adaptions of individual organisms.
Moreover, no one I know thinks that selective environments (or environments in general) are static, and in a thread some weeks ago I responded to warren's strange belief that biological adaptive processes - genetic changes in the biology of an organisms - must somehow directly account for adaptation on every time grain from seconds to millenia. There are perfectly respectable disciplines that study adaptation on short time grains. Moreover, evolutionary biology and associated disciplines are interested in how those short-term adaptive strategies themselves evolve. As Gedanken correctly points out, that environments vary has not gone unnoticed by biologists and other life scientists.
Another place warren goes astray is his apparent assumption that organisms process every bit of information flowing in their environments. As over a century of research on sensory and perceptual systems shows, they don't. The sensory and perceptual apparatus of various organisms has itself evolved so as to severely constrain the amount of environmental 'information' organisms must process. Moreover, an enormous amount of the actual processing is hard-wired, so to speak, in the structures of the sensory systems. To sketch just one illustration, a human retina has approximately 125m receptor cells - rods and cones. By the time visual information has reached the ganglion cells that send it to the lateral geniculate nucleus of the thalamus that number has been reduced to approximately 1m. A whole lot of 'processing' is built into the parallel architecture of the retina, into the way the retina is wired and cross-wired from layer to layer, and that structure - that architecture - is itself the product of an evolutionary process. So to say that biology ignores those sorts of problems is simply and flatly false.
RBH
[ 25. January 2003, 12:14: Message edited by: RBH ]
IP: Logged
|
|
|
Printer-friendly view of this topic