Re: Re: For Kim G. S. Øyhus
- From: kim@xxxxxxxxxxx (Kim G. S. Øyhus)
- Date: Sun, 15 Jan 2006 23:10:33 +0000 (UTC)
In article <1137348907.367711.254810@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
Seanpit <seanpitnospam@xxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote:
>
>Kim G. S. Øyhus wrote:
>
>< snip >
>
>
>> >> >Neither of
>> >> >you have shown how a flagellar motility system can work as a motility
>> >> >system with significantly fewer than 30,000 bp of genetic real estate.
>> >>
>> >> Hershey has shown well enough that it can. Check his posts for evidence.
>> >
>> >Which post in particular is most convincing to you? I've not read any
>> >of Howard's posts where he shows how a flagellar system can be built
>> >using significantly less than 30,000bp of genetic real estate. It
>> >seems to me that you're just saying stuff, hoping that it will fly. You
>> >know Howard never made such a post.
>>
>> I leave it to them to answer that, since they are so good at it.
>
>That figures. You baldly assert that Howard has shown, in various
>posts, how a flagellar system of motility can be built with
>significantly less than 30,000bp of genetic real estate, but somehow
>you can't provide a link to such a post? ; )
Why do you discuss this with me instead of them? Are you just trying
to quarrel?
Anyway, my mathematical arguments are plenty strong enough to falsify
you without resorting to detailed examinations of the flagellar
system. You do not understand this, but that is because you are
ignorant and arrogant.
>> >And, from the beginning, we have been talking about environments that
>> >favor the more complex system. Can such a complex system actually
>> >evolve even in a *favorable* environment? That's always been the
>> >question.
>>
>> But you write only about point mutations, which is NOT the only method
>> of evolution by far. And you still have not calculated the real
>> complexity or a reasonable complexity. If you had, those would be evidences.
>
>As I've written over and over again, you can use any type of mutation
>you want. Nowhere do I argue for a limit use of only point mutations.
>That's a clear strawman representation of my true position.
Why do you present a strawman of your position instead of your position?
You write almast all the time of point differences in sequence space,
without considering other options.
>Beyond this, it is very clear that some types of functional systems,
>like a flagellar motility system, require far greater amounts of
>genetic real estate, at minimum, than do other types of functional
>systems. I don't see how anyone can really argue against this concept
>and expect to be taken seriously.
It is quite frustrating how you have misunderstood people through so
many posts. Try to understand this:
1. Some functions need more information to be described. Some concepts
need more data to be described. Some programs need more instructions
to be implemented. Some bitstrings need larger Turing machines to be
generated. ALL this are variants of the same thing, and is what
Kolmogorov Complexity is about.
This can also be stated in this form: There are a minimal size of
the DNA necessary to generate a biological system having a particular
well defined task.
You have misunderstood us, so you believe we disagree with this.
2. Flagellums are NOT a function, not a problem, not a program. They
are a physical system, a construction, matter.
Since evolution selects ONLY according to fitness and reproduction,
flagellums are NOT selected. However, the FUNCTION of the flagellums
can be selected, since the function is motion, and motion often affect
fitness and reproduction.
3. There are NOT a minimum complexity for motion, since motion is
present anyway, through the laws of nature, through brownian
motion. Thus, it wrong to claim that flagellums is the simplest way of
motion.
4. It is provably impossible to know where the minimum complexity is
for infinitely many functions, so you cannot know anything of this.
>< snip >
>
>
>> >I never said that evolution selects based on the type of system.
>>
>> I see from what other people answered you that other people beside me
>> understood you to mean that evolution selects based on the type of
>> system.
>
>Regardless, this is not what I said nor what I meant to say. I've
>specifically said, over and over again, that evolution selects based on
>functional differences regardless of the type of sequence or system
>involved.
Well then, what IS the functional difference that you claim flagellums
has the minimal complexity for? It cannot be motion, because that is
too simple.
>< snip >
>
>
>> >> You keep attacking the unrealistic system of pure sequential point
>> >> mutation from zero. You attack this instead of real evolution.
>> >
>> >I do not require a sequential build up of the flagellar system from
>> >zero.
>>
>> A lie. You did.
>
>No, I didn't. If you think otherwise, prove it. I've specifically
>written detailed descriptions concerning this argument - if you care to
>actually read them.
So, you do not remember what you yourself write.
>< snip >
>
>
>> >The word "complex" used in Kolmogorov Complexity, actually means
>> >"random" or "chaotic".
>>
>> The word "complex" used in Kolmogorov Complexity, actually means
>> "Length of shortest program which makes the object."
>
>Right - which has to do with randomness. A truly random string is not
>compressible. However, this sort of non-compressible "complexity" has
>nothing to do with the meaning or function of a sequence or system.
You are completely wrong in your belief that Kolmogorov Complexity is
only about non-compressible strings. It is Chaitin who is focused on
those. Kolmogorov, Martin Lof, Solomonoff, and the other creators of
Algorithmic Information Theory work in the region of COMPRESSIBLE
strings, the region of minimal complexity, NOT in the region of maximal
compressibility, which gives randomness.
You have totally missed this understanding. Instead of listening to
me, or reading and studying Kolmogorov Compleixity, you have accused
me falsely of not knowing the field, and tried to denigrate me, when
it is your ignorant arrogance which has made you say these lies.
>> >It is impossible to prove that a sequence of
>> >characters is truly random or chaotic. It is not impossible, however,
>> >to prove that a functional system has a certain minimum size and
>> >specificity requirement. Functional or system complexity is not the
>> >same thing as Kolmogorov Complexity Kim. They are quite different.
>>
>> You are mistaken. From Wikipedia:
>> "More formally, the complexity of a string is the length of the
>string's shortest
>> description in some fixed description language."
>>
>> "Shortest description" correspond do "certain minimum size"
>> "language" correspond to "biology"
>> "string" correspond to "function"
>
>The term "complexity" used here is not the same thing as functional
>complexity - as noted by the Baranger reference listed below.
So, you claim that computers are not able to calculate fractals and
chaos. Well, that IS the meaning of what you just wrote. You just mess
around, not understanding this subject.
>Also, the
>term "language" as used here, is not the same thing as meaningful words
>or sentences or meaningful systems like the English language system.
>There is a difference between determining degrees of randomness or
>chaotic complexity and functional complexity. They are not the same
>thing and have no direct "correspondence" to each other.
You are completely wrong at this, since computers ARE able to
calculate fractals and chaos.
>> >For example, consider an interesting excerpt from a paper written by
>> >Michel Baranger, a physicist from Cambridge:
>> >
>> >"When you look at an elementary mathematical fractal, it may seem to
>> >you very 'complex', but this is not the same meaning of complex as when
>> >saying 'complex systems'. The simple fractal is chaotic, it is not
>> >complex. Another example would be the simple gas mentioned earlier: it
>> >is highly chaotic, but it is not complex in the present sense. We
>> >already saw that complexity and chaos have in common the property of
>> >nonlinearity. Since practically every nonlinear system is chaotic some
>> >of the time, this means that complexity implies the presence of chaos.
>> >But the reverse is not true. Chaos is a very big subject. There are
>> >many technical papers. Many theorems have been proved. But complexity
>> >is much, much bigger. It contains lots of ideas, which have nothing to
>> >do with chaos. . . " (Ref)
>> >
>> >Note that Baranger draws a distinction between the terms complexity, as
>> >used to denote chaotic- or random-type complexity and complexity used
>> >to describe an interacting interdependent system. They are different
>> >things. They are not the same. Consider that the use of the term
>> >complexity "in the present sense" is not the same as the use of the
>> >term complexity in the KC sense.
>>
>> I am becoming quite tired of your arrogant delusions. You are again
>> wrong, and do not understand KC at all.
>> The fractal consists of infinitely many parts, but has a simple
>> mathematical description, which is directly equivalent to a short
>> Turing machine generating the fractal at increasing resolutions.
>> But you are not able to understand why this means that they are
>> the same use of complexity, because you do not understand KC.
>
>Do you disagree with Baranger when he says, "The simple fractal is
>chaotic, it is not complex."?
I just WROTE that it has a simple description! Do you not understand
that "simple" is "not complex"?
I see I was completely right when I stated that you are not able to
understand.
>The use of the term complexity is not
>the same, despite your assertion, when describing KC or fractals or
>randomness or chaos verses a "complex" interacting system of function.
>
>http://necsi.org/projects/baranger/cce.pdf
You do not understand.
>< snip >
>
>> >If the ratio is 1 in 1e2000, the probability of a
>> >particular neighbor being beneficial is quite low. This means that the
>> >probability that a particular beneficial island is in fact an island is
>> >quite high. You have the whole problem turned on its head.
>>
>> You confuse density with granularity. You simply do not understand
>> what you write about.
>
>If a beneficial sequence is completely surrounded, on all sides, by
>non-beneficial sequences, how is this beneficial sequence not an
>island?
A low density of sequence do not imply they make only islands. You
again show that you do not understand what you write about. Here is a
counterexample to your claim:
Lay the 20 amino acids on a line, and consider a space of a little
more than 2000 dimensions of these lines, and construct a sierpinski
trangle in all these dimensions, and consider the elements of the
sierpinski triangle as the beneficial sequences. Now, since the
sierpinski triangle is not an island, but instead a holey structure,
and can have the density/ratio of 1e-2000 you mention, it is a
counterexample to your statement.
But you are of course not going to understand any of this.
>> >> And I am NOT writing about the number of potentially beneficial
>> >> sequences in sequence space. I am instead writing about the number of
>> >> potentially beneficial seqences neighboring beneficial sequences.
>> >
>> >Which has to do with the actual number of beneficial sequences in
>> >sequence space.
>>
>> No. You are totally wrong, and it is easy to show that you are totally
>> wrong: Just consider 1 arbitrary sequence in 2 different sequence
>> spaces, where there are the same number of neighbors, but where the
>> first sequence space has no further functional sequences, while the
>> second space is filled with functional sequences.
>
>The first sequence space has a very low ratio of beneficial vs.
>non-beneficial sequences while the second space has a very high ratio.
Yes.
>> The first space will have a density of -> 0, while the second space will
>> have a density of -> 1, while both spaces have the same number of
>> functional neighbors to the arbitrary sequence.
>
>You just said that the first space only had 1 functionally beneficial
>sequence - that all the other surrounding sequences are non-beneficial.
Yes.
> How then can you say that this sequence has the same number of
>beneficial neighbors as the second scenario where all the sequences in
>the sequence space are defined as beneficial?
Because it HAS the same number of beneficial neighbors. That is how
the scenarios are constructed. It even works if it has no beneficial
neighbors.
Here is an example of 2 such scenarios, with 0 beneficial neighbors,
for a sequence of 3 base pairs:
Scenario 1.
Main sequence: AAA
Other sequences: none
Scenario 2.
Main sequence: AAA
Other sequences:
ATT ATC ATG ACT ACC ACG AGT AGC AGG
TAT TAC TAG TTA TTT TTC TTG TCA TCT TCC TCG TGA TGT TGC TGG
CAT CAC CAG CTA CTT CTC CTG CCA CCT CCC CCG CGA CGT CGC CGG
GAT GAC GAG GTA GTT GTC GTG GCA GCT GCC GCG GGA GGT GGC GGG
As you can check for yourself, the main sequence in both scenarios has
no neighbors, but the density is low in 1, and high in 2.
>> Thus, this construction of sequence spaces with maximally different
>> densities but with the same number of beneficial neighbor sequences
>> shows that there is no strict relation between these 2 numbers.
>
>This makes absolutely no sense to me.
True.
I hope the tables above help. But they only do if you actually
check them instead of just believing stuff about them.
>> But I am very sure you will not understand a bit of this, since you
>> are too arrogant and ignorant.
>
>I'm not too arrogant to admit a mistake that I can actually understand.
> Ignorant perhaps. It might be helpful if you could get someone else
>to help you explain this notion of yours in a way I can actually
>understand. So far, what you are saying here truly does make no sense
>at all to me.
I hope you are truthful. All you need do now, is check the tables above.
You will do that if you are truthful.
>< snip >
>
>> >> You are completely mistaken. Here is a counterexample with high
>> >> specifity and high density:
>> >>
>> >> aaaaaaaaaaaaaaaaa......aaaaaaaaaaaaaaaaaaa
>> >>
>> >> That is supposed to represent 30,000 a's.
>> >> Change any of them, and it does not work any more.
>> >> In other words, it has maximal possible specificity.
>> >
>> >This has high specificity, but very low density since only 1 such
>> >sequence exists in all of sequence space. Finding such a sequence
>> >would be very difficult, on average. Doesn't this make any sense to
>> >you?
>>
>> I had not finished the description at this point. As I made clear
>> below, this sequence space is almost filled with functional sequences,
>> thus having a density of -> 1.
>
>If the sequence space is filled with beneficial sequences, then the
>density will be very high because of the total number of beneficial vs.
>non-beneficial sequences is very high.
Yes.
>Therefore, the density of
>beneficial sequences in sequence space is in fact dependent upon the
>total number of beneficial sequences in sequence space.
Of course it is, but it is not dependent on the specificity, on the
number of beneficial neighbor sequences.
>> >> But, change more than 1 of them, and it is functional agains.
>> >
>> >What function does it have?
>>
>> Unspecified, and irrelevant. The only relevant thing is that is has a
>> function.
>
>Ok . . .
Good.
>> >> In other words: All sequences except 30,000*19 are functional.
>> >
>> >Ah, but this is a set up of your own creation.
>>
>> Yes, it is. And if you were right, it should not have been possible
>> for me to create such a sequence.
>
>How so? You set up the sequence space so that there is a very high
>number of beneficial sequences creating a very high ratio of beneficial
>vs. non-beneficial sequences. Given this setup, the odds of a
>particular sequence having many beneficial neighbors is extremely good.
> However, given a very low ratio of beneficial vs. non-beneficial
>sequences, the odds are not so good.
The 2 scenarios were constructed with the SAME number of neighbors,
but very DIFFERENT number of beneficial sequences. They had the same
specificity, but different density.
>> >In real life
>> >language/information systems, it doesn't work like this. The vast
>> >majority of sequences at such a high level will be non-beneficial. Only
>> >a very tiny relative number will actually have a potentially beneficial
>> >function/meaning.
>>
>> You are wrong, as proven by Kolmogorov Complexity.
>
>Again, KC doesn't say anything about functional vs. non-functional
>sequences. Beyond this, you really can't reasonable argue that
>beneficial sequences do not decrease, exponentially, at higher and
>higher levels of functional complexity - relative to non-beneficial
>sequences.
You repeat your misunderstanding of KC.
>> >> Thus, this hypothetical example is a counterexample to your claim.
>> >
>> >But, your "hypothetical" doesn't mirror what actually happens in real
>> >life.
>>
>> You are wrong, as proven by Kolmogorov Complexity.
>
>Funny how real life demonstrations show that potentially beneficial
>sequences, in all language/information systems, do indeed decrease
>exponentially as a total fraction of all sequences, with increasing
>size and/or specificity requirements. Even those on your own side in
>this forum, like Leonid, tend to agree with this statement.
You are totally wrong, and you do not understand K.C. at all, and you
misrepresent K.C. and I gave you an example for "A New Kind of
Science" which shows that you are wrong.
I can give a popularized explanation for why you are wrong: Short
sequences can be part of long sequences, but long sequences cannot be
part of short sequences. Those short functional sequences can be
functional parts of long sequences, but functional long sequences
cannot be functional part of short sequences. In other words: Long
functional sequences may have redundant parts. They may be short
sequences with added garbage. This means that once a sequence is long
enough to do a particular function, even longer sequence will also be
able to do that particular function. Thats the reason why long
sequences have the same ratio of beneficial sequences instead of
decreasing exponentially. However, since longer sequences can do stuff
that shorter cannot, the ratio must increase.
>> >> >> And Kolmogorov Complexity do indeed have something to say about this:
>> >> >> If the minimum functional complexity density is 1e-30 in 100
>> >> >> a.a. sequences, then the corresponding density in 2000 a.a. sequences
>> >> >> would be about the same, 1e-30, or higher, definitely not lower.
>> >> >
>> >> >You're wrong. If the specificity of each stretch of 100aa were 1e-30
>> >> >sequences the specificity of a 2000aa stretch would be 1e-30^20 or
>> >> >1e-600.
>> >>
>> >> I am not wrong. I supplied evidence above, from Wolframs book. And
>> >> you assume that specificity is approximately constant per base, per
>> >> sequence, which you have no evidence for, and which is not true.
>> >
>> >There are several published papers suggesting that such numbers are
>> >true. (See my discussions about papers by Yockey and Sauer). Beyond
>> >this, your math is way off.
>>
>> And this from a man who do not even know how to combine independent
>> probabilities.
>
>Look, if the specificity for 100aa is 1e-30, and a 200aa stretch has
>this same specificity for each 100aa stretch, what is the overall
>specificity for the 200aa sequence?
Then it is indeed 1e-30 as you say. BUT: you CANNOT split a minimally
into 2 pieces, because if you could, it would not be minimal, the 2
pieces would be smaller each. You would have 2 minimal sequences, not 1
twice as long.
>It is not 1e-30 as you suggest.
>It is, rather, 1e-30 x 1e-30 = 1e-60 as the odds of picking such a
>200aa sequence, at random, out of 200aa sequence space. If you still
>think I'm wrong, get Leonid to explain how I'm wrong here (Hint: Leonid
>made this particular calculation himself).
Your calculations can only be right if you do not couple them to the
demand of being minimally complex.
If you specify specificity without regard to the complexity, your
calculations can be right. But when you consider complexity as well,
you cannot subdivide or split into parts the problem any more, because
the impossibility of simplifying is a part of being minimally complex.
>As another example, consider a slot machine with the 26 letters of the
>alphabet on one of the wheels. If only the latter "A" where the
>"winning" letter, what would the odds be of success with a specificity
>of 1 in 26? Of course, the odds would be 1 in 26 spins of the wheel.
Yes.
>Now, given a specificity o f 1 in 26 per wheel, what would the odds of
>success be of getting 3 A's on 3 wheels with one pull of the slot
>machine handle? The odds would be 1 in 26^3 - right? Now, what if 13
>of the 26 letters where "winners"? What would the odds of success be
>for 3 wheels to all hit "winning" letters at the same time? The odds
>would be 1 in 2^3 - right?
Yes.
Why did you not apply that type of calculations to the probability of
neighbors being functional? You would have gotten the right formula,
like I did.
>< snip >
>
>> >> >Ok, for argument's sake, let's just say that there are two beneficial
>> >> >sequences in all of 2000aa sequence space (i.e., 1e-2600 sequences).
>> >> >If randomly distributed in sequence space, how far will they be (i.e.,
>> >> >how many character differences) on average, from each other? Use your
>> >> >formula to answer this question and lets see if your notions of
>> >> >sequence space dimensionally and how this affects the closeness of
>> >> >beneficial sequences are correct.
>> >>
>> >> That formula does not answer that question.
>> >>
>> >> The average difference would be about 2000*19/20=1900 with a standard
>> >> deviation of about 40 amino acids.
>> >
>> >The average number of character difference would be only 40aa? This is
>> >wrong.
>>
>> So, you do not know what a standard deviation is either.
>
>I was reading to fast on this one and somehow read that your answer was
>40aa instead of 1900aa. Sorry.
Excuse Accepted!
:-)
>However, since you did actually get the right answer here, consider
>that given a very low density of only 2 beneficial sequences in
>sequence space, where all the other sequences in sequence space are
>defined as non-beneficial, the 2 beneficial sequences would indeed be
>highly isolated. They would in fact be islands - surrounded on all
>sides by vast numbers of non-beneficial sequences.
Yes.
>So, you see, given very low ratios of beneficial sequences, the
>creation of islands is not only possible, but quite likely.
When the ratio is so low that only 1 or a very few sequence exist,
then it is true.
However, with a very low ratio, but still so high that there are
hundreds of beneficial sequences, non-island structures can exist,
like toruses, sierpinski triangles, swiss cheese, branchings, etc.
Kim0
.
- Follow-Ups:
- Re: For Kim G. S. Øyhus
- From: Seanpit
- Re: For Kim G. S. Øyhus
- References:
- For Kim G. S. Øyhus
- From: Seanpit
- Re: For Kim G. S. Øyhus
- From: Seanpit
- Re: Re: For Kim G. S. Øyhus
- From: Kim G. S. Øyhus
- Re: For Kim G. S. Øyhus
- From: Seanpit
- For Kim G. S. Øyhus
- Prev by Date: Re: Opinion: Intelligent Design Proponents Distance Themselves from Creationists
- Next by Date: Re: Keeping up with the mental retardation of Richard Dawkins
- Previous by thread: Re: For Kim G. S. Øyhus
- Next by thread: Re: For Kim G. S. Øyhus
- Index(es):
Relevant Pages
|
