Missing genetic information refutes neo-Darwinism

[wuschel]

You are assuming that an infinite number of influencing factors add up to an infinite amount of influence.

The weird thing about invoking infinity is that you get some weird results.

The only way an infinite number of influential factors will not add up to infinite influence is when an infinite number of them is not influential -> in which case they are not influential factors.

An infinite amount of very little - regardless of how little - still yields infinity.

 

[Yiab]

That is only a problem if you assume that this 750MB is a compressed description of the phenotype, which it is not. A complete description of the organism may not be infinite, but for all intents and purposes it might as well be, because a complete description would necessarily include all environmental influences no matter how small. If it is not infinite, it is at the very least as large as the Universe.

Not even as large as the universe.
At the very most you can get a complete description of an organism by having all information about everything in the "past" section of its light-bubble (fancy talk for everything that exists in a bubble expanding backwards in time at the speed of light). For example, something that is 1 light year away cannot have influenced an organism which came into existence two days ago.

A better analogy could be this one:

Genetic alphabet, 4 letters
Genetic dictionary, 64 words with 21 different meanings
A human's genome, 20'500 recipes​

...Including recipes which can alter the frequency at which other recipes are implemented and including recipes of sufficient complexity to, given the right combinations of ingredients sitting around, duplicate the recipes themselves.

From the ontogensis of birds I conclude that information outside the developing egg/embryo cannot have a relevant impact. How should a human embryo scan its environment and profit from this information? The same is valid for the umbilical cord.

http://www.pptox.dk/Consensus/tabid/72/Default.aspx

Fetal life and early infancy are periods of remarkable susceptibility to environmental hazards.

I have no idea how reliable this source is, but a few very cursory google searches turned up a wide variety of things indicating that environment influences the developing fetus in significant ways, but I didn't see one which looked like a good overview.

fetal alcohol syndrome

http://en.wikipedia.org/wiki/Phocomelia:

The condition may be inherited or occurs sporadically. It is also connected with prenatal exposure to the anti-nausea drug thalidomide.

(emphasis mine)

I vaguely recall hearing about hormone imbalances in the womb during particular stages of pregnancy causing a genetically female fetus to become physiologically male or vice versa, but I can't think of anything to type into google for this that wouldn't get rule8'd.

Now, I would like to re-quote you to remind you of the claim you actually made here and why the things I've just said are very relevant.

information outside the developing egg/embryo cannot have a relevant impact. ... The same is valid for the umbilical cord.

How can you claim this?

"The Levinthal paradox is a thought experiment in the theory of protein folding dynamics. In 1969 Cyrus Levinthal noted that, because of the very large number of degrees of freedom in an unfolded polypeptide chain, the molecule has an astronomical number of possible conformations. ... If the protein is to attain its correctly folded configuration by sequentially sampling all the possible conformations, it would require a time longer than the age of the universe to arrive at its correct native conformation." (Wiki)

Make it fold any other way.
The whole reason that these genes survived over others is because the proteins they coded for naturally curled into these useful shapes. If there were any other shape more natural (remembering that it is constructed by reading alleles and attaching amino acids by use of an intermediate protein) these genes wouldn't have developed.

The "information" in use in protein folding is primarily particle physics.

The weird thing about invoking infinity is that you get some weird results.

This is definitely true. For example...

The only way an infinite number of influential factors will not add up to infinite influence is when an infinite number of them is not influential -> in which case they are not influential factors.

An infinite amount of very little - regardless of how little - still yields infinity.

This is false. You'd expect that adding together infinitely many non-zero things will always give an infinite result, but it doesn't have to. I'd suggest you look at the works of Zeno (for the wrong conclusions), Newton and Weierstrass to start with (I'm still amazed at how many people are unaware of the existence and/or contents of calculus in this day and age).

The simple fact is that you can take infinitely many influences of ever-decreasing but non-zero influence and add them up to a finite influence. To oversimplify:
[latex]$\sum\limits_{n=0}^{\infty}2^{-n} = 2 < \infty$[/latex]

 

[Earthborn]

For example, something that is 1 light year away cannot have influenced an organism which came into existence two days ago.

An organism that was born two days ago is still influenced by something that is 1 light year away, because its parents, or its grandparents or its great grandparents may have existed a year ago, and the environment in which it lives may have been influenced by it a year ago.

Also, an object 1 light year away may influence an organism right now through an influence that object has sent out a year ago. In fact, I am an organism who has been influenced by objects millions of lightyears away because those things sent photons that caused chemical changes in my eyes, and I am fairly sure I did not exist millions of years ago.

 

[Taffer]

For example, something that is 1 light year away cannot have influenced an organism which came into existence two days ago.

I believe you are referring to the light cone^WP. Different from "1 light year".

 

[wuschel]

The simple fact is that you can take infinitely many influences of ever-decreasing but non-zero influence and add them up to a finite influence. To oversimplify:
[latex]$\sum\limits_{n=0}^{\infty}2^{-n} = 2 < \infty$[/latex]

Thought someone would bring that up. In an universe of finite observable size, the influence anything can have on anything else (given there is such influence to begin with) cannot become arbitrarily small without becoming zero, in which case there is no - as in "zero" - influence.

 

[wogoga]

In other words: suppose a Star Trek like transporter scans an entire human being and places all necessary information to rebuild a human in its buffer. You would need a whole lot more memory space than the 750 MB needed to store the human being's DNA sequence, even when compressed. If you believe (a view wogaga attributes to "neo-Darwinists") in the misleading analogy that DNA is a sort of computer program to build a human, you might get the impression that

TransporterBuffer - 750 MB = MissingInformation​

Of course, the information is not really missing. It is just that nobody in his right mind argued that it was to be found in DNA.

A human body of 100 kg consists of around 10^28 atoms. 5 bit are enough to specify the atom by its atomic number, and 93 bit are needed to give an unambiguous reference number to each of the 10^28 atoms. Let us ignore gaseous regions e.g. in the lungs and assume that every atom has direct neighbours, on average 14. This results in 10^28 * 14/2 pairwise connections. Also a few bit are probably needed in order to specifiy the connection or the chemical bond.

This results in a "TransportBuffer" of around 10^30 bytes, and is obviously not an adequate starting point to estimate the missing genetic information. In order to do that, we must deal with the number of freedom degrees of the relevant properties. We can also ask, how many degrees of freedom have for instance bolts and nuts, and how much information is used in the production process of a bolt or a nut. If bolts and nuts belong together, they can share some information. Sharing information however can also entail new problems, e.g. undesired dependencies.

If we take seriously our close relationship with chimpanzees, we must conclude that a continuous spectrum of bodies ranging from chimpanzees to humans has existed, all of which with a genetic information of around 750 MB (or less than 15 MB, if we exclude non-functional DNA). The difference between a chimpanzee body and a human body is quite substantial. Let us only deal with the macroscopic level:

"The human skeleton consists of both fused and individual bones supported and supplemented by ligaments, tendons, muscles and cartilage." "At birth a newborn baby has approximately 300 bones, whereas on average an adult human has 206 bones (these numbers can vary slightly from individual to individual)." "There are approximately 650 skeletal muscles within the typical human. However, the exact number is difficult to define because different sources group muscles differently." (Wiki)

How many degrees of freedom has one single bone, one single muscle? How many ways exist to combine the bones, ligaments, tendons, muscles and cartilages with each other?

The environment of a developing egg/embryo cannot provide relevant (constructive) information concerning the ontogenetic design of a primate. And the genetic information present in the ferilized egg is obviously not enough to determine all the relevant degrees of freedom. So why should the hypothesis that non-material information could be involved not at least be taken into consideration?

Cheers, Wolfgang


The upright gait was only one of many traits which had to evolve in us after our separation from chimps. For that to happen, the structures of bones, of muscules and of tendons had to gradually change. Let us ignore that in fact the bone structure (involved in the upright-gait evolution) alone consists of several bones with each several traits.

So let us make the completely unrealistic simplification that one 'progressive' single-step mutation in the genetic factor of each (i.e. bone, muscle and tendon) structure is enough to entail a relevant increase in fitness.

Let us further assume that the probability of such progressive mutations in newborns is each as high as 10^-5. So we conclude that among 10^15 newborns (i.e. a billion newborns of a million generations), only one indivudual will carry all three necessary mutations.

Because a change in only one or two of the three involved structures cannot lead to a relevant increase in fitness (rather the contrary), it becomes obvious that the upright gait cannot have evolved in a neo-Darwinian way.

 

[cyborg]

So why should the hypothesis that non-material information could be involved not at least be taken into consideration?

I ask again: where is the information in Rule 30 coming from?

 

[Taffer]

And the genetic information present in the ferilized egg is obviously not enough to determine all the relevant degrees of freedom.

Evidence?

 

[Yiab]

An organism that was born two days ago is still influenced by something that is 1 light year away, because its parents, or its grandparents or its great grandparents may have existed a year ago, and the environment in which it lives may have been influenced by it a year ago.

Also, an object 1 light year away may influence an organism right now through an influence that object has sent out a year ago. In fact, I am an organism who has been influenced by objects millions of lightyears away because those things sent photons that caused chemical changes in my eyes, and I am fairly sure I did not exist millions of years ago.

You're right, I should have been more specific. What I meant was an organism one light year away which came into existence just before now cannot have influenced an organism which came into existence two days ago. Things which influenced the former may also have influenced the latter, of course, but that is a far cry from direct influence.

I believe you are referring to the light cone^WP. Different from "1 light year".

Yes, I am referring to that, but I wasn't sure if that was standard terminology so I made an attempt to describe it in temporally-grounded terms. Essentially what I am saying is that one does not need to take into account the entire universe, merely the past light cone in order to have complete description. Since the past section of an organism's light cone is not, in fact, the entire universe, we cannot use this light cone as functionally infinite since there is a larger finite entity which is relevant - the entire universe.

Thought someone would bring that up. In an universe of finite observable size, the influence anything can have on anything else (given there is such influence to begin with) cannot become arbitrarily small without becoming zero, in which case there is no - as in "zero" - influence.

While I agree with your conclusion, the finite size of the universe is an insufficient condition to establish it - one must also have quantization of "influence", which is effectively provided by quantum mechanics. Of course, this quantization of influence has as a necessary conclusion of its adoption in this sort of scenario: that the entire universe can be accurately viewed as a single gargantuan quantum finite state automaton with gravity somehow included, meaning that there are only finitely many possible quantum states the universe can have and so there cannot be infinitely many anythings, let alone influences on a particular organism.

 

[Taffer]

Yes, I am referring to that, but I wasn't sure if that was standard terminology so I made an attempt to describe it in temporally-grounded terms. Essentially what I am saying is that one does not need to take into account the entire universe, merely the past light cone in order to have complete description. Since the past section of an organism's light cone is not, in fact, the entire universe, we cannot use this light cone as functionally infinite since there is a larger finite entity which is relevant - the entire universe.

I've not been following along with this side discussion, but I'd agree with you. The only environmental factors which can possibly have effected an organism exist within its historic lightcone.

 

[Yiab]

A human body of 100 kg consists of around 10^28 atoms. 5 bit are enough to specify the atom by its atomic number, and 93 bit are needed to give an unambiguous reference number to each of the 10^28 atoms. Let us ignore gaseous regions e.g. in the lungs and assume that every atom has direct neighbours, on average 14. This results in 10^28 * 14/2 pairwise connections. Also a few bit are probably needed in order to specifiy the connection or the chemical bond.

This results in a "TransportBuffer" of around 10^30 bytes, and is obviously not an adequate starting point to estimate the missing genetic information. In order to do that, we must deal with the number of freedom degrees of the relevant properties.

Your calculations are misleading in precisely the way that you ignore the environment of the developing fetus.

The amount of information you are calculating is the information it would take to build a fully adult human being with experiences and memories from an external supply of individual atoms - this is not what happens during gestation. There are twenty common amino acids available during protein-building from genes and two more extremely rare ones, plus an extremely large number of them which are not coded for in genes. Inside the cell, only the twenty (or twenty-two) necessary amino acids are going to be present, so we've already got a humongous reduction in information complexity here - a reduction from "a vast number" to 22.
Each amino acid consists of the same 9 atoms arranged in an identical pattern joined to a side chain different for each amino acid which has from 1 (glycine) to 18 (triptophan) additional atoms with an average of 10.2 additional atoms. This tells us that each codon actually codes for a specific configuration of about 19 atoms. The order of the codons combined with particle physics determines protein folding patterns and concentration of RNA polymerase determines how frequently genes get read.
So each codon consists of 64 possibilities and so it is only 6 bits of information, but due to the environment of the genes inside the cell, these 6 bits of information codes for, according to your scheme, 95 bits (or 11 bytes, 1 nibble and 3 bits) of information even if we ignore your completely superfluous "reference numbers". None of this process is mysterious or requires information from any other source, it is gene patterns + environment within the cell + particle physics + statistical mechanics.

Nothing that I've said has even begun to talk about compression algorithms in any computer-scientific sense, this is all about using a common reference frame - the same way that I could condense over 167,000 characters into 9: "King Lear" (http://www.gutenberg.org/dirs/etext97/1ws3310.txt).

 

[wogoga]

Apart from how vague and unscientific the "more than 20 000 properties" claim is, one would at least need to establish that these claimed "more than 20 000 properties" are indeed "more than 20 000" independent degrees of freedom, because otherwise it looks just like the apparent "Complexity" of the decimal expansion of "PI", which, in fact, is the result of a rather simple algorithm.

It is assumed that the 20'000 genes give rise to more than 20'000 different types of enzymes. And should we not attribute to each of these enzymes at least one property respectively "independent degree of freedom"? The HIV-1 protease for instance cleaves the viral polyprotein at eight different sites during the maturation process of the virus. HIV-1 proteases also have learned defend to themselves from protease inhibitors. We may nevertheless dispute whether 20'000 properties are enough to determine all human enzymes. However, we can be sure that 20'000 properties are not enough to determine the ontogenesis of a human at all levels.

"It is relevant to point out that a gene on average consists of a thousand base pairs, thereby supplying many bits of information. However most of this supply of information is a mirage. The vast bulk of a protein is devoted to folding up into the right shape. The region of interest is the hot spot which only consists of a handful of amino acids. It should also be noted that a fair percentage of the genome is devoted to house-keeping machinery for the eukaryote cell." groups.google.com/group/sci.skeptic/msg/9b3b2aeb97dfba48

The main problem with information increase by means of algorithms or 'elementary cellular automaton rules' is the complete lack of degrees of freedom. Whereas a given phenotype has a continuous spectrum of neighbours, fractals and the decimal expansion of the square root of prime numbers have no neighbours.

Cheers, Wolfgang

"Reductionists have resolved the problem of the missing genetic information by an argument, which is based on a famous philosophical error. They confuse the category of information size with the category of combination number. If some information is condensed to a 1 million bit, one needs a million bit to store it. It is nonsense to argue that 20 bits are enough because 20 bits give more than a million (2^20) combinations. One must clearly distinguish between the information size and the number of combinations that can be stored in a memory of a given size." members.lol.li/twostone/E/psychon.html#final

 

[wuschel]

[latex]$\sum\limits_{n=0}^{\infty}2^{-n} = 2 < \infty$[/latex]

Revisiting your argument, I suppose once infinity is asserted, you need to assert it all the way:

[latex]$\sum\limits_{n=-\infty}^{\infty}2^{-n} = ???$[/latex]

In an universe with an infinite number of possible things that can happen, why would you put an upper bound on the effect size at "1"?

 

[wuschel]

It is assumed that the 20'000 genes give rise to more than 20'000 different types of enzymes.

I would have assumed the 20.000+ properties were meant with respect to the phenotype.

The main problem with information increase by means of algorithms or 'elementary cellular automaton rules' is the complete lack of degrees of freedom. Whereas a given phenotype has a continuous spectrum of neighbours, fractals and the decimal expansion of the square root of prime numbers have no neighbours.

Look up "Chaitin's incompleteness"! What this comes down to is that from just looking at the data (without knowing where it originated) for data above a certain size you cannot know that it is complex - because the problem is in NP.

That, or a Nobel Prize for you!

As far as the assertion of a continuous phenotype: How would one design an experiment to test this assumed continuity? How could it be falsified? How do you know, then?

 

[drkitten]

Look up "Chaitin's incompleteness"! What this comes down to is that from just looking at the data (without knowing where it originated) for data above a certain size you cannot know that it is complex - because the problem is in NP.

That, or a Nobel Prize for you!

Actually, there would probably be a Nobel Prize -- more accurately, a Turing Award -- if you could show that calculating Kolomogorov complexity was in NP.

Because it's not. It can't be calculated by a Turing Machine at all (reduction from the halting problem).

 

[drkitten]

Revisiting your argument, I suppose once infinity is asserted, you need to assert it all the way:

[latex]$\sum\limits_{n=-\infty}^{\infty}2^{-n} = ???$[/latex]

In an universe with an infinite number of possible things that can happen, why would you put an upper bound on the effect size at "1"?

You've now crossed over into the happy magical land of "this isn't even wrong."

 

[wuschel]

Actually, there would probably be a Nobel Prize -- more accurately, a Turing Award -- if you could show that calculating Kolomogorov complexity was in NP.

Because it's not. It can't be calculated by a Turing Machine at all (reduction from the halting problem).

You haven't proven yet that the phenotype is continuous, yet you assert a human to be a Turing machine of infinite size?

For your enlightenment: The halting problem is in principle decidable for any finite size TM.

Glad you asked!

 

[wuschel]

You've now crossed over into the happy magical land of "this isn't even wrong."

I'm afraid your argument was a little to concise for me to understand how you arrived at that conclusion. Please bear with me undeserving mortal and clue me in on your ... "infinite" wisdom!

 

[cyborg]

The main problem with information increase by means of algorithms or 'elementary cellular automaton rules' is the complete lack of degrees of freedom. Whereas a given phenotype has a continuous spectrum of neighbours, fractals and the decimal expansion of the square root of prime numbers have no neighbours.

You say this as if the concept of a continuum was an aberration unto the discrete.

I say a continuum is just infinite discrete. Hell - I'll further say that this continuum you claim to exist isn't supported by physics - so it is discrete anyway. As if all the information you are quoting as significant even were.

 

[Yiab]

Revisiting your argument, I suppose once infinity is asserted, you need to assert it all the way:

[latex]$\sum\limits_{n=-\infty}^{\infty}2^{-n} = ???$[/latex]

In an universe with an infinite number of possible things that can happen, why would you put an upper bound on the effect size at "1"?

Because we're not talking about summing all possible influences simultaneously on a particular thing.
You made a universally quantified statement: adding up infinitely many things always gives an infinite answer.
I refuted your universally quantified statement by giving a counterexample.
The existence of infinite series with infinite sums is not in question.

Besides, working in 2-adic arithmetic, [latex]$\sum\limits_{n=-\infty}^{\infty}2^{-n} = 1 + \sum\limits_{n=0}^{\infty}1*2^n$[/latex] is a perfectly well-defined number, though I am not familiar enough with the p-adics to express it in a nicer-looking way, though I have the bizarre notion that it might be zero.

Besides Besides, if you want to try asserting infinity "all the way" (whatever that means) we have to go into the category of ordinals:
[latex]$\sup\limits_{\kappa\in Ord}\sum\limits_{\lambda<\kappa}2^{\lambda}$[/latex] which is, ultimately, a useless and fairly meaningless expression, though it is a well-formed one. Welcome to set theory.