|
Author
|
Topic: Towards a simple definition of CSI
|
Bruce Fast
Member
Member # 924
|
posted 16. March 2006 18:44
quote:
quote:
The next question, however, is once CSI organisms exist, can they, through the mechanisms of RM+NS generate more CSI?
I would suggest that the answer is NO
As I think we cross-posted, are you suggesting that CSI as I defined it, now CBI, cannot be increased by a CBI orgainism through RM+NS?
IP: Logged
|
|
Irving
Member
Member # 535
|
posted 16. March 2006 19:25
quote:
As I think we cross-posted, are you suggesting that CSI as I defined it, now CBI, cannot be increased by a CBI orgainism through RM+NS?
Yes, I think we cross-posted... ![[Smile]](smile.gif)
I would suppose that a better definition of blueprint as an "independent detachable pattern," would give an indication if RM & NS could increase CBI. As CSI uses those terms ![[Razz]](tongue.gif) , I'd say no.
quote:
I want a term which places DNA and computer software into the "in" group, but places the entire abiotic world in the "out" group. I want this because there seems to be an incredible confusion about what life is, an attempt to believe that some life is "simple".
Two sides of the same coin? I prefer defining CSI outside of RM & NS and then seeing if life contains it, while you prefer defining DNA and then seeing if RM & NS can produce it. Right?
In your approach, CSI (now CBI), is restricted to DNA. Defined against DNA, CBI would likely not apply to Mt. Rushmore, or say a transmitted squent of prime numbers.
Whereas the CSI approach provides greater utility. I'd the concept of Specified Complexity to be refined to the point where a robot could be programmed to recognize it. Such a program placed aboard a space probe, or planetary rover could then auto-search for signs of Intelligence and signal home when a likely artifact has been found. Consider this effort at Alien Space Station Research
quote:
I was of the understanding that Dembski's CSI was exactly that. I am dissappointed that it does not seem to be. I also have a lot more respect for those who say "I don't believe that Dembski's CSI exists."
I would suggest that CSI could be that which places DNA & Computer Software into the "in" group, with abiotic world in the "out" group. The Jury is still out however.
I would though, be cautious regarding computer software. In some camps, things like The Halting Problem suggest that Computer Software will never achieve true Intelligence...But then if your just trying to determine if the nature of Computer Software (not it's execution) is beyond RM & NS, then you might have success.
[ 16. March 2006, 19:26: Message edited by: Irving ]
IP: Logged
|
|
Bruce Fast
Member
Member # 924
|
posted 17. March 2006 00:01
quote:
Two sides of the same coin? I prefer defining CSI outside of RM & NS and then seeing if life contains it, while you prefer defining DNA and then seeing if RM & NS can produce it. Right?
This really isn't the point. A struggle I have when discussing the ID/evolution thing is in defining what is special about life compared to non-life. I really want to clarify, and provide terminology, that says "DNA is special, it has this characteristic that the abiotic world doesn't have. Further, that specialness is the same kind of special that a computer program has." There is lots of discussion in the ID/evo discussion, we are only at the beginning. Once we get here, the challenge of first life becomes much clearer. Many will put away their soup-pots. Further, even if RM+NS can produce some CBI, that hardly establishes that all of the DNA is merely the product of RM+NS. I believe, for instance, that the maximum reasonable pace of RM+NS developing CBI could be established and is likely rather slow. I then suspect that events such as the cambrian explosion would be seen as unattainable via RM+NS. However, we still need some terminology to work with in these discussions.
quote:
CBI would likely not apply to Mt. Rushmore, or say a transmitted squent of prime numbers.
I don't exactly see how Mt. Rushmore is information. Mt. Rushmore is the product of information, and information can be gathered from it (you can measure lots of stuff) but it isn't the information, the blueprint, itself.
As to a transmitted sequence of prime numbers, well, as I have defined complex (amount of data after maximal compression is applied, even when the nature of the data is understood), well, a list of prime numbers is extremely compressable, which is to say, once the 10th prime has been detected, the value of the 11th number is totally predictable.
quote:
I would though, be cautious regarding computer software. In some camps, things like The Halting Problem suggest that Computer Software will never achieve true Intelligence...But then if your just trying to determine if the nature of Computer Software (not it's execution) is beyond RM & NS, then you might have success.
I develop software, I have spent my fair share of time tinkering with AI software. Computers don't think! I would never suggest that they do. Computer software, however, is an order of machine expressions that can be isolated, and that creates specific characteristics in something outside of itself. Ie, it meets the B part of CBI. I actually think that anything greater than maybe the famous "Hello World" program would qualify as complex (beyond UPB) and it certainly is information.
IP: Logged
|
|
Stephen Wright
Member
Member # 195
|
posted 17. March 2006 10:15
quote:
“A struggle I have when discussing the ID/evolution thing is in defining what is special about life compared to non-life. I really want to clarify, and provide terminology, that says "DNA is special, it has this characteristic that the abiotic world doesn't have. Further, that specialness is the same kind of special that a computer program has." - Bruce
Autopoiesis & dissipative structures are the best terms we have to define the differences, at this time. My own version, which would delineate the contrasting elements, points to logical functions indwelling the information object associated with the physical object. This works well when thinking in terms of computer programs, when they are defined as object-oriented. There is usually a “theme” to them that is a purpose aimed at a target state or states. It is the lack of this logic-driven function that is not in inanimate systems and makes them random in nature. The guidance is exactly its energetic vectors and no more.
A living entity can instruct its vectors with guidance from environmental feedback.
[ 17. March 2006, 10:51: Message edited by: Stephen Wright ]
IP: Logged
|
|
Bruce Fast
Member
Member # 924
|
posted 17. March 2006 12:08
Hmmm, new vocabulary.
Autopoiesis - the process whereby an organization produces itself.
This term separates out most computer software. While it does include all DNA based life, it does not necessarily disclude non-information-based chemical reactions.
Dissipative structures - A system that exits far from thermodynamic equilibrium, hence efficiently dissipates the heat generated to sustain it, and has the capacity of changing to higher levels of orderliness.
Again, a term that clearly discludes most computer software, but may include non-information-based chemical reactions.
As there is something very "computer softwarelike" about DNA, and as abiogeneis somehow requires the move from non-information-based to information based, I really think that another term is called for. for there to be an abiogenesis theory (as opposed to hypothesis) there must be an established pathway from non-information-based chemical reactions to an information base such as DNA or RNA. A term, a term that is not hidden in the obscurity of the scientific elete, is needed. That word should not focus on "reproduction", but upon that complex order "information" that is DNA, and also is computer software.
IP: Logged
|
|
Micah Sparacio
Member
Member # 6
|
posted 17. March 2006 13:35
Bruce, there is something wierd about both DNA and software code. They both demand an explanation for their formation, but their formation isn't the only interesting fact about them. What's even more interesting about them is that what they do (their functional role) seems to depend critically on how they were formed. And if I'm representing Dembski's views correctly, the functional role of DNA and computer software is what gives both things their specificaiton.
Anyway, what I'm trying to get at is the fact that the interesting thing about both DNA and computer code isn't just the objects themselves, but their capacities/abilities. Such capacities/abilities are not common and they depend on very precise formations. They wouldn't be nearly as interesting if they didn't do anything and just existed as rocks exist.
As a way to elucidate my thoughts here (which I'm working out as I type), take the recently announced existence of a cosmic double helix formation. Now that's somewhat interesting, but let's ask why. First, it has some pleasing visual properties. There's also the "interesting" or "aha, neat coincidence" response that one might have when associating that double helix formation with the one we're much more familiar with. As one of those passing, curious, nonrigorous thoughts, one might also consider "maybe there's a common cause between micro double helix formations and cosmic double helix formations."
But, for the time being at least, it seems that the there is a crucial difference between the double helix of DNA and the double helix in the pictures from NASA. DNA does some functionally significant things. So does software code. But I doubt that the cosmic double helix has this property (outside of merely existing and containing some physical objects).
[ 17. March 2006, 13:38: Message edited by: Micah Sparacio ]
IP: Logged
|
|
Stephen Wright
Member
Member # 195
|
posted 17. March 2006 13:55
quote:
“there must be an established pathway from non-information-based chemical reactions to an information base such as DNA or RNA.”
Bruce, I would ask to ask you to reconsider chemistry and physics pathways as being non-information based. Today, in these two fields, so much technology is added to our knowledgebase through virtual processes - that I wouldn’t hesitate to call virtual chemistry the preferred way to work. Sure, the test tubes and pipettes have to go to work under the hood to manifest the theoretical, but the lab ain’t what it used to be. You will find many of the techs pounding the keys.
www.chem.ox.ac.uk/vrchemistry/default.html
At the LiveChem link on this website – you can mix cobalt (Co+2) and concentrated HCl - on screen - and make cobalt blue salt right in the test tube.
Science is really observation of the physical – being virtually recreated into the data category of information. The predictive ability of science is derived from the facts being bound to the logical premises of causality. The laws of science are information objects, IMHO. Where it is different, is the next level up – in the hierarchy of information processes. That is where mutual information comes into play.
I think what you are saying, works in these terms – “there must be an established pathway from the individual facts of information-based chemical reactions to the usable mutual information-based objects such as DNA or RNA.” Facts are isolated – but mutual information is what characterizes what is useful or is communicative. Sender and receiver must have some mutual information they share, to be able to code and decode. After a message – more mutual information is created at the receiver. Guidance comes from instructions being communicated. The receiver must have enough mutual information to decode and apply what is in the message. The new mutual information can then be “understood” and vector a purposeful function.
IP: Logged
|
|
David L. Hagen
Member
Member # 323
|
posted 17. March 2006 13:59
Thanks Stephen for "Autopoiesis" and "Dissipation".
Thanks too Bruce - you beat me to the post.
For those like me who could not find "autopoiesis" in McGraw-Hill etc.
Autopoiesis - Wikipedia
quote:
Autopoiesis literally means "auto (self)-creation" (from the Greek: auto - for self- and poiesis - for creation or production) and expresses a fundamental complementarity between structure and function. The term was originally introduced by Chilean biologists Francisco Varela and Humberto Maturana in 1973:
"An autopoietic machine is a machine organized (defined as a unity) as a network of processes of production (transformation and destruction) of components which: (i) through their interactions and transformations continuously regenerate and realize the network of processes (relations) that produced them; and (ii) constitute it (the machine) as a concrete unity in space in which they (the components) exist by specifying the topological domain of its realization as such a network." (Maturana, Varela, 1973, p. 78)
"[…] the space defined by an autopoietic system is self-contained and cannot be described by using dimensions that define another space. When we refer to our interactions with a concrete autopoietic system, however, we project this system on the space of our manipulations and make a description of this projection." (Maturana, Varela, 1973, p. 89)
The canonical example of an autopoietic system, and one of the entities that motivated Varela and Maturana to define autopoiesis, is the biological cell. . . .
quote:
More generally, the term autopoiesis refers to the dynamics of a non-equilibrium ( or non-equilibrium thermodynamic (NET) (Dyke, Charles, 1988, ch. 9)) system; that is, organized states (sometimes also called dissipative structures) that remain stable for long periods of time despite matter and energy continually flowing through them. . . .
Autopoiesis - Google
I am using "recursive" as a more popular term in defining biotic life.
--------------------------
Bruce
Re: Specified Info vs Mt Rushmore:
1) Recommend considering both “encoded” information (such as a blueprint, template, CNC code etc.) and “embodied” information. (such as the manufactured product or Mt. Rushmore busts.)
Reverse Engineering can typically recover "embodied" information and convert it to "encoded" design information. (Anyone have a better word that "embodied"?)
To help move the discussion, I propose the following Design Principles for the transformation of Design Information or CSI (popularly “hypothesis” to become a “theory” and then “law”)
2) Transformation of Information:
Complex Specified Information can be transformed among encoded and embodied forms, but transformation does not increase the total amount of Complex Specified Information.
3) CSI vs Shannon Info
Complex Specified Information is less than or equal to Shannon Information
4) Design Information vs CSI
Design Information is a subset of Complex Specified Information
I am working to a precise definition of “Design Information” as a subset of CSI to quantify how it is different from and less than or equal to “Shannon Information” (which only gives you the channel capacity, not what goes through the channel.)
5) Parallels between Computer/Manufacturing and Cells
Similar to Micah’s thoughts, in defining your “Blueprint”, may I propose identifying the main components of a designing/computing/manufacturing process that parallel the cell function as a way to further develop your Complex Blueprint Information ideas. E.g. what are the components and processes that parallel the cell.
1) Origin of macromolecular construction information (To be modeled)
2) Genome = 3d pattern sequence (encoded construction information)
3) Nucleotide pairs (physical embodiment of code alphabet.)
4) Codons (Physical embodiment of triplex block code)
5) DNA sequence (physical embodiment of pattern sequence encoding 3D construction information)
6) mRNA (read code, decode and transcribe)
7) Protein sequence (components in assembly process)
8) Folded Protein (assembled product)
---------------------
[ 17. March 2006, 15:14: Message edited by: David L. Hagen ]
IP: Logged
|
|
Irving
Member
Member # 535
|
posted 18. March 2006 01:13
quote:
I am using "recursive" as a more popular term in defining biotic life.
The first word I thought of as well...
quote:
Complex Specified Information is less than or equal to Shannon Information
Hmmm...Dembski's response to Matt Young.
quote:
There is no equivocation in my work over information. What I do is define a type of information--specified complexity--that enriches Shannon information but at the same is not reducible to it.
IP: Logged
|
|
David L. Hagen
Member
Member # 323
|
posted 18. March 2006 09:30
Appreciate the clarifying citation Irving. Lets see if I can make it more precise without defining it yet
The numerical measure of Complex Specified Information is less than or equal to the numerical measure of the Shannon Information (information channel bandwith or information entropy).
See Discussion on CSI & Shannon Info. Is the 2nd law a subset of the 4th law? ISCID
Shannon Information - Wikipedia
[ 18. March 2006, 09:36: Message edited by: David L. Hagen ]
IP: Logged
|
|
Irving
Member
Member # 535
|
posted 18. March 2006 15:40
quote:
The numerical measure of Complex Specified Information is less than or equal to the numerical measure of the Shannon Information (information channel bandwith or information entropy).
Much better, though in looking at a fuller context of the quote I provided earlier...
quote:
There is no equivocation in my work over information. What I do is define a type of information--specified complexity--that enriches Shannon information but at the same is not reducible to it. What gives specified complexity its traction in detecting design is a coincidence of two things: (1) an event that under a chance hypothesis has small probability and therefore high information content in the first of Young’s senses; and (2) a pattern that is objectively given and complexity-theoretically tractable, and yet matches the event.
I wonder if your "numerical measure" includes the coincidence of part (2). I suspect it's restricted to (1). A full measure of CSI could exceed Shannon, if the part (1) portion is fully compressed.
I know that the trend is in using Kolmogorov complexity in tieing CSI and computer software together; but, given that the Specification portion of CSI is focused upon meaningful information, and that meaningful must necessarily include information in it's context...I wonder if someting like McCabe's Cyclomatic Complexity may not be an appropriate software tie-in?
IP: Logged
|
|
David L. Hagen
Member
Member # 323
|
posted 18. March 2006 18:53
Good question Irving. At least category 1. I was thinking of measuring the probabilty and thus "information" in bits. For Design Information, I am probably thinking of category 2. i.e., specifying the objective pattern, as well as methods to measure its probability and thus category 1.
Interesting comment that CSI could exceed Shannon Information. I had the impression that Shannon Info was the maximum possible.
Exploring, I see that using a nonuniform alphabet results in less efficient coding efficiency, resulting in fewer possible configurations and thus lower Shannon Information. Yockey (2005) Sect. 4.1; Appendix p 205-209.
As I understand it, the probability is one in M^N for an alphabet of M characters e.g.,(20)^N for genetic coding; vs 2^(NH) where H (equation 4.5) is the "Shannon entropy of the sequence of events." H = -(sum on i) p(i) log2 p(i).
Yockey calls the difference in probability "junk" sequences. I would prefer to call them Coding Inefficiency.
"Coding Inefficiency" is the loss in usable information per character between encoding with a uniform code and encoding with a non-uniform code. CI = M-H
The Information Inefficiency for a sequence of N characters using an code of M characters is:
Ii=N(M - H) where H is the code's Shannon Entropy.
I like Yockey's use of "Shannon Entropy" to reduce the popular confusion over "Information."
If both Shannon Entropy (Information) and CSI are measured in the same code or alphabet do you still see CSI can be greater than Shannon Entropy?
Or do you see CSI as specified with a uniform code vs a compressible code where compression would give a lower Shannon Entropy?
I would welcome any examples you can give or reasons why you see the difference between them?
(I am working up to include Random (or Chaotic) information and Order Information (natural law), and then a number of categories of Design Inromation.)
[ 18. March 2006, 19:13: Message edited by: David L. Hagen ]
IP: Logged
|
|
Irving
Member
Member # 535
|
posted 18. March 2006 19:20
quote:
If both Shannon Entropy (Information) and CSI are measured in the same code or alphabet do you still see CSI can be greater than Shannon Entropy?
Probably not. Is that ambiguous enough? ![[Big Grin]](biggrin.gif)
I think the trouble may be that they cannot share the same code or alphabet...at least not as we may currently formulate them.
I see CSI as Information about Information. Thus a sequence of prime numbers has a certain Shannon Entropy, but it is only CSI when in a context that is normally independent and detachable. Thus,
CSI = Shannon Entropy + Physical Process Independence (complexity?)
As I mentioned before, I see the use of Kolmogorov complexity is determining a physical selection "Best Case." If Natural Selection is an IF - THEN decision tree, then Cyclomatic Complexity might be considered the "Actual Case." Certainly proposed evolutionary pathways would be subject to such quantification, not just in best-case, but the entire probability space.
IP: Logged
|
|
Stephen Wright
Member
Member # 195
|
posted 19. March 2006 14:59
quote:
“So, what's all the fuss about specified complexity? The actual term specified complexity is not original with me. It first occurs in the origin-of-life literature, where Leslie Orgel used it to describe what he regards as the essence of life. That was thirty years ago. More recently, in 1999, surveying the state of origin-of-life research, Paul Davies remarked: "Living organisms are mysterious not for their complexity per se, but for their tightly specified complexity" (The Fifth Miracle, p. 112). Orgel and Davies used specified complexity loosely. In my own research I've formalized it as a statistical criterion for identifying the effects of intelligence.” - William A. Dembski Talk delivered at the American Museum of Natural History, 23 April 2002
To separate the concept of CSI from a biotic context will obscure the focused thinking, which created the term. I support a naïve view of CSI that would help keep it from the attacks by those individuals who would try to link this valuable concept to sources outside of pragmatic analysis. The concept has its origin firmly within the research of life and observable physical properties. At issue is the linkage of intelligence with the products resulting from its application. Materialistic science has problems classifying purposeful behavior and CSI is a good definition of the real attributes that are regularly observed.
Intelligence is a biotic phenomenon. The ID conjecture is based approximately on the syllogism:
* intelligent activity, as human thought, is directly causal in functional design
* ontogenetic processes result in functional design
* intelligent activity is causal within ontogenetic processes
Complexity in information theory refers to structural relations in the virtual domain that have corresponding physical representation. To look for pathways of manifest events in the virtual domain, will take the “edge” off the focus of the debate about CSI. I am a strong proponent of Wheeler’s “it from bit”. But to view the material progression of transforms in any other way than being governed by received scientific methods, will reduce the viability of any argument for the role of intelligent activity.
[ 19. March 2006, 15:00: Message edited by: Stephen Wright ]
IP: Logged
|
|
Bruce Fast
Member
Member # 924
|
posted 20. March 2006 17:45
Wow, when you have a good thread going, and you get busy for a few days, the stuff to respond to becomes significant. However, let me try.
David L. Hagen, March 2006 13:59
quote:
Bruce
Re: Specified Info vs Mt Rushmore:
1) Recommend considering both “encoded” information (such as a blueprint, template, CNC code etc.) and “embodied” information. (such as the manufactured product or Mt. Rushmore busts.)
Reverse Engineering can typically recover "embodied" information and convert it to "encoded" design information. (Anyone have a better word that "embodied"?)
I appreciate the term "embodied" information, yet I radically reject the "reverse engineering can typically recover" concept (when you do not permit access to the "encoded" information that may linger witin the embodied info.). Consider the following:
1 - If you give a computer programmer the challenge of making a "funcionally equivelant" Microsoft Word program, the result could be very functionally equivelant, but the source code will be wildly different.
2 - If you examine my phenotype in every detail (restricted to not examining the DNA or RNA, examine the proteins all you want) you cannot recover my DNA. For instance, my recessive genes leave no mark in my phenotype (as I understand it).
3 - Even an examination, and reverse engineering of Mt. Rushmore will produce blueprints, but those blueprints will likely carry significant differences to the originals, even if the resultant blueprints could do an excellent job of reproducing the president's faces.
quote:
2) Transformation of Information:
Complex Specified Information can be transformed among encoded and embodied forms, but transformation does not increase the total amount of Complex Specified Information.
Transformation of CSI between encoded forms can regularly be achieved without loss, the latter is usually not the case when transformed to the embodied case -- see above.
quote:
3) CSI vs Shannon Info
Complex Specified Information is less than or equal to Shannon Information
If my understanding of Shannon is correct, this is so.
quote:
4) Design Information vs CSI
Design Information is a subset of Complex Specified Information
I am working to a precise definition of “Design Information” as a subset of CSI to quantify how it is different from and less than or equal to “Shannon Information” (which only gives you the channel capacity, not what goes through the channel.)
Is your "Design Information" going to be what I am seeking, a definition that encompasses DNA, and computer software, that rejects all abiotic, and non-intelligently-designed products, and also is not the "embodied", only the "encoded" information? I believe that such is the definition that is called for.
quote:
5) Parallels between Computer/Manufacturing and Cells
I don't do manufacturing, I develop software. Let me stay within my expertise on this. Though I see some paralells between the character of software and the character of DNA based organisms, the paralells are really quite loose.
Nuclotide pairs approximately equals a byte.
Codons approximately equals a word.
Gene approximately equals a subroutine (procedure/function)
Genome approximately equals a program.
Beyond that, the paralells become harder to find. The most critical reality that I still see, however, is that the order of events -- what codon follows what codon, what word of compiled code follows what word makes the difference between a functioning entitity and a "bug".
Irving, 18. March 2006 01:13
quote:
Hmmm...Dembski's response to Matt Young.
quote:
There is no equivocation in my work over information. What I do is define a type of information--specified complexity--that enriches Shannon information but at the same is not reducible to it.
This seems to be a question of symantics. Shannon, if I understand correctly, is all about data compresion, and "total amount of real information". Specified complexity has something to do with the "kind" of information, or the nature of that information. Therefore, there is some concepts that are part of Specified Complexity that are not covered by Shannon (Shannon is more general). However, Shannon is good a separating "real information" from "noise", and at quantifying that "real information". Although some "real information" is not SC, SC is "real information". Therefore, as SC is a subset of Shannon, the total amount of SC will always be less than or equal to Shannon. However, if we know that some information is SC, we know characteristics about that information that we would not know if we only knew what Shannon teaches us about that information. This is one of those cases of "less is more".
IP: Logged
|
|
|
Printer-friendly view of this topic