|
Author
|
Topic: Information creation and transcendence
|
Mark Elkington
Member
Member # 120
|
posted 02. January 2003 20:13
ID postulates that there is a type of information which cannot be created in significant quantity by purely naturalistic processes, e.g. the information present within organisms. Call it "information created by an intelligent designer," Id. The following is an attempt to unpack the implications of this.
For simplicity of argument, assume the process of human reproduction to be purely naturalistic.
Scenario
The Earth is as we know it, but without any life; its total information content, Id, equals E.
God creates plant and animal life, including Adam and Eve; Id = E + L.
Eve gives birth to Cain and Abel. Has Id increased? Shannon information has, but Id has not, by definition: Whatever information change is associated with Eve’s offspring it is not Id-type information, since it was produced by the naturalistic process of human reproduction.
Extending this logic, the Earth with a human population multiplied to 6 billion still has Id = E + L.
The output of human history is not only people but also art, science, literature, technology, music etc. We are designers, and this legacy is our collective design work. Taking this into account, what then is the total Id for the Earth?
Possibilities
1. Id <= E + L
Human creative endeavour does not create the kind Id-type information we are interested in. Humans are therefore not "intelligent designers" of the kind required by ID. On what basis then do we claim to be able to identify and analyse the work of an intelligent designer?
Further, this implies that the intelligent design embodied in living things is qualitatively different to that produced by human invention. How would such a distinction defined? Measures of complexity, functional density, etc. could be used for quantitative distinctions.
OR
2. Id > E + L
2.1 Human life and work is composed of purely naturalistic processes
2.1.1 Id has increased through purely naturalistic processes
2.1.1.1 There is no minimum complexity threshold for naturalistic Id generation
Therefore in principle there is no universal informational objection to other naturalistic processes creating Id, e.g. unguided Darwinian evolution. By induction, arbitrarily complex natural structures and processes can generate Id. The only question then is a secondary one, i.e. Can this Id be naturalistically accumulated and concentrated indefinitely?
OR
2.1.1.2 There is a minimum complexity threshold for naturalistic Id generation
This threshold may be the human brain. That is, such a structure could not have evolved naturalistically (no bootstrapping), but once created is able to generate new Id.
OR
2.1.2 Id has increased with divine help
We can't create Id by definition, but it is present in our art and science, therefore God has been directly intervening and injecting Id throughout history.
OR
2.2 Humans are partially transcendent
Human life and work includes supernaturalistic processes. In some way we transcend the material universe, enabling us to supernaturally create Id. It seems we must conclude that our minds are not purely material, but have an active component and existence beyond neurones.
Conclusion
All intelligent designers must be transcendent for the information argument of ID to work. Either way this has interesting implications for the status of human designers. But is this flowchart correct and complete?
Even if humans are bona fide generators of Id, is there a conservation law which says that we can never create Id greater than or even approaching our own embodied Id? That is, we could never build a computer smarter than ourselves.
Mark
[ 02. January 2003, 21:27: Message edited by: Mark Elkington ]
IP: Logged
|
|
Daniel Gustaf Swedlund
Member
Member # 633
|
posted 16. January 2003 22:20
The whole scenario starts at a given point that lacks in its axiomity. Entering the postulate from the premise that E=Earth already exists, makes the further scenario strictly Sci-Fi inclined.
Given that ID= information created by an intelligent designer, as you stated,obviously refers to a creator that in this case might be called god.
So the initial ID is by definition outsourced by the ID property of god,and therefore the whole discussion is turned metaphysical.
Seriously speaking, the nature of the whole quest does by its presentation more likely be an input for a new computer game. Here your given reply might be: Of course, what did you think? And the answer is that I’m only giving you new leverage for expanding your ideas. ![[Wink]](wink.gif)
/Rutherford
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 17. January 2003 07:45
There's a basic point there that needs to be addressed, though:
Can humans create complex specified information?
If humans are natural, and natural processes can't create CSI, then humans can't--or at least we can't do anything fundamentally different than what the rest of nature can do.
But if this is the case, we cannot non-circularly use human works to determine what the hallmarks of design are, if we take these hallmarks to include qualities that cannot be produced by nature.
Does, therefore, the idea of design presuppose that humans are (in some way) transcendent?
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 17. January 2003 10:26
Somewhere I read a very interesting suggestion: That the NFL theorems would also limit the capacity of “search” that we humans do in finding answers to problems (e.g. “design”).
If you think about it, you realize that in our mental design processes, we go through possibilities for working systems that we have “thought” of (e.g. came up with by a process of modification of things we have thought of before or searching lists of modifications of things from our experience), then we mentally simulate those ideas to test their quality. When we find an idea that seems to have some merit, we focus greater effort on either mental simulation (including even modeling in physical or computer) and then test the idea in the real world. (The only real difference from this and evolution is that in evolution the new pattern is constructed first, then tested immediately rather than using “forethought” to test the new pattern.)
So if NFL theorems did limit the process of evolution, why does it not limit the process of evolution of designs? Alternately, does this not conclusively show that the possibilities of new configurations, both of devices and systems of human design and of biological complexity are very broad, such that the possibilities of new complexity working is “reasonably” within grasp?
This is a very important question being asked here. Is human thought somehow beyond the capacity of physical systems? (For if any physical system can perform the equivalent thought processes in “design” or “information creation”, then that demonstrates that “information creation” is a capacity of physical (and therefore understandable) systems.
Since this can be observed in the near term as Artificial Intelligence research progresses, we will possibly see many answers to the questions of “transcendence” in our lifetimes. (In other words important but ignored implications of ID claims will be tested.)
Now when asking about whether human thought is “beyond” (or “transcendent” of) physical systems, the obvious question of “displacement” of the source of that capacity will come up. What is extremely important here is that this question, as posed, is not a question of the “source” of the capacity! (We are not discussing the question of where that capacity came from. We are not asking whether or not the “CSI” has been “smuggled” into the system, or by whom. Assume for purposes of this discussion that whatever is of interest has been “smuggled” into the system in question.) The question is of whether a physical system can have that capacity, and this will be answered as we observe the design of AI systems evolve. The question is of the mechanisms by which systems retain “information” that will be later realized by the system, and of the implications of those mechanisms to other ID claims.
Now of course there will be the inevitable challenge that AI systems cannot come up with anything new and thus are not directly creating “new information”. As I have pointed out in various threads, this becomes a problem of definitions. If the AI system appears to be learning from its environment, and appears to create new designs, and appears to produce concepts that have not previously been “thought” of before -- is that “new information” or not? This is a question of definition of what is meant by “new information”.
For indeed if an AI system should start regularly producing statements (or other transmittable thought or concept representations like images) that have never been thought of before, and we claim that those are not new information then we have simply defined the concepts of information and “new information” so that they where imbedded in the original system. If this is our definition of those terms, then how do we distinguish the claim that microevolution is not producing any “new information”, or that extended construction of new forms by evolutionary processes would constitute “new information”?
How does the concept of “displacement” of a source of information actually work (if it is indeed a meaningful concept)? If a computer system creates apparent “new information”, yet the source of that “new information” is displaced to a prior source, that implies that the “new information” was embodied in the computer system for a period of time. How can it be embodied in the computer system (or other physical system such as a proposed human intelligence operating by physical processes only) before the new concepts to be output by that system are actually realized?
I propose that the computer AI system being considered for an example is not limited by strict deterministic algorithms, but rather that it embodies a true quantum random number generator. Thus at any point that a programmer would have requested a pseudorandom number generator, that the actual true quantum number generator is substituted by the hardware and compiler, so that the computer system is not limited to strict deterministic algorithms constraints and is non-deterministic. So we are not proposing that there is no “new information” simply because of deterministic output limitations of repeatability. With that, I believe that ID advocates still regularly claim that the AI system’s output is “displaced” to the creator of the programming. So the question of the last paragraph should be taken in this context, how does a non-deterministic process that generates apparent “new information” do so if it does not by definition already contain that unrealized potential information?
Does this not then conflict with the concepts of information as realization of potential?
When Dembski says “The fundamental intuition underlying information is not, as is sometimes thought, the transmission of signals across a communication channel, but rather, the actualization of one possibility to the exclusion of others,” is not that “actualization” of one possibility out a choice of selections made by the AI system somehow a creation of “new information”? And if it is not, how are we defining our terms so that the AI system already contains that “information”? (Remember we are potentially using a quantum random number generator and the AI algorithm is not deterministic, so the question must be asked as if the program being observed were recompiled on such a machine.) And if the AI or other physical system does not already contain that information, is the AI system therefore “transcendent”?
When the claim is made that the “information” is thus already contained in the physical system that will eventually output the apparent new information, how does this differ from the physical system simply being the “communication channel” through which the signals are transmitted? If information can be contained in such a system (as the “displacement” concept implies), then the system simply is equivalent to a model container of information -- or a communication channel -- and Dembski’s claim is turned on its head by his own descriptions!
Now change the system being considered above to a simple present day genetic algorithm performing a design task. But include the true quantum random number generator. Clearly these algorithms are creating what we would call “new designs” and thus apparent “new information”. Ask the same questions about how the computer system thus embodied contains all the potential output that it can generate over all time in its original boundaries -- and the implications of that claim or definition of terms.
If a system does not already contain all the information it (potentially) will output, then there are serious failures of the concept that “information” being considered (or defined) is actually quantifiable. If information is not quantifiable, it is not measurable, and fails as a scientific concept in the way that it is being presented in the ID community.
I think that the concept of a physical system already containing all the information that it can potentially output is extremely important to understand! And the implications for this thread are direct and profound.
I personally think that ID is a theory in crisis, because it cannot find consistent answers to these questions.
Please be careful and consistent in how you define and use terms in this thread!
[ 17. January 2003, 14:28: Message edited by: gedanken ]
IP: Logged
|
|
Micah Sparacio
Member
Member # 6
|
posted 17. January 2003 16:23
Hey Rex,
Welcome to Brainstorms!
I wanted to poke my nose in and say that, ever since reading No Free Lunch, the question that you raise has been lingering in the back of mind...and I've yet to come to terms with it.
I think I've raised the question in public before...not sure, but this is where I see the problem:
How do we know that the information trail (that Dembski wants us to follow in genetic algorithms) doesn't stop at the human being as source. How do we know that we're not just a complex algorithm in which all the information that comes out has been programmed in?
If we can't know that, then what can the foundation of CSI be?
My line of thinking on the issue is relfected, to a certain degree, in my latest paper ( Mental Realism )
I try not to think of the issue in the typical Natural versus Non-Natural or Physical versus non-Physical manner. Rather, I argue for an expanded physical ontology in which the mental is a 2nd order physical property which constrains and channels physical causal chains towards target end states. Now here is what I'm thinking about CSI: first order properties and their relations, left to their own devices, will not readily produce what Dembski is calling CSI. However, if we expand our physical ontology to include irreducible, 2nd order physical properties, one of which is the mental, then we can understand CSI as being the property of certain end-states of complex first-order physical systems which have been guided towards their end-states by a particular 2nd order mental property to satisfy an end goal.
What the above does is distinguish between physical properties: those which can create CSI and those which can not. A purely reductive physical ontology (everything is reduced to first-order physical properties), on my reading of the world, will see no CSI. If we expand our physical ontology to include irreducible 2nd order physical properties, then we can come to terms with the mental and its unique causal role in the world.
[ 17. January 2003, 16:25: Message edited by: Micah Sparacio ]
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 17. January 2003 17:16
Gedanken,
I am in complete agreement that the effectiveness/success of ID or design science depends on a precise and workable concept/definition of information. I would also agree that a workable definition of information should address information from human and non-human systems in a logically consistent manner.
IMO, there are three key issues associated with defining information. The first issue is the fundamental nature of information. The second issue is the question of quantifying information and information generating/transfer capacity. The third issue is the question of who has the authority to develop and impose a specific definition.
On the specific nature of information, there are at least two major options. Information can be defined as some type of a yet undiscovered particle or form or energy. Second, information can be defined as a mathematical/logical concept which is useful in the analysis of biological systems (and possibly other phenomena). My opinion is that there is currently no evidence for a information as a form or energy. Information is thus a definable mathematical logical concept, and the definition used should be the definition which is most useful for scientific analysis.
Since I view information as a mathematical concept used in modeling, I believe the quantification of information is based on what is useful in analysis. The first feature to be quantified is the volume of information at location at a point in time. I define this point in time volume of information in terms of the teleological complexity of causal relationships. For a specified causal relationship this is the ratio of total number of possible forms N to teleological forms Nf. Note that this is a relative rather than absolute measure. (relative to the causal relationship being evaluated and relative to the test being performed.)
The second feature of information I quantify is the volume of processing required to produce a transformation between the volumes of information at two points in time. Typically, the volume of processing would be defined in terms of a search process where the processing volume is in units of number of ‘variation-selection’ cycles required using a random search routine. The measure of processing volume would be the same even if the actual processing is based on something other than a random search process.
The measures of information and information processing capacity I use are based on computer or logic machine search routines. The measures as defined are used to evaluate the amount of information and or the amount of information processing required for simulations to match or exceed the performance of a biological system with respect to a specific operation. As explained elsewhere, these are the types of measures needed to evaluate predictive theories of the general type "Under ideal conditions, the biological design process X will produce design the design Y which optimizes the chances of achieving goal G."
The final issue, IMO, is who has the authority to define information. IMO, the individuals developing and applying a type of analysis have the right and responsibility to develop definitions of information relevant to the applications being considered.
SUMMARY
If any one is interested in what I call the design science approach to defining information I will be glad expand on the above outline. I believe my approach is both mathematically and scientifically sound and I believe it is useful. These definitions are intended for use with predictive theories based on the dynamic and teleological interpretation of scientific determinism. The definitions appear to be useful in analyzing evolutionary processes, information processing in nervous systems, and in defining and analyzing human behavior including human creative behavior.
Definitions are precise and meaningful in the context or framework of how they are used. The definitions of information and information processing capacity (intelligence) have been problematic, IMO, because formulating mathematical, predictive theories has been a problem. The definitions I offer are, I believe, precise and meaningful in the context of theories representing dynamic and teleological causal relationships.
IP: Logged
|
|
Irving
Member
Member # 535
|
posted 17. January 2003 18:12
Gedanken,
quote:
If you think about it, you realize that in our mental design processes, we go through possibilities for working systems that we have “thought” of (e.g. came up with by a process of modification of things we have thought of before or searching lists of modifications of things from our experience), then we mentally simulate those ideas to test their quality. When we find an idea that seems to have some merit, we focus greater effort on either mental simulation (including even modeling in physical or computer) and then test the idea in the real world. (The only real difference from this and evolution is that in evolution the new pattern is constructed first, then tested immediately rather than using “forethought” to test the new pattern.)
IMHO this slightly misses the point by focusing on the immediate. The issue is to distinguish "Intelligent" design, from "ad hoc" design, or "Natural" design, et. al. The accent is not on design, but on what is "Intelligent." What is it that distinguishes intelligence from the natural world, and how is it manifested.
The natural word deals with the immediate, intelligence can envision a complex future state not yet realized. Evolution adapts, Intelligence plans. An intelligent design is one that anticipates future states, not one that reacts to the current state.
IP: Logged
|
|
Irving
Member
Member # 535
|
posted 17. January 2003 18:27
Gedanken,
Further thoughts on your AI issue... Let's take a look a simple piece of predicate logic:
a) Moses is a Man.
b) All men Die.
The AI system can deduce:
c) Moses will die.
Now, is this "new" information, or a new insight into existing information? All the information was already in a & b, it just wasn't explicitly stated until assembled into c.
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 18. January 2003 00:18
Before I reply further, could readers give consideration to Mark Elkington’s questions, and specifically to Rex Kerr’s question “Can humans create complex specified information?” (By Rex’s question, I would understand it to mean “new” CSI that is not already embodied in the human -- in other words do humans disobey the “conservation of information” law that is being considered?)
My bottom line is a question of the “displacement” problem. The only way for a human (or other natural system) to generate any “new information” or CSI, is for that human to be “transcendent,” and specifically for the aspect of generation of that information to be “transcendent”. Specifically, each decision that is part of “new information” must depend on that “transcendence”, and cannot be a decision that is made by the physical (or understandable) aspect of our mind, or else the “displacement” problem applies, and that information must have already been part of (contained in) our system, it must be part of the information content of our system.
We learn more and more about making AI systems produce complex operations, including “planning” operations.
We also learn more and more about how human nervous systems work, and how the brain works. To the extent that we gain further understanding of how specific aspects of thought occur in the physical brain, are you willing to accept that the “displacement” problem applies, and that the information of those plans and decisions and “designs” are thus required to already be present as “information” that is present in the human system?
Or are you willing to accept that even those aspects of the human brain (as a physical natural system) that become understood are simultaneously “transcendent” (and can violate “conservation of information”)? If so, would computer systems or other natural systems also have aspects of “transcendence” (and violate “conservation of information”)?
And for those who wish to not think of the human decision making capacity as a capacity of the brain and of a physical system, why is our decision making capacity affected by alcohol and drugs?
(I am in no way denying the soul or any other transcendent aspect of consciousness -- I’m just saying that large components of our brain and decision making capacity have known physical aspects and that we will understand those physical aspects in greater detail over time. Therefore the CSI that comes out of that physical aspect must obey the constraints of the “displacement” problem if “conservation of information” is to hold. It is the consequences of the “conservation of information” and “displacement” problems that I am interested in. And “conservation of information” is the consequence of the statement that “no new information” is generated -- either the information must already be present in the system that outputs it, or else it is “new”.)
Give these issues some consideration and discussion. I will respond later when I have more time.
[ 18. January 2003, 01:26: Message edited by: gedanken ]
IP: Logged
|
|
Irving
Member
Member # 535
|
posted 18. January 2003 01:11
quote:
2.2 Humans are partially transcendent
Human life and work includes supernaturalistic processes. In some way we transcend the material universe, enabling us to supernaturally create Id. It seems we must conclude that our minds are not purely material, but have an active component and existence beyond neurones.
My current choice...though I'm not sure what the author exactly means by the final clause.
quote:
Can humans create complex specified information?
If humans are natural, and natural processes can't create CSI, then humans can't--or at least we can't do anything fundamentally different than what the rest of nature can do.
But if this is the case, we cannot non-circularly use human works to determine what the hallmarks of design are, if we take these hallmarks to include qualities that cannot be produced by nature.
Does, therefore, the idea of design presuppose that humans are (in some way) transcendent?
Yes, humans can create CSI. I don't think the issue at this time is whether or not natural processes can create CSI. The point advanced by Dembski is that even if natural processes could, there hasn't been enough time.
quote:
We learn more and more about making AI systems produce complex operations, including “planning” operations.
You'll need to explain this one a little more...
quote:
Or are you willing to accept that even those aspects of the human brain (as a physical natural system) that become understood are simultaneously “transcendent”? If so, would computer systems or other natural systems also have aspects of “transcendence”?
If life has been Intelligently Designed, then it is not a "natural" system. It may be manufactured via "natural" processes just as all human creations are created via natural processes (chemistry, physics, etc...). Ergo, computer systems are not "natural" systems.
A question for gedanken:
Is Lawrencium (periodic element 103) natural? If so, can it be created via a natural process?
IP: Logged
|
|
gedanken
Member
Member # 594
|
posted 18. January 2003 02:13
Irving:
“I don't think the issue at this time is whether or not natural processes can create CSI. The point advanced by Dembski is that even if natural processes could, there hasn't been enough time.”
But is this by definition, or is it something that has to be demonstrated with evidence from other principles? I believe that according to Dembski’s terminology it is by definition, because he defines CSI as something that cannot be created by a natural system with any sufficient probability. In fact the low “probability” is precisely what is required to fit the definition of CSI, so of course any CSI, by definition cannot be created in a reasonable time period by a “natural process” due to the low probability. I also agree this is not an issue (because it is defined to not be an issue and I won’t argue with whatever definitions you want to use for your terms) -- rather I just want to know if these are the definitions you proceed with, as I will intend to discuss the consequences of those definitions..
“You'll need to explain this one [AI planning] a little more...”
Here, I just suggest reading the journals and conferences on artificial intelligence “planning”: Proceedings of the Artificial Intelligence Planning and Scheduling Conference. Also search for any associated journals -- I will not plan to discuss this further.
By this, and prior posts, I assume you mean that a computer program is not a “natural process”. Since natural processes cannot produce CSI, the only systems that can produce CSI are those that are not “natural processes”. Furthermore (by this and prior posts) I assume that you accept that a computer program and other systems are limited, that they cannot produce more CSI than is embedded in them by their creator -- in other words “conservation of information”. All the CSI that a computer program (not a “natural process”) can output is that which is encoded by the intelligent agent that encodes the information into the non-natural process. Very specifically, a given program contains as part of its embodiment all the CSI that it can produce, and only that specific CSI, and that the “information content” of that program must include all the CSI that it can (even potentially) produce as an output.
“Is Lawrencium (periodic element 103) natural? If so, can it be created via a natural process?”
Of course Lawrencium can by synthesized using high energy techniques in a human laboratory. But I would expect that Plutonium and Boron are plentiful in the Earth’s Sun. Clearly there are a lot of alpha particles in the reactions in the Sun. So I would expect Curium would be produced naturally in the Sun. If that is so, there could be small amounts of Californium created, since there very clearly is a lot of high energy helium in the sun. At least one isotope of Californium has a half life of 898 years, and another 2.6 years, so I would expect there to be Californium available. With that, if the Boron is at high energies I would then expect Lawrencium to be produced naturally in the Sun. Of course we don’t see any Lawrencium in any of the “spinoffs” of the Sun or other suns (as most planets are), because its half life is at most 3.6 hours and would have been gone long ago. The other heavy elements (the ones with long half-lives) are still present on Earth. If it is not produced in Earth’s Sun, it is surely so in some other higher energy sun, as most of the heavy elements near to those numbers are clearly so produced. In any case, to see it on Earth it must be synthesized in a high energy physics lab -- but of course when the conditions are set up the Lawrencium naturally occurs.
But taken this way, how do we know that the Earth itself is a “natural” system? It could have been part of a beyond-natural event long ago (just as is being considered for life itself), even as far back as the creation of the universe. (Just as you take life, that can be seen to reproduce by natural means, must have been created by a non-natural process in some far back point in time.) Can you dispute that? So if the Earth is not known to be “natural” (any more than you consider human life), then what are the restrictions on what information the Earth and its environment can produce? (And of course, what are the restrictions on what a system that can reproduce by natural processes can create?)
So is your point that once a system is created by an intelligence (say the first life, or even a human life) then that system that was created by intelligence may be capable of itself creating new CSI beyond what it was initially endowed with? “Yes, humans can create CSI.” (And by that I take that the CSI was not simply already part of the information contained in the human, because in that case the human would not be “creating” the CSI.)
But now your argument is inconsistent! That’s my point. (That there are inherent contradictions in what ID is saying, and what we can observe, and that there are inherent contradictions in the different aspects of ID claims.)
Because if an intelligence can create another system, and that other system can then create CSI that is not already embodied in that system, then a computer program (created by an intelligent human) can create new CSI that is not embodied in the computer program initially! Or the reverse, just because the computer program was created by a human, if it cannot create any more CSI than was placed into it by the human, then what is the reasoning that a human (even created by a prior intelligent agent) can create any information that is not already inherent in the human to start with?
What I really want is for people to think through their definitions and what they are saying carefully.
[ 18. January 2003, 03:24: Message edited by: gedanken ]
IP: Logged
|
|
Irving
Member
Member # 535
|
posted 18. January 2003 08:53
Gedanken,
quote:
What I really want is for people to think through their definitions and what they are saying carefully.
Great Point, though things should be read carefully also.
quote:
By this, and prior posts, I assume you mean that a computer program is not a “natural process”.
So far so good.
quote:
Since natural processes cannot produce CSI, the only systems that can produce CSI are those that are not “natural processes”.
Possibly correct. If that is how CSI is defined, then it is axiomatic. However, I thought the working defintion was:
Specified complexity is displayed by any object or event that has an extremely low probability of occurring by chance, and matches a discernable pattern.
Therefore CSI treats information as an Object and can't be removed from probability and chance in it's definition.
quote:
Furthermore (by this and prior posts) I assume that you accept that a computer program and other systems are limited, that they cannot produce more CSI than is embedded in them by their creator
Here's where we go wrong. In this and prior posts I don't believe I ever said this. Remember, I consider computers Non-natural systems. I also consider that Non-natural systems might be able to produce CSI. One of threads in this forum is about the ev program, and whether or not it produces CSI. I'd have to look at each computer implementation on a case by case basis. Hence my statement that I'd have to look at what you are specifically referring to as AI to make a comment. While no expert, I've worked on and developed a few AI applications and studied many a journal and conference proceedings related to the topic. As a forum on Complexity, Information, & Design, I'd love to investigate this topic further.
In another thread, I'm investigating the idea that uniformity in an environment susceptible to randomness is an indicator of design. Computers are inherently non-random. Therefore a computer acting in a random environment could create CSI in that environment.
Since the original post limited Id = E, I asked the Lawrencium question in the context of Id = E. This discussion seems a very metaphysical and semantic discussion, and I was attempting to solidify a definition of the term "natural."
Thanks!
[ 18. January 2003, 08:55: Message edited by: Irving ]
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 18. January 2003 09:36
Gedanken encouraged participation in this thread by suggesting that the failure to properly define information constituted a serious problem for ID. Since he introduced the subject, let us recognize that the failure to properly and precisely information and information processing capacity is a serious problem for science as it is practiced today and especially for the biological sciences.
From work with computers we now have reasonably reliable methods of defining and measuring how much information and how much information processing capacity are required to perform certain types of tasks. We also know that there are a vast array of techniques which can reduce the amount of raw processing capacity required.
Our knowledge of biological systems is far from complete, but we do know that biological systems involve operations that require very, very large volumes of information and information processing capacity (intelligence).
We don’t know exactly how biological systems perform certain complex types of information processing, like painting the Mona Lisa, or writing a Shakespearean play, but we do know of certain raw processing techniques which could produce these complex designs (like the monkeys with typewriters). Based on these crude techniques, we know that the amounts of information and the amounts of raw information processing capacities required for these operations are vast.
Despite well established knowledge of the complexity of the information processing performed by biological systems, we have scientists claiming to offer serious explanations of these phenomena in terms of processes with minuscule processing capacities (Darwin’s one cycle per life time and the Maxwell demon with the ability to generate one bit of information every 55 lives). We have scientists suggesting that the change from genetics to the full grown organism is a simple transformation, when we have abundant evidence that such transformations require vast amounts of information processing/information generation.
If anyone is interested in a serious discussion of information and information processing in biological systems, then 1)you have to recognize the vast body of knowledge accumulated on the subject from work with computers, and 2)you have to accept that metaphysical concepts like ‘information as energy’ or ‘Darwinian processing’ or transcendence are not viable explanations for the complex information processing that is known to occur.
Information and information processing is mathematics. If you ignore the mathematics, particularly if you ignore realistic estimates of the volumes of information and information processing associated with biological systems, you have no basis for a productive discussion.
IP: Logged
|
|
Irving
Member
Member # 535
|
posted 18. January 2003 11:43
Warren,
quote:
If anyone is interested in a serious discussion of information and information processing in biological systems, then 1)you have to recognize the vast body of knowledge accumulated on the subject from work with computers, and 2)you have to accept that metaphysical concepts like ‘information as energy’ or ‘Darwinian processing’ or transcendence are not viable explanations for the complex information processing that is known to occur.
I tend to agree with you here. Would you consider much of the problem is with Shannon's use of the term entropy to describe the average informtation content of a message stream?
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 19. January 2003 01:47
Irving stated that humans can create CSI. Are there any examples where a human-created work has been shown to create CSI under Dembski's framework, with all components of the computation given due diligence?
Warren said
quote:
let us recognize that the failure to properly and precisely information and information processing capacity is a serious problem for science as it is practiced today and especially for the biological sciences.
I agree, in that there is much valuable predictive information that is simply thrown out because biologists do not, on the whole, approach problems quantitatively by instinct.
However, that doesn't mean that there is a vast disparity between the information in a genome and the computational power of a planetwide evolutionary algorithm running for 3e9 years. If we take your numbers, we get about 60,000,000 bits of information after a 3e9 generations (most of that time we had multiple generations per year, but anyway). The vast majority of mutations occur in coding regions in the human genome, despite the fact that the vast majority of the genome is non-coding. This suggests that the coding regions contain the bulk of the region. There are about 30k predicted proteins of an average size of 400 amino acids, for a total of 12,000,000 amino acids. At 4.3 bits per amino acid (a bit less, actually, given the frequency of each residue), that's about 50,000,000 bits of information if we think every single amino acid is crucial (and, of course, very many mutations do nothing).
So, anyway, yes, paying attention to information processing capacity is important, but evolution tends not to fail the back-of-the-envelope calculations. If there is a failure, it's more subtle than that.
(Incidentally, where is the one bit per 55 lives from? I'd not heard that figure before.)
IP: Logged
|
|
|
Printer-friendly view of this topic