Information in biology

I've been hammering the creationist argument about information a lot. It gets brought up over and over, and the never seem to acknowledge that people have addressed their concern before.

Check out the blog:

http://aigbusted.blogspot.com

-Ryan

proponents of “Intelligent Design” creationism appeal to information theory to make their arguments look more rigorous

I think I have a different perspective. Specified information, which was defined originally by Leslie Orgel (as William Dembski himself says) is perfectly meaningful once we agree on what scale to rank individuals (and the relevant one is fitness). Dembski's critics point to his vagueness about what the scale is -- but defining it as fitness is the way to rescue that part of his argument. The rest of his argument, that deterministic and random functions can't create it, is where he is wrong, as I explain in my article in the current issue of Reports of the NCSE (the issue that isn't quite up on their web page yet).

The idea that random stuff has lots of information is mistaken. It has lots of things that need explaining but that is different from saying that it has a lot of specified information. It has none. If I ask you for some good hot tips on the coming horse races, and you say sure and send me random numbers, I ought to be royally ticked off, and should not congratulate you on sending me so much information. Because what you sent me does not help me at all.

If I ask you for some good hot tips on the coming horse races, and you say sure and send me random numbers, I ought to be royally ticked off, and should not congratulate you on sending me so much information.

That’s the full extent of ID’s appeal to information theory, take the negative base 2 logarithm of a probability.

Joe, the whole concept of "specified information" is nonsense. A 'specification' means the object conforms to a pattern, i.e., is easy to describe. But 'information' in the sense it is understood by mathematicians and computer scientists measures to what extent an object is hard to describe; that it, how much it does not conform to a pattern.

I recommend reading Kolmogorov complexity and its applications by my colleagues Li and Vitanyi, or my long paper with Elsberry debunking the notion of 'specified complexity'.

Semantic content. I think that's in a way what WAD is trying to say, that life has meaning. But that's a philosophical argument, and he's trying to make it sound scientific by calling it 'information' instead of 'meaning'... or 'intent'... or 'purpose'...or 'design'. Circular and pointless.

JuliaL: Why the negative base 2 logarithm of a probability?

The Stanford Encyclopedia of Philosophy also has an article on "Creationism," which contains the following statement in its conclusion:

"Scientifically Creationism is worthless, philosophically it is confused, and theologically it is blinkered beyond repair. The same is true of its offspring, Intelligent Design Theory."

I forget--how did Dembski claim to specify information? It is so easy to look at some thing, and see a pattern in it. How much complex specified information can I detect in an ink blot, or in clouds? I can see patterns in biology, some that require a lot of words to describe, but many are easily explained by common descent or contingency.

mark: I forget--how did Dembski claim to specify information? It is so easy to look at some thing, and see a pattern in it. How much complex specified information can I detect in an ink blot, or in clouds? I can see patterns in biology, some that require a lot of words to describe, but many are easily explained by common descent or contingency.

Ye Filter.

Dembski's No Free Lunch explains csi in a way that I could eventually understand. It was quite convincing and I think most of you have missed it.

On an unrelated issue: Dawkin's could not give an example of a mutation that increased information (by any definition). Is there an example of a mutation that improves the function of an enzyme or makes a new structure or is evolution in a positive direction?

Merlin

I forget–how did Dembski claim to specify information? It is so easy to look at some thing, and see a pattern in it.

Merlin Perkins: Dembski's No Free Lunch explains csi in a way that I could eventually understand. It was quite convincing and I think most of you have missed it. On an unrelated issue: Dawkin's could not give an example of a mutation that increased information (by any definition). Is there an example of a mutation that improves the function of an enzyme or makes a new structure or is evolution in a positive direction? Merlin

On an unrelated issue: Dawkin’s could not give an example of a mutation that increased information (by any definition). Is there an example of a mutation that improves the function of an enzyme or makes a new structure or is evolution in a positive direction?

Dembski’s No Free Lunch explains csi in a way that I could eventually understand. It was quite convincing and I think most of you have missed it.

Why the negative base 2 logarithm of a probability?

In other words, why not the positive base 3? Why not divide by 112 before taking the negative base 2 logarithm? Why not just add 42?

The idea that random stuff has lots of information is mistaken....Because what you sent me does not help me at all.

PG nails it.

* NO SEMANTIC CONTENT IN INFORMATION THEORY

* INFORMATION THEORY IS USELESS TO ID and
everything else he said on this too.

Surprisingly there are even a couple
articles on one of the IDist websites
saying the same thing. It makes me think
the author was trying to get through to
someone. It obviously did'nt work but
that's what happens when people have
concrete in their heads.

I have not been impressed with Dembski's work with regards to CSI. Behe's idea about IC seems more a plausible angle, if it could be positively demonstrated that there is no chemical pathways to certain structures. I ask the crowd here for your thoughts: if design is present in nature, is there any way to detect it, excepting for blatant messages to us in the DNA?

Muldoon: You'd have to give us an exact definition fo "design" before anyone could objectively verify whether or not it is "detected" in nature. "It looks designed (to me), therefore it is designed" doesn't cut it, especially for an artsy type like me who sees "design" of sorts in ice crystals.

Some of the confusion and difficulties associated with attempting to apply information theory to biology can be illustrated with a simple analogy to dendritic growth (e.g., icicles, stalactites, or other mineral or neuronal growth). Initially, before a specified branch of a dendrite has developed, the probability of its development would appear to be very low. However, this misconception arises because the branch is specified, and it ignores the contingencies that could just as easily have produced others.

In reality, millions of possibilities are available before a particular branch develops, and once a branch starts (because of contingencies existing in the system or its environment), the probabilities of its continuation may be quite large. This is just another way of saying that the probability of reaching certain states is dependent on what nearby states are available. Singling out a particular branch and asserting that its probability is low relative to the background states from which it developed (and therefore special in some way) is misleading because millions of other branches could just as easily have developed.

It may be meaningful in some way to calculate the probability (or, equivalently, the entropy or “information”) of that particular branch relative to nearby states, but that doesn’t take into account the millions of other branches that could have but didn’t develop. Given the right conditions and energy throughput, dendritic growth may be inevitable for a given system; it’s just the particular configuration that appears improbable if one ignores the broader picture.

Similarly, singling out particular features of an organism and suggesting that they are improbable (and therefore must be designed) ignores the record of evolutionary history. The fact that there are so many varied life forms that exist, and have existed, suggests that a large number of life forms are possible within the energy ranges and conditions on this planet. Just because other forms didn’t develop doesn’t mean that, given suitable contingencies, they couldn’t have developed. The variety of life on this planet appears to be in a constant state of flux. And the evolutionary tree does appear to be much like constantly changing dendritic growth. Species and their characteristics arise against a background of contingencies, and once they are established and selected, they form a template for further development along a particular branch (at least temporarily until further contingencies wipe out the branch).

When Dembski asserts something is improbable (and hides this assertion in a negative logarithm to base 2), he is arrogantly assuming he knows specifically how the current state of an organism was achieved and that there were no alternative organisms that could have developed. It’s like he gets dealt a 52-card hand from a shuffled deck of cards and concludes that the amount of information contained in his hand is minus log2(1/52!).

I think some of your terminology is muddled muldoon however I think your question if I am interpreting it correctly has some merit. The problem lies in proving their are no viable mutation pathways from point A to point B. This is not impossible just incredibly difficult to do in a truly rigorous sense. You have to account for an impractical amount of possible contingencies if your initial set of conditions fail to be viable as well as the wide variety of possible pathways. On the other hand in cases where researchers have tried to fill in potential intermediates they have found a fair amount of success suggesting that in practice we will find molecular intermediates given a reasonable amount of time and resources for a wide variety of systems. It gets much dicer if you want to talk about any gene because many of them are not nearly as well suited to experimental characterization.

Behe’s idea about IC seems more a plausible angle, if it could be positively demonstrated that there is no chemical pathways to certain structures.

There is more information inherent in a random collection of molecules than a highly ordered one. Using water as an example, given the position of one molecule in an ice cube of a given size, you could very specify the position of all the other water molecules with a minimum amount of information that specified the parameters of the crystal structure and distance between molecules. Given the same amount of water in a steam vapor and the position of one molecule, it would require much more information to specify the position of any other water molecule. You could encode a much more complex message using the position of the water molecules in the vapor than in the ice crystal. Similarly, there is more total information in a random collection of six billion nucleotides in a beaker than there is in my genome. The only difference is that I have RNA polymerases and ribosomes to decode what little information remains. As I said, I have no background in information theory, so please let me know if I have missed something.

if design is present in nature, is there any way to detect it,

The problem lies in proving their are no viable mutation pathways from point A to point B.

And all this time I'd been saddled with the impression that the word "specified" within the expression "CSI" had been put in there by the usual suspects in order to indirectly introduce the concept of a "designer" at the ground floor of the whole gallimaufry.

After all, something can't be "specified" unless some specifier has engaged in specifying. The CSI in the bacterial genome that produces a flagellum is clearly exactly equal, functionally, to a CNC tape for a milling machine or a set of cards for a Jacquard loom. Both of those are complex, informative, and would have been specified by some engineer or designer.

Henry J.

I’d think that to be relevant, the information has to be at least somewhat persistent. Positions of molecules in a vapor change constantly, so there’s no persistence there.

This does not argue against my point that there is a wealth of information there, no matter how useful you may find it.

If, however, you insist on persistance, then just change my analogy to a pile of 1000 bricks stacked in a 10 X 10 X 10 array versus the same number of bricks randomly scattered over a lawn where they will persist in this arrangement until serious effort is expended to move them.

If, however, you insist on persistance, then just change my analogy to a pile of 1000 bricks stacked in a 10 X 10 X 10 array versus the same number of bricks randomly scattered over a lawn where they will persist in this arrangement until serious effort is expended to move them.

Why is someone who is so ignorant of information theory that he can say “The idea that random stuff has lots of information is mistaken” writing articles about it for Reports of the NCSE?

Joe, the term "information" has precise meaning for mathematicians and computer scientists under the rubric of Shannon or Kolmogorov information. If you want to call the "Whatchamacallit" in the genome "information", you are going to confuse the rest of us who already have a different definition, established for many years, in mind. Why not give it a another name entirely?

Jeffrey Shallit: Joe, the term "information" has precise meaning for mathematicians and computer scientists under the rubric of Shannon or Kolmogorov information. If you want to call the "Whatchamacallit" in the genome "information", you are going to confuse the rest of us who already have a different definition, established for many years, in mind. Why not give it a another name entirely?

If someone says that our genome contains a lot of information making us well-adapted, is that an unreasonable thing to say?

Our genomes undoubtedly have evolved so that a genotype is in a very small fraction of all possible sequences, one which is in the far upper tail of the distribution of fitnesses.

Jeffrey Shallit: Joe, the term "information" has precise meaning for mathematicians and computer scientists under the rubric of Shannon or Kolmogorov information. If you want to call the "Whatchamacallit" in the genome "information", you are going to confuse the rest of us who already have a different definition, established for many years, in mind. Why not give it a another name entirely?

Mike Elzinga: Organisms develop and are selected by contingencies within the environment. Thousands of arrangements may be possible, (in fact there is usually a distribution of traits within a reproducing population of organisms), but they get sorted, with some being favored more than others. Change either the environment or the distribution of traits, and something else falls out. Singling out the thing that falls out as somehow being special is at the root of the confusion. It ignores the history of evolution which shows that an enormous range of possibilities exists and that some of these persist for a time because they are just robust enough to survive in their current environments. [Emphasis added]

Why is someone who is so ignorant of information theory that he can say "The idea that random stuff has lots of information is mistaken" writing articles about it for Reports of the NCSE?

Thanks, Carl. That ought to do a good job of covering up the fact that I am an ignoramus. Check's in the mail.

Without getting into the subject of how much of an ignoramus I might be, let me just explain why I might take a different view as to whether it is reasonable to use the phrase Specified Information.

On the view of Popper’s Ghost, it would be unreasonable – for that genome would contain, in PG’s view, (almost) exactly the same information content as a random string of nucleotides of the same length.

Up until the sentence quoted, Joe is writing about Specified Information, not Shannon information. His claim is correct about specified information, and about any number of other senses of the word information. The claim would be incorrect only if Joe had written “Shannon information” or comparable – and he did not.

Joe has been publishing solid work on the relationship between information theory and evolutionary biology since before many of the posters to this thread were born – and thus may know more about information theory than some here seem to have surmised.

It’s going to be hard to find a different one, because “information” is also a word used in colloquial English as well as in the languages of Mathematics and of Computer Science.

Yeah, colloquially “information” means data that’s useful to somebody. A subjective judgment if ever there was one.

Thanks, Carl. That ought to do a good job of covering up the fact that I am an ignoramus. Check’s in the mail.

What is "Specified"? This seems to be that the 'event' has a specific pattern. In the book he runs around with a lot of vague definitions. "Specification depends on the knowledge of subjects. Is specification therefore subjective? Yes." Page 66 That means that if only if Dembski says something is specified, it is. That's pretty useless of course....

What is "Information"? Dembski mentions Shannon uncertainty (p. 131) without objection as the average of the 'surprisal'. He accepts my use of the Rsequence measure too (p. 212-218). So I will take information to be the Shannon information. ...

Joe is aware of, and perhaps alluding to, our recent work on the relation between information and natural selection, which suggests surprising connections between information and Darwinian fitness. This starts to bring relate semantic aspects of meaning to information-theoretic quantities, and suggests that perhaps fitness may even be viewed as a type of information measure.

Popper's Ghost: The claim that “The idea that random stuff has lots of information is mistaken” is not correct for any formally definable notion of information, whether you call it "specified information" or anything else.

If we apply Shannon’s criteria to random stuff, the result seems to be just the opposite. Random stuff is just noise without any information contents.

Following Nyquist and Hartley, it is convenient to use a logarithmic measure of information.... We will use the base 2 and call the resulting units binary digits or bits.

Joe:

You say "It’s going to be hard to find a different one, because “information” is also a word used in colloquial English as well as in the languages of Mathematics and of Computer Science. "

Sorry, I find this objection silly. The words "field" and "group" also have meanings in colloquial English, but if you talk about "elements in a group" or "field extensions", no one is going to think you are talking about their colloquial meanings.

If you want to give a mathematical definition of information that differs from the standard, call it something else.

Carl Bergstrom:

Dembski's "specified information" is completely bogus. Read my long paper with Elsberry.

Popper's Ghost: By "information contents", what you really mean is signal, as in "signal-to-noise ratio". Shannon's concept of information is an objective measure which doesn't distinguish between contentful (signal) and non-contentful (noise).

And "noise" and "random" are in no way synonyms; the noise can be highly ordered -- think punk rock or any other kind of music you don't like :-), an interfering radio station, etc.

Note that noise introduces uncertainty, and information reduces uncertainty. Thus the more noise, the more information must be transmitted over the channel. Channels without noise don't need parity bits or other redundancies.

The notion of randomness relates (via algorithmic information theory) to compressibility. More compressible sequences require fewer bits; random sequences, being uncompressible, require the maximum number of bits. Bits are Shannon's measure of information. See http://www.stanford.edu/class/ee104/shannonpaper.pdf

Thanks PvM, that was an interesting article.

If I may review it from my admittedly dim understanding (not using any of the two common information theories in my work), Godfrey-Smith has extensively covered "information" as used in biology. He is correctly noting the obvious problem of accepting Shannon's description of a physical observable, yet speculate in formulations that mistakes information for a physical object.

This philosophical description is however not how a scientist would have described a scientific area. We have two different models of information (Shannon theory and algorithmic information theory) that would need to be covered. Godfrey-Smith doesn't mention the later, even when discussing coding issues. Those two models have already limited applications in physics, which would warrant a description in such a review, and it would be surprising if biology would find information a better tool.

Finally, I think Godfrey-Smith may have less of a grip on signal theory. In its purest form signal theory isn't about what information (correlations) a signal codes of a source system, but how to detect and decode a predetermined signal among noise. This can even be done in a model-less fashion.

Source system identification (modelling), whether as a black box or as a physically informed model, is the domain of systems and control theory. I think that may be a point of the Developmental Systems Theory Godfrey-Smith mentions, but you wouldn't recognize it from his description. I have no idea of how successful systems identification can be of biological systems, with their likely distributed, changing (during growth, repair, sickness, et cetera), and contingent components. It is hard enough to reverse-engineer when you have systems with reasonably static and localizable but unobservable internal nodes, say IC circuits.

I think the later type of review would have prepared the reader for an informed discussion along the lines of Shallitt, Elzinga and PG. Each has obviously done some thinking about the applicability of information in biology, and I'm enjoyed about the opportunity to learn more.

But what environment is a random string of nucleotides a message about? Not ours. Rather, if one were to decode that string, one would end up with an unpredictable environment incapable of producing viable organisms – it’s unpredictability requires that a much larger amount of information would be required to encode it.

... the whole concept of “specified information” is nonsense. A ‘specification’ means the object conforms to a pattern, i.e., is easy to describe. But ‘information’ in the sense it is understood by mathematicians and computer scientists measures to what extent an object is hard to describe; that it, how much it does not conform to a pattern.

This is so very wrong that it demonstrates a total lack of comprehension of what K-C theory is about, how it measures information, or what it says about anything. No one who actually understands K-C theory would ever make a statement like Dembski's quote above. No one. But to make matters worse - this statement explicitly invalidates the entire concept of specified complexity. What this statement means - what it explicitly says if you understand the math - is that specification is the opposite of complexity. Anything which posesses the property of specification by definition does not posess the property of complexity. In information-theory terms, complexity is non-compressibility. But according to Dembski, in IT terms, specification is compressibility. Something that possesses "specified complexity" is therefore something which is simultaneously compressible and non-compressible. [Emphasis mine.]

Dawkin’s could not give an example of a mutation that increased information (by any definition).

But it still remains true that natural selection is a narrowing down from an initially wider field of possibilities, including mostly unsuccessful ones, to a narrower field of successful ones. This is analogous to the definition of information with which we began: information is what enables the narrowing down from prior uncertainty (the initial range of possibilities) to later certainty (the "successful" choice among the prior probabilities). According to this analogy, natural selection is by definition a process whereby information is fed into the gene pool of the next generation. If natural selection feeds information into gene pools, what is the information about? It is about how to survive. Strictly it is about how to survive and reproduce, in the conditions that prevailed when previous generations were alive. To the extent that present day conditions are different from ancestral conditions, the ancestral genetic advice will be wrong. In extreme cases, the species may then go extinct. To the extent that conditions for the present generation are not too different from conditions for past generations, the information fed into present-day genomes from past generations is helpful information. Information from the ancestral past can be seen as a manual for surviving in the present: a family bible of ancestral "advice" on how to survive today. We need only a little poetic licence to say that the information fed into modern genomes by natural selection is actually information about ancient environments in which ancestors survived.

Random genetic error (mutation), sexual recombination and migratory mixing, all provide a wide field of genetic variation: the available alternatives. Mutation is not an increase in true information content, rather the reverse, for mutation, in the Shannon analogy, contributes to increasing the prior uncertainty. But now we come to natural selection, which reduces the "prior uncertainty" and therefore, in Shannon's sense, contributes information to the gene pool. In every generation, natural selection removes the less successful genes from the gene pool, so the remaining gene pool is a narrower subset. The narrowing is nonrandom, in the direction of improvement, where improvement is defined, in the Darwinian way, as improvement in fitness to survive and reproduce.

“If evolution is a car, then reproduction is the engine, mutation is the fuel, and natural selection provides the steering wheel.”

If natural selection feeds information into gene pools, what is the information about? It is about how to survive. Strictly it is about how to survive and reproduce, in the conditions that prevailed when previous generations were alive. To the extent that present day conditions are different from ancestral conditions, the ancestral genetic advice will be wrong. In extreme cases, the species may then go extinct. To the extent that conditions for the present generation are not too different from conditions for past generations, the information fed into present-day genomes from past generations is helpful information.

Does motivation of the questioner somehow make the question not legitimate?

Dawkins did not answer the question because he could not. And when he tried, he answered a different question. Various of you have at least tried to answer my question. Good.

Intuitively, (what one relies on when one is ignorant), one knows specification when he see it. (Wasn't it Dawkins who acknowledged this when he told his students that what they were seeing is only the appearance of design?) And intuitively one know what information is, and intuitively, YOU know that you cannot make random changes to any highly complex, functioning system and expect improvement. Would you go into an automated factory, make a few random changes and go to the shipping dock to await the new products? If you do, take a book. You'll be there a while.

Now, why should I doubt my intuition? If there were good examples of where it was wrong, I would begin to doubt. (The lactase example was not convincing. The Milano one was better.) Or if there were a good model of how NS could weed out all the bad mutations and keep the ones that increased information/improved function/created new structures, then you might be on to something. But along with what might be improved function/increased information there is clearly a lot of the opposite. Five thousand plus know genetic diseases? Wouldn't any increased information/improved function be swamped by the garbage?

Merlin

The claim that “The idea that random stuff has lots of information is mistaken” is not correct for any formally definable notion of information, whether you call it “specified information” or anything else.

That we don’t find organisms with random strings of nucleotides is precisely because our genome reflects the predictable environment in which we live, but that certainly doesn’t mean that a random sequence would have less information; random sequences aren’t relevant, and talking about them sows confusion.

There is more information inherent in a random collection of molecules than a highly ordered one.

The idea that random stuff has lots of information is mistaken. It has lots of things that need explaining but that is different from saying that it has a lot of specified information. It has none.

When people say that natural selection builds lots of information into our genome, they must be talking about something other than Shannon information. They are implicitly talking about information that conveys joint information about something (such as what the organism’s phenotype should be to be highly fit). That is what Dembski was trying to get at with his advocacy of specified information, and what Orgel was getting at when he originated talking about specification.

Carl Bergstrom and I are almost alone among posters here in thinking that the idea of specified information is not useless or mistaken. The articles that have been cited here as proving that SI is wrong are ones that point out (correctly) that Dembski did not have a clear definition of what the genome was supposed to be conveying information about. (And they also pointed out that he brought in Kolmogorov information in a backwards way that doesn’t help). But underlying Dembski’s argument is his belief that the adaptive information contained in our genome cannot have been put there by natural selection. He is wrong about that, it can be put there by selection.

So, yes, Popper’s Ghost, my statement about Tex’s posting was mistaken and yes, in that respect I was wrong. Crow away.

It’s going to be hard to find a different one, because “information” is also a word used in colloquial English as well as in the languages of Mathematics and of Computer Science.

If I need to have some idea which of 8 horses will win a race, and you are the inside tipster and give me the word that the race is fixed and horse number 3 will definitely win, and it does, maybe I should say “thanks for the Whatchamacallit, it was Totally Tubular!” Or maybe …

Definition: The value of information associated with a cue or signal C is defined as the difference between the maximum expected payoff or fitness that a decision-maker can obtain by conditioning on C and the maximum expected payoff that could be obtained without conditioning on C.

Definition: The value of a signal S is defined as the difference between the maximum expected payoff or fitness that a decision-maker can obtain by conditioning on S and the maximum expected payoff that could be obtained without conditioning on S.

When people say that natural selection builds lots of information into our genome, they must be talking about something other than Shannon information.

— Joe

If all one is trying to do with the idea of “information” is to make some quantitative assessment about survivability or fitness in a given environment, then I don’t see anything that is either profound or necessary in the idea, or that isn’t already available in other, more obvious kinds of calculation such as simple probability or change-of-entropy calculations.

Physical systems are full of examples in which improbable states are more easily reached by way of nearby states (crystal growth, catalysis and catalytic substrates, optical pumping, bootstrap processes, pressure changes, temperature changes, etc.). It is then possible, at least in principle, to calculate the probabilities of reaching and surviving in various nearby states from the current state (where the current state itself may be a result of the system’s history). Presumably, if one could nail down the probabilities of reaching successive stages in an evolving biological system, one could model theoretical evolutionary systems far out into their future by bootstrapping off successive states of development

But one can’t go back too far in the history of the system in order to do such a calculation on a late state because earlier states may not lead to many (or any) of the later states. That has to be a matter of system history. I think Torbjörn’s quote of Dawkins (comment #139317) is another way of stating these ideas in the case of gene pools.

BTW, they themselves slip into the language of my alternative definition, e.g., "the fitness value of an informative cue". Of course "informative" is redundant; a cue is "informative" precisely to the degree that it has value (as defined).

So given the definition of what we care about (the horse race), SI is definable and works

P.S. how do you define a horse race? Perhaps you meant "specification of what we care about" ... how do you specify it, in a way that "works"?

I failed to emhasize Dawkin’s description of what information means in such a Shannon analogy that he comtemplated. (Which btw PG here also describes.)

Not really, all we are saying is that there exists a correlation between the environment and the genetic coding. See the work by Adami or Schneider.

According to Shannon-Weaver information theory, random noise maximizes information. This is not just playing word games. The random variation that mutations add to populations is the variation on which selection acts. Mutation alone will not cause adaptive evolution, but by eliminating nonadaptive variation, natural selection communicates information about the environment to the organism so that the organism becomes better adapted to it. Natural selection is the process by which information about the environment is transferred to an organism's genome and thus to the organism (Adami et al. 2000).

Does this mean that Shannon´s concept is applicable only, when there is a transmitter (of a message) and a receiver to interpret that message? Then, Shannon´s equation would not be useful in estimating “information content” of a system that we do not expect to contain a message?

And “noise” and “random” are in no way synonyms; the noise can be highly ordered – think punk rock or any other kind of music you don’t like :-), an interfering radio station, etc. We are now discussing “noise” in its colloquial meaning. I have equated (maybe incorrectly) noise with randomness, (white) noise being a pure example of randomness.

...the Mozart would be the noise.

According to Shannon-Weaver information theory, random noise maximizes information.

According to Shannon-Weaver information theory, random noise maximizes information.

It maximizes entropy not information.

Nah; Mozart turns everything else into noise. :-)

It maximizes entropy not information.

Hmmm ... actually, while the talkorigins page I linked to attributes the statement to (Adami et al. 2000), nothing of the sort appears in that paper.

It is also interesting to notice that Dembski equivocates between specification and function. Other error is that he calculates complexity with regards to uniform complexity and from scratch, ignoring natural contingency of biological systems.

Thus, it is trivial to show that when we discovered that HIV evolved VPU unit (thanks to Abbie Smith), HIV virus is CSI. I is complex (because Dembski ignores previous steps), it has a function so it is specified thus HIV with VPU (HIV-1?) is CSI and it is obviously designed. By definition.

Thus CSI fails because if you refuse to model according to evolutionary models (all organisms are descending one from another, so are their gene pools) you can make everything as CSI. In other words, all mutations that change function are designed mutations.

Err, instead of "with regards to uniform complexity" it should be "with regards to uniform probability". Also, "It[HIV] is complex" not I :o

Popper's Ghost:

It maximizes entropy not information.

Argue with Adami; it's his statement.

Merlin said: Now, why should I doubt my intuition?

Merlin: Or if there were a good model of how NS could weed out all the bad mutations and keep the ones that increased information/improved function/created new structures, then you might be on to something. But along with what might be improved function/increased information there is clearly a lot of the opposite. Five thousand plus know genetic diseases? Wouldn't any increased information/improved function be swamped by the garbage?

Intuitively, (what one relies on when one is ignorant), one knows specification when he see it. (Wasn’t it Dawkins who acknowledged this when he told his students that what they were seeing is only the appearance of design?) And intuitively one know what information is, and intuitively, YOU know that you cannot make random changes to any highly complex, functioning system and expect improvement. Would you go into an automated factory, make a few random changes and go to the shipping dock to await the new products? If you do, take a book. You’ll be there a while.

Or if there were a good model of how NS could weed out all the bad mutations and keep the ones that increased information/improved function/created new structures, then you might be on to something. But along with what might be improved function/increased information there is clearly a lot of the opposite. Five thousand plus know genetic diseases? Wouldn’t any increased information/improved function be swamped by the garbage?

Merlin: ... intuitively, YOU know that you cannot make random changes to any highly complex, functioning system and expect improvement.

Thanks all, but particularly to Mike Elzinga and PG, for enlightening examples of contingencies in evolutionary history as pertaining to Dawkins et al descriptions of information in variation and selection.

Dawkins did not answer the question because he could not. And when he tried, he answered a different question.

if there were a good model of how NS could weed out all the bad mutations and keep the ones that increased information/improved function/created new structures, then you might be on to something.

Wouldn’t any increased information/improved function be swamped by the garbage?

Is it? Can you show me where in his 2000 article he makes this claim? I’d say he makes exactly the opposite one… Perhaps your ‘reference’ was wrong? It always helps reading the original literature, it’s a mistake commonly reserved for creationists though

Wouldn’t any increased information/improved function be swamped by the garbage?

intuitively, YOU know that you cannot make random changes to any highly complex, functioning system and expect improvement.

Popper's Ghost:

Is it? Can you show me where in his 2000 article he makes this claim? I’d say he makes exactly the opposite one… Perhaps your ‘reference’ was wrong? It always helps reading the original literature, it’s a mistake commonly reserved for creationists though

It helps to read ahead before responding ... like, as far as the next message where I noted -- after reading the original literature -- that the statement I quoted from talkorigins seems to have misrepresented its source.

Joe Felsenstein

A quick search gave me Kimura's neutral theory of mutation. Is that the one? It seems to me that it all depends on the ratios of deleterious, neutral and beneficial mutations. I have heard estimate of one out of 100, 1 out of 1000, and fewer beneficial mutations to all others. What are the current estimates?

Of course, deleterious mutations TEND to get weeded out, and the worst are weeded out the best, but with new ones arising continually and beneficial one rare, the net effect would seem to be negative. Could this be resolved by a computer model?

There is still the problem of getting new structures/new functions by mutations.

On some creationist or ID website, I read a discussion of an article claiming that the accumulation of "neutral" mutations had a negative effect. If I see it again I'll let you know.

PvM and Popper's Ghost

"Specification is simple: it's called function in biology"

Yes, however, the thing commonly specified is a polypeptide sometimes thousands of amino acids long that has to fold in a certain way. Lots of room for error. Not much for improvement. (Sex spreads the errors) Are rates of genetic diseases decreasing? Have they been decreasing throughout the existence of humans?

Torbjorn Larsson

Beneficial and a trivial increases in information, perhaps. And you may be correct on everything else if you assume the evolution that I am questioning. We observe genetic diseases. Where are the new structures/new functions?

....

Demski, using work by Douglas Axe, describes small islands of functioning genes surrounded by a vast ocean of non-functioning possible sequences of amino acids. The implication being that you cannot get from one function to another by descent with modification. Is this a fair picture?

Demski, using work by Douglas Axe, describes small islands of functioning genes surrounded by a vast ocean of non-functioning possible sequences of amino acids. The implication being that you cannot get from one function to another by descent with modification. Is this a fair picture?

Excellent. Sorry for not seeing the subsequent posting.

Of course, deleterious mutations TEND to get weeded out, and the worst are weeded out the best, but with new ones arising continually and beneficial one rare, the net effect would seem to be negative.

Demski, using work by Douglas Axe, describes small islands of functioning genes surrounded by a vast ocean of non-functioning possible sequences of amino acids.

The implication being that you cannot get from one function to another by descent with modification.

Merlin: A quick search gave me Kimura's neutral theory of mutation. Is that the one? It seems to me that it all depends on the ratios of deleterious, neutral and beneficial mutations. I have heard estimate of one out of 100, 1 out of 1000, and fewer beneficial mutations to all others. What are the current estimates? Of course, deleterious mutations TEND to get weeded out, and the worst are weeded out the best, but with new ones arising continually and beneficial one rare, the net effect would seem to be negative. Could this be resolved by a computer model?

Kimura, M. 1962. On the probability of fixation of mutant genes in a population.   Genetics   47: 713-719.

I believe somewhere in the related posts is also a note by Hawks that the observed rate short-term violates Haldane’s limits by a factor 10 - 100, making a joke of another creationist scam claim.

What about the Haldane limit? I thought such rapid evolution was impossible! J. B. S. Haldane famously estimated that the substitution cost of new alleles in humans limited the rate of adaptive evolution. In his estimate, the slow rate of human reproduction limited substitutions to one every 300 generations. This became known as the "Haldane limit". Motoo Kimura used Haldane's argument as a reason why selection could not explain the substitution rate, and he asserted that as support for his neutral theory of molecular evolution. We do not challenge neutralism, but it is clear that Haldane's limit is a problem, since every estimate says that humans have been evolving at many times that rate. Maynard Smith (1968, Nature) showed that Haldane's argument depended on the unrealistic assumption of independence among all selected loci, so that the substitution load depends critically on the fitness of the optimal genotype among all selected loci. If selection on many loci is non-independent, then a very large number of genes may be selected with the same substitution load as a single gene under Haldane's assumptions. Later, Ewens (e.g., 1972, Am Naturalist) made a similar argument. Ewens (2004, Mathematical Population Genetics) reviewed this problem, pointing out an additional weakness of Haldane's argument: it depends solely on mortality selection, while many genes may be under fertility selection. These considerations show that Haldane's limit does not constrain the adaptive substitution rate in humans to 1/300 generations, and our estimated rate of 13 per generation is not excluded. Moreover, considering the high infant and juvenile mortality evidenced in Neolithic and later populations, much of that death rate resulting directly from disease and dietary deficiencies, the number of selective deaths available to drive substitutions has clearly been high.

And you may be correct on everything else if you assume the evolution that I am questioning.

In population genetics, linkage disequilibrium is the non-random association of alleles at two or more loci, not necessarily on the same chromosome. ...
Linkage disequilibrium is generally caused by genetic linkage and the rate of recombination; mutation rate; random drift or non-random mating; and population structure.

Where are the new structures/new functions? ….

Torbjörn Larsson, OM:

I believe somewhere in the related posts is also a note by Hawks that the observed rate short-term violates Haldane’s limits by a factor 10 - 100, making a joke of another creationist scam claim.

Ehrm, always check before posting. [Hangs head in shame.] Hawks et al work blows Haldane's limit out of the water:

What about the Haldane limit? I thought such rapid evolution was impossible! J. B. S. Haldane famously estimated that the substitution cost of new alleles in humans limited the rate of adaptive evolution. In his estimate, the slow rate of human reproduction limited substitutions to one every 300 generations. This became known as the "Haldane limit". Motoo Kimura used Haldane's argument as a reason why selection could not explain the substitution rate, and he asserted that as support for his neutral theory of molecular evolution. We do not challenge neutralism, but it is clear that Haldane's limit is a problem, since every estimate says that humans have been evolving at many times that rate.

Torbjorn Larsson and Joe Felsenstein

I will try to get thru the materials you have referenced. Time and the math may be a problem. Thanks. I can see that I am not well enough informed to discuss this with you. None the less, the problem is getting new information/structures/function, by whatever definition. I understand that one can design a good radio antenna using a evolutionary algorithm, but one will never happen upon a bicycle wheel using the same algorithm.

Mike Elzinga

The fossil record shows little change, once a species shows up. Isn't this why the theory of punctuated equilibrium was proposed; to explain the stasis?

Popper's Ghost

Dembski, of course. Thanks, my mistake, and I should have said amino acids when I said genes. It is in No Free Lunch. For a functioning enzyme that has 1000 amino acids, there are 20 to the 1000th possible sequences of that same length. Sequences that vary slightly from the functioning one, will perform the same function, but as more changes are made, function drops off and ceases but the number of the remaining possible sequences is so huge that it is impossible to get to another functioning sequence by chance.

I understand that one can design a good radio antenna using a evolutionary algorithm, but one will never happen upon a bicycle wheel using the same algorithm.

Merlin: Dembski, of course. Thanks, my mistake, and I should have said amino acids when I said genes. It is in No Free Lunch. For a functioning enzyme that has 1000 amino acids, there are 20 to the 1000th possible sequences of that same length. Sequences that vary slightly from the functioning one, will perform the same function, but as more changes are made, function drops off and ceases but the number of the remaining possible sequences is so huge that it is impossible to get to another functioning sequence by chance.

I understand that one can design a good radio antenna using a evolutionary algorithm, but one will never happen upon a bicycle wheel using the same algorithm.

Merlin: Torbjorn Larsson and Joe Felsenstein I will try to get thru the materials you have referenced. Time and the math may be a problem. Thanks. I can see that I am not well enough informed to discuss this with you. None the less, the problem is getting new information/structures/function, by whatever definition. I understand that one can design a good radio antenna using a evolutionary algorithm, but one will never happen upon a bicycle wheel using the same algorithm.

The fossil record shows little change, once a species shows up. Isn’t this why the theory of punctuated equilibrium was proposed; to explain the stasis?