Can mutations increase information in the genome? [Archive I]

RBH

May 22, 2007, 01:15 AM

Well I wasn't aware of the difference.....I thought it was obvious what was meant by "information in the genome" as in the genetic sequences....It makes a big difference if you want to measure something. In order to answer a question like "Do mutations increase information in the genome", you have to be able to measure "information" before and after mutations in order to answer it. And what was being asked is whether you, in asking the question, knew what "information" means in terms that one can measure. As noted, there are two main ways, Shannon information and Kolmogorov-Chaitin (algorithmic) information. And yes, mutations can increase both. As noted, gene duplication is a frequent kind of mutation and increases information in both senses. Subsequent divergent evolution almost certainly further increases K-C, and may also further increase Shannon info.

RBH

spamandham

May 22, 2007, 01:19 AM

... and also where creationists get the idea that mutations can't increase the information in the genome...

They pull it from thin air as an unsupportable assertion, and then claim it's proven by the 2nd Law of Thermodynamics. It's worse than junk science, it's abject bullshitery.

But it uses scientific sounding jargon, and those who want to believe it will just blindly accept it on the authority of theologians who wear lab coats in public (do you really need to wear a lab coat on a sound stage or while giving a televized lecture?) and calling themselves "Dr."

"Well gee Ethel, that there creation scientist says that some law of science proves evolution is impossible, and I believe it, because I hope it's true, because I have faith, because I know it's true, because he said so and he's wearin' a lab coat and people call him "Dr.", so that means he's purty smart."

Dr.GH

May 22, 2007, 01:27 AM

Richard Dawkins gets puzzled by creationist - http://youtube.com/watch?v=zaKryi3605g

So the question is how do evolutionists respond to this claim by creationists that mutations have never been shown to increase information in the genome (and thus collapsing the entire modern evolution theory)Well the obvious answer is that creationists have their collective heads too far up eachother's *special dark warm spots* to know what information is, and their brains have all died from lack of oxygen.

Shall we take a simple version of information? This could be the transmission rate of a string. As an example, take (00000000...000000000).

can be prefectly transmitted (in an imaginary world) for ever and there is never any increase or decrease in the information of the string. But in biology we are dealing with billions of individuals so now the requirement is that billions of strings;
(0000000...0000000000)
.
.
.
(0000000...0000000000)

be replicated billions of times. We know that the chemistry of the cell cannot achive this feat- it is chemically impossible. A single instance of (1000000...0000000000) has demonstrated an increase of information. The really significant point is that every biological population the replication (generation) of all those many billions of bits will result in many millions of replication errors (mutations). Of these, some few millions will be deadly, some fewer will be beneficial, and the vast majority will do nothing at all. But all are increases in information of the genome.

Anyone who does not know this should be ashamed by their ignorance of biology, genetics, and information theory. ((Edited to add: OK SO not if they are just little kids. OH Or if they were home schooled- their parents fucked them up and now they have to get better. That isn't their fault unless they persist in being stupid))

Godless Dave

May 22, 2007, 01:48 AM

If no information is ever increased then how can macroevolution occur?

Microevolution needs an "information" increase just as much as macroevolution, so I'm not sure why you're making a distinction.

Janus

May 22, 2007, 01:53 AM

Unbelievable. The team that filmed Dawkins were frauds in the first place, for trying to pass themselves off as unbiased journalists, but this is even worse. The answer that we see Dawkins giving after the interruption isn't his answer to the question about information, it's his answer to a previous, completely different question. This video has been made with full knowledge of this despicable act of dishonesty. But well, what do you expect, creationists must be liars to remain creationists.

http://www.skeptics.com.au/articles/dawkins.htm

Janus

May 22, 2007, 01:54 AM

Oh and, funny that ratings for that video have been disabled, and all comments must be approved before appearing.

Godless Dave

May 22, 2007, 01:57 AM

Thanks....this greatly answers the question....which is gene duplication.....so now I am puzzled as to why Richard Dawkins didn't just say gene duplication

For all we know, he did. As Janus points out, the video was edited.

and also where creationists get the idea that mutations can't increase the information in the genome...

Some liar just started claiming it. It might have been Behe or Demski, I'm not sure.

Janus

May 22, 2007, 01:59 AM

Thanks....this greatly answers the question....which is gene duplication.....so now I am puzzled as to why Richard Dawkins didn't just say gene duplication and also where creationists get the idea that mutations can't increase the information in the genome...

As the article I linked to says:

In September 1997, I allowed an Australian film crew into my house in Oxford without realising that their purpose was creationist propaganda. In the course of a suspiciously amateurish interview, they issued a truculent challenge to me to "give an example of a genetic mutation or an evolutionary process which can be seen to increase the information in the genome." It is the kind of question only a creationist would ask in that way, and it was at this point I tumbled to the fact that I had been duped into granting an interview to creationists - a thing I normally don't do, for good reasons. In my anger I refused to discuss the question further, and told them to stop the camera. However, I eventually withdrew my peremptory termination of the interview as a whole. This was solely because they pleaded with me that they had come all the way from Australia specifically in order to interview me. Even if this was a considerable exaggeration, it seemed, on reflection, ungenerous to tear up the legal release form and throw them out. I therefore relented.

My generosity was rewarded in a fashion that anyone familiar with fundamentalist tactics might have predicted. When I eventually saw the film a year later 1, I found that it had been edited to give the false impression that I was incapable of answering the question about information content 2. In fairness, this may not have been quite as intentionally deceitful as it sounds. You have to understand that these people really believe that their question cannot be answered! Pathetic as it sounds, their entire journey from Australia seems to have been a quest to film an evolutionist failing to answer it.

With hindsight - given that I had been suckered into admitting them into my house in the first place - it might have been wiser simply to answer the question. But I like to be understood whenever I open my mouth - I have a horror of blinding people with science - and this was not a question that could be answered in a soundbite. First you first have to explain the technical meaning of "information". Then the relevance to evolution, too, is complicated - not really difficult but it takes time. Rather than engage now in further recriminations and disputes about exactly what happened at the time of the interview (for, to be fair, I should say that the Australian producer's memory of events seems to differ from mine), I shall try to redress the matter now in constructive fashion by answering the original question, the "Information Challenge", at adequate length - the sort of length you can achieve in a proper article.

Oolon Colluphid

May 22, 2007, 04:42 AM

... and of course, even if Dawkins had genuinely been flummoxed and unable to answer the question in the interview, he does explain it (very well, as usual) in the above-linked article. Even if the lack of response were genuine, it means absolutely nothing. It would mean he was at a loss when interviewed, not that he could not answer at all, let alone that there is no answer.

If someone asks you to explain the evolution of the Lesser Spotted Surinam Tiger Frog's croak, you may well pause. (Given that I just made it up, I'd be surprised if you didn't ;).) Then you go look it up, and come back later with an answer. If someone filmed you pausing, looking perplexed, reaching for and fumbling through Cochran's Living Amphibians of the World... and then declared that "evolutionists have no explanation for the Lesser Spotted Surinam Tiger Frog's croak!"... well you wouldn't exactly be impressed.

And they wouldn't exactly be honest.

But, that's not even what happened here. This episode isn't just dishonest, it is moronic.

Dean Anderson

May 22, 2007, 05:00 AM

Something has occurred to me.

Creationists often use the "Small things evolving into big things breaks the second law of thermodynamics" argument, and this is easily refuted by pointing out that if it were true then single ova growing into adult humans would also break the second law of thermodynamics in the same way.

Can a similar refutation be used against the "mutation cannot increase information content" argument?

After all, there is much more information (whichever way you measure it) in a newborn baby than there is in an ovum. There are all the developing neural pathways and so on for a start - plus (if the baby is female) already ova in her ovaries that each contain just as much information as her original ovum did.

So if a creationist asks incredulously how cell-division with occasional mutation can increase information over time, can we simply point out that if it couldn't then we could not grow from ova to babies?

Ezkerraldean

May 22, 2007, 05:32 AM

if "increasing information" means "increasing the number of genes" then yes, mutations can cause this. if a gene (or a whole chromosome) is replicated, then the extra material is mutated, then there are more genes in the genome than there were before.

"complexity" (whatever that is) in an organism isn't linked to genome size that strongly. the genome of Amoeba dubia is 200 times larger than the human genome.

d0t

May 22, 2007, 05:39 AM

What kind of information?
Shannon information: easy to show an increase after gene duplication + single mutation, but may not be a very useful measure in biology because it attains a maximum at pure randomness.
There doesn't seem to be another measure that will just give a number, no way to say "drosophila DNA has 42 millihumans of information". Creationists use this to move the goalpost by rejecting every single definition of 'information' you can think of.

Also see the Egnor vs PZ & Friends posts like this one (http://scienceblogs.com/pharyngula/2007/02/michael_egnor_comes_back_for_a.php) or this one (http://scienceblogs.com/goodmath/2007/02/a_math_geek_on_dr_egnors_evasi_1.php).

travc

May 22, 2007, 06:13 AM

One more time... with feeling (as they say):

"Information" only has meaning with respect to something else

Setting aside special cases such as coding theory (channel capacity) and such... Information in a technical sense is best defined as the reduction in the uncertainty of the condition/state of thing A given you know the the condition/state of thing B.
I(A:B) = H(A) - H(A|B)

Eg: A map contains no information unless the lines on it correspond to an actual place. Or say, a map of Paris contains no information with respect to New York, but a lot with respect to Paris.

The information which the process of evolution creates is (roughly) with respect to creating an organism which can successfully reproduce. Or in a gene-centric view, a gene which will persist and/or increase in frequency over the course of generations.

Here is a paper which goes through all this pretty clearly and also defines "complexity" in a useful (IMO obviously) way:
http://arxiv.org/pdf/quant-ph/0301075
And a paper which specifically addresses the "no new information" BS using an alife experiment and compares with long term E. coli data:
http://myxo.css.msu.edu/papers/nature2003/Nature03_Complex.pdf
(I made figure 2 ;))

BTW: Tom Schneider has the "site of ultimate doom" when it comes to molecular information theory.
http://www-lmmb.ncifcrf.gov/~toms/

Oolon Colluphid

May 22, 2007, 06:27 AM

Trav, what does H stand for there? Other than that, I was following you.

The Nature paper is awesome.

travc

May 22, 2007, 07:06 AM

Trav, what does H stand for there? Other than that, I was following you.

The Nature paper is awesome.

H is uncertainty. Typically expressed in the number of bits it takes to define the state of something. H(A|B) is the uncertainty of A given the state of B.

H(X) = - sum( p_i log2(p_i) )
where p_i is the probability of X being in state "i"... sum over all possible states X could be in.
Note: if you change the base of the log, you are still calculating uncertainty... just in different units. Sometimes people use "nats" by using natural log, aka ln, aka log_e.

This is also called Shannon Entropy (or just entropy by people who annoy the hell out of me by IMO misusing the term "entropy" (which should really be reserved for thermodynamics) in a very confusing way). It is also the maximum possible amount of information something can contain... though it does not mean it actually contains that much information or what that information is with respect to.

A key to understanding information theory is to realize you are always talking about things that can be in different states. That could be a string of characters "abc" vs "bca", a DNA sequence, or a collection of 10 coloured marbles drawn from a bag.

For biology, H is normally (or at least should be) for a *collection* of different genetic sequences... not a single sequence! Think "the uncertainty as to which sequence you get when you pick one out of the collection". The "collection" is often a population of organisms.

If something can be in 4 different states (all occurring with equal probability), it has an uncertainty of 2 bits.
If instead it could be one of 8 different states with equal probability, that would be 4 bits.
For 4 possible states occurring with probabilities .97, .1, .1, and .1... you get H = 1.039 (because it will almost always be in the first state, so less uncertain).

A lot of people get confused by applying coding theory to genetic sequences and other stuff where it really doesn't apply. There is a lot of good theories and formulas about encoded information being passed along a noisy communications channel from a sender to a receiver (very important stuff for say, computers). Saying that Shannon Entropy = Information is a mistake borne out of this naive misapplication of results from coding theory.

Ok, I should get off my soapbox and go to sleep. Lots of work to do tomorrow.

Oolon Colluphid

May 22, 2007, 07:13 AM

Many thanks! :notworthy:

Howay the Toon

May 22, 2007, 07:29 AM

Uhm....what do you mean by this? The genome...the information...why do you ask for a definition?

How can you possibly discuss whether a particular quality is increasing or decreasing without defining that quality in some way it can be measured.

If you don't do this, you are just spouting nonsense. Which is probably why creationists don't it. Spouting nonsense seems normal to them

travc

May 22, 2007, 07:51 AM

Many thanks! :notworthy:

Thanks for letting me vent... this is a subject I am a wee bit opinionated on ;)

One thing I should mention:

The definition of information I gave (and used in those papers) leads to what may seem a counter-intuitive result.

If there is no variation in a sequence/gene within a population, it contains not information (with respect to natural selection/evolution)!

You may say, that can't be right... say there is a particular allele (version of a gene) which is so favorable that it get fixed in the population (ie. everyone has that version of the gene.) That gene obviously contains some important information about how to successfully survive and reproduce.

Well, from the point of view of natural selection, if there is no variation within a population, there can be no selection. So a fixed allele really does have no effect (and therefore no information with respect to natural selection). Odd, but true.

Weirder still, imagine when that great new allele first arises (by mutation). Whatever change occurred will be counted as added information, but everyone else's information content for that gene will drop! The environment which the "survive and reproduce" part takes place in has now changed since there is a new competitor (the new great allele), and the old alleles don't encode for information about how to survive and reproduce in that new environment (since they are inferior and any sequence containing that allele will eventually die out).

These effects are seen in more classical measures too... like the selection coefficient of an allele or trait, and pop up in QTL all the time.

I guess my punchline is that a meaningful definition of information in relation to evolution/natural selection actually reveals some interesting things about how natural selection actually works. Though it is different from the everyday notion of "information".

Oolon Colluphid

May 22, 2007, 08:09 AM

On the 'Information Challenge' incident, it seems that the Youtube version linked above is different in some interesting respects to the original one. Over at the RD boards, Richard himself points some out:
http://richarddawkins.net/forum/viewtopic.php?t=14255

Barbarian

May 22, 2007, 08:17 AM

SophistiCat

May 22, 2007, 10:12 AM

The definition of information does not even matter. Since every genome is reachable from every other genome via mutations, if information, whatever kind of information, would not be able to increase, then it could not decrease either (because we could simply look at the same path from the opposite direction, when it would be increasing information). Therefore, the only kind of information in genome unable to increase is the constant, independent-from-the-genome information.

If creationists claimed that theoretical paths increasing information can exist but are impossible in nature for other reasons, I'd listen. But it is painfully obvious to me that they regard mutations as damage and the "no increase in information" soundbite as a savant way of stating this.

This is not a very good argument, or at least it needs to be elaborated. Replace "impossible" with "unlikely" - and it loses its punch. In a stochastic system, such as the distribution of air molecules in a room, any macro-state is theoretically reachable from any other state, but some macro-states are far more probable than others. While it is possible for the molecules to bunch up in one corner of the room, it is far to remote a possibility to seriously consider, let alone a long series of such states of ever decreasing probability.

Creationists would say that, like thermodynamic entropy, information can fluctuate from some equilibrium state (corresponding to the created kind perhaps), but the larger the fluctuation, the less probable it is. (Hey, did I just invent a new creationist "argument"? :eek:) Yes, I know that this analogy doesn't map to evolutionary transitions, but creationists wouldn't know that, and your explanation doesn't make it clear why it's disanaloguous.

Artie Jessup

May 22, 2007, 11:59 AM

RAFH

May 22, 2007, 03:23 PM

There have been some very good answers on information theory and how it pertains to the complexity of the genome. Here are some examples of gene duplication in nature:

good overview (http://www.umich.edu/~zhanglab/publications/2003/Zhang_2003_TIG_18_292.pdf)

heat shock proteins (http://www.pubmedcentral.nih.gov/picrender.fcgi?artid=1206269&blobtype=pdf)

microbial genomes (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10978127)

mosquito insecticide resistance (http://mbe.oxfordjournals.org/cgi/content/abstract/24/4/1056)

primate colour vision (http://www.genome.org/cgi/content/abstract/9/7/629)

gene duplication and vertebrate evolution (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?itool=abstractplus&db=pubmed&cmd=Retrieve&dopt=abstractplus&list_uids=12200472)

human evolution (http://biology.plosjournals.org/archive/1545-7885/2/7/pdf/10.1371_journal.pbio.0020207-L.pdf)

Now for the bad news.... Dawkins was likely BSing about the "information question" interview....here's (http://www.trueorigin.org/ca_gb_01.asp) why (http://telicthoughts.com/why-ed-brayton-is-on-my-blogroll/) I doubt his story.

They claim to have a tape (supposedly accidentally made because someone forgot to switch off the camera when asked to), they claim the tape vindicates their side of the story. They claim slander (actually it would be libel) in the assertions there was 'foul play'. Fine, let them release the orginal tape for review and analysis.

Frankly, I don't see what the big deal is anyway. 11 seconds, who cares. I take as much time as I like to answer questions. It means nothing, particularly with regard to the validity of the Theory of Evolution. To make a big deal about it is ridiculous.

Febble

May 22, 2007, 03:59 PM

They claim to have a tape (supposedly accidentally made because someone forgot to switch off the camera when asked to), they claim the tape vindicates their side of the story. They claim slander (actually it would be libel) in the assertions there was 'foul play'. Fine, let them release the orginal tape for review and analysis.

Frankly, I don't see what the big deal is anyway. 11 seconds, who cares. I take as much time as I like to answer questions. It means nothing, particularly with regard to the validity of the Theory of Evolution. To make a big deal about it is ridiculous.

And to answer a really stupid question, first you have to figure out what the hell they want to know.

wnope

May 22, 2007, 03:59 PM

That depends on the definition of information.
Care to provide any? Would you like to use Shannon information? Kolmogorov (sp?) information? what kind?
=Uncool-

It would seem relatively easy to increase complexity/information in a Kolmogorov paradigm.

From what I see, if you have a repeating sequence of 0 and 1s for 20 bits, it can be described as "a string of repeating 0s and 1s for 20 bits" but if you change any of the digits from 1 to 0 or 0 to 1 you cannot describe the sequence without adding much more information (i.e. the 20 numbers in full).

Granted, my knowledge on Kolmogorov started about 40 seconds ago.

espritch

May 22, 2007, 04:35 PM

lee_merrill

May 22, 2007, 04:57 PM

Can a similar refutation be used against the "mutation cannot increase information content" argument?

After all, there is much more information (whichever way you measure it) in a newborn baby than there is in an ovum.
Not if both can be derived from comparable DNA sequences, I believe is the point. That is, a blueprint for a house A and and blueprint for house B could be said to have comparable amounts of information if the houses are comparable (even if they are not exactly the same). So then 50 houses of this sort with their accompanying blueprints would not have more essential information than the one blueprint for house A, that is, mere aggregation of information does not somehow make more information in the sense it seems to be meant in the question of interest here.

show_no_mercy

May 22, 2007, 05:11 PM

Probably not the best example, but a person with Down Syndrome has more "information" than a normal person.

Sven

May 22, 2007, 05:21 PM

Not if both can be derived from comparable DNA sequences, I believe is the point. That is, a blueprint for a house A and and blueprint for house B could be said to have comparable amounts of information if the houses are comparable (even if they are not exactly the same). So then 50 houses of this sort with their accompanying blueprints would not have more essential information than the one blueprint for house A, that is, mere aggregation of information does not somehow make more information in the sense it seems to be meant in the question of interest here.
This would only be true if there wre no input from the environment during development. Since different phenotypes can result from the sam genotype due to different environments, there's obviously an input of information from the environment. IOW, it's entirely possible to have an increase in information during development.

RBH

May 22, 2007, 06:13 PM

Not if both can be derived from comparable DNA sequences, I believe is the point. That is, a blueprint for a house A and and blueprint for house B could be said to have comparable amounts of information if the houses are comparable (even if they are not exactly the same). So then 50 houses of this sort with their accompanying blueprints would not have more essential information than the one blueprint for house A, that is, mere aggregation of information does not somehow make more information in the sense it seems to be meant in the question of interest here.The problem with this analogy is the blueprint metaphor for DNA sequences. DNA is not the sole determinant of phenotypes (houses), and as Sven pointed out, inputs from the environment can alter the developmental sequences for which DNA is a recipe (not blueprint), and thus the "information" (however defined) in DNA is an incomplete specification of the "information" (defined the same way, whatever that way is) of the phenotype (house).

RBH

lee_merrill

May 23, 2007, 11:15 AM

The problem with this analogy is the blueprint metaphor for DNA sequences. DNA is not the sole determinant of phenotypes ...
However, the question at hand is about increase of information in the genome.

Sven

May 23, 2007, 11:26 AM

However, the question at hand is about increase of information in the genome.
No. Your claim was that there's no increase in information during growing up. The question at hand was not about increase of information in the genome.

Just admit that you were wrong instead of changing the topic.

Febble

May 23, 2007, 11:31 AM

However, the question at hand is about increase of information in the genome.

But it's directly relevant. The environment is the source of the information. If the environment and the phenotype are a good fit, the genotype will be replicated, together with any mutations it bears. If the environment and the phenotype are not a good fit, the genotype, with any mutations it bears, is less likely to be replicated. This is absolutely fundamental. The environment is the filter that "keeps" the genotypes that lead to phenotypes that suit it, and "rejects" the genotypes that lead to phenotypes that don't.

That's how the milk gets into the coconut, i.e. the information about the environment into the genome.

RBH

May 23, 2007, 01:24 PM

Someone -- Dawkins? Sagan? -- once explicitly characterized evolution as a process that transfers information about the selective environment into the genome over generations.

In Unweaving the Rainbow Dawkins wroteThe information that … [evolution] packs away is information about ancestral environments and how to survive them.RBH

lee_merrill

May 23, 2007, 01:26 PM

No. Your claim was that there's no increase in information during growing up. The question at hand was not about increase of information in the genome.
"So the question is how do evolutionists respond to this claim by creationists that mutations have never been shown to increase information in the genome (and thus collapsing the entire modern evolution theory)" (opening post)

And my claim was not just what you were saying, please read my posts more carefully...

Someone ... once explicitly characterized evolution as a process that transfers information about the selective environment into the genome over generations.
"So the question is how do evolutionists respond to this claim ... that mutations have never been shown to increase information in the genome ..."

So this is a derailment, but I hope we can get on topic again...

RBH

May 23, 2007, 01:35 PM

"So the question is how do evolutionists respond to this claim ... that mutations have never been shown to increase information in the genome ..."

So this is a derailment, but I hope we can get on topic again...See Febble's post immediately preceding mine for context. Mutations are part of the process by which that information is transferred; the other (main) part is natural selection. A gene duplication all by itself is an increase in genomic information, whether "information" is measured as Shannon info or K-C (algorithmic) info. Natural selection operates to preserve those that increase reproductive success.

RBH

lee_merrill

May 23, 2007, 02:36 PM

A gene duplication all by itself is an increase in genomic information, whether "information" is measured as Shannon info or K-C (algorithmic) info.
I don't see how a duplication increases K-C information, though. Making two copies of an algorithm is not another algorithm, and certainly not a more complex or capable one. It's fine if natural selection winnows them, but this winnowing is not increasing information, it's the mutations that have to do that, and the winnowing then selects those that have a good increase, which point is in dispute.

May it be said that I agree that information can increase by mutation--I only want people here to more carefully consider their arguments, ill-considered posts I find rather trying, especially when it seems there is a veritable stream of misunderstandings. Trav's posts, may it be said, were splendid, not that I understand them thoroughly, tho.

JoeNothin

May 23, 2007, 02:43 PM

See Febble's post immediately preceding mine for context. Mutations are part of the process by which that information is transferred; the other (main) part is natural selection. A gene duplication all by itself is an increase in genomic information, whether "information" is measured as Shannon info or K-C (algorithmic) info. Natural selection operates to preserve those that increase reproductive success.

RBH
So, if I'm reading you correctly, you're saying that the assertion "... that mutations have never been shown to increase information in the genome ..." is simply false on it's face. I remember reading studies on talkorigins.com that provide examples of "increased information" in the genome, so I think I know what you mean. It's pretty cool stuff -- and so easy to find!

RBH

May 23, 2007, 02:48 PM

I don't see how a duplication increases K-C information, though. Making two copies of an algorithm is not another algorithm, and certainly not a more complex or capable one. It's fine if natural selection winnows them, but this winnowing is not increasing information, it's the mutations that have to do that, and the winnowing then selects those that have a good increase, which point is in dispute.A-B-B'-C-D (where B' is the new duplicate) is equal (in K-C info) to A-B-C-D? Nope. The most compact description of the former is guaranteed to be longer than the most compact description (in the same language) as the latter, and that's an increase in K-C information.

Remember, the question is about information increase in the genome, not "complexity" or "capability". Let's keep the goalposts in one place here. The "point in dispute" is not whether there is a "good" (i.e., adaptive) change, but merely that there is a change in information. Whether that change produces increased reproductive advantage relative to unchanged organisms is another issue.

RBH

Sven

May 23, 2007, 05:17 PM

"So the question is how do evolutionists respond to this claim by creationists that mutations have never been shown to increase information in the genome (and thus collapsing the entire modern evolution theory)" (opening post)

And my claim was not just what you were saying, please read my posts more carefully...

Please stop claiming things known to be false. You quoted Dean Anderson's post (#20) in your response, which was only about the development of newborns and their increase in information.

If you wanted to talk about something different (like the OP), you clearly had no reason to quote him

travc

May 24, 2007, 02:14 AM

Someone -- Dawkins? Sagan? -- once explicitly characterized evolution as a process that transfers information about the selective environment into the genome over generations.

The first article I linked defines the math for that actually. You can used some channel coding theory stuff if you take a population going from one generation to the next as the information transimission, mutation as a noise source, and selection as a filter. Assuming discrete generations makes the math pretty easy actually.

Of course, this is not what ID types are thinking of (if they are thinking at all) when they talk about information and complexity.

travc

May 24, 2007, 02:33 AM

May it be said that I agree that information can increase by mutation--I only want people here to more carefully consider their arguments, ill-considered posts I find rather trying, especially when it seems there is a veritable stream of misunderstandings. Trav's posts, may it be said, were splendid, not that I understand them thoroughly, tho.

I'm blushing ;) Though it must be said that my posts are kindof a clever sidestep to the OP, since they are all about defining a biologically/evolutionarily relevant meaning for "information" (while still conforming to the fundamental math of info theory). The idea is to have a definition of information that is actually useful in predicting what has/is/may happen in the real biological world.

Anyways, I'm no expert on K-C complexity (not really familiar with the term), but Kolmogorov complexity is a nice concept in some ways, but has one huge flaw. Unless I'm horribly mistaken, is the minimum sized machine it takes to generate a sequence. Unfortunately, determining what that machine is for an arbitrary finite sequence is not computationally tractable. (Chomsky would be the canonical reference for this I think.)

Worse yet, it is irrelevant to genetic sequences IMO. The information/complexity which is biologically relevant isn't just the predictability/autocorrelation/description length of the sequence, but instead is how much potential information in the sequence actually corresponds to the fundamental task of successfully reproducing. Anything else in the sequence is meaningless from the POV of evolution.

Kolmogorov complexity is a great concept for other things though. If you want to build a learner/model (a machine) which can predict sequences learning from a subset this is a great notion. If you think about it, it is basically Occum's Razor. It is also useful when trying to pin down just how complex (in the Chomsky computational sense) things which generate sequences have to be... for example, just what sort of computational machinery is required in a songbird's brain given it can generate a song with a particular sequence of notes.
http://www.zebrafinch.org/ <- random link I just found

Febble

May 24, 2007, 05:36 AM

May it be said that I agree that information can increase by mutation

Excellent.

You consider then, that the information on a genome is the information required to build an organism, whether a successful one, or an unsuccessful one.

We can then turn to the question of how a genome comes to carry the specific information required to build an organism that has a high probability of being be successful in its environment.

Do agree that the source of this information is the environment itself?

lee_merrill

May 24, 2007, 09:54 AM

A-B-B'-C-D (where B' is the new duplicate) is equal (in K-C info) to A-B-C-D? Nope. The most compact description of the former is guaranteed to be longer than the most compact description (in the same language) as the latter, and that's an increase in K-C information.
Not if the minimum K-C program involves "do a B 1 time", and this new version then is "do a B 2 times", then the program sizes are the same.

Unfortunately, determining what that machine is for an arbitrary finite sequence is not computationally tractable.
Yes, I agree.

Worse yet, it is irrelevant to genetic sequences IMO. The information/complexity which is biologically relevant isn't just the predictability/autocorrelation/description length of the sequence, but instead is how much potential information in the sequence actually corresponds to the fundamental task of successfully reproducing.
Which I think is well described algorithmically, which would then be K-C complexity?

You consider then, that the information on a genome is the information required to build an organism, whether a successful one, or an unsuccessful one.
Yes, the blueprint for a bluebird, though I expect RBH will object to this!

Do agree that the source of this information is the environment itself?
The environment is an essential part of the success of this organism, but that I think is not the point at issue. Volume "T" of the Encyclopedia Brittanica is not notably successful in the wild, yet it has a lot of real information in there, and I think that is the sort of information meant, in the case of DNA, "How do we make a froggie"?

We might also consider the probability that a bird will be captured for its feathers by a poacher, but that is a different sort of information as well...

Regards,
Lee

Febble

May 24, 2007, 10:00 AM

Yes, the blueprint for a bluebird, though I expect RBH will object to this!

Well, I prefer the metaphor of a recipe myself. What gets cooked depends not only on the recipe but on the availability and quality of ingredients, fluctuations in the temperature of the oven etc. But apart from that, fine.

The environment is an essential part of the success of this organism, but that I think is not the point at issue. Volume "T" of the Encyclopedia Brittanica is not notably successful in the wild, yet it has a lot of real information in there, and I think that is the sort of information meant, in the case of DNA, "How do we make a froggie"?

Regards,
Lee

This is not a response to my question at all. Let me repeat it:

We can then turn to the question of how a genome comes to carry the specific information required to build an organism that has a high probability of being be successful in its environment.

Do agree that the source of this information is the environment itself?

lee_merrill

May 24, 2007, 10:16 AM

Well, I prefer the metaphor of a recipe myself. What gets cooked depends not only on the recipe but on the availability and quality of ingredients, fluctuations in the temperature of the oven etc.
Right, but the recipe contains information which may be considered in and of itself. We are not looking for warts on toads here, just "How to make a froggie."

This is not a response to my question at all. Let me repeat it: Do agree that the source of this information is the environment itself?
Yes, the source of the information of which you spoke includes the environment (the environment is not the total source), yet I believe this is not the information that is in view here.

So your recipe analogy shows how to generate new information in the genome! Duplications and insertions and changes of letters can change a recipe for a biscuit into a recipe for a cake, the cake recipe having more information than the biscuit one.

And once we acknowledge extra functions arriving in bacteria, novel useful functions which do appear, then we have the same thing happening in concrete examples, in biology. Especially if you grant (as I do) that birds came from fishes...

Sven

May 24, 2007, 10:17 AM

The environment is an essential part of the success of this organism, but that I think is not the point at issue. Volume "T" of the Encyclopedia Brittanica is not notably successful in the wild, yet it has a lot of real information in there, and I think that is the sort of information meant, in the case of DNA, "How do we make a froggie"?
You dodged the question with an irrelevant analogy: encyclopedias don't fuck and reproduce, they don't have genes which get mutated.

Principia

May 24, 2007, 10:28 AM

Not if the minimum K-C program involves "do a B 1 time", and this new version then is "do a B 2 times", then the program sizes are the same.
LOL, no. "do a B 1 time" cannot be the shortest algorithm, because one could simply have it "do a B" without specifying one time.

Furthermore, "do a B 1 time" is in fact shorter than "do a B 2 time". What is the minimially complex representation of 1 compared to 2? Do you know?

IanC

May 24, 2007, 10:35 AM

Not if the minimum K-C program involves "do a B 1 time", and this new version then is "do a B 2 times", then the program sizes are the same.
No, the minimum size to say B once is different from saying it 2 times.

JonF

May 24, 2007, 10:58 AM

Not if the minimum K-C program involves "do a B 1 time", and this new version then is "do a B 2 times", then the program sizes are the same.
Others have already pointed out that this is wrong, but not exactly why.

Let's say "Do B" requires N bits. Then "Do B 1 time" requires N+1 bits, and "Do B 2 times" requires N+2 bits. 1 can be represented with one bit, but 2 requires two bits.

Didn't you once claim to know something about this stuff?

lee_merrill

May 24, 2007, 11:00 AM

Febble

May 24, 2007, 11:07 AM

Right, but the recipe contains information which may be considered in and of itself. We are not looking for warts on toads here, just "How to make a froggie."

At this point yes. But then we consider the next part:

Yes, the source of the information of which you spoke, which I believe is not the information that is in view here.

Yes it is. Listen carefully:

You start with a recipe for a walnut and raisin cake. You bake it. It turns out OK. You bake it again, but this time, you don't have any raisins so you use cranberries.. Everyone loves it. So next time you use the altered recipe with cranberries. Until one day you don't have any cranberries so you use turnips. Everyone hates it, so you don't use turnips again. You stick with cranberries until one day you are out of sugar so you use honey. Everyone loves the honey, so you stick with honey and cranberries. One day you leave out the walnuts. People prefer it without walnuts. So you end up with a recipe for cranberry and honey cake, and people love it. Your walnut and raisin recipe has evolved. It now contains information not just how to make any old cake but a cake that the people you bake for really love. Information about the kind of cake your family really likes has been encoded into the recipe.

That's how the information from the environment gets into the genome so that it contains the information as to how to build an organism that will be a success in that environment.

So your recipe analogy shows how to generate new information in the genome! Duplications and insertions and changes of letters can change a recipe for a biscuit into a recipe for a cake, the cake recipe having more information than the biscuit one.

Maybe, maybe not. The point is that your evolved recipe will contain information as to how to bake something that your family will enjoy, not something they throw in the trash.

And once we acknowledge extra functions arriving in bacteria, novel useful functions which do appear, then we have the same thing happening in concrete examples, in biology. Especially if you grant (as I do) that birds came from fishes...

Not sure what this is about, but sure, birds came from fishes.

Principia

May 24, 2007, 11:14 AM

Yet a programming language that has a special version of an instruction for each possible number for each operand is not going to be very minimal. I would say that the proper statement is "duplication does not always increase information".

But this is all nit-picky.
Well, it is "nit-picky" only if it suits you, eh? You are still wrong. KC information does not care about the programming language used to implement the algorithm. Do you know why? And if you don't, why are you posturing as an expert in information theory?

Duplication always increases KC information. Do you know why?

Jet Black

May 24, 2007, 11:19 AM

You start with a recipe for a walnut and raisin cake. You bake it. It turns out OK. You bake it again, but this time, you don't have any raisins so you use cranberries.. Everyone loves it. So next time you use the altered recipe with cranberries. Until one day you don't have any cranberries so you use turnips. Everyone hates it, so you don't use turnips again. You stick with cranberries until one day you are out of sugar so you use honey. Everyone loves the honey, so you stick with honey and cranberries. One day you leave out the walnuts. People prefer it without walnuts. So you end up with a recipe for cranberry and honey cake, and people love it. Your walnut and raisin recipe has evolved. It now contains information not just how to make any old cake but a cake that the people you bake for really love. Information about the kind of cake your family really likes has been encoded into the recipe.
.

and yet your sister, who inherited the original recipie from mother as well, but who'se husband hates cranberries is still using the original recipie.

lee_merrill

May 24, 2007, 11:19 AM

Duplication always increases KC information. Do you know why?
No, I don't, would you tell me why, please? If you consider this both on-topic, and important...

lee_merrill

May 24, 2007, 11:21 AM

Information about the kind of cake your family really likes has been encoded into the recipe.
Well, fine, so yes, this information in the recipe, and thus in the genome can increase.

Febble

May 24, 2007, 11:36 AM

Well, fine, so yes, this information in the recipe, and thus in the genome can increase.

Well, more than that. Specifically, information from the environment is encoded into the genome so that the information specifies the kind of custom-specified froggie that will be built - not just any old froggie.

It's why custom-specified froggies are predicted by the ToE. And few other fancy items.

lee_merrill

May 24, 2007, 11:48 AM

Well, more than that.
Well, by "this" I meant what you were saying.

Why is it that people have to dispute every statement I make, even when I agree with them?

It seems the requirement in any discussion is that I have to be wrong! :Cheeky:

Febble

May 24, 2007, 12:24 PM

Well, by "this" I meant what you were saying.

Why is it that people have to dispute every statement I make, even when I agree with them?

It seems the requirement in any discussion is that I have to be wrong! :Cheeky:

Not wrong, unclear. Lots of us have remarked that the referents for your pronouns are often unclear.

OK, so we are agreed as to how it happens that organisms evolve to be so beautifully "designed" for their environment? That the ToE actually works?

Febble

May 24, 2007, 12:25 PM

and yet your sister, who inherited the original recipie from mother as well, but who'se husband hates cranberries is still using the original recipie.

Yup. And there are still monkeys.

JonF

May 24, 2007, 12:31 PM

Duplication always increases KC information. Do you know why?
No, I don't, would you tell me why, please? If you consider this both on-topic, and important...
Not reading again, hum?

Duplication always requires at least one more bit to specify the duplicate item.

As I pointeds out already in message 60 (http://www.iidb.org/vbb/showpost.php?p=4481506&postcount=60).

JonF

May 24, 2007, 12:39 PM

Let's say "Do B" requires N bits. Then "Do B 1 time" requires N+1 bits, and "Do B 2 times" requires N+2 bits. 1 can be represented with one bit, but 2 requires two bits.
Yet a programming language that has a special version of an instruction for each possible number for each operand is not going to be very minimal. I would say that the proper statement is "duplication does not always increase information".
Irrelevant.

A minimal programming language in which there is no special version of an instruction for each possible number for each operand can specify "Do B' in, let's say, M bits. Then "Do B twice" takes M bits plus however many bits it takes to specify "do it again", minimum one bit.

A minimal programming language in which there is a special version of an instruction for each possible number for each operand can specify "Do B' in, let's say, P bits. But if it's minimal, it can't specify "Do B twice" because that's not necessary to specify "Do B". Therefore, to specify "Do B twice" you need to extend the language, which takes more bits.

Therefore K-C information increases with any duplication.

Sheesh.

lee_merrill

May 24, 2007, 12:52 PM

Not wrong, unclear. Lots of us have remarked that the referents for your pronouns are often unclear.
"Information about the kind of cake your family really likes has been encoded into the recipe." (Febble)

"Well, fine, so yes, this information in the recipe, and thus in the genome can increase." (Lee)

But how is this unclear? You simply must dispute every statement I make, I then conclude.

OK, so we are agreed as to how it happens that organisms evolve to be so beautifully "designed" for their environment? That the ToE actually works?
Certainly, but not to the degree that naturalists say.

Duplication always requires at least one more bit to specify the duplicate item.
:confused: Let's consider 2 and then 3 here.

Then "Do B twice" takes M bits plus however many bits it takes to specify "do it again", minimum one bit.
Alas, poor Yorick...

Dean Anderson

May 24, 2007, 12:58 PM

and yet your sister, who inherited the original recipie from mother as well, but who'se husband hates cranberries is still using the original recipie.

Yes, but that proves nothing! They're still both of the "dessert" kind...

JonF

May 24, 2007, 01:11 PM

Originally Posted by JonF
Duplication always requires at least one more bit to specify the duplicate item.
Let's consider 2 and then 3 here.
Not clear what you mean. If you mean duplicates and triplicates, then specifying "Do B" takes M bits. Then specifying a duplicate takes at least one bit more, for M+1 K-C information. But specifying a trplicate takes at least two bits more, either two "do it again" bits or a binary number for the number of repetitions, for M+2 K-C information. A duplicate is more K-C information, another duplicate is yet more K-C information. You ain't gonna change this; people smarter and more knowledgable that you or I have studied it in depth, and this is a fundamental result.

Rather than trying to poke holes in something you obviously don't understand, you might try (for once) learning something about the subject before pontificating. I'm not an expert on information theory, but stuff like this is basic and intuitively obvious to the casual observer.

Then "Do B twice" takes M bits plus however many bits it takes to specify "do it again", minimum one bit.
Alas, poor Yorick...
Well, that's certainly a pointless and meaningless comment.

lee_merrill

May 24, 2007, 01:26 PM

Not clear what you mean. If you mean duplicates and triplicates, then specifying "Do B" takes M bits. Then specifying a duplicate takes at least one bit more, for M+1 K-C information.
Nope, think about it! 2 in binary is 10, 3 is 11, the same number of bits.

Rather than trying to poke holes in something you obviously don't understand, you might try (for once) learning something about the subject before pontificating.
Alas, poor Yorick! You are needlessly shooting yourself in ye foote...

Sven

May 24, 2007, 01:30 PM

You start with a recipe for a walnut and raisin cake. You bake it. It turns out OK. You bake it again, but this time, you don't have any raisins so you use cranberries.. Everyone loves it. So next time you use the altered recipe with cranberries. Until one day you don't have any cranberries so you use turnips. Everyone hates it, so you don't use turnips again. You stick with cranberries until one day you are out of sugar so you use honey. Everyone loves the honey, so you stick with honey and cranberries. One day you leave out the walnuts. People prefer it without walnuts. So you end up with a recipe for cranberry and honey cake, and people love it. Your walnut and raisin recipe has evolved. It now contains information not just how to make any old cake but a cake that the people you bake for really love. Information about the kind of cake your family really likes has been encoded into the recipe.
With each time you use this analogy, it becomes better and better! :notworthy:

Sven

May 24, 2007, 01:33 PM

"Information about the kind of cake your family really likes has been encoded into the recipe." (Febble)

"Well, fine, so yes, this information in the recipe, and thus in the genome can increase." (Lee)

But how is this unclear? You simply must dispute every statement I make, I then conclude.
Not at all. I also wondered what you meant by this information.

xunzian

May 24, 2007, 01:34 PM

No one has mentioned things like ribosome shifting in viruses. You want a great way to increase the amount of information in a gene? Great, have a frameshift induced into the ORF and, BOOM, new polypeptide.

Heck, you don't even need ribosome shifting, you can just have multiple start-sites.

Any given piece of DNA can contain 3 different polypeptides, based on the reading frame. And that isn't even considering things like domain-swapping, post-translational modification, binding partners, ect.

Sven

May 24, 2007, 01:34 PM

Rather than trying to poke holes in something you obviously don't understand, you might try (for once) learning something about the subject before pontificating.
Don't forget that you are talking to Lee Merrrill, the expert about ancient languages, archeology, the bible back to front and front to back, information theory, biochemistry, etc. etc.

He does not have to learn anything!

RAFH

May 24, 2007, 01:42 PM

Yes, but that proves nothing! They're still both of the "dessert" kind...

Not if the third sister is married to a guy that likes what most people consider dessert for dinner, because he's that kind of guy. So they have a variation of the walnut-raisin cake for dinner a lot. Except because its for dinner, sister 3 makes very large batches, which means she runs out of the ingredients that much faster and has to substitute other ingredients more often. Which means her dinner version of the walnut-raisin cake changes faster and has generated more versions and, consequently, more failed version which ended up in the scrap heap the future paleontologist mines with great interest.

Febble

May 24, 2007, 02:45 PM

"Information about the kind of cake your family really likes has been encoded into the recipe." (Febble)

"Well, fine, so yes, this information in the recipe, and thus in the genome can increase." (Lee)

Febble: Well, more than that. Specifically, information from the environment is encoded into the genome so that the information specifies the kind of custom-specified froggie that will be built - not just any old froggie.

It's why custom-specified froggies are predicted by the ToE. And few other fancy items.

Lee: But how is this unclear? You simply must dispute every statement I make, I then conclude.

I didn't dispute it. I simply amplified it ("more than that") and will do so still further, in case you have missed my point - that it is not simply that the genome can increase in quantity (however we measure the quantity of information) - actually the information on the genome might be reduced (it might have the instructions to add walnuts deleted) - the important point is that the genome, whether it contains MORE information or LESS information, now contains RELEVANT information as to how to make a specific kind of cake YOUR FAMILY WILL LIKE.

travc

May 24, 2007, 02:59 PM

You start with a recipe for a walnut and raisin cake. You bake it. It turns out OK. You bake it again, but this time, you don't have any raisins so you use cranberries.. Everyone loves it. So next time you use the altered recipe with cranberries. Until one day you don't have any cranberries so you use turnips. Everyone hates it, so you don't use turnips again. You stick with cranberries until one day you are out of sugar so you use honey. Everyone loves the honey, so you stick with honey and cranberries. One day you leave out the walnuts. People prefer it without walnuts. So you end up with a recipe for cranberry and honey cake, and people love it. Your walnut and raisin recipe has evolved. It now contains information not just how to make any old cake but a cake that the people you bake for really love. Information about the kind of cake your family really likes has been encoded into the recipe.

That's how the information from the environment gets into the genome so that it contains the information as to how to build an organism that will be a success in that environment.

:notworthy: Love the analogy. This fit in perfectly with what I've been saying about information only having a meaning in relation to something else, but much more clearly illustrates how selection can play that role.

Can I steal it :)

lee_merrill

May 24, 2007, 03:00 PM

I didn't dispute it. I simply amplified it ("more than that") and will do so still further, in case you have missed my point - that it is not simply that the genome can increase in quantity ...
However, I was wondering how my pronoun referent could have been unclear.

Don't forget that you are talking to Lee Merrrill, the expert about ancient languages, archeology, the bible back to front and front to back, information theory, biochemistry, etc. etc.
I freely admit I know little biology or chemistry or archeology. I have learned some Hebrew though (is this impossible?) and have studied computer science.

travc

May 24, 2007, 03:07 PM

I didn't dispute it. I simply amplified it ("more than that") and will do so still further, in case you have missed my point - that it is not simply that the genome can increase in quantity (however we measure the quantity of information) - actually the information on the genome might be reduced (it might have the instructions to add walnuts deleted) - the important point is that the genome, whether it contains MORE information or LESS information, now contains RELEVANT information as to how to make a specific kind of cake YOUR FAMILY WILL LIKE.

One of the results we demonstrated in the Journal of Theo Bio paper is that under idealized conditions (infinite population size, static environment, mutation rate below a threshold), evolution will always increase the amount of information relevant to selection in the genome.

Drift (requires finite population size) can lead to reductions in relevant information. And every time the environment changes, it can change what is relevant to survival and what isn't... so that too can reduce it. There is also a mutational threshold above which things meltdown, but that is pretty high for a large population.

Febble

May 24, 2007, 03:47 PM

However, I was wondering how my pronoun referent could have been unclear.

Well, in case it wasn't I clarified it. I didn't say you were wrong. OK?

Febble

May 24, 2007, 03:48 PM

:notworthy: Love the analogy. This fit in perfectly with what I've been saying about information only having a meaning in relation to something else, but much more clearly illustrates how selection can play that role.

Can I steal it :)

I'd be honoured!

Dr.GH

May 24, 2007, 04:01 PM

I'd be honoured!
Feel doubly honored then, because I am going to steal it too. :D And the addendum by Jet Black to respond to the "Why are there still monkeys?" question.:notworthy: to you both.

Vicious Love

May 24, 2007, 07:09 PM

Nope, think about it! 2 in binary is 10, 3 is 11, the same number of bits.

Oh, come on. Yes, 2 and 3 are both 2-bit numbers. But 1 in binary is 1, and "no number specified" in binary is . That's 1 and zero bits, respectively. Did you honestly overlook this, or are you simply trying to save face?

JonF

May 24, 2007, 07:31 PM

Not clear what you mean. If you mean duplicates and triplicates, then specifying "Do B" takes M bits. Then specifying a duplicate takes at least one bit more, for M+1 K-C information.
Nope, think about it! 2 in binary is 10, 3 is 11, the same number of bits.
Nope, think about it! We have complete freedom with the language for each problem, independent of other possible problems. We specify that any added bit(s) specifies the number of copies to make, so specifying one duplicate takes one bit (meaning "make one copy") for M+1 bits total. Specifying two total copies as 102 is M+2, and is not minimal per my previous sentence, and therefore is not the K-C complexity of one duplicate.

lee_merrill

May 25, 2007, 12:09 AM

We specify that any added bit(s) specifies the number of copies to make...
You are not being minimal, then, and no, we don't have freedom to choose any representation and call what we measure with that, K-C complexity.

But 1 in binary is 1, and "no number specified" in binary is . That's 1 and zero bits, respectively. Did you honestly overlook this...
As I said before, having special instructions for each possible value of the operands is not minimal. Did you, erm, honestly overlook this?

And why is every point so excruciating? I do think it's because I cannot be acknowledged to make a valid point, thus every point must be disputed, even the simplest ones, and the plainest.

Dr.GH

May 25, 2007, 01:14 AM

If you think that writing in some high ordered instruction set is representative of the real machine operations, you are an idiot.

I don't specifiy "who" is an idiot. That is open to anyone who satisfied the criteria.

Jet Black

May 25, 2007, 05:04 AM

And the addendum by Jet Black to respond to the "Why are there still monkeys?" question.:notworthy: to you both.

what was that? I've forgotten :)

Febble

May 25, 2007, 05:19 AM

what was that? I've forgotten :)

Here (http://www.iidb.org/vbb/showpost.php?p=4481833&postcount=70)

JonF

May 25, 2007, 08:08 AM

We specify that any added bit(s) specifies the number of copies to make...
You are not being minimal, then, and no, we don't have freedom to choose any representation and call what we measure with that, K-C complexity.
Adding one bit is not being minimal? In any representation there must be some specification of when the pattern stops and what the information after that point means; that can be the same for all representations and can therefore be ignored.

We do have freedom to choose any representation we wish but not freedom to assign that representation as the K-C complexity. That's not what I'm doing. I'm comparing representations.

If representation A (mine) is smaller than representation B (yours) then representation B is not the K-C complexity of any situation which both A and B represent, and any situation which requires representation B has more K-C complexity than the situation represented by A. That's true even if A is not the K-C compexity of the situation which can be represented by both A and B. As it is with your use of "10" to specify two copies total; since "1" can be be used to represent the number of copies to create for a total of two copies, your representation is not the K-C complexity of two total copies. Since there's no way to specify three total copies without using two bits, three copies must have a larger minimal representation

And why is every point so excruciating? I do think it's because I cannot be acknowledged to make a valid point, thus every point must be disputed, even the simplest ones, and the plainest.
It's mostly because you keep making false claims about areas in which you are obviously ignorant.

lee_merrill

May 25, 2007, 09:01 AM

Adding one bit is not being minimal?
Right, it's not best to count on your fingers. Why should this even be in dispute?

If representation A (mine) is smaller than representation B (yours) then representation B is not the K-C complexity of any situation which both A and B represent, and any situation which requires representation B has more K-C complexity than the situation represented by A. That's true even if A is not the K-C compexity of the situation which can be represented by both A and B.
:confused: Representation of a given number in A can be smaller, and yet the B algorithm can be smaller than the A algorithm.

It's mostly because you keep making false claims about areas in which you are obviously ignorant.
It's been a while since I did any algorithmics, but I actually have a graduate degree in computer science.

But in any case! I think we have resolved this question, yes, mutation can increase information.

Principia

May 25, 2007, 09:02 AM

Representations are irrelevant in Kolmogorov complexity because it handles representation invariance with the neat little theorem, K_U(x) \leq K_V(x) + C, for any string x and different representations U and V. If lee_merrill claims to know of a language V such that K_V("1") = K_V("2"), then there exists a more universal language U such that the above relationship holds. The reason is that lee's language V would simply require a larger translation to U (i.e. a bigger dictionary). So it's a no-win situation for lee. He is simply wrong.

Principia

May 25, 2007, 09:07 AM

I think we have resolved this question, yes, mutation can increase information.No, we resolved that mutations may increase K-C and Shannon information. Who knows what crazy shit information concept you creationists can come up with next.:huh:

lee_merrill

May 25, 2007, 10:12 AM

Representations are irrelevant in Kolmogorov complexity because it handles representation invariance with the neat little theorem, K_U(x) \leq K_V(x) + C, for any string x and different representations U and V.
But the constant depends on the languages! Thus for any string of real interest, this constant will probably swamp the length of the algorithm, for this constant factor is said to be (http://en.wikipedia.org/wiki/Kolmogorov_complexity) an implementation of the entire other language L2 in language L1. So representation is a factor, and we even read here (http://en.wikipedia.org/wiki/Kolmogorov_complexity): "The length of any integer constant n which occurs in the program P is the number of bits required to represent n...", which would be log-base-2 of n, contra JonF...

The reason is that lee's language V would simply require a larger translation to U (i.e. a bigger dictionary). So it's a no-win situation for lee. He is simply wrong.
And going from U to V would require a possibly large constant added to the representation in U!

But maybe it's about time to close this thread...

Sven

May 25, 2007, 10:17 AM

I freely admit I know little biology or chemistry or archeology.
For some reason, I don't remember you admitting this in the train wrecks in BC&H. Or, if you indeed admitted it (adn I just forgot), this did not hinder you from making authoritative claims about archeology.

lee_merrill

May 25, 2007, 10:20 AM

... this did not hinder you from making authoritative claims about archeology.
Conclusions is what I make, and quote authorities. I have not dug in the deserts of Arabia myself.

Making a conclusion does not involve a posture as an expert, otherwise you would be claiming expertise too, and to be an authority on any subject you have an opinion on.

Principia

May 25, 2007, 10:38 AM

And going from U to V would require a possibly large constant added to the representation in U!You have no clue what you are talking about.

lee_merrill

May 25, 2007, 10:43 AM

You have no clue what you are talking about.
I believe my statement is in fact, correct.

Principia

May 25, 2007, 10:46 AM

It doesn't matter what you believe. It only matters what you can demonstrate. And you have not demonstrated understanding.

In point of fact you argue against yourself. U is more universal than your ad hoc language V. So yeah, you have to add a lot of extra baggage to U to get V. So V does not produce a minimal description. U does.

You are still wrong.

SophistiCat

May 25, 2007, 10:49 AM

Since there's no way to specify three total copies without using two bits, three copies must have a larger minimal representation

Actually, there may be a way. If your string abounds with triplets of sequences, it may be more economical to indicate a triplet with a single bit (or even zero bits), and a single or a doublet with more bits. This idea is actually used in efficient coding, such as Morse code and compression algorithms, where more frequent elements (e.g. characters 'e' and 't' in English texts) are coded with shorter bit strings.

Principia

May 25, 2007, 10:49 AM

So lee, are you going to make up your own definition of representation invariance and Kolmogorov complexity? Just like you made up your own definition of the Delta dirac? The pattern of your flat-out lying is just appalling.

So have you figured out the last one yet? Is Delta dirac a function or not? Do you know how to generate all PDFs from a Delta dirac?

Sven

May 25, 2007, 10:50 AM

Conclusions is what I make, and quote authorities. I have not dug in the deserts of Arabia myself.

Making a conclusion does not involve a posture as an expert, otherwise you would be claiming expertise too, and to be an authority on any subject you have an opinion on.
*sigh*

I invite everyone here not familar with the threads I mentioned to dig them up (if you have much time and strong nerves) to see what I mean.

ETA: I just visited the old Tyre thread again to dig up specific examples, but after skimming about three posts, I just could not bear it. Sorry, folks, you have to look for yourself.