|
Author
|
Topic: Towards a simple definition of CSI
|
Bruce Fast
Member
Member # 924
|
posted 12. March 2006 17:46
Reading Doug Wedel's recent post "The Origin of Biological Information" I was frustrated by the lack of a concise definition of CSI. My sense is that the definition is, "read the book" (where the book seems to be "No Free Lunch".) Further to the lack of a concise definition of CSI, the feeling that the thread left me with was that any definition of CSI was carefully loaded to make sure that RM+NS could not possibly produce a product that meets the definition.
To my sense of fair-play these two concepts: read the book, and careful loading of a definition, are unfair. I therefore request that this thread be used to communicate a concise definition of CSI, a definition that will stand even if the only percieved cause of CSI falls.
IP: Logged
|
|
Micah Sparacio
Member
Member # 6
|
posted 12. March 2006 20:53
It might be best for those who are requesting a precise definition of CSI to show some good faith by giving a "best shot" description of the concept, as far as they understand it at this point.
Then clarification can proceed.
All concepts can be expressed simply in precise definitional form. But to fully understand such concepts, it might take the hard work of elaboration: sometimes of article or book length.
It might be best for Bruce and Doug to start off by describing which aspects of CSI they think contribute to the confusion.
Complexity (Universal Probability Bound)
Specification (detached informational pattern)
Information (the information describing this particular object or event)
In precise form, CSI = the property of their being some event or object J, with complexity K conforming to an independent pattern L.
K must meet some high threshold, normally the UPB.
IP: Logged
|
|
Irving
Member
Member # 535
|
posted 13. March 2006 22:19
quote:
Further to the lack of a concise definition of CSI, the feeling that the thread left me with was that any definition of CSI was carefully loaded to make sure that RM+NS could not possibly produce a product that meets the definition.
I thought that was the point?
To detect design without any knowledge the designer, one is left to attempt a design inferrence purely through knowledge of Nature. Which essentially gives a two step process.
1. Define what is impossible for Nature to produce.
2. Then look to see if there are objects that meet that defintion. If so, design!
IP: Logged
|
|
David L. Hagen
Member
Member # 323
|
posted 14. March 2006 00:15
Common Examples of CSI:
1) Cathedrals and Mt. Rushmore.
2) The Encyclopedia Britanica.
3) The global collection of patents.
4) Powered automobiles and aeroplanes.
5) Computers encrypting messages with error correction.
Is there any question that nature cannot produce these?
If you say that evolution by Random Mutation and Natural Selection (RM/NS) resulted in human beings that make these, then we need to bore down to the origin of life.
IP: Logged
|
|
Bruce Fast
Member
Member # 924
|
posted 14. March 2006 00:55
Micah Spericio, you are right, I should take a whack at presenting my understanding of CSI. So let me do so.
There are three terms that need defining. Complex, Specified and Information.
We've gotta start at the beginning, with information.
Let me take a whack at not so much defining as describing information.
1 - Information is in a separate paradime to matter and energy. Which is to say, information does not abide by all of the same laws as matter and energy do.
2 - Unlike matter and energy, information can be created, destroyed, and copied.
3 - There are two kinds of information, analog and digital.
3a - Analog information is stable, and precise beyond measure (though with time, it may wander significantly.) Consider, for instance, the position of a needle in an analog guage. Though that needle will virtually never accurately reflect the source of information (it will not be an exact copy) the position of the needle at any moment in time is exact, down to quanta. Analog information has one peculiar characteristic, though it can be approximately copied, and can be approximated with digital information, it is virtually impossible to make an exact copy of analog information. The fact that analog information cannot be exactly copied may, in fact, remove it from consideration as a mechanism of the necessary information transmission in life.
3b - Digital information can be copied exactly. It is possible to make multiple digital copies which experience zero information loss. The process of copying increases the robustness of that copied information. All copies must be destroyed for the information to be destroyed.
4 - Information requires a medium (matter or energy) upon which it is rendered. However, the process of copying can involve a change of medium. Information can be copied between any two information carrying mediums. Digital information can always be moved to from any digital information carrying medium to any other with no loss of information.
4a - Therefore, if it can be recorded by standard means on a computer hard drive, it is digital information. If it cannot be so stored, it is not digital information.
5 - Information has "volume". Digital information's volume can always be measured in bits. However, information has no physical volume, no mass. Though it must be carried on matter or energy, the amount of matter or energy required to carry it is dependant upon the method of encoding, not directly upon the "volume" of information. Ie, you cannot determine the number of bits of information simply by measuring the volume of the carried medium.
6 - Information is amenable to compression. This is a can of worms, but suffice it to say if a compression method permits precision engineering (there are lossy compression methods), and if the compression method will render the information with less "volume" then that information is amenable to that compression method.
6a - Not all compression methods are as effective, or are effective at all with all types of information.
Now for "complex",
I believe that Doug Wedel has provided an excellent definition of "complex". The complexity of information can be simply measured by the size of the information packed when "maximally compressed". Maximally compressed is defined as being compressed by the most effective method, knowing the background and nature of the data. This definition, obviously contains the weekness that the guy down the hall may conjur up a more effective compression scheme than yours. If so, he has just proven that there is less information in the information packet as there was thought to be.
Specified -- from specification. Specified information is information that provides all details necessary to assemble some specific thing.
There seem to me to be two varants of this term, which I will described as prescriptive and descriptive specification. Descriptive specification is a specification gleaned from the 'thing' that is being specified. An example would be pi. The value of pi does not produce pi, it only describes it. An error in the value of pi does not change the value of pi. Prescriptive specification is the specification which is actually applied to produce a thing. In this case, an error in the prescription will produce a change in the 'thing' being produced. (Note that if an error in the specification does not produce a change in the 'thing' being produced, then that bit of information is not "specifying" information.
I beleive that where CSI as applied to the nature of life is concerned, it is only the prescriptive definition of specification that we care about. Therefore, unlike what was decided in Doug Wedel's thread, I conclude that the value of pi is not, well, prescriptive CSI.
Given the above definitions (feel free to debate 'em) CSI is complex information (cannot be totally compressed) which prescriptively specifies a thing.
There, Micah Spericio, is my best attempt to come up with a concise definition of CSI.
IP: Logged
|
|
Doug Wedel
Member
Member # 1901
|
posted 14. March 2006 10:57
quote:
Unlike matter and energy, information can be created, destroyed, and copied.
Would Professor Dembski agree with this? I thought I read in one of his publications that he believed that information like energy was "conserved". It is a fundamental point.
I enjoyed your discussion of analog, digital, the medium, volume, compression. I would add that the most important aspect of information is its functional role in biology -- how it affects organisms' survival probabilities. I would also add that information is a "composite", or a compound, built up from input, data processing, and correlation with values or valences.
IP: Logged
|
|
Doug Wedel
Member
Member # 1901
|
posted 14. March 2006 11:24
quote:
It might be best for Bruce and Doug to start off by describing which aspects of CSI they think contribute to the confusion.
Let's say I play the Powerball lottery. I always play my own 9-digit Social Security number. One day I win $100 million. What can we say about this event in terms of CSI?
Well, the probability here is straightforward; it's precisely the odds of one particular 9-digit number winning the Powerball lottery.
The specification here is also pretty straightforward; it's my Social Security number.
Finally, the information in this problem is also straightforward -- it's what came out of the lottery machine.
Now, the difficulty with Complex Specified Information is that while it _seems_ to offer us a set of definitions and a way of reasoning that would enable us to draw inferences about the simple problem I just gave -- it doesn't. Is the number generated by the lottery machine CSI because it matches my SS#? I'm not sure. I don't think so though. But I'm just not sure ...
IP: Logged
|
|
Micah Sparacio
Member
Member # 6
|
posted 14. March 2006 11:36
Doug - 1 in 1 billion is above the UPB. So the complexity criterion is not met.
[ 14. March 2006, 11:37: Message edited by: Micah Sparacio ]
IP: Logged
|
|
Bruce Fast
Member
Member # 924
|
posted 14. March 2006 12:19
Doug,
This is a fundimental point.
quote:
quote:
Unlike matter and energy, information can be created, destroyed, and copied.
Would Professor Dembski agree with this? I thought I read in one of his publications that he believed that information like energy was "conserved". It is a fundamental point.
Other perspectives are welcome. Is information subject to the laws of conservation? When I put my thinking cap on, it is clear that to store information requires energy. However, the amount of energy required is more dependant on the medium holding the energy than on the amount of information. In fact, the kind of energy I use to create CSI is intellectual energy -- intelligence.
IP: Logged
|
|
Doug Wedel
Member
Member # 1901
|
posted 14. March 2006 13:52
quote:
Doug - 1 in 1 billion is above the UPB. So the complexity criterion is not met.
Can you _simply_ explain the UPB?
quote:
When I put my thinking cap on, it is clear that to store information requires energy. However, the amount of energy required is more dependant on the medium holding the energy than on the amount of information. In fact, the kind of energy I use to create CSI is intellectual energy -- intelligence.
I totally agree with your intuition that information can increase without limit, and that, as you imply, it can be done for _arbitrarily small_ amounts of energy.
IP: Logged
|
|
Bruce Fast
Member
Member # 924
|
posted 14. March 2006 14:06
Micah Sparicio, quote:
Doug - 1 in 1 billion is above the UPB. So the complexity criterion is not met.
UPB is an arbitrary threshold. It is useful in determining whether something could reasonably have ever occurred by chance, but offers little in the debate of what complexity is. As a threshold, it implies that complexity is on a continuum from not very complex to infinitely complex.
(Doug, UPB, the universal probability bound, is a value calculated by Dembski which basically is the number of atoms in the universe or something like that. His suggestion simply is that if the likelihood of an event happening is less than UPB, then it didn't happen by chance.)
Doug,
Information clearly holds to a law of entropy, however the kind of entropy that information experiences is different than that of the physical world. Again, we are dealing with a different paradyme with different rules. Because the amount of energy (of the matter/energy sort) required to store information is dependant more on the method of storage than it is on the volume of information, I propose that the required energy is just an artifact of the storage medium, that information itself does not require any energy (of the matter/energy sort) to create.
Information, on the other hand, does require energy to create, but the energy is "of a different sort, a different paradyme." The best tool we have for creating information is, well, intelligence.
We may attempt to define the elusive "intelligence" as the ability to create information. It actually isn't a bad definition. However, if that is our working definition, one must honestly consider that, by that definition, RM+NS is intelligent.
Ie: if you put a filter onto a random generator, the result is greater order than the random input. Therefore, by this definition, any filter is an intelligent agent. That would include the bug screen on my window.
IP: Logged
|
|
Doug Wedel
Member
Member # 1901
|
posted 14. March 2006 14:58
quote:
UPB is an arbitrary threshold. It is useful in determining whether something could reasonably have ever occurred by chance, but offers little in the debate of what complexity is. As a threshold, it implies that complexity is on a continuum from not very complex to infinitely complex...UPB...is a value calculated by Dembski which basically is the number of atoms in the universe or something like that
Thanks -- with that clarification I can now formally restate my lottery question. This is a "thought experiment" not possible in reality (think Einstein flying alongside a beam of light) but intended to check the logical constructs of CSI.
1. Assume there are N players in a lottery.
2. Further assume that N = 2 * UPB (N is 2 * the number of atoms in the universe -- big lottery).
3. Each player is assigned a unique lottery number.
4. The winner is chosen.
Now let's look at this in terms of Micah's categories:
quote:
Complexity (Universal Probability Bound)
Specification (detached informational pattern)
Information (the information describing this particular object or event)
In the thought experiment:
COMPLEXITY clearly exceeds the UPB
SPECIFICATION is each player's lottery number.
INFORMATION is the number picked by the machine.
Now, what does CSI tell us about this? Can we draw any kind of inferences about the process? Did the winner win by luck, or can we infer nonluck by the tenets of CSI?
IP: Logged
|
|
Irving
Member
Member # 535
|
posted 14. March 2006 15:20
quote:
SPECIFICATION is each player's lottery number.
Perhaps...probably need a reading on this...
ISCID has a short description of Specified Complexity in the ISCID Encyclopedia of Science and Philosophy located under research tools.
Specified Complexity
The area of concern being...
quote:
The idea behind specificity is that not only must an event be unlikely (complex), it must also conform to an independently given, detachable pattern.
While the lottery tickets are independently given, are the numbers they represent detachable?
By this I mean, the function of the Power Ball machine is to produce numbers. So in that context (a number generating process) is a numerical specification a "detached" specification?
Consider this lottery example of CSI posted on Dembski's Blog. Here the specification is independent & detached since the specification involves siblings, not numbers...
IP: Logged
|
|
|
|
Bruce Fast
Member
Member # 924
|
posted 14. March 2006 16:57
Irving, thanks for the link to the ISCID Encyclopedia. This is helpful. Interestingly, the definiton there simply avoids the issue of "information" altogether.
I must admit that the term specified, in the encyclopedia's definition is painfully weak by me. I think the idea that a "thing" is made from the specification, as I describe, offers clearer insight. When the "thing" that is made is capable of producing intricate product (replicating DNA for instance) the fact that the specification is non-random seems rather obvious to me.
I fail to see how any thinking person can reject the general definition of specified complexity: complex (having limited compressability even knowing the nature of the data) information which is used to generate a "thing" which is capable of performing intricate tasks.
However, I still question the conclusion brought forth here. While it may be establishable that sepecified complexity can only be brought about by intelligence, this would seriously depend on your definition of intelligence.
I propose that "intelligence" be defined as the process of generating (creating) information.
If so, it seems that any filter applied to random input creates information. With this definition in mind, however, I suspect that the only real resistance that the evolutionary community would have to it is that they may fear that there is some subtle "hidden meaning" that they are not able to see.
However, if Dembski suggests CSI as proof of intelligent design, I must challenge that by a reasonable definition of intelligence RM+NS is an agent of intelligence in and of itself.
IP: Logged
|
|
|
Printer-friendly view of this topic