The Pitman CSI Formula
R. Baldwin
Complex Specified Information (CSI). Here is the Pitman CSI formula:
Given a string A and a reference string B
CSI = X^n - (X^hd)
Where:
n = the number of symbols in string A
X = the number of symbols in the symbol set
hd = the Hamming Distance between A and B (i.e., the number of symbols
that do not match)
Reference his post of July 15, 2007:
news:1184513686.4...@g12g2000prg.googlegroups.com
So.. how well does this work?
Let's try a few examples.
Case 1:
A = 0000000000
B = 3141592653
X = 10
n = 10
hd = 10
CSI = 10^10 - 10^10
CSI = 0
10 digits of zero have zero CSI with respect to the first 10 digits of pi.
This seems reasonably intuitive.
Case 2:
A = 3141592653
B = 0000000000
X = 10
n = 10
hd = 10
CSI = 10^10 - 10^10
CSI = 0
The first 10 digits of pi have zero CSI with respect to 10 repeating digits
of zero. This seems less intuitive, but since B is supposed to be the
specification, we'll let it pass.
Case 3:
A = 3141592653
B = 3141592653
X = 10
n = 10
hd = 10
CSI = 10^10 - 10^0
CSI = 9,999,999,999
This is OK. The digits of pi look reasonably complex, and they match the
specification, so we get the maximum value.
Case 4:
A = 0000000000
B = 0000000000
X = 10
n = 10
hd = 0
CSI = 10^10 - 10^0
CSI = 9,999,999,999
Whoops! This is a surprisingly big number, and not at all intuitive, because
a string of zeros is not intuitively complex. Now we see the first serious
problem with the Pitman CSI formula: Any string, no matter what its
contents, no matter what pattern is found therein, has the maximum possible
CSI value with respect to itself. As long as A and B match, the CSI for A is
X^n - 1, where X is the number of symbols in the alphabet, and n is the
length of A.
Case 5:
A = 010110111
B = 101001000
X = 2
n = 10
hd = 10
CSI = 2^10 - 2^10
CSI = 0
Here's another problem. When A is the exact complement of B, it has zero
CSI., even though a trivial computation changes A into B.
Case 6:
A = 1245340028
B = 8200435421
X = 10
n = 10
hd = 8
CSI = 10^10 - 10^8
CSI = 9,900,000,000
Two new problems: First, when A is B in reverse order, whether CSI turns out
high or low is hit or miss. This is not a very intuitive result, having CSI
depend so drastically on whether you got the reference pattern forwards or
backwards. Second, the CSI for a string with a large Hamming Distance is not
much smaller than the CSI for a string that matches the reference pattern
exactly.
Looking at the case for X = 10, n = 10:
hd = 0 CSI = 9,999,999,999 10 left-hand 9's, 0 right hand 0's
hd = 1 CSI = 9,999,999,990 9 left-hand 9's, 1 right hand 0's
hd = 2 CSI = 9,999,999,900 8 left hand 9's, 2 right hand 0's
hd = 3 CSI = 9,999,999,000 7 left hand 9's, 3 right hand 0's
hd = 4 CSI = 9,999,990,000 6 left hand 9's, 4 right hand 0's
hd = 5 CSI = 9,999,900,000 5 left hand 9's, 5 right hand 0's
hd = 6 CSI = 9,999,000,000 4 left hand 9's, 6 right hand 0's
hd = 7 CSI = 9,990,000,000 3 left hand 9;s, 7 right hand 0's
hd = 8 CSI = 9,900,000,000 2 left hand 9's, 8 right hand 0's
hd = 9 CSI = 9,000,000,000 1 left hand 9, 9 right hand 0's
hd = 10 CSI = 0 0 left hand 9's, 10 right
hand 0's
Now lets try X = 2, n = 4:
hd = 0 CSI = 15 = 1111B 4 left hand 1's, 0 right hand 0's
hd = 1 CSI = 14 = 1110B 3 left hand 1's, 1 right hand 0
hd = 2 CSI = 12 = 1100B 2 left hand 1's, 2 right hand 0's
hd = 3 CSI = 8 = 1000B 1 left hand 1, 3 right hand 0's
hd = 0 CSI = 0 = 0000B 0 left hand 1, 4 right hand 0's
From this we see:
1. CSI is independent of the contents of a string, except with regard to its
Hamming Distance from some reference pattern.
2. CSI for a given string depends heavily on the choice of reference string,
since that determines the Hamming Distance.
3. CSI may be seen as simply the Hamming Distance expressed as a variety of
unary numeral system. For CSI in radix X, the number of zeros on the right
hand of the number is equal to the Hamming Distance (that is, of course, an
artifact of formulae having the form X^a - X^b).
4. CSI does not have very good dynamic range.
5. CSI does not have very good granularity. There are only n+1 possible
values it can take, if it is defined.
5. CSI does not provide an intuitive understanding of the "complexity" of a
string.
Case 7:
A = 3141592653
B = 31415926535
X = 10
n = 10
hd = undefined
CSI = undefined
Even though Sean was aware of modified Hamming Distance as one possible
metric to compare the match between two strings of different length, he
chose not to use it. He specifically selected, for reasons known only to
himself, Hamming Distanct. That means CSI can only exist for a string if it
is compared to a reference of exactly the same length. But it does lead to
unintuitive results here.
Given all this, it is rather difficult to see how the Pitman CSI formula
will tell us anything useful about a string. It is highly unlikely to make a
design predictor. But, kudos to Sean for at least giving it a try. That's
more than most ID-ists have done.
Bobby Bryant
"R. Baldwin" <res0...@nozirevBACKWARDS.net> writes:
> Given all this, it is rather difficult to see how the Pitman CSI
> formula will tell us anything useful about a string. It is highly
> unlikely to make a design predictor. But, kudos to Sean for at least
> giving it a try. That's more than most ID-ists have done.
The whole idea is bunkum, whether you take Sean's version or not.
Suppose you set up a random number generator -- a real one based on
thermal noise or radioactive decay rather than an algorithmic pseudo-
RNG -- to generate a string of a million binary digits and write them
directly into the memory of your computer. The resulting string would
be genuinely random, and thus not have any Dembski-style CSI because
it would not conform to "an independently given pattern". Right?
Now you make a copy of the string in your computer's memory, and you
find that the copy has incredible amounts of CSI, because it's a
million digits long and it conforms exactly to an independently given
pattern, namely the first string.
So what's going on here? The original doesn't have any CSI, but the
exact copy has vast amounts? CSI isn't a property of the string
itself?
Now suppose you offer both strings for others to examine. Can they
tell which has CSI and which doesn't? Only if they know which is the
independently given pattern and which is the copy. How is it that you
can only tell which is intelligently designed if you know the "causal
history" of the two strings? What's the point in a CSI calculation,
if you have to know which of the comparanda was intelligently designed
beforel you can apply it?
Interestingly, you can tell the examiners which string is which by
giving them a single bit of information. An easy way to to that would
be to append a bit to the head of the strings, with the convention
that "0" means "original" and "1" means "copy". Now the examiners can
easily determine which has CSI and which doesn't.
But they have two strings 1,000,001 bits long, which differ only by
a single bit in a distinguished position, and one of those strings has
vast amounts of CSI whereas the other has almost none. That one bit
seems to carry a mighty lot of CSI along with it.
FWIW, Sean's method, CSI = X^n - (X^hd), gives a CSI of 2^million -1
for *both* strings, even though one of them was generated by a random
process. If you compare the strings with the prepended bit, that
number is reduced to a mere 2^million -2, again for both strings, even
though one of them only has a single bit set by an intelligent
designer.
It's absurd to think you can prove God exists by means of an algorithmic /
mathematical analysis of the description of a found object. CSI certainly
isn't going to do that for you, no matter how you define it.
--
Bobby Bryant
Reno, Nevada
Remove your hat to reply by e-mail.
Mark VandeWettering
> Sean Pitman has been good enough to provide his mathematical formula for
> Complex Specified Information (CSI). Here is the Pitman CSI formula:
>
> Given a string A and a reference string B
>
> CSI = X^n - (X^hd)
>
> Where:
>
> n = the number of symbols in string A
> X = the number of symbols in the symbol set
> hd = the Hamming Distance between A and B (i.e., the number of symbols
> that do not match)
>
> Reference his post of July 15, 2007:
> news:1184513686.4...@g12g2000prg.googlegroups.com
>
> So.. how well does this work?
Well, it is at least defined, but it doesn't have very many useful
properties. For one thing, it doesn't actually have a definition unless
you actually think that you have a string to compare something against.
Indeed, it can only be measured against a single, prespecified string.
I'm not certain what the utility of that might be. In fact, it seems
rather obviously pointless.
> Let's try a few examples.
>
> Case 1:
> A = 0000000000
> B = 3141592653
> X = 10
> n = 10
> hd = 10
>
> CSI = 10^10 - 10^10
> CSI = 0
> 10 digits of zero have zero CSI with respect to the first 10 digits of pi.
> This seems reasonably intuitive.
Intuitive only because we have a preconception about what probability
distributions are drawn from.
Indeed. This seems rather pointless.
Pointless blather.
Mark
R. Baldwin
news:slrnf9m3jv.2...@fishtank.brainwagon.org...
> On 2007-07-16, R. Baldwin <res0...@nozirevBACKWARDS.net> wrote:
>> Sean Pitman has been good enough to provide his mathematical formula for
>> Complex Specified Information (CSI). Here is the Pitman CSI formula:
>>
>> Given a string A and a reference string B
>>
>> CSI = X^n - (X^hd)
>>
>> Where:
>>
>> n = the number of symbols in string A
>> X = the number of symbols in the symbol set
>> hd = the Hamming Distance between A and B (i.e., the number of
>> symbols
>> that do not match)
>>
>> Reference his post of July 15, 2007:
>> news:1184513686.4...@g12g2000prg.googlegroups.com
>>
>> So.. how well does this work?
>
> Well, it is at least defined, but it doesn't have very many useful
> properties. For one thing, it doesn't actually have a definition unless
> you actually think that you have a string to compare something against.
> Indeed, it can only be measured against a single, prespecified string.
> I'm not certain what the utility of that might be. In fact, it seems
> rather obviously pointless.
It is pointless, to anyone with a fair knowledge of math. The metric itself
is so badly flawed that it was not necessary to go into what makes one
reference string better than any other.
[snip rest]
richardal...@googlemail.com
wrote:
I fail to see how that definition is of any use at all to anyone, even
if it gave a useful measure of CSI - which, as you have so clearly
demonstrated, it doesn't.
It all depends of having a reference string. What reference string can
we use when referring to proteins, or genetic code?
Not that Sean will address this problem, of course. He'll simply
insist that everyone else is too stupid to understand the predictive
power of his formula.
RF
TomS
<gwDmi.39628$Um6....@newssvr12.news.prodigy.net>, Bobby Bryant stated..."
This seems to be very problematic for any objective measure of
CSI.
At least, I can't think of any response. I'd very much like to hear
of any.
--
---Tom S.
"There was a lot more to magic, as Harry quickly found out, than waving your
wand and saying a few funny words."
JK Rowling, Harry Potter and the Sorcerer's Stone, Chapter VIII, page 133
_Arthur
wrote:
> Sean Pitman has been good enough to provide his mathematical formula for
> Complex Specified Information (CSI). Here is the Pitman CSI formula:
>
> Given a string A and a reference string B
>
> CSI = X^n - (X^hd)
>
> Where:
>
> n = the number of symbols in string A
> X = the number of symbols in the symbol set
> hd = the Hamming Distance between A and B (i.e., the number of symbols
> that do not match)
Pitman's CSI formula clearly shows that bird's ability to fly exhibit
high CSI, while pig's ability to fly has low CSI, and elephant's
flight an even lower CSI, except for Dumbo.
Here the string B, the reference string is "Designer wanted creatures
with feathers and beaks to fly in the sky"
and string A is "Birds are flying creatures with feathers and beaks"
You can measure how low the Hamming distance is between A and B, and
how sky-high the CSI is.
The measure of CSI for fruit bats is left as an exercise to the
reader. Feel free to come up with your own Reference String, as long
it faithfully carries The Designer intent.
hersheyhv
wrote:
> Complex Specified Information (CSI). Here is the Pitman CSI formula:
>
> Given a string A and a reference string B
>
> CSI = X^n - (X^hd)
>
> Even though Sean was aware of modified Hamming Distance as one possible
> metric to compare the match between two strings of different length, he
> chose not to use it. He specifically selected, for reasons known only to
> himself, Hamming Distanct. That means CSI can only exist for a string if it
> is compared to a reference of exactly the same length. But it does lead to
> unintuitive results here.
Gee. When I point out that whatever math or model he uses, he always
implies that one is going from a sequence of length x to another
sequence of length x in a long series of single letter changes, he
accuses me of presenting a "strawman" version of his 'real' argument.
So maybe he will point us to his 'real' argument sometime?
> Given all this, it is rather difficult to see how the Pitman CSI formula
> will tell us anything useful about a string. It is highly unlikely to make a
> design predictor. But, kudos to Sean for at least giving it a try. That's
> more than most ID-ists have done.
For the sequences he is most interested in, this mathematical
description (absurd though it is) means that he has abandoned all
pretense of his number having any relationship to the *function* of
the protein in a cell. It is quite obvious to *anyone* with knowledge
of proteins and how they function, that one can have large hamming
distances between sequences while the proteins encoded by those
sequences have basically the same *structure* and *function*. *And*
one can change a single aa and completely disrupt either or both
*structure* and *function*. So, as long as he includes the hamming
distance in his calculation, there is no possible way to infer
*functional* design by use of that value of CSI.
John Harshman
Even more importantly, I think, Pitman's formula requires knowledge of
the target or reference sequence. How, in any biological case, does he
determine the reference sequence against which the real sequence is to
be compared? Until he explains this, his formula is operationally
useless, even if it means something in theory. What is the reference
sequence for hemoglobin?
T Pagano
<res0...@nozirevBACKWARDS.net> wrote:
>Sean Pitman has been good enough to provide his mathematical formula for
>Complex Specified Information (CSI). Here is the Pitman CSI formula:
>
>Given a string A and a reference string B
>
> CSI = X^n - (X^hd)
>
>Where:
>
> n = the number of symbols in string A
> X = the number of symbols in the symbol set
> hd = the Hamming Distance between A and B (i.e., the number of symbols
>that do not match)
>
CSI is an acronym for Complex Specified Information which is made up
of two separate measurable attributes----at least as far as Dembski's
theory is concerned. Both complexity and specified information have
defined criteria which must individually be met before any inference
can be made and the inference is NOT deductive.
Atheists in this forum have consistently raised a straw ID theory
which asserts that CSI is
(1) a single measure,
(2) can be arrived at mechanically (that is, algorithmically), and
(3) produces a result which is used to conclusively deduce whether or
not an intelligent agent is required.
All three are false. And it's hard to tell if Ptiman is on to
something with his derived value which he also labels "CSI" or whether
he is lending aid and comfort to the false atheist ID straw man.
Probability theory alone doesn't provide any tools for determining
degree of randomness of a bit string. Chaitin, Kolmogorov, and
Solomonoff offered a theory whereby they argued that the
compressibility of some bit string given some reference class was such
a measure. The less compressible a string then the more random it
was. In other words, the collection of algorithmically compressible
(and therefore non random) sequences has small probability among the
totality of sequences, so that observing such a sequence is reason to
look for explanations other than chance. I'm guessing that Pitman is
attempting to use this outcome as a means of deriving some single
measure. Dembski does no such thing.
Lastly, creationists and other like-minded people should avoid
engaging ID challenges as currently formulated. They are largely
sterile (that is, fruitless) because they are divorced from any real
world problem situation. Atheists can't help themselves in this
regard. Abiogenesis and neoDarwinism can only flourish when divorced
from real world biology and biochemistry. Forcing the atheist to
produce a real world problem will undoubtedly end the challenges
completely.
Regards,
T Pagano
Bobby Bryant
T Pagano <not....@address.net> writes:
> Atheists in this forum have consistently raised a straw ID theory
> which asserts that CSI is
> (1) a single measure,
> (2) can be arrived at mechanically (that is, algorithmically), and
Actually, I suspect that most of us take the position that it can't be
arrived at _at all_. Dembski has certainly never "arrived at" a
rigorous CSI calculation for any biological organism or structure.
Back when Dembski was actively peddling CSI, some of us took the
position that the only way the "C" part could be calculated is if you
knew the actual mechanism that produced the thing under consideration.
You can't very well evaluate how improbable something is merely by
looking at it.
As for the "S" part, in his NFL book he pretended to satisfy that
condition by claiming that the famous bacterial flagellum is like
a motorboat motor. Is he claiming that designs can be manufactured
before the design exists, or does he think motorboats have been
around longer than bacterial flagella?
Dembski was a clown long before he started making fart animations.
snex
> On Sun, 15 Jul 2007 20:37:38 -0700, "R. Baldwin"
>
i have offered a real world challenge pagano. you havent even
attempted to answer it. all i want you to do is tell me whether some
set of bits picked up on a radio frequency is intelligently designed
or just random noise. the security of our nation could depend on it.
John Harshman
> On Sun, 15 Jul 2007 20:37:38 -0700, "R. Baldwin"
> <res0...@nozirevBACKWARDS.net> wrote:
>
>
>>Sean Pitman has been good enough to provide his mathematical formula for
>>Complex Specified Information (CSI). Here is the Pitman CSI formula:
>>
>>Given a string A and a reference string B
>>
>> CSI = X^n - (X^hd)
>>
>>Where:
>>
>> n = the number of symbols in string A
>> X = the number of symbols in the symbol set
>> hd = the Hamming Distance between A and B (i.e., the number of symbols
>>that do not match)
>>
>
>
> CSI is an acronym for Complex Specified Information which is made up
> of two separate measurable attributes----at least as far as Dembski's
> theory is concerned. Both complexity and specified information have
> defined criteria which must individually be met before any inference
> can be made and the inference is NOT deductive.
I would like to know how one can determine rigorously that the criteria
have been met.
It has also been claimed by various creationists that CSI is a
quantitative measure, i.e. it's not either present or absent, but is an
amount that can increase or decrease. But Pagano's statement above seems
to be claiming otherwise, and that CSI is a binary, yes/no proposition.
If that's not the case, I would like to know how it can be quantified,
and how you can tell whether one item has more or less CSI than another.
Sean's formula at least has this virtue, which is much more than any
other creationist has attempted. (And it still requires comparison to a
known reference string, which we are unlikely to have in any real case.)
Pagano, of course, relies entirely on armwaving and obfuscation.
> Atheists in this forum
Starting out with a falsehood is, unfortunately, Pagano's modus
operandi. No matter how many times it's explained to him, he persists in
equating evolution with atheism. And that's a good index of his
intellectual honesty.
> have consistently raised a straw ID theory
> which asserts that CSI is
> (1) a single measure,
> (2) can be arrived at mechanically (that is, algorithmically), and
> (3) produces a result which is used to conclusively deduce whether or
> not an intelligent agent is required.
>
> All three are false.
Then perhaps Pagano can tell us what is correct. Is CSI two measures?
How are these measures performed, if not algorithmically? And if there
is a result, which is unclear from what Pagano says, what good is that
result?
> And it's hard to tell if Ptiman is on to
> something with his derived value which he also labels "CSI" or whether
> he is lending aid and comfort to the false atheist ID straw man.
I think Sean is showing great courage in presenting a real formula. This
is much more than Pagano has ever done, and Brave Sir Tony is being a
real shit for dumping on him without presenting any alternative or even
a real critique. But I think Sean also lends aid and comfort to the
false atheist ID straw man, as is Pagano. Sean shows that CSI is vacuous
by presenting a formula that does nothing useful, while Pagano shows it
by presenting nothing at all.
[snip more tendentious nonsense]
Inez
> Atheists in this forum have consistently raised a straw ID theory
> which asserts that CSI is
> (1) a single measure,
> (2) can be arrived at mechanically (that is, algorithmically), and
> (3) produces a result which is used to conclusively deduce whether or
> not an intelligent agent is required.
I think you've got this exactly backwards or at least mostly
backwards. I don't know about point number 1, but I assert that it is
cannot be arrived at at all, mechanically or otherwise, and does not
conclusively deduce anything.
Mark Nutter
Sure there's plenty of real-world challenges you could try to address
by ID. Here's a few more:
1) How would you use ID to determine whether or not Mt. Rushmore is
intelligently designed? The Matterhorn?
2) Is the Intelligent Designer intelligently designed?
3) Are fads the product of intelligent design? Election results?
Complex weather patterns? Why or why not?
m
_Arthur
wrote:
> T Pagano wrote:
> > have consistently raised a straw ID theory
> > which asserts that CSI is
> > (1) a single measure,
> > (2) can be arrived at mechanically (that is, algorithmically), and
> > (3) produces a result which is used to conclusively deduce whether or
> > not an intelligent agent is required.
>
> > All three are false.
Prayer, Faith and Scriptures are essential tools to measure CSI.
Ray Martinez
> On Sun, 15 Jul 2007 20:37:38 -0700, "R. Baldwin"
>
I think you are right; that Pitman has accepted Atheist straw men or
even worse: bought their presuppositions which have no correspondence
to reality. Historically, since Paley, ID is an observation and not a
theory. Once ID is positively identified by observation - game over.
Atheist presuppositions start by denying the reality of ID. Once their
presuppositions are accepted no amount of evidence can overcome said
supposition since it is not eligible to be explained apart from their
starting presuppositions.
Ray
snex
design is only observed when you see somebody design something. nobody
saw anybody design life.
jensp...@hotmail.com
> Sean Pitman has been good enough to provide his mathematical formula for
> Complex Specified Information (CSI). Here is the Pitman CSI formula:
>
> Given a string A and a reference string B
>
> CSI = X^n - (X^hd)
>
This looks rather phatetic. Looks like an idea from someone without no
real clue about what he is dooing.
Now if you want an quantificatio nof the difference between strings,
then a lot of very usuable ideas is already in use out there.
Like e.g. the editing distance. I.e. the smallest number of editing
operations needed to turn on string into the other. A very usefull
measure when it comes to fuzzy-searches on computers.
The editing distance is however a not a very well defined measure. So
imstead you might use the algorithmic complexity of the difference
between the two strings. I.e. the size of the smallest general Turing
machine that can turn one string into the other. Which ofcause means
that actual number can only be aproximated for any two strrings.
J.O.
Ray Martinez
>
> - Show quoted text -
Did you see anyone design your computer?
Your car?
Your camera?
Your television?
Surely this is not your point?
Ray
Seanpit
wrote:
That's right. The question is one of producing strings that match
each other even though each is the product of independent processes or
origins - like a mathematical algorithm vs. a radiosignal from outer
space. A radiosignal that matches a mathematical algorithm perfectly,
over time, becomes less and less likely to be the result of a truly
random process - regardless of the original algorithm(s) chosen for
comparison purposes.
> Case 5:
> A = 010110111
> B = 101001000
> X = 2
> n = 10
> hd = 10
>
> CSI = 2^10 - 2^10
> CSI = 0
>
> Here's another problem. When A is the exact complement of B, it has zero
> CSI., even though a trivial computation changes A into B.
It is only a trivial computation given the proper algorithm - which
itself is a "match" and therefore has maximum CSI.
> Case 6:
> A = 1245340028
> B = 8200435421
> X = 10
> n = 10
> hd = 8
>
> CSI = 10^10 - 10^8
> CSI = 9,900,000,000
>
> Two new problems: First, when A is B in reverse order, whether CSI turns out
> high or low is hit or miss. This is not a very intuitive result, having CSI
> depend so drastically on whether you got the reference pattern forwards or
> backwards. Second, the CSI for a string with a large Hamming Distance is not
> much smaller than the CSI for a string that matches the reference pattern
> exactly.
Again, this is an algorithm issue. Pick the correct algorithm, and
the match becomes clear. Don't pick the correct algorithm and the
match is not so clear. The apparent match is only clear to you
because you already have access to the algorithm, in your mind, that
allows you to quickly recognize that A is simply the reverse of B.
> Looking at the case for X = 10, n = 10:
> hd = 0 CSI = 9,999,999,999 10 left-hand 9's, 0 right hand 0's
> hd = 1 CSI = 9,999,999,990 9 left-hand 9's, 1 right hand 0's
> hd = 2 CSI = 9,999,999,900 8 left hand 9's, 2 right hand 0's
> hd = 3 CSI = 9,999,999,000 7 left hand 9's, 3 right hand 0's
> hd = 4 CSI = 9,999,990,000 6 left hand 9's, 4 right hand 0's
> hd = 5 CSI = 9,999,900,000 5 left hand 9's, 5 right hand 0's
> hd = 6 CSI = 9,999,000,000 4 left hand 9's, 6 right hand 0's
> hd = 7 CSI = 9,990,000,000 3 left hand 9;s, 7 right hand 0's
> hd = 8 CSI = 9,900,000,000 2 left hand 9's, 8 right hand 0's
> hd = 9 CSI = 9,000,000,000 1 left hand 9, 9 right hand 0's
> hd = 10 CSI = 0 0 left hand 9's, 10 right
> hand 0's
>
> Now lets try X = 2, n = 4:
> hd = 0 CSI = 15 = 1111B 4 left hand 1's, 0 right hand 0's
> hd = 1 CSI = 14 = 1110B 3 left hand 1's, 1 right hand 0
> hd = 2 CSI = 12 = 1100B 2 left hand 1's, 2 right hand 0's
> hd = 3 CSI = 8 = 1000B 1 left hand 1, 3 right hand 0's
> hd = 0 CSI = 0 = 0000B 0 left hand 1, 4 right hand 0's
>
> From this we see:
> 1. CSI is independent of the contents of a string, except with regard to its
> Hamming Distance from some reference pattern.
Yep . . .
> 2. CSI for a given string depends heavily on the choice of reference string,
> since that determines the Hamming Distance.
Yep . . .
> 3. CSI may be seen as simply the Hamming Distance expressed as a variety of
> unary numeral system. For CSI in radix X, the number of zeros on the right
> hand of the number is equal to the Hamming Distance (that is, of course, an
> artifact of formulae having the form X^a - X^b).
> 4. CSI does not have very good dynamic range.
I don't see the relevance . . .
> 5. CSI does not have very good granularity. There are only n+1 possible
> values it can take, if it is defined.
Don't see your point . . .
> 5. CSI does not provide an intuitive understanding of the "complexity" of a
> string.
The only reason a string has "intuitive" complexity or randomness is
when you don't know any simple algorithm for producing it. The fact
is that there could always be such an algorithm. The fact that you
haven't yet found it doesn't mean it isn't there. So, the best you
can say is that a particular string looks "intuitively" random - to
YOU. It may not look so random to someone else who knows how it was
actually produced.
> Case 7:
> A = 3141592653
> B = 31415926535
> X = 10
> n = 10
> hd = undefined
>
> CSI = undefined
>
> Even though Sean was aware of modified Hamming Distance as one possible
> metric to compare the match between two strings of different length, he
> chose not to use it. He specifically selected, for reasons known only to
> himself, Hamming Distanct. That means CSI can only exist for a string if it
> is compared to a reference of exactly the same length. But it does lead to
> unintuitive results here.
Modified Hamming Distance would work just fine - as I've already
pointed out to you before. Just use the size of larger sequence space
in the calculation.
> Given all this, it is rather difficult to see how the Pitman CSI formula
> will tell us anything useful about a string. It is highly unlikely to make a
> design predictor.
It isn't a design predictor, by itself. It is a bias detector and can
be used to predict the future nature of the bias of a growing string.
A hypothesis concerning the origin of bias requires additional
knowledge that goes beyond CSI.
> But, kudos to Sean for at least giving it a try. That's
> more than most ID-ists have done.
Not really . . . The concept has been there all along. Despite your
assertion to the contrary, bias can be detected without any knowledge
of the source of the string. This bias can then be used to predict
the future appearance of the string as it continues to grow.
Determining the actual source of the bias requires additional
information.
Sean Pitman
www.DetectingDesign.com
Seanpit
wrote:
That's right. The question is one of producing strings that match
each other even though each is the product of independent processes or
origins - like a mathematical algorithm vs. a radiosignal from outer
space. A radiosignal that matches a mathematical algorithm perfectly,
over time, becomes less and less likely to be the result of a truly
random process - regardless of the original algorithm(s) chosen for
comparison purposes.
> Case 5:
> A = 010110111
> B = 101001000
> X = 2
> n = 10
> hd = 10
>
> CSI = 2^10 - 2^10
> CSI = 0
>
> Here's another problem. When A is the exact complement of B, it has zero
> CSI., even though a trivial computation changes A into B.
It is only a trivial computation given the proper algorithm - which
itself is a "match" and therefore has maximum CSI.
> Case 6:
> A = 1245340028
> B = 8200435421
> X = 10
> n = 10
> hd = 8
>
> CSI = 10^10 - 10^8
> CSI = 9,900,000,000
>
> Two new problems: First, when A is B in reverse order, whether CSI turns out
> high or low is hit or miss. This is not a very intuitive result, having CSI
> depend so drastically on whether you got the reference pattern forwards or
> backwards. Second, the CSI for a string with a large Hamming Distance is not
> much smaller than the CSI for a string that matches the reference pattern
> exactly.
Again, this is an algorithm issue. Pick the correct algorithm, and
the match becomes clear. Don't pick the correct algorithm and the
match is not so clear. The apparent match is only clear to you
because you already have access to the algorithm, in your mind, that
allows you to quickly recognize that A is simply the reverse of B.
> Looking at the case for X = 10, n = 10:
> hd = 0 CSI = 9,999,999,999 10 left-hand 9's, 0 right hand 0's
> hd = 1 CSI = 9,999,999,990 9 left-hand 9's, 1 right hand 0's
> hd = 2 CSI = 9,999,999,900 8 left hand 9's, 2 right hand 0's
> hd = 3 CSI = 9,999,999,000 7 left hand 9's, 3 right hand 0's
> hd = 4 CSI = 9,999,990,000 6 left hand 9's, 4 right hand 0's
> hd = 5 CSI = 9,999,900,000 5 left hand 9's, 5 right hand 0's
> hd = 6 CSI = 9,999,000,000 4 left hand 9's, 6 right hand 0's
> hd = 7 CSI = 9,990,000,000 3 left hand 9;s, 7 right hand 0's
> hd = 8 CSI = 9,900,000,000 2 left hand 9's, 8 right hand 0's
> hd = 9 CSI = 9,000,000,000 1 left hand 9, 9 right hand 0's
> hd = 10 CSI = 0 0 left hand 9's, 10 right
> hand 0's
>
> Now lets try X = 2, n = 4:
> hd = 0 CSI = 15 = 1111B 4 left hand 1's, 0 right hand 0's
> hd = 1 CSI = 14 = 1110B 3 left hand 1's, 1 right hand 0
> hd = 2 CSI = 12 = 1100B 2 left hand 1's, 2 right hand 0's
> hd = 3 CSI = 8 = 1000B 1 left hand 1, 3 right hand 0's
> hd = 0 CSI = 0 = 0000B 0 left hand 1, 4 right hand 0's
>
> From this we see:
> 1. CSI is independent of the contents of a string, except with regard to its
> Hamming Distance from some reference pattern.
Yep . . .
> 2. CSI for a given string depends heavily on the choice of reference string,
> since that determines the Hamming Distance.
Yep . . .
> 3. CSI may be seen as simply the Hamming Distance expressed as a variety of
> unary numeral system. For CSI in radix X, the number of zeros on the right
> hand of the number is equal to the Hamming Distance (that is, of course, an
> artifact of formulae having the form X^a - X^b).
> 4. CSI does not have very good dynamic range.
I don't see the relevance . . .
> 5. CSI does not have very good granularity. There are only n+1 possible
> values it can take, if it is defined.
Don't see your point . . .
> 5. CSI does not provide an intuitive understanding of the "complexity" of a
> string.
The only reason a string has "intuitive" complexity or randomness is
when you don't know any simple algorithm for producing it. The fact
is that there could always be such an algorithm. The fact that you
haven't yet found it doesn't mean it isn't there. So, the best you
can say is that a particular string looks "intuitively" random - to
YOU. It may not look so random to someone else who knows how it was
actually produced.
> Case 7:
> A = 3141592653
> B = 31415926535
> X = 10
> n = 10
> hd = undefined
>
> CSI = undefined
>
> Even though Sean was aware of modified Hamming Distance as one possible
> metric to compare the match between two strings of different length, he
> chose not to use it. He specifically selected, for reasons known only to
> himself, Hamming Distanct. That means CSI can only exist for a string if it
> is compared to a reference of exactly the same length. But it does lead to
> unintuitive results here.
Modified Hamming Distance would work just fine - as I've already
pointed out to you before. Just use the size of larger sequence space
in the calculation.
> Given all this, it is rather difficult to see how the Pitman CSI formula
> will tell us anything useful about a string. It is highly unlikely to make a
> design predictor.
It isn't a design predictor, by itself. It is a bias detector and can
be used to predict the future nature of the bias of a growing string.
A hypothesis concerning the origin of bias requires additional
knowledge that goes beyond CSI.
> But, kudos to Sean for at least giving it a try. That's
> more than most ID-ists have done.
Not really . . . The concept has been there all along. Despite your
snex
no, and thats why these objects are not "observed design." however, i
do see car factories, camera factories, and television factories --
and factories to make parts for those factories. ive never seen any of
these objects reproduce themselves the way i see biology reproduce
itself. ive also never seen a biology factory.
>
> Ray
Tiny Bulcher
So tell us how and by what characteristics one distinguishes design from
non-design?
Seanpit
> In article <139lpvrqfjvu...@news.supernews.com>,
> "R. Baldwin" <res0k...@nozirevBACKWARDS.net> writes:
>
> > Given all this, it is rather difficult to see how the Pitman CSI
> > formula will tell us anything useful about a string. It is highly
> > unlikely to make a design predictor. But, kudos to Sean for at least
> > giving it a try. That's more than most ID-ists have done.
>
> The whole idea is bunkum, whether you take Sean's version or not.
>
> Suppose you set up a random number generator -- a real one based on
> thermal noise or radioactive decay rather than an algorithmic pseudo-
> RNG -- to generate a string of a million binary digits and write them
> directly into the memory of your computer. The resulting string would
> be genuinely random, and thus not have any Dembski-style CSI because
> it would not conform to "an independently given pattern". Right?
It would be unlikely to conform to an independently derived pattern,
but it would not be impossible. That's the problem. The "randomness"
of a finite or infinite string can never be proven.
> Now you make a copy of the string in your computer's memory, and you
> find that the copy has incredible amounts of CSI, because it's a
> million digits long and it conforms exactly to an independently given
> pattern, namely the first string.
All you have done here is shown that the copied string was copied in a
very non-random way. My version of CSI is all about detecting likely
bias in the production of a string. Two identical copies of a million-
character string would strongly support the hypothesis of biased
production.
< snip rest >
Sean Pitman
www.DetectingDesign.com
steve...@hotmail.com
>
> - Show quoted text -
Nope. I've never seen anyone design a car. I have never once
observed a car being designed. (If you didn't notice, observed is the
key word).
I have been taught history, and seen people design other things, like
houses, and seen webpages and ads and hundreds of other things that
imply cars being designed. From this I assume that people design
cars, instead of cars coming fully formed and full of gas from the
magical Wand of SUV Making.
The only standard we have to decide if something has been designed or
not is our own creations. Until we see something that is Not Human
design and create we have no ablility to judge if things we didn't
design where designed or not. We can observe and theorize but without
seeing design and creation take place, weither its aliens or a
Designer or Allah or whatever we have no proof of Design.
Ray Martinez
The womb is a ID factory; the eye is formed in the womb.
Ray
Seanpit
>
> > Sean Pitman has been good enough to provide his mathematical formula for
> > Complex Specified Information (CSI). Here is the Pitman CSI formula:
>
> > Given a string A and a reference string B
>
> > CSI = X^n - (X^hd)
>
> > Where:
>
> > n = the number of symbols in string A
> > X = the number of symbols in the symbol set
> > hd = the Hamming Distance between A and B (i.e., the number of symbols
> > that do not match)
>
> > Reference his post of July 15, 2007:
> >news:1184513686.4...@g12g2000prg.googlegroups.com
>
> > So.. how well does this work?
>
> Well, it is at least defined, but it doesn't have very many useful
> properties. For one thing, it doesn't actually have a definition unless
> you actually think that you have a string to compare something against.
> Indeed, it can only be measured against a single, prespecified string.
> I'm not certain what the utility of that might be. In fact, it seems
> rather obviously pointless.
The comparison is not so pointless if the pre-specified strings are
all produced by algorithms that are much simpler in comparison with
the strings themselves. For example, the formula for pi is very
simple, yet is capable of producing a string of infinite size. Any
match of a sequence in question to any such algorithmically produced
string would be a significant indicator of a biased non-random origin
of the analyzed string.
Also, a match to absolutely any single string or pattern can be quite
useful as far as identifying a biased non-random pattern. This
feature becomes very useful in evaluating the symmetry of a finite
object - like a highly symmetrical granite cube or snowflake. The
snowflake, in particular, can show an almost infinite variety of
patterns. Yet, a single snowflake often shows very high degrees of
point/reflective/rotational symmetry of one half compared to the other
half. This is a very strong indication of a biased non-random process
in the formation of one-half of the snowflake vs. the other half.
That information can be quite "useful". It isn't worthless or
meaningless.
> > Let's try a few examples.
>
> > Case 1:
> > A = 0000000000
> > B = 3141592653
> > X = 10
> > n = 10
> > hd = 10
>
> > CSI = 10^10 - 10^10
> > CSI = 0
>
> > 10 digits of zero have zero CSI with respect to the first 10 digits of pi.
> > This seems reasonably intuitive.
>
> Intuitive only because we have a preconception about what probability
> distributions are drawn from.
Exactly . . .
Rather, such a match would be a very good clue as to the non-biased
origin or production of the match. You'd call that information
"pointless"? Really?
Just because something is simple or basic doesn't mean it is pointless
or worthless.
< snip >
Sean Pitman
www.DetectingDesign.com
snex
we can observe a factory being designed. nobody has observed a womb
being designed. in fact, wombs are also formed in the womb, we can
watch it happen just as the eye is formed in the womb. strange thing,
as we watch, there is no designer doing anything.
>
> Ray
Seanpit
You yourself use this very same concept when you build your
phylogenetic trees. You assume evolutionary relationships based on
the closeness of matches between various genetic sequences. In other
words, you assume that a very close match was not the result of some
sort of random production of each sequence - but that there is in fact
some non-random biased relationship. And, I'd agree. There most
certainly is a biased relationship.
That is basically all that my CSI definition does. It helps one
detect a biased relationship between different sequences.
> RF
Sean Pitman
www.DetectingDesign.com
Ray Martinez
Atheist philosophy dismissing evidence.
Ray
Ken Rode
The formation of eyes is described through developmental biology, Ray.
Is it your contention, then, that everything we believe we know about
developmental biology is wrong? Do you believe that God assembles each
one of us individually, or do you believe that He might have put some
process in place whereby an organism could be constructed
automatically? If the former, then do you also believe that God
continues to assemble us as we continue to grow?
Absolute statements like "the eye is formed in the womb" are trivially
false, Ray. A significant number of multicellular species of animals
have eyes of some form or other, and relatively few of these species
have wombs. What did you really mean to say?
Rolf
"Ray Martinez" <pyram...@yahoo.com> wrote in message
news:1184609920.1...@o61g2000hsh.googlegroups.com...
What is observed?
> Atheist presuppositions start by denying the reality of ID.
Entirely wrong, atheism has nothing with acceptance of the ToE to do.
Denying the reality of ID is the what comes off the vacuity of ID - no
positive evidence for ID has ever been shown. All ID has shown so far is
that it is a scam aimed at teaching creationism in schools.
If it had anything with science to do, it would act in scientific fora and
not in courtrooms. Behe even refuse to do research because he thinks it is
useless. As far as I understand science, it always do research - to prove or
disprove a particular hypothesis. Bot not Behe. He is satisfied with saying
'science is wrong, and that is that.' Science is never satisfied- that is
waht sets it apart from creationism. Science corrects itself, while
creationism is elevated to religious faith.
Once their
> presuppositions are accepted no amount of evidence can overcome said
> supposition since it is not eligible to be explained apart from their
> starting presuppositions.
>
But you are objcetive and have no starting presuppositions, is that it? I
really think you believe that yourself, and that is proof that you are
(expletives deleted.)
> Ray
>
>
Vend
http://groups.google.com/group/talk.origins/msg/777086cd56999e73
On Jul 16, 7:20 pm, Seanpit <seanpitnos...@naturalselection.
0catch.com> wrote:
> On Jul 15, 2:24 pm, Mark Isaak <eci...@earthlink.net> wrote:
> > >> Units or no, you yourself have implicitly but repeatedly said CSI cannot
> > >> be measured.
> > >> If it can be measured, then measure it:
> > >> 11121123344432992877777777551111121
> > > You haven't provided enough information to calculate CSI. You've only
> > > given the size and specific characters per position. You haven't given
> > > any other "target" or reference string for comparison. The
> > > calculation of CSI, like symmetry, requires a comparison to be made
> > > between strings Q and Z.
> > > CSI = X^n - (X^hd)
> > > Where:
> > > X = the number of possible characters per position
> > > n = the size of the sequence
> > > hd = the Hamming Distance between strings Q and Z
Say I have this sequence: 11121123344432992877777777551111121 and I
want to know whether it's designed. How should I choose the reference
string? How should I interpet the resulting CSI value?
By the way I think that your formula does funny things:
A: 1 1 0 0 1 0 0 1 0 0 0 0 1 1 1 1 (first 16 binary digits of pi)
B: 0 0 1 1 0 1 1 0 1 1 1 1 0 0 0 0 (not A)
CSI(A, B) = 0
C: 1 1 1 0 0 1 0 0 1 0 0 0 0 1 1 1 (A right-rotated by one)
CSI(A, C) = 65472
Two unrelated strings of n randomly generated bits (independent with
uniform probability) have an average hamming distance of n/2.
Therefore their average CSI is:
2^n - 2^(n/2) = 2^n - 2^(n - n/2) = 2^n * (1 - 2^(-n/2)) = 2^n * (1 -
sqrt(1 / 2^n))
The limit as n approaches infinity is infinity.
Moreover the CSI of a sequence with itself is 2^n.
The difference with the previously calculated formula is: sqrt(2^n)
The relative difference is: sqrt(2^n) / 2^n = 1 / sqrt(2^n)
Whose limit as n approeaches infinity is zero.
If I did my calculations well, this means that with your definition of
CSI as the length increases it becomes more and more difficult to
distinguish two completely unrelated random sequences from two
completely correlated (one the copy of the other) ones.
This makes me wonder what your CSI is good for.
Rolf
By strictly natural, chemical processes. No designer there
> Ray
>
>
>
>
>
Ernest Major
Martinez <pyram...@yahoo.com> writes
then Ray, all argue (unintentionally?) against "intelligent design" as a
non-naturalistic cause of life.
--
alias Ernest Major
snex
what evidence? do you have actual video footage of a designer in a
womb? how about pictures? how about your own personal testimony of
seeing one? how about knowing a guy who saw one?
if you have any of the above, answer this one question: what color is
the designer?
>
> Ray
Vend
wrote:
> > And it's hard to tell if Ptiman is on to
> > something with his derived value which he also labels "CSI" or whether
> > he is lending aid and comfort to the false atheist ID straw man.
>
> I think Sean is showing great courage in presenting a real formula. This
> is much more than Pagano has ever done, and Brave Sir Tony is being a
> real shit for dumping on him without presenting any alternative or even
> a real critique. But I think Sean also lends aid and comfort to the
> false atheist ID straw man, as is Pagano. Sean shows that CSI is vacuous
> by presenting a formula that does nothing useful, while Pagano shows it
> by presenting nothing at all.
I think that probably most ID/Creation "science" active supporters are
less intellectually honest than Sean Pitman. They pretend to have a
scientific theory but they take great care never to provide some
definition that even slightly reseambles something that could be used
to carry out a test.
Sean's CSI isn't a good example: it's just the (rewritten) hamming
distance with some undefined reference string. As the reference string
can be anything we are still left unable to calculate the CSI.
But I agree that at least this is a try.
T Pagano
<jharshman....@pacbell.net> wrote:
>T Pagano wrote:
>
>> On Sun, 15 Jul 2007 20:37:38 -0700, "R. Baldwin"
>> <res0...@nozirevBACKWARDS.net> wrote:
>>
>>
>>>Sean Pitman has been good enough to provide his mathematical formula for
>>>Complex Specified Information (CSI). Here is the Pitman CSI formula:
>>>
>>>Given a string A and a reference string B
>>>
>>> CSI = X^n - (X^hd)
>>>
>>>Where:
>>>
>>> n = the number of symbols in string A
>>> X = the number of symbols in the symbol set
>>> hd = the Hamming Distance between A and B (i.e., the number of symbols
>>>that do not match)
>>>
>>
>>
>> CSI is an acronym for Complex Specified Information which is made up
>> of two separate measurable attributes----at least as far as Dembski's
>> theory is concerned. Both complexity and specified information have
>> defined criteria which must individually be met before any inference
>> can be made and the inference is NOT deductive.
>
>I would like to know how one can determine rigorously that the criteria
>have been met.
I suggest with no less rigor than Harshman can muster in his
story-telling about the forearms of some population of dinosaur
transforming over time into avian wings via the neoDarwinian
mechanism. Here Harshman is not using the criteria of "rigor" to
illuminate the the matter at hand, but as an entrace requirement to
consideration. ID theory must hurdle the bar of "rigor" before
consideration yet his story-telling about the purported emergence of
the avian wing is not likewise exposed to the hurdle.
Nonetheless a good example of the rigor with which CSI can be
determined is the from the movie "Contact." A time slice of
electromagetic radiation is received from the direction of the star
Vega. What is the mode of causation of the 1126 bit pattern: law,
chance, law-chance, or the result of an intelligent agent? If we
observe complex specified information we can infer intelligent agency.
What is the algorighmic process that is followed:
1. Is the bit pattern required by any laws of nature? No,
therefore the pattern is contingent.
2. Is the bit pattern complex? Yes and the probability of such a
bit pattern by chance is smaller than 10^-150 which exceeds the
probablistic resources of the universe. Probability and complexity
are inversely related.
3. Is the bit pattern specified? Yes, the SETI researchers recognized
the first 100 prime numbers in order.
4. One could infer that all the probabilistic resources of the entire
universe could not have generated this specified pattern and therefore
is the result of an intelligent agent.
more to follow if time permits
>
>It has also been claimed by various creationists that CSI is a
>quantitative measure, i.e. it's not either present or absent, but is an
>amount that can increase or decrease. But Pagano's statement above seems
>to be claiming otherwise, and that CSI is a binary, yes/no proposition.
>
>If that's not the case, I would like to know how it can be quantified,
>and how you can tell whether one item has more or less CSI than another.
>Sean's formula at least has this virtue, which is much more than any
>other creationist has attempted. (And it still requires comparison to a
>known reference string, which we are unlikely to have in any real case.)
>Pagano, of course, relies entirely on armwaving and obfuscation.
>
>> Atheists in this forum
>
>Starting out with a falsehood is, unfortunately, Pagano's modus
>operandi. No matter how many times it's explained to him, he persists in
>equating evolution with atheism. And that's a good index of his
>intellectual honesty.
>
>> have consistently raised a straw ID theory
>> which asserts that CSI is
>> (1) a single measure,
>> (2) can be arrived at mechanically (that is, algorithmically), and
>> (3) produces a result which is used to conclusively deduce whether or
>> not an intelligent agent is required.
>>
>> All three are false.
>
>Then perhaps Pagano can tell us what is correct. Is CSI two measures?
>How are these measures performed, if not algorithmically? And if there
>is a result, which is unclear from what Pagano says, what good is that
>result?
>
>> And it's hard to tell if Ptiman is on to
>> something with his derived value which he also labels "CSI" or whether
>> he is lending aid and comfort to the false atheist ID straw man.
>
>I think Sean is showing great courage in presenting a real formula. This
>is much more than Pagano has ever done, and Brave Sir Tony is being a
>real shit for dumping on him without presenting any alternative or even
>a real critique. But I think Sean also lends aid and comfort to the
>false atheist ID straw man, as is Pagano. Sean shows that CSI is vacuous
>by presenting a formula that does nothing useful, while Pagano shows it
>by presenting nothing at all.
>
>[snip more tendentious nonsense]
snex
what about the bit patterns i presented? are they law, chance, law-
chance, or an intelligent agent? get crackin.
T Pagano
<jharshman....@pacbell.net> wrote:
>T Pagano wrote:
>
>> On Sun, 15 Jul 2007 20:37:38 -0700, "R. Baldwin"
>> <res0...@nozirevBACKWARDS.net> wrote:
>>
>>
>>>Sean Pitman has been good enough to provide his mathematical formula for
>>>Complex Specified Information (CSI). Here is the Pitman CSI formula:
>>>
>>>Given a string A and a reference string B
>>>
>>> CSI = X^n - (X^hd)
>>>
>>>Where:
>>>
>>> n = the number of symbols in string A
>>> X = the number of symbols in the symbol set
>>> hd = the Hamming Distance between A and B (i.e., the number of symbols
>>>that do not match)
>>>
>>
>>
>> CSI is an acronym for Complex Specified Information which is made up
>> of two separate measurable attributes----at least as far as Dembski's
>> theory is concerned. Both complexity and specified information have
>> defined criteria which must individually be met before any inference
>> can be made and the inference is NOT deductive.
>
>I would like to know how one can determine rigorously that the criteria
>have been met.
>
>It has also been claimed by various creationists that CSI is a
>quantitative measure, i.e. it's not either present or absent, but is an
>amount that can increase or decrease. But Pagano's statement above seems
>to be claiming otherwise, and that CSI is a binary, yes/no proposition.
Complex Specified Information (CSI) is not a single measurable
quantity. The "observable component" is the SPECIFIED PATTERN
detachable from some object, system or event under investigation. The
"Complexity component" is calculated after the specified pattern is
found and is the probability of the SPECIFIED PATTERN given the
reference class of possibilities. Remember that probability is
inversely related to complexity.
The probablity tells us whether the observed SPECIFIED PATTERN is
sufficiently complex to be outside the probablisitic resources
available to the universe. Dembski determined any probability smaller
than 10^-150 is outside the unverse's law-chance reach. Harshman is
certainly entitled to show otherwise.
Whether some specified pattern increases depends on the event under
investigation. In the case of the movie "Contact" it took the 1126
bits (the first 100 primes in order) some finite time period to arrive
at the SETI receivors. Obviously the SETI pattern sieve couldn't
recognize the 100 primes in order until a sufficient time period
passed. In this problem situation the specified pattern increased
over time.
However, we aren't always investigating the mode of causation of time
varying events. When investigating objects like the bacterial
flagellum the SPECIFIED PATTERN is fixed. Here the motor is the
SPECIFIED PATTERN recognized from our background knowledge of motor
design.
more to follow if time permits
>
Seanpit
wrote:
> Even more importantly, I think, Pitman's formula requires knowledge of
> the target or reference sequence. How, in any biological case, does he
> determine the reference sequence against which the real sequence is to
> be compared? Until he explains this, his formula is operationally
> useless, even if it means something in theory. What is the reference
> sequence for hemoglobin?
Beneficial biosystem functions from a given perspective (i.e., a
particular kind of life form) are the "targets". Different types of
functional systems have different minimum structural threshold
requirements that consist of both size and specificity elements.
Those systems that require greater size and/or specificity minimum
thresholds occupy a larger sequence space of potential options. The
location of the beneficial targets within sequence/structure space is
unknown. Theoretically though, changing an existing genetic sequence
into a sequence that matches any one of the potential target sequences
would produce maximum CSI for that mutating sequence (relative to an
as yet unfound potentially beneficial target).
In this sense, then, any beneficial molecule or structural biosystem,
like hemoglobin, is the reference "target". It has a beneficial
address in sequence and structure space, which is known to be
relatively rare. Therefore, any match to this address or any other
beneficial address with equivalent minimum structural threshold
requirements is unlikely to be the product of a truly random process.
In other words, some sort of biasing agent was likely involved (like a
pre-existing sequence that was already close to this address, etc).
This hypothesis becomes even more reliable with a decreasing ratio of
potentially beneficial targets vs. non-targets.
Sean Pitman
www.DetectingDesign.com
John Harshman
Actually, I have never told any stories about the forearms, etc. You may
have me confused with someone else. Now, what can be shown quite easily
is that birds are nested within dinosaurs, and this is done by a nice,
rigorous, and satisfyingly mathematical process called phylogenetic
analysis. If birds are nested within dinosaurs, if follows that a
dinosaur forelimb became transformed into a bird wing, and indeed we
know of many intermediate forms along the way, if Pagano would care to
look. While I suspect that "the neoDarwinian mechanism" was involved, I
have no way of determining whether this is true.
> Here Harshman is not using the criteria of "rigor" to
> illuminate the the matter at hand, but as an entrace requirement to
> consideration. ID theory must hurdle the bar of "rigor" before
> consideration yet his story-telling about the purported emergence of
> the avian wing is not likewise exposed to the hurdle.
Let's skip ahead to the part where you actually explain something.
> Nonetheless a good example of the rigor with which CSI can be
> determined is the from the movie "Contact." A time slice of
> electromagetic radiation is received from the direction of the star
> Vega. What is the mode of causation of the 1126 bit pattern: law,
> chance, law-chance, or the result of an intelligent agent? If we
> observe complex specified information we can infer intelligent agency.
Note that this is just what Sean denies. CSI, to Sean, does not imply
intelligent agency. You and Sean seem to differ radically in your notion
of just what CSI is. But go on.
> What is the algorighmic process that is followed:
>
> 1. Is the bit pattern required by any laws of nature? No,
> therefore the pattern is contingent.
What bit pattern would have been required by any laws of nature?
> 2. Is the bit pattern complex? Yes and the probability of such a
> bit pattern by chance is smaller than 10^-150 which exceeds the
> probablistic resources of the universe. Probability and complexity
> are inversely related.
There seems to be a ghost of a definition of complexity here, but I bet
he backs away from it if asked for details.
> 3. Is the bit pattern specified? Yes, the SETI researchers recognized
> the first 100 prime numbers in order.
Hard to say what "specified" means here. Would 123412341234 also have
been specified? How do you decide if a signal is specified or not?
> 4. One could infer that all the probabilistic resources of the entire
> universe could not have generated this specified pattern and therefore
> is the result of an intelligent agent.
Actually, I don't see all that much complexity in the bit pattern, nor
does complexity seem necessary in order for the researchers to conclude
that the signal was intelligent. But never mind. This seems to have
nothing to do with either defining CSI or determining that anything in
particular displays it. Or with anything that can be used in a
biological setting, which is what we're supposedly here for.
> more to follow if time permits
Prediction: time will not permit. Pagano is about due to disappear from
this thread.
hersheyhv
> In article <apagano-qevm93df8mtc08flcror7a0je310qsc...@4ax.com>,
> T Pagano <not.va...@address.net> writes:
>
> > Atheists in this forum have consistently raised a straw ID theory
> > which asserts that CSI is
> > (1) a single measure,
> > (2) can be arrived at mechanically (that is, algorithmically), and
>
> arrived at _at all_. Dembski has certainly never "arrived at" a
> rigorous CSI calculation for any biological organism or structure.
Actually, I suspect that most of us think that the reason that CSI
cannot be arrived at by any mathematical process is that the
declaration that something has CSI is pulled out of the declarer's
arse whenever they want.
> Back when Dembski was actively peddling CSI, some of us took the
> position that the only way the "C" part could be calculated is if you
> knew the actual mechanism that produced the thing under consideration.
> You can't very well evaluate how improbable something is merely by
> looking at it.
>
> As for the "S" part, in his NFL book he pretended to satisfy that
> condition by claiming that the famous bacterial flagellum is like
> a motorboat motor. Is he claiming that designs can be manufactured
> before the design exists, or does he think motorboats have been
> around longer than bacterial flagella?
>
> Dembski was a clown long before he started making fart animations.
>
> --
> Bobby Bryant
> Reno, Nevada
>
> Remove your hat to reply by e-mail.
Vend
I don't think I really understood this post of yours, but I think that
you were asked to provide a reference sequence for hemoglobin, and it
appears that you are proposing hemoglobin itself as the reference. Are
you proposing to compare a sequence to itself or am I missing
something?
Moreover it may be worth pointing out that there are no beneficial
molecules or structures.
There are beneficial mutations.
John Harshman
In that it is not single, it is not measurable, and it's not a quantity?
> The "observable component" is the SPECIFIED PATTERN
> detachable from some object, system or event under investigation. The
> "Complexity component" is calculated after the specified pattern is
> found and is the probability of the SPECIFIED PATTERN given the
> reference class of possibilities. Remember that probability is
> inversely related to complexity.
How do you derive the reference class of possibilities?
> The probablity tells us whether the observed SPECIFIED PATTERN is
> sufficiently complex to be outside the probablisitic resources
> available to the universe. Dembski determined any probability smaller
> than 10^-150 is outside the unverse's law-chance reach. Harshman is
> certainly entitled to show otherwise.
First I would have to know what you mean by pretentious gobbledigook
like "law-chance reach". How about if I agree that anything with a
probability of 10^-150 is unlikely to occur?
> Whether some specified pattern increases depends on the event under
> investigation. In the case of the movie "Contact" it took the 1126
> bits (the first 100 primes in order) some finite time period to arrive
> at the SETI receivors. Obviously the SETI pattern sieve couldn't
> recognize the 100 primes in order until a sufficient time period
> passed. In this problem situation the specified pattern increased
> over time.
Was that intended to communicate something? A pattern increased over
time? In this case, a sequence of bits got longer. Is that what you mean?
> However, we aren't always investigating the mode of causation of time
> varying events. When investigating objects like the bacterial
> flagellum the SPECIFIED PATTERN is fixed. Here the motor is the
> SPECIFIED PATTERN recognized from our background knowledge of motor
> design.
That's absurd. A motor isn't a specified pattern at all. But go ahead.
Derive the reference class of possibilities and perform whatever other
calculations you need to show that a bacterial flagellum has CSI. Show
your work.
Am I to suppose that CSI is either present or absent, and is not
quantitative?
> more to follow if time permits
I doubt it, but we'll see.
I like the way you ignore all of what I say below, by the way.
Seanpit
> On Jul 16, 11:59 am, bdbry...@wherever.ur (Bobby Bryant) wrote:
>
> > In article <apagano-qevm93df8mtc08flcror7a0je310qsc...@4ax.com>,
> > T Pagano <not.va...@address.net> writes:
>
> > > Atheists in this forum have consistently raised a straw ID theory
> > > which asserts that CSI is
> > > (1) a single measure,
It is hard to know what you mean by "single measure". Do you have an
example of the multiple measures of your version of CSI?
> > > (2) can be arrived at mechanically (that is, algorithmically)
If you can't calculate or algorithmically derive or evaluate CSI at
all, how then can it be a scientific tool? How then can it be a
"measure" of anything?
> > Actually, I suspect that most of us take the position that it can't be
> > arrived at _at all_. Dembski has certainly never "arrived at" a
> > rigorous CSI calculation for any biological organism or structure.
Yet, my CSI calculation would given a non-random match between many
genetic sequences and biosystems - supporting the hypothesis of common
origin over that of random independent production.
> Actually, I suspect that most of us think that the reason that CSI
> cannot be arrived at by any mathematical process is that the
> declaration that something has CSI is pulled out of the declarer's
> arse whenever they want.
Not when it comes to my CSI calculation and means of evaluating the
potentially biased nature of a sequence or pattern.
Sean Pitman
www.DetectingDesign.com
Vend
0catch.com> wrote:
> On Jul 16, 1:07 pm, Vend <ven...@virgilio.it> wrote:
>
>
>
> > > > > You haven't provided enough information to calculate CSI. You've only
> > > > > given the size and specific characters per position. You haven't given
> > > > > any other "target" or reference string for comparison. The
> > > > > calculation of CSI, like symmetry, requires a comparison to be made
> > > > > between strings Q and Z.
>
> > > > > CSI = X^n - (X^hd)
>
> > > > > Where:
> > > > > X = the number of possible characters per position
> > > > > n = the size of the sequence
> > > > > hd = the Hamming Distance between strings Q and Z
>
> > Say I have this sequence: 11121123344432992877777777551111121 and I
> > want to know whether it's designed. How should I choose the reference
> > string? How should I interpet the resulting CSI value?
>
> pattern was or was not designed. CSI only tells you if a pattern was
> the likely result of some sort of bias. Once the bias is detected, one
> can then try to hypothesize about the origin of the bias. However,
> this type of hypothesis requires additional information about the
> material which carries the pattern as well as some history with the
> material in question as it relates to various forces of nature -
> intelligent and non-intelligent.
It seems to me that your CSI isn't capable of detecting bias in a
sequence.
It's just a way of comparing two sequences.
> > By the way I think that your formula does funny things:
>
> > A: 1 1 0 0 1 0 0 1 0 0 0 0 1 1 1 1 (first 16 binary digits of pi)
> > B: 0 0 1 1 0 1 1 0 1 1 1 1 0 0 0 0 (not A)
> > CSI(A, B) = 0
>
> > C: 1 1 1 0 0 1 0 0 1 0 0 0 0 1 1 1 (A right-rotated by one)
> > CSI(A, C) = 65472
>
> > Two unrelated strings of n randomly generated bits (independent with
> > uniform probability) have an average hamming distance of n/2.
> > Therefore their average CSI is:
> > 2^n - 2^(n/2) = 2^n - 2^(n - n/2) = 2^n * (1 - 2^(-n/2)) = 2^n * (1 -
> > sqrt(1 / 2^n))
> > The limit as n approaches infinity is infinity.
>
> > Moreover the CSI of a sequence with itself is 2^n.
> > The difference with the previously calculated formula is: sqrt(2^n)
> > The relative difference is: sqrt(2^n) / 2^n = 1 / sqrt(2^n)
> > Whose limit as n approeaches infinity is zero.
>
> > If I did my calculations well, this means that with your definition of
> > CSI as the length increases it becomes more and more difficult to
> > distinguish two completely unrelated random sequences from two
> > completely correlated (one the copy of the other) ones.
> > This makes me wonder what your CSI is good for.
>
> As the length of a sequence increases, the CSI for two independently
> produced random strings will approach: (1/X) (X^n - X^hd). Compare
> this to the CSI for two completely correlated sequences which equals
> X^n.
Sorry but I don't understand where you get that (1/X) from.
Moreover, if we are talking about random strings, then hd is a random
variable (binomially distributed) thus CSI is also a random variable.
I was considering expectations.
Vend
> On Jul 16, 8:25 am, T Pagano <not.va...@address.net> wrote:
>
>
>
> > On Sun, 15 Jul 2007 20:37:38 -0700, "R. Baldwin"
>
> > >Sean Pitman has been good enough to provide his mathematical formula for
> > >Complex Specified Information (CSI). Here is the Pitman CSI formula:
>
> > >Given a string A and a reference string B
>
>
> > >Where:
>
> > > X = the number of symbols in the symbol set
> > > hd = the Hamming Distance between A and B (i.e., the number of symbols
> > >that do not match)
>
> > CSI is an acronym for Complex Specified Information which is made up
> > of two separate measurable attributes----at least as far as Dembski's
> > theory is concerned. Both complexity and specified information have
> > defined criteria which must individually be met before any inference
> > can be made and the inference is NOT deductive.
>
> > which asserts that CSI is
> > (1) a single measure,
> > (3) produces a result which is used to conclusively deduce whether or
> > not an intelligent agent is required.
>
> > something with his derived value which he also labels "CSI" or whether
> > he is lending aid and comfort to the false atheist ID straw man.
>
Historically, people "observed" that lightnings were caused by Zeus,
Thor or <enter your preferred thunder deity />.
Were they correct?
_Arthur
> Once ID is positively identified by observation - game over.
And, in your upcoming paper, will you positively observe Design,
Ray ?
Where ? In the Pampas of Patagonia ? In the rainforests of Congo ?
In the steppes of Khazkstan ?
In the Perfection of Man ?
Isn't your Great Paper due Real Soon Now ?
I think your brother pre-announced it some time ago.
John Harshman
All that is very nice, but it doesn't answer the question. We apparently
don't know the reference sequence, and apparently there are a great many
reference sequences to be taken into account, including but not limited
to all known functional sequences -- in fact, the reference sequence is
the universe of all possible sequences that have any function in some
context. I don't see any way to apply your formula as given to the
situation you have described. Do you?
Seanpit
wrote:
> Actually, I have never told any stories about the forearms, etc. You may
> have me confused with someone else. Now, what can be shown quite easily
> is that birds are nested within dinosaurs, and this is done by a nice,
> rigorous, and satisfyingly mathematical process called phylogenetic
> analysis. If birds are nested within dinosaurs, if follows that a
> dinosaur forelimb became transformed into a bird wing, and indeed we
> know of many intermediate forms along the way, if Pagano would care to
> look. While I suspect that "the neoDarwinian mechanism" was involved, I
> have no way of determining whether this is true.
Many nested patterns are clearly demonstrated in various biosystems -
to a very high degree of my CSI going well beyond what would be
expected by random chance alone. Clearly some sort of biased common
origin must have been responsible. The question is, was this common
origin deliberate or non-deliberate in nature?
The only way to answer that question is to study the material in
question (biosystems in this case) as they relate to both deliberate
and non-deliberate natural forces. One must observe to see if there
is a pattern of limitation of one potential explanation vs. the
other. If there isn't, there is no way to tell the actual origin of
this clearly observed very predictable bias.
As it turns out, there is a very clear pattern of limitation when it
comes to the proposed mechanism of random mutation and function-based
(i.e., natural) selection that falls well shy of the known and
theoretical limits of what intelligent deliberate forces of nature can
achieve with the same material.
< snip rest >
Sean Pitman
www.DetectingDesign.com
Victor Eijkhout
> For example, the formula for pi is very
> simple, yet is capable of producing a string of infinite size.
Actually, the formula for pi involves infinite sums, so it's not all
that simple.
Victor.
--
Victor Eijkhout -- eijkhout at tacc utexas edu
Seanpit
wrote:
SETI scientists have the same problem. They get a radiosignal from
outer space. The very best they can do to evaluate the non-random
nature of the signal compare what they are getting to a known class of
non-random patterns. Is it therefore possible that SETI scientist
could miss a truly biased pattern because they simply didn't recognize
it at biased? Sure it is. In fact, this possibility is impossible to
overcome. This does not mean, however, that those signals that do
match a known class of biased patterns cannot be adequately recognized
as biased.
The same thing is true of biosystems. The match between hemoglobin
and other known biosystem proteins is clearly biased. The various
matches carry greater CSI than would be expected were the match the
result of independent random generation.
Sean Pitman
www.DetectingDesign.com
Free Lunch
_Arthur <Art...@sympatico.ca> wrote in
<1184624650.2...@w3g2000hsg.googlegroups.com>:
I thought it was his grandfather.
shel...@msn.com
wrote:
> T Pagano wrote:
snip
> > However, we aren't always investigating the mode of causation of time
> > varying events. When investigating objects like the bacterial
> > flagellum the SPECIFIED PATTERN is fixed. Here the motor is the
> > SPECIFIED PATTERN recognized from our background knowledge of motor
> > design.
>
> That's absurd. A motor isn't a specified pattern at all.
Explain this apparent absurdity. A pattern is a design, a structure, a
blueprint. A specified pattern is one that is understood,
recognizable. Identifying a system as a "motor" isn't a recognition of
a specified pattern? Or do you have a problem with regarding a
flagellum as a "motor"?
hersheyhv
According to his formula, self-reference maximizes CSI. That is, if
you don't have to change anything in the source to get the target, you
have CSI = X^n-(X^hd) = X^n - X^0 = X^n. Anything larger than 0 for
hd and CSI decreases.
Oh, and do note that X^n is the total sequence space. And Sean has
repeatedly said that he never meant that there had to be a search
through total sequence space to find functional sequences. I do fail
to see how his numerology reflects his words. But then, I think his
formula for CSI smells like a cattle feed-lot.
jensp...@hotmail.com
> This makes me wonder what your CSI is good for.
For anyone that understands just a bit of mathematics and information
theory is plainly obvious, that it's good for absolutely nothing. It's
an idea comming for a completely clueless person.
J.O.
Seanpit
There would be no beneficial mutations unless mutations were in fact
able to produce novel beneficial molecules or molecular systems.
Sean Pitman
www.DetectingDesign.com
Seanpit
> Seanpit <seanpitnos...@naturalselection.0catch.com> wrote:
> > For example, the formula for pi is very
> > simple, yet is capable of producing a string of infinite size.
>
> Actually, the formula for pi involves infinite sums, so it's not all
> that simple.
Relative to the size of the output it is . . . especially when you are
talking about the basic formula (i.e., the circumference of a circle
divided by its diameter). But, if you don't like pi, then use any
other simple algorithm - like the square root of 2.
> Victor.
Sean Pitman
www.DetectingDesign.com
John Harshman
> On Jul 16, 2:55 pm, John Harshman <jharshman.diespam...@pacbell.net>
> wrote:
>
>
>>Actually, I have never told any stories about the forearms, etc. You may
>>have me confused with someone else. Now, what can be shown quite easily
>>is that birds are nested within dinosaurs, and this is done by a nice,
>>rigorous, and satisfyingly mathematical process called phylogenetic
>>analysis. If birds are nested within dinosaurs, if follows that a
>>dinosaur forelimb became transformed into a bird wing, and indeed we
>>know of many intermediate forms along the way, if Pagano would care to
>>look. While I suspect that "the neoDarwinian mechanism" was involved, I
>>have no way of determining whether this is true.
>
> Many nested patterns are clearly demonstrated in various biosystems -
> to a very high degree of my CSI going well beyond what would be
> expected by random chance alone. Clearly some sort of biased common
> origin must have been responsible. The question is, was this common
> origin deliberate or non-deliberate in nature?
No, the first question is what you mean by "common origin". What sort of
common origin results in a nested hierarchy? I submit that the only
mechanism that fits is descent with modification and branching. If you
would like to claim that the particular changes that happened along this
tree of descent were not random but were chosen on purpose by the
mysterious designer, I have no way to show you wrong. (Though nothing in
the nature of those changes seems to require intervention either.)
> The only way to answer that question is to study the material in
> question (biosystems in this case) as they relate to both deliberate
> and non-deliberate natural forces. One must observe to see if there
> is a pattern of limitation of one potential explanation vs. the
> other. If there isn't, there is no way to tell the actual origin of
> this clearly observed very predictable bias.
Once more you have managed to confuse two questions: common descent and
the causes of mutations. Don't you ever tire of this?
> As it turns out, there is a very clear pattern of limitation when it
> comes to the proposed mechanism of random mutation and function-based
> (i.e., natural) selection that falls well shy of the known and
> theoretical limits of what intelligent deliberate forces of nature can
> achieve with the same material.
This is your claim. I have very little interest in discussing it with
you. If you want to argue, let's argue about common descent, which is my
field.
Seanpit
> > > > You haven't provided enough information to calculate CSI. You've only
> > > > given the size and specific characters per position. You haven't given
> > > > any other "target" or reference string for comparison. The
> > > > calculation of CSI, like symmetry, requires a comparison to be made
> > > > between strings Q and Z.
>
> > > > CSI = X^n - (X^hd)
>
> > > > Where:
> > > > X = the number of possible characters per position
> > > > n = the size of the sequence
> > > > hd = the Hamming Distance between strings Q and Z
>
> Say I have this sequence: 11121123344432992877777777551111121 and I
> want to know whether it's designed. How should I choose the reference
> string? How should I interpet the resulting CSI value?
As I said before, CSI, by itself, does not tell you if the string's
pattern was or was not designed. CSI only tells you if a pattern was
the likely result of some sort of bias. Once the bias is detected, one
can then try to hypothesize about the origin of the bias. However,
this type of hypothesis requires additional information about the
material which carries the pattern as well as some history with the
material in question as it relates to various forces of nature -
intelligent and non-intelligent.
> By the way I think that your formula does funny things:
>
> A: 1 1 0 0 1 0 0 1 0 0 0 0 1 1 1 1 (first 16 binary digits of pi)
> B: 0 0 1 1 0 1 1 0 1 1 1 1 0 0 0 0 (not A)
> CSI(A, B) = 0
>
> C: 1 1 1 0 0 1 0 0 1 0 0 0 0 1 1 1 (A right-rotated by one)
> CSI(A, C) = 65472
>
> Two unrelated strings of n randomly generated bits (independent with
> uniform probability) have an average hamming distance of n/2.
> Therefore their average CSI is:
> 2^n - 2^(n/2) = 2^n - 2^(n - n/2) = 2^n * (1 - 2^(-n/2)) = 2^n * (1 -
> sqrt(1 / 2^n))
> The limit as n approaches infinity is infinity.
>
> Moreover the CSI of a sequence with itself is 2^n.
> The difference with the previously calculated formula is: sqrt(2^n)
> The relative difference is: sqrt(2^n) / 2^n = 1 / sqrt(2^n)
> Whose limit as n approeaches infinity is zero.
>
> If I did my calculations well, this means that with your definition of
> CSI as the length increases it becomes more and more difficult to
> distinguish two completely unrelated random sequences from two
> completely correlated (one the copy of the other) ones.
> This makes me wonder what your CSI is good for.
As the length of a sequence increases, the CSI for two independently
produced random strings will approach: (1/X) (X^n - X^hd). Compare
this to the CSI for two completely correlated sequences which equals
X^n.
Sean Pitman
www.DetectingDesign.com
John Harshman
I have no idea what you're computing when you compute CSI here. Are you
using one protein as a reference sequence for another, or are you doing
some kind of meta-computation in which you compare the entire pattern of
sequence similarities among proteins to some reference pattern of
similarities? Or are you merely agreeing that computation of CSI by your
formula is not feasible in a biological situation?
If all you're saying is that homologous sequences resemble each other
more than expected by chance, that's nice but hardly interesting. I'm
not even sure of you definition of "biased"; it seems to be something
similar to "unlikely by chance". Again, that's hardly interesting. We
knew that. Similar sequences do not arise by chance, and nested
hierachies of similar sequences do not arise by chance. They demand
explanation. But we know the explanation: descent with modification and
branching. Rarely, the explanation for some similarity is not common
descent but common function, as with a few important residues in the
lysozymes of some folivores. Don't see where this gets you, though.
Seanpit
> > > Say I have this sequence: 11121123344432992877777777551111121 and I
> > > want to know whether it's designed. How should I choose the reference
> > > string? How should I interpet the resulting CSI value?
>
> > As I said before, CSI, by itself, does not tell you if the string's
> > pattern was or was not designed. CSI only tells you if a pattern was
> > the likely result of some sort of bias. Once the bias is detected, one
> > can then try to hypothesize about the origin of the bias. However,
> > this type of hypothesis requires additional information about the
> > material which carries the pattern as well as some history with the
> > material in question as it relates to various forces of nature -
> > intelligent and non-intelligent.
>
> It seems to me that your CSI isn't capable of detecting bias in a
> sequence. It's just a way of comparing two sequences.
Yes - a comparison where if the two sequences were produced
independently by truly random processes, they would not have maximum
CSI. Rather, they would have the maximum CSI divided by the number of
possible characters per position. Therefore, extremely low and
extremely high CSI values are strong indicators of bias.
> > > If I did my calculations well, this means that with your definition of
> > > CSI as the length increases it becomes more and more difficult to
> > > distinguish two completely unrelated random sequences from two
> > > completely correlated (one the copy of the other) ones.
> > > This makes me wonder what your CSI is good for.
>
> > As the length of a sequence increases, the CSI for two independently
> > produced random strings will approach: (1/X) (X^n - X^hd). Compare
> > this to the CSI for two completely correlated sequences which equals
> > X^n.
>
> Sorry but I don't understand where you get that (1/X) from.
> Moreover, if we are talking about random strings, then hd is a random
> variable (binomially distributed) thus CSI is also a random variable.
> I was considering expectations.
Actually not. The Hamming Distance of two randomly produced strings
will tend toward a match between the strings that is 1/X or the
maximum possible distance divided by the number of possible characters
per position.
For example, if you flip a coin 10 times and then do it again and
compare the two sequences, how many of the positions will match, on
average? The answer is 5 or the maximum Hamming Distance of 10
divided by the number of possible characters per position (i.e., 2),
which equals 5 matches. If the experiment produced 10 matches or 0
matches, that would be out of line with a prediction of truly random
flips of the coin. Over time, if the flips are truly random, the
number of matches will fit this formula to a greater and greater
degree.
Sean Pitman
www.DetectingDesign.com
John Harshman
Well, it certainly contains a motor, if a motor is defined as a system
that transforms electrical or chemical energy to rotary motion. But I
would consider this too general to be considered a pattern. The motor of
a 1972 Pinto is a pattern. The bacterial flagellar motor doesn't operate
like that Pinto motor, or like any other motor we know of. It has a
unique design and a unique structure, and would have a unique blueprint
if indeed it had a blueprint, which it doesn't. Pagano is trying to say
simply "this is a motor, therefore this is designed", and I think he's
skipping all the important steps.
Vend
0catch.com> wrote:
> There would be no beneficial mutations unless mutations were in fact
> able to produce novel beneficial molecules or molecular systems.
A beneficial mutation is a mutation that increases the fitness of the
oranism bearing it.
What is a beneficial molecule or molecular system?
Seanpit
wrote:
> > SETI scientists have the same problem. They get a radiosignal from
> > outer space. The very best they can do to evaluate the non-random
> > nature of the signal compare what they are getting to a known class of
> > non-random patterns. Is it therefore possible that SETI scientist
> > could miss a truly biased pattern because they simply didn't recognize
> > it at biased? Sure it is. In fact, this possibility is impossible to
> > overcome. This does not mean, however, that those signals that do
> > match a known class of biased patterns cannot be adequately recognized
> > as biased.
>
> > The same thing is true of biosystems. The match between hemoglobin
> > and other known biosystem proteins is clearly biased. The various
> > matches carry greater CSI than would be expected were the match the
> > result of independent random generation.
>
> I have no idea what you're computing when you compute CSI here. Are you
> using one protein as a reference sequence for another, or are you doing
> some kind of meta-computation in which you compare the entire pattern of
> sequence similarities among proteins to some reference pattern of
> similarities? Or are you merely agreeing that computation of CSI by your
> formula is not feasible in a biological situation?
>
> If all you're saying is that homologous sequences resemble each other
> more than expected by chance, that's nice but hardly interesting.
It is actually very interesting. It is the basis for your notion of
common ancestry. I also use this evidence as the basis for my belief
in the common origin of such biosystems. The only thing we differ on
is the nature of the origin; not the fact that that origin was common.
> I'm
> not even sure of you definition of "biased"; it seems to be something
> similar to "unlikely by chance".
Biased means not random.
> Again, that's hardly interesting.
I guess what's "interesting" is in the eye of the beholder . . .
> We knew that.
The question is, how did you know that? You had to have some basis
for "knowing" that the origin was most likely biased . . .
> Similar sequences do not arise by chance, and nested
> hierachies of similar sequences do not arise by chance.
How do you know? The fact is that they could have arisen by purely
random "chance" as you put it. It is possible. It just isn't
statistically likely.
> They demand
> explanation. But we know the explanation: descent with modification and
> branching. Rarely, the explanation for some similarity is not common
> descent but common function, as with a few important residues in the
> lysozymes of some folivores. Don't see where this gets you, though.
There is a difference between detecting bias and detecting the actual
origin of that bias. We agree on the bias part. What we don't agree
on is the origin of the bias.
The real problem comes with the fact that those systems with
increasing "irreducible complexity", as a measure of the minimum
structural threshold requirements needed to produce the functions in
question, show less and less CSI (as I defined the term) compared to
anything else in the gene pool - as a fraction of the maximum possible
CSI. It is the significant decrease in CSI relative to each increase
in IC that gets the ToE into real trouble. Increasing the CSI at
higher levels of minimums structural threshold requirements is a real
challenge for the proposed mechanism of random mutation and function-
based selection that rapidly reaches the point of incredulity.
Sean Pitman
www.DetectingDesign.com
Seanpit
wrote:
< snip >
> > As it turns out, there is a very clear pattern of limitation when it
> > comes to the proposed mechanism of random mutation and function-based
> > (i.e., natural) selection that falls well shy of the known and
> > theoretical limits of what intelligent deliberate forces of nature can
> > achieve with the same material.
>
> This is your claim. I have very little interest in discussing it with
> you. If you want to argue, let's argue about common descent, which is my
> field.
And, it is your assumption that you can adequately hypothesize common
descent over extended periods of time by simply looking at the pattern
in question - without any additional knowledge as to how the material
you are considering actually relates to various forces of nature over
time.
Your problem is that you assume common descent over common design even
though you do not claim any basis in mechanism for your assumption.
You simply know that "common descent" is the answer based pretty much
on your philosophical position that "No intelligent designer would
have done it that way". How do you know this? Without reference to a
viable mechanism for your common descent when it comes to its
functional elements beyond extremely low levels, what real scientific
basis beyond some philosophical assumptions do you have?
Sean Pitman
www.DetectingDesign.com
Seanpit
wrote:
> Well, it certainly contains a motor, if a motor is defined as a system
> that transforms electrical or chemical energy to rotary motion. But I
> would consider this too general to be considered a pattern. The motor of
> a 1972 Pinto is a pattern. The bacterial flagellar motor doesn't operate
> like that Pinto motor, or like any other motor we know of. It has a
> unique design and a unique structure, and would have a unique blueprint
> if indeed it had a blueprint, which it doesn't. Pagano is trying to say
> simply "this is a motor, therefore this is designed", and I think he's
> skipping all the important steps.
The match isn't really that "general". There are a suprising number
of similarities between the flagellar motor and certain types of human-
designed motors. A lot of the parts look very similar and function in
a similar capacity. The "match" to the independently derived systems
is quite uncanny.
Sean Pitman
www.DetectingDesign.com
Vend
0catch.com> wrote:
> On Jul 16, 3:18 pm, Vend <ven...@virgilio.it> wrote:
>
> > > > Say I have this sequence: 11121123344432992877777777551111121 and I
> > > > want to know whether it's designed. How should I choose the reference
> > > > string? How should I interpet the resulting CSI value?
>
> > > As I said before, CSI, by itself, does not tell you if the string's
> > > pattern was or was not designed. CSI only tells you if a pattern was
> > > the likely result of some sort of bias. Once the bias is detected, one
> > > can then try to hypothesize about the origin of the bias. However,
> > > this type of hypothesis requires additional information about the
> > > material which carries the pattern as well as some history with the
> > > material in question as it relates to various forces of nature -
> > > intelligent and non-intelligent.
>
> > It seems to me that your CSI isn't capable of detecting bias in a
> > sequence. It's just a way of comparing two sequences.
>
> Yes - a comparison where if the two sequences were produced
> independently by truly random processes, they would not have maximum
> CSI.
Ok.
> Rather, they would have the maximum CSI divided by the number of
> possible characters per position.
Don't think so.
> Therefore, extremely low and
> extremely high CSI values are strong indicators of bias.
They indicate that the two sequences are correlated.
But they don't indicate that one sequence has been generated by a
biased process.
> > > As the length of a sequence increases, the CSI for two independently
> > > produced random strings will approach: (1/X) (X^n - X^hd). Compare
> > > this to the CSI for two completely correlated sequences which equals
> > > X^n.
>
> > Sorry but I don't understand where you get that (1/X) from.
> > Moreover, if we are talking about random strings, then hd is a random
> > variable (binomially distributed) thus CSI is also a random variable.
> > I was considering expectations.
>
> Actually not. The Hamming Distance of two randomly produced strings
> will tend toward a match between the strings that is 1/X or the
> maximum possible distance divided by the number of possible characters
> per position.
>
> For example, if you flip a coin 10 times and then do it again and
> compare the two sequences, how many of the positions will match, on
> average? The answer is 5 or the maximum Hamming Distance of 10
> divided by the number of possible characters per position (i.e., 2),
> which equals 5 matches. If the experiment produced 10 matches or 0
> matches, that would be out of line with a prediction of truly random
> flips of the coin. Over time, if the flips are truly random, the
> number of matches will fit this formula to a greater and greater
> degree.
The Hamming distance of two independently uniformly randomly generated
sequences of length n over an alphabet of X symbols is distributed
according to a binomial distribution whose expected value is n/X.
According to your formula CSI = X^n - X^hd the expected CSI of the two
random sequences is:
X^n - X^(n / X)
Which is not (1/X) (X^n - X^hd).
By the way, if you are averageing over hd, how can hd appear in the
final formula?
hersheyhv
0catch.com> wrote:
> On Jul 16, 8:03 am, John Harshman <jharshman.diespam...@pacbell.net>
> wrote:
>
> > Even more importantly, I think, Pitman's formula requires knowledge of
> > the target or reference sequence. How, in any biological case, does he
> > determine the reference sequence against which the real sequence is to
> > be compared? Until he explains this, his formula is operationally
> > useless, even if it means something in theory. What is the reference
> > sequence for hemoglobin?
>
> Beneficial biosystem functions from a given perspective (i.e., a
> particular kind of life form) are the "targets".
So *function* is the target? Not *sequence* (or even *structure*)?
And what would be the source *function* from which the target
*function* is reached if one starts with, say, the gamma-G form of
beta globin of hemoglobin and wanted to produce the gamma-A form of
beta globin of hemoglobin (if we did this, the CSI would be pretty
close to maximal, only identity of the two would produce a larger
CSI)? Or the reverse, since it is not known which is ancestral. Or
should we start with the putative ancestor of the delta and beta beta
globins? Or the putative ancestor between epsilon beta globin and the
gamma-A and gamma-G? Or the putative ancestor to all the beta
globins? Or the common ancestor to the beta and alpha globins
(knowing that there are chordates with only alpha globin in their
hemoglobin)? Or, are you starting from some random site in the total
sequence space you measure by X^n? Curious minds want to know?
> Different types of
> functional systems have different minimum structural threshold
> requirements that consist of both size and specificity elements.
How do you measure the specificity elements? All I ever see is size
and the measurement of total sequence (not structure or function)
space for that size.
> Those systems that require greater size and/or specificity minimum
> thresholds occupy a larger sequence space of potential options.
But NOT, according to you, total sequence space. Unless you want to
imply that total sequence space must be considered for some reason
related to how a particular sequence is found. And you keep saying
that you are not proposing a random search of total sequence space
from some random site elsewhere in total sequence space.
> The
> location of the beneficial targets within sequence/structure space is
> unknown.
Actually, all the beneficial targets that have ever *actually* been
found are contained in the small contiguous fraction of total
structure space indicated in Fig2 (or for different size subsets, Fig
4c) of the Choi and Kim paper. So the location of all the beneficial
targets one is actually interested in are known. Moreover, as Choi
and Kim point out, if one wants to find a protein with a particular
*function*, the place to look is near the proteins with similar
functions; they will typically have the similar *structures*
(regardless of size). [Sequence is a much poorer predictor of
function than structure is.]
> Theoretically though, changing an existing genetic sequence
> into a sequence that matches any one of the potential target sequences
> would produce maximum CSI for that mutating sequence (relative to an
> as yet unfound potentially beneficial target).
Can you tell me how one identifies the "mutating sequence" and the
"target"? And are you saying that if all you need to do is change one
aa to, say, convert a beta globin gamma-G into a gamma-A, that that
produces the same amount of CSI as converting an ancestral alpha-like
globin into one of the many different sequences (but nearly identical
structures) of modern adult or fetal or embryonic beta globin? Notice
that the *initial* change to convert an alpha into a beta need not
involve a large number of aa's, as there isn't even that much
structural difference between alpha and beta, but that subsequent
*neutral* change (and a little tweaking of the binding sites to
produce the separate embryonic, fetal, and adult forms) accounts for
most of the differences in the modern adult beta globins from
different species.
> In this sense, then, any beneficial molecule or structural biosystem,
> like hemoglobin, is the reference "target". It has a beneficial
> address in sequence and structure space, which is known to be
> relatively rare.
Well, actually the *structure* space nearby is filled with myoglobins
and other heme-binding globin proteins that are even more distantly
separated. In fact, even the *sequence* space for many of the
specific binding sites in these proteins are similar. As is the
structure of the genes (such as intron placement). So it looks like
the *structure* space near hemoglobin is actually filled by other
molecules with clearly related *structures* *and* *functions*, and to
a lesser extent, even similar *sequences*.
> Therefore, any match to this address or any other
> beneficial address with equivalent minimum structural threshold
> requirements is unlikely to be the product of a truly random process.
As in, say, the "strawman" idea of starting at some random site in
total sequence space and randomly searching one aa at a time until
one accidently lands on *the* island of sequence?
> In other words, some sort of biasing agent was likely involved (like a
> pre-existing sequence that was already close to this address, etc).
And *that* is exactly what the evidence from Choi and Kim shows.
> This hypothesis becomes even more reliable with a decreasing ratio of
> potentially beneficial targets vs. non-targets.
All that is needed is a search of *nearby* structure space for
structures with modified or 'novel' useful function.
>
> Sean Pitmanwww.DetectingDesign.com
John Harshman
It's a useless statement of some small fraction of the basis of my
notion of common ancestry.
> I also use this evidence as the basis for my belief
> in the common origin of such biosystems. The only thing we differ on
> is the nature of the origin; not the fact that that origin was common.
And this is your major obfuscation here, intentional or not. You are
confusing the origin of variation with the origin of the nested
hierarchy that contains that variation. This notion of "common origin"
is not useful, largely because of that conflation.
>>I'm
>>not even sure of you definition of "biased"; it seems to be something
>>similar to "unlikely by chance".
>
> Biased means not random.
Not as I understand it. The roll of a loaded die is biased, but it's
still random. The random distribution happens to differ from that of a
fair die.
>>Again, that's hardly interesting.
>
> I guess what's "interesting" is in the eye of the beholder . . .
>>We knew that.
>
> The question is, how did you know that? You had to have some basis
> for "knowing" that the origin was most likely biased . . .
Your use of language is bizarre. The basis is simple enough, and it's
simple probability, under any distribution you may prefer. No need to go
into anything like CSI.
>>Similar sequences do not arise by chance, and nested
>>hierachies of similar sequences do not arise by chance.
>
> How do you know? The fact is that they could have arisen by purely
> random "chance" as you put it. It is possible. It just isn't
> statistically likely.
It's unlikely enough that we can afford to ignore the possibility. On
this we are all agreed, so I can't imagine why you belabor the point.
>>They demand
>>explanation. But we know the explanation: descent with modification and
>>branching. Rarely, the explanation for some similarity is not common
>>descent but common function, as with a few important residues in the
>>lysozymes of some folivores. Don't see where this gets you, though.
>
> There is a difference between detecting bias and detecting the actual
> origin of that bias. We agree on the bias part. What we don't agree
> on is the origin of the bias.
>
> The real problem comes with the fact that those systems with
> increasing "irreducible complexity", as a measure of the minimum
> structural threshold requirements needed to produce the functions in
> question, show less and less CSI (as I defined the term) compared to
> anything else in the gene pool - as a fraction of the maximum possible
> CSI. It is the significant decrease in CSI relative to each increase
> in IC that gets the ToE into real trouble. Increasing the CSI at
> higher levels of minimums structural threshold requirements is a real
> challenge for the proposed mechanism of random mutation and function-
> based selection that rapidly reaches the point of incredulity.
Once again you conflate common descent, which is what I'm interested in,
with the mechanism of change, which is what you are apparently
interested in. I have very little interest in arguing about whether
mutation and selection are sufficient to account for adaptation. I'm
talking about the cause of nested hierarchy, which has nothing to do
with selection except as selection is one underlying cause of fixation.
Seanpit
> On Jul 17, 12:35 am, Seanpit <seanpitnos...@naturalselection.
>
> 0catch.com> wrote:
> > There would be no beneficial mutations unless mutations were in fact
> > able to produce novel beneficial molecules or molecular systems.
>
> A beneficial mutation is a mutation that increases the fitness of the
> oranism bearing it.
A beneficial mutation is a mutation that produces a beneficial
molecule or physicla system that wasn't there before.
> What is a beneficial molecule or molecular system?
Any system added to an organism that provides a survival/reproductive
advantage - regardless of how such a system was obtained - i.e., via
mutation, horizontal or vertical transfer, insertion by deliberate
human or non-human design, etc.
Sean Pitman
www.DetectingDesign.com
Vend
> On Jul 16, 4:02 pm, Vend <ven...@virgilio.it> wrote:
>
> > On Jul 17, 12:35 am, Seanpit <seanpitnos...@naturalselection.
>
> > 0catch.com> wrote:
> > > There would be no beneficial mutations unless mutations were in fact
> > > able to produce novel beneficial molecules or molecular systems.
>
> > A beneficial mutation is a mutation that increases the fitness of the
> > oranism bearing it.
>
> A beneficial mutation is a mutation that produces a beneficial
> molecule or physicla system that wasn't there before.
Actually, it doesn't have to be so.
A beneficial mutation can also change some structure that existed
previously or even remove a structure.
> > What is a beneficial molecule or molecular system?
>
> Any system added to an organism that provides a survival/reproductive
> advantage - regardless of how such a system was obtained - i.e., via
> mutation, horizontal or vertical transfer, insertion by deliberate
> human or non-human design, etc.
Thus anything obtained through a beneficial mutation.
John Harshman
> On Jul 16, 3:40 pm, John Harshman <jharshman.diespam...@pacbell.net>
> wrote:
>
> < snip >
>
>>>As it turns out, there is a very clear pattern of limitation when it
>>>comes to the proposed mechanism of random mutation and function-based
>>>(i.e., natural) selection that falls well shy of the known and
>>>theoretical limits of what intelligent deliberate forces of nature can
>>>achieve with the same material.
>>
>>This is your claim. I have very little interest in discussing it with
>>you. If you want to argue, let's argue about common descent, which is my
>>field.
>
> And, it is your assumption that you can adequately hypothesize common
> descent over extended periods of time by simply looking at the pattern
> in question - without any additional knowledge as to how the material
> you are considering actually relates to various forces of nature over
> time.
Yes. Feel free to argue against that assumption by showing how it can
fail under any realistic conditions.
> Your problem is that you assume common descent over common design even
> though you do not claim any basis in mechanism for your assumption.
Once again you conflate independent questions. You don't even seem to
notice when you do it.
> You simply know that "common descent" is the answer based pretty much
> on your philosophical position that "No intelligent designer would
> have done it that way". How do you know this?
No, what I know is that common descent is expected to produce a nested
hierarchy. It can't help doing so. And I know of no other mechanism that
would be expected to do that. Sure, a designer could do it, but a
designer could produce any pattern at all. Absent some particular reason
to expect a designer to produce nested hierarchy, I feel free to ignore
that question, just as I feel free to ignore the idea that any given
snowflake was produced by a designer. I know a mechanism that is capable
of producing snowflakes, and will do so if left to operate. "I had no
need of that hypothesis".
> Without reference to a
> viable mechanism for your common descent when it comes to its
> functional elements beyond extremely low levels, what real scientific
> basis beyond some philosophical assumptions do you have?
I don't know why this is so difficult for you. Let's agree for the sake
of argument that only divine intervention can move a sequence across the
large "neutral gaps" that you claim exist between sequences. I don't
care. All I care is that the differences among species are arranged in a
nested hierarchy. What caused those differences to arise is not
relevant. I can infer from the pattern that there is a tree, that it has
a particular shape, and which branches of that tree particular
differences arose on. I can thus say that the differences, whatever
their purposes if any, arose at different times and in common ancestors
of multiple species. If you want to come up with an alternative that
explains that particular hierarchy better than the inferred tree, please
do so. Just saying "common design" is not saying anything.
John Harshman
I would be happy to see you expand on that claim. What human-designed
motors? What parts?
T Pagano
<jharshman....@pacbell.net> wrote:
>T Pagano wrote:
>
>> On Sun, 15 Jul 2007 20:37:38 -0700, "R. Baldwin"
>> <res0...@nozirevBACKWARDS.net> wrote:
>>
>>
>>>Sean Pitman has been good enough to provide his mathematical formula for
>>>Complex Specified Information (CSI). Here is the Pitman CSI formula:
>>>
>>>Given a string A and a reference string B
>>>
>>> CSI = X^n - (X^hd)
>>>
>>>Where:
>>>
>>> n = the number of symbols in string A
>>> X = the number of symbols in the symbol set
>>> hd = the Hamming Distance between A and B (i.e., the number of symbols
>>>that do not match)
>>>
>>
>>
>> CSI is an acronym for Complex Specified Information which is made up
>> of two separate measurable attributes----at least as far as Dembski's
>> theory is concerned. Both complexity and specified information have
>> defined criteria which must individually be met before any inference
>> can be made and the inference is NOT deductive.
>
>I would like to know how one can determine rigorously that the criteria
>have been met.
>
>It has also been claimed by various creationists that CSI is a
>quantitative measure, i.e. it's not either present or absent, but is an
>amount that can increase or decrease. But Pagano's statement above seems
>to be claiming otherwise, and that CSI is a binary, yes/no proposition.
>
>If that's not the case, I would like to know how it can be quantified,
>and how you can tell whether one item has more or less CSI than another.
>Sean's formula at least has this virtue, which is much more than any
>other creationist has attempted. (And it still requires comparison to a
>known reference string, which we are unlikely to have in any real case.)
>Pagano, of course, relies entirely on armwaving and obfuscation.
A SPECIFIED PATTERN is not a scalar quatitiy (nor is it likely to be
reducible to one), but a pattern which is non ad hoc and independent
of the event it describes. It restricts the reference class of
possible events. For a pattern to also be "specified" it is not
enough to identity it simply by inspecting an event and noting its
features, it must be suitably" independent" of events.
In the movie "Contact" merely reading off the "1s" and "0s" from the
incoming stream of electro magnetic radiation does not make the
subsequent pattern "specified" nor does it make it independent of that
event. However, recognizing that the bit stream of 1126 bits
represented the first 100 prime numbers in order makes it both non ad
hoc and suitably independent of the event.
Let's consider the bacterial flagellum. Irreducibly complex
biological systems like the flagellum always satisfy independently
given functional requirements. In the case of the flagellum humans
developed outboard rotary motors long before they figured out that the
flagellum was such a machine. Our backgound knowledge of outboard
rotary motors makes recognition of the motor function of the flagellum
both non ad hoc and independent of the system as a whole. I suspect
no biologist would question whether functional components that arise
in biology are specified. The specified pattern would be the building
blocks for the flagellum and their well-matched connection giving
function.
Finally it can be seen that SPECIFIED PATTERNS cannot be given scalar
quantities to satisfy the atheist straw man. However once the
SPECIFIED PATTERN is observed and recognized its complexity may be
calcuable (by calculating its probability) and result in a scalar
quantity. For the flagellum the probability of such a SPECIFIED
PATTERN could be the probability of originating the building blocks
required for the functional system, multiplied times the probability
of locating them in one place once the building blocks are given,
multiplied times the probability of configuring them once the building
blocks are given and located in one place. Anyone think that
probability isn't smaller than Dembski's upper probability bound of
10^-150?
more to follow if time permits
>
>> Atheists in this forum
>
>Starting out with a falsehood is, unfortunately, Pagano's modus
>operandi. No matter how many times it's explained to him, he persists in
>equating evolution with atheism. And that's a good index of his
>intellectual honesty.
>
>> have consistently raised a straw ID theory
>> which asserts that CSI is
>> (1) a single measure,
>> (2) can be arrived at mechanically (that is, algorithmically), and
>> (3) produces a result which is used to conclusively deduce whether or
>> not an intelligent agent is required.
>>
>> All three are false.
>
>Then perhaps Pagano can tell us what is correct. Is CSI two measures?
>How are these measures performed, if not algorithmically? And if there
>is a result, which is unclear from what Pagano says, what good is that
>result?
>
>> And it's hard to tell if Ptiman is on to
>> something with his derived value which he also labels "CSI" or whether
>> he is lending aid and comfort to the false atheist ID straw man.
>
>I think Sean is showing great courage in presenting a real formula. This
>is much more than Pagano has ever done, and Brave Sir Tony is being a
>real shit for dumping on him without presenting any alternative or even
>a real critique. But I think Sean also lends aid and comfort to the
>false atheist ID straw man, as is Pagano. Sean shows that CSI is vacuous
>by presenting a formula that does nothing useful, while Pagano shows it
>by presenting nothing at all.
>
>[snip more tendentious nonsense]
Nic
Well, it has a stator, a rotor, some bearings, a curved piece, an
adaptor, and the flagellum itself. You have to have a stator and
rotor, and you need enough bearings to constrain the motion of the
rotor and relieve it of any axial or radial forces. Such bearings may
also have to act as glands. Really, it is only the adaptor where a
designer may have had any choice at all - the need for one could have
been designed out in the first place.
I don't think this flagellum is a good example for your purposes at
all. Something like the heart is much more engine-like.
> Sean Pitmanwww.DetectingDesign.com
Mark Nutter
> A SPECIFIED PATTERN is not a scalar quatitiy (nor is it likely to be
> reducible to one), but a pattern which is non ad hoc and independent
> of the event it describes.
Is the man in the moon a specified pattern? Why or why not?
m
T Pagano
>> Atheists in this forum
>
>Starting out with a falsehood is, unfortunately, Pagano's modus
>operandi. No matter how many times it's explained to him, he persists in
>equating evolution with atheism. And that's a good index of his
>intellectual honesty.
Harshman never reads too carefully which is his usual undoing. And
Harshman couldn't find any quotes from my 1800+ google history to
prove I equate "atheism" and evolutionism." These two doctrines share
at least one critical metaphysical doctrine and although one could
argue that one might depend upon the other they are quite different.
I do---more or less---equate the labels "atheist" and "evolutionist."
I don't really see a dimes worth of difference between the beliefs and
arguments of these two classes of people in this forum. Would
Harshman disagree?
more to follow if time permits.
Nic
Before outboard rotary motors, there were paddle steamers. So if
anything, the order in which humans invented things displays
inadequate design direction. It is not a point in favour of a
designer to say that evolution followed or didn't follow a similar
course to human invention.
> Our backgound knowledge of outboard
> rotary motors makes recognition of the motor function of the flagellum
> both non ad hoc and independent of the system as a whole.
Bacteria also move by a method known as gliding. Biologists were able
to recognise this without having seen anything similar in human
technology.
However, that may well be what you mean by non ad hoc.
> I suspect
> no biologist would question whether functional components that arise
> in biology are specified. The specified pattern would be the building
> blocks for the flagellum and their well-matched connection giving
> function.
>
> Finally it can be seen that SPECIFIED PATTERNS cannot be given scalar
> quantities to satisfy the atheist straw man. However once the
> SPECIFIED PATTERN is observed and recognized its complexity may be
> calcuable (by calculating its probability) and result in a scalar
> quantity. For the flagellum the probability of such a SPECIFIED
> PATTERN could be the probability of originating the building blocks
> required for the functional system, multiplied times the probability
> of locating them in one place once the building blocks are given,
> multiplied times the probability of configuring them once the building
> blocks are given and located in one place. Anyone think that
> probability isn't smaller than Dembski's upper probability bound of
> 10^-150?
It didn't happen all at once like that. You must know that?
John Harshman
OK, stop right there. So you disagree with those creationist who claim
that CSI is quantitative. Why not just come out and say so?
> In the movie "Contact" merely reading off the "1s" and "0s" from the
> incoming stream of electro magnetic radiation does not make the
> subsequent pattern "specified" nor does it make it independent of that
> event. However, recognizing that the bit stream of 1126 bits
> represented the first 100 prime numbers in order makes it both non ad
> hoc and suitably independent of the event.
> Let's consider the bacterial flagellum. Irreducibly complex
> biological systems like the flagellum always satisfy independently
> given functional requirements. In the case of the flagellum humans
> developed outboard rotary motors long before they figured out that the
> flagellum was such a machine. Our backgound knowledge of outboard
> rotary motors makes recognition of the motor function of the flagellum
> both non ad hoc and independent of the system as a whole. I suspect
> no biologist would question whether functional components that arise
> in biology are specified. The specified pattern would be the building
> blocks for the flagellum and their well-matched connection giving
> function.
I would, depending on just what you mean by that. Functional components
(if you mean "proteins") are specified in some ways -- they are
constrained by physics and chemistry, they are exactly coded by genes,
and they must have some structural features that make them functional.
They are unspecified in other ways: we can't say in advance what form an
adaptation may take, since there are many possible solutions to most
problems, and many possible sequences that will lead to similar solutions.
> Finally it can be seen that SPECIFIED PATTERNS cannot be given scalar
> quantities to satisfy the atheist straw man.
Who is this atheist straw man, exactly? Does he hang out with the
agnostic tin woodsman and the hindu cowardly lion? Sean Pitman has
claimed that CSI can be measured. Is he the atheist straw man?
> However once the
> SPECIFIED PATTERN is observed and recognized its complexity may be
> calcuable (by calculating its probability) and result in a scalar
> quantity. For the flagellum the probability of such a SPECIFIED
> PATTERN could be the probability of originating the building blocks
> required for the functional system, multiplied times the probability
> of locating them in one place once the building blocks are given,
> multiplied times the probability of configuring them once the building
> blocks are given and located in one place. Anyone think that
> probability isn't smaller than Dembski's upper probability bound of
> 10^-150?
I think it's a ridiculous calculation that has nothing to do with how
biology or evolution really works, so who cares?
> more to follow if time permits
Don't bother. You're just saying the same thing over and over.
Mark VandeWettering
> On Jul 15, 11:18 pm, Mark VandeWettering <wetter...@attbi.com> wrote:
>>
>> > Sean Pitman has been good enough to provide his mathematical formula for
>> > Complex Specified Information (CSI). Here is the Pitman CSI formula:
>>
>> > Given a string A and a reference string B
>>
>> > CSI = X^n - (X^hd)
>>
>> > Where:
>>
>> > n = the number of symbols in string A
>> > X = the number of symbols in the symbol set
>> > hd = the Hamming Distance between A and B (i.e., the number of symbols
>> > that do not match)
>>
>> >news:1184513686.4...@g12g2000prg.googlegroups.com
>>
>> > So.. how well does this work?
>>
>> Well, it is at least defined, but it doesn't have very many useful
>> properties. For one thing, it doesn't actually have a definition unless
>> you actually think that you have a string to compare something against.
>> Indeed, it can only be measured against a single, prespecified string.
>> I'm not certain what the utility of that might be. In fact, it seems
>> rather obviously pointless.
>
> The comparison is not so pointless if the pre-specified strings are
> all produced by algorithms that are much simpler in comparison with
> the strings themselves. For example, the formula for pi is very
> simple, yet is capable of producing a string of infinite size. Any
> match of a sequence in question to any such algorithmically produced
> string would be a significant indicator of a biased non-random origin
> of the analyzed string.
I have no idea what you mean, and I'm willing to bet that it is because
you don't have any idea what you mean either. Since I don't know what
you mean, I'll instead just try to fill in some of the misconceptions
that you have.
In no particular order:
1. You make it sound as if there is only one formula for pi. In fact,
there are a very large number of independent ways to determine the
value of pi, and each way is could lead to a different algorithm.
For instance:
a) Pi is the ratio of the circumference of a circle to its diameter.
We know that it must lie between the values of the permiter of
inscribed and circumscribed regular polygons, so we can bound
the value of pi to any specified degree of certainty by computing
perimeters. Archimedes used this to determine that pi must be
between 223/71 and 22/7.
b) It's the limit of the Gregory-Leibniz series.
Pi = 4/1 - 4/3 + 4/5 - 4/7...
This convertes rather slowly, and is based upon the Taylor series
for the arctan(1).
c) A similar formula was proposed by John Machin, which finds that
pi = 4 * atan(1/5) - atan(1/239), which is a much faster series.
d) If you had uniformly distributed random numbers in the unit square
ranging from [0:1], then pi/4 of them are within distance one of
the origin.
e) The Wallis Product
Product(((n+1)/n)^(-1^n-1), n=1 to infinity)
is pi/2.
f) The BBP formula for pi finds that
Sum(1/16^k * (4/(8k+1) - 2/(8k+4) - 1/(8k+5) -1/(8k+6))) = pi
This is a paricularly interesting formula, because it allows you
to compute individual hex digits of pi without computing the
digits before it. This was a very surprising result.
g) The integral of exp(-x^2) from -infinity to infinity is sqrt(pi).
h) and this goes on and on and on...
It's hard to see how any argument about "the" formula for pi relates
in anyway to information. And, of course your description of how to
compute CSI doesn't even try to do so, so I'm not sure what your objection
might be...
2. Pi is, itself, not "computable". Algorithms must, by their definitions,
actually halt, so while we can compute ever closer approximations to pi
using algorithms, it is hard to actually develop any intuition about
pi by doing so. For instance, it is not known whether the digits of pi
are normal. Again, since you didn't talk about approximations or any
such thing, it's hard to see how this is relevent to yuor discussion.
3. Concluding that any match of a "random" number to the digits of pi is a
significant indicator of a biased-non random origin is absurd. As an
obvious counterexample, here are 9 random decimal digits:
263243223
If you look at the 185,403,654 digits past the decimal point in pi, you
will find these precise digits. This is not particularly surprising.
(These decimal digits were generated from data produced by the HotBits
web service).
> Also, a match to absolutely any single string or pattern can be quite
> useful as far as identifying a biased non-random pattern. This
> feature becomes very useful in evaluating the symmetry of a finite
> object - like a highly symmetrical granite cube or snowflake. The
> snowflake, in particular, can show an almost infinite variety of
> patterns. Yet, a single snowflake often shows very high degrees of
> point/reflective/rotational symmetry of one half compared to the other
> half. This is a very strong indication of a biased non-random process
> in the formation of one-half of the snowflake vs. the other half.
Meaningless babble, and has nothing to do with CSI as you've formulated it
above.
> That information can be quite "useful". It isn't worthless or
> meaningless.
>
>> > Let's try a few examples.
>>
>> > Case 1:
>> > A = 0000000000
>> > B = 3141592653
>> > X = 10
>> > n = 10
>> > hd = 10
>>
>> > CSI = 10^10 - 10^10
>> > CSI = 0
>>
>> > 10 digits of zero have zero CSI with respect to the first 10 digits of pi.
>> > This seems reasonably intuitive.
>>
>> Intuitive only because we have a preconception about what probability
>> distributions are drawn from.
>
> Exactly . . .
You act as if that is good. It is not.
>> > Case 2:
>> > A = 3141592653
>> > B = 0000000000
>> > X = 10
>> > n = 10
>> > hd = 10
>>
>> > CSI = 10^10 - 10^10
>> > CSI = 0
>> > The first 10 digits of pi have zero CSI with respect to 10 repeating digits
>> > of zero. This seems less intuitive, but since B is supposed to be the
>> > specification, we'll let it pass.
>>
>> > Case 3:
>> > A = 3141592653
>> > B = 3141592653
>> > X = 10
>> > n = 10
>> > hd = 10
>>
>> > CSI = 10^10 - 10^0
>> > CSI = 9,999,999,999
>> > This is OK. The digits of pi look reasonably complex, and they match the
>> > specification, so we get the maximum value.
>>
>> > Case 4:
>> > A = 0000000000
>> > B = 0000000000
>> > X = 10
>> > n = 10
>> > hd = 0
>>
>> > CSI = 10^10 - 10^0
>> > CSI = 9,999,999,999
>> > Whoops! This is a surprisingly big number, and not at all intuitive, because
>> > a string of zeros is not intuitively complex. Now we see the first serious
>> > problem with the Pitman CSI formula: Any string, no matter what its
>> > contents, no matter what pattern is found therein, has the maximum possible
>> > CSI value with respect to itself. As long as A and B match, the CSI for A is
>> > X^n - 1, where X is the number of symbols in the alphabet, and n is the
>> > length of A.
>>
>> Indeed. This seems rather pointless.
>
> Rather, such a match would be a very good clue as to the non-biased
> origin or production of the match. You'd call that information
> "pointless"? Really?
Yes. In fact, the designation seems so obvious I wonder what your
objection could be.
> Just because something is simple or basic doesn't mean it is pointless
> or worthless.
It's not pointless because it is simple. It is pointless because it is
pointless.
Mark
>
>< snip >
>
> Sean Pitman
> www.DetectingDesign.com
>
>
_Arthur
What's the CSI of a granite cube ? Is it higher or lower than the CSI
of a peeble ? Higher or lower than the CSI of the Rosetta Stone ?
John Harshman
Thank you, then, for providing one below.
> These two doctrines share
> at least one critical metaphysical doctrine and although one could
> argue that one might depend upon the other they are quite different.
>
> I do---more or less---equate the labels "atheist" and "evolutionist."
> I don't really see a dimes worth of difference between the beliefs and
> arguments of these two classes of people in this forum.
Fascinating. He doesn't equate "atheism" and "evolutionism", but he
equates the beliefs of atheists and evolutionists. If the beliefs in
question are not atheism and evolutionism respectively, I don't know
what the words mean. Of course the arguments in this forum are similar
between theist and atheist evolutionists, because the existence of God
is not what's being argued about, but the question of evolution. Atheist
and theists who are evolutionists tend to agree on evolution, but
disagree on the existence of God. Simple, really.
> Would
> Harshman disagree?
Certainly. Your inability to see differences is not relevant. An atheist
is a person who doesn't believe God exists. An evolutionist is a person
who believes that species are descended from common ancestors. They are
not synonymous. One can easily be both or neither or one and not the
other. Unless you're going to try out the "no true Scotsman" ploy, there
are a great many people posting here who are both evolutionists and not
atheists. The reason you can't tell the difference is that you think
theism is synonymous with ignorant fundyism, which it is not. The
previous pope, for example, was both a theist and an evolutionist; not
sure about the current one.
Mark VandeWettering
What part of your insight above suggests the preference of yours?
>> I'm
>> not even sure of you definition of "biased"; it seems to be something
>> similar to "unlikely by chance".
>
> Biased means not random.
That's hardly clearing anything up.
>> Again, that's hardly interesting.
>
> I guess what's "interesting" is in the eye of the beholder . . .
>
>> We knew that.
>
> The question is, how did you know that? You had to have some basis
> for "knowing" that the origin was most likely biased . . .
>
>> Similar sequences do not arise by chance, and nested
>> hierachies of similar sequences do not arise by chance.
>
> How do you know? The fact is that they could have arisen by purely
> random "chance" as you put it. It is possible. It just isn't
> statistically likely.
You can't talk about the probability of anything unless you know something
about the probability distribution from which it is drawn. Your formulation
doesn't mention such things, so it is impossible for it to make any claims
about the likelihood of anything.
>> They demand
>> explanation. But we know the explanation: descent with modification and
>> branching. Rarely, the explanation for some similarity is not common
>> descent but common function, as with a few important residues in the
>> lysozymes of some folivores. Don't see where this gets you, though.
>
> There is a difference between detecting bias and detecting the actual
> origin of that bias. We agree on the bias part. What we don't agree
> on is the origin of the bias.
You've presented no mechanism, or even rationale, for your claims of
common design.
> The real problem comes with the fact that those systems with
> increasing "irreducible complexity", as a measure of the minimum
> structural threshold requirements needed to produce the functions in
> question, show less and less CSI (as I defined the term)
Defined in what sense? None of the formulas above mention anything
resembling a "strutural threshold requirement": where does that come from?
> compared to
> anything else in the gene pool - as a fraction of the maximum possible
> CSI. It is the significant decrease in CSI relative to each increase
> in IC that gets the ToE into real trouble. Increasing the CSI at
> higher levels of minimums structural threshold requirements is a real
> challenge for the proposed mechanism of random mutation and function-
> based selection that rapidly reaches the point of incredulity.
Babble.
Mark
>
> Sean Pitman
> www.DetectingDesign.com
>
Mark VandeWettering
>> In article <139lpvrqfjvu...@news.supernews.com>,
>> "R. Baldwin" <res0k...@nozirevBACKWARDS.net> writes:
>>
>> > Given all this, it is rather difficult to see how the Pitman CSI
>> > formula will tell us anything useful about a string. It is highly
>> > unlikely to make a design predictor. But, kudos to Sean for at least
>> > giving it a try. That's more than most ID-ists have done.
>>
>> The whole idea is bunkum, whether you take Sean's version or not.
>>
>> Suppose you set up a random number generator -- a real one based on
>> thermal noise or radioactive decay rather than an algorithmic pseudo-
>> RNG -- to generate a string of a million binary digits and write them
>> directly into the memory of your computer. The resulting string would
>> be genuinely random, and thus not have any Dembski-style CSI because
>> it would not conform to "an independently given pattern". Right?
>
> It would be unlikely to conform to an independently derived pattern,
> but it would not be impossible. That's the problem. The "randomness"
> of a finite or infinite string can never be proven.
>
>> Now you make a copy of the string in your computer's memory, and you
>> find that the copy has incredible amounts of CSI, because it's a
>> million digits long and it conforms exactly to an independently given
>> pattern, namely the first string.
>
> All you have done here is shown that the copied string was copied in a
> very non-random way. My version of CSI is all about detecting likely
> bias in the production of a string. Two identical copies of a million-
> character string would strongly support the hypothesis of biased
> production.
Since you think that "biased" means simply "non-random", in what way
does this help your preference for design over descent?
Mark
>
>< snip rest >
>
> Sean Pitman
> www.DetectingDesign.com
>
>
T Pagano
>> have consistently raised a straw ID theory
>> which asserts that CSI is
>> (1) a single measure,
>> (2) can be arrived at mechanically (that is, algorithmically), and
>> (3) produces a result which is used to conclusively deduce whether or
>> not an intelligent agent is required.
>>
>> All three are false.
>
>Then perhaps Pagano can tell us what is correct. Is CSI two measures?
>How are these measures performed, if not algorithmically? And if there
>is a result, which is unclear from what Pagano says, what good is that
>result?
>
>> And it's hard to tell if Ptiman is on to
>> something with his derived value which he also labels "CSI" or whether
>> he is lending aid and comfort to the false atheist ID straw man.
>I think Sean is showing great courage in presenting a real formula.
Oh Pleeeze.....It doesn't take any courage to post in this forum. It
does take courage to stake your personal reputation and career by
publishing works openly for the press and the scientific community to
attack. I'd say Behe who is on the teaching staff at Lehigh
University and is sticking his neck out fits that bill. Pitman risks
nothing.
Dembski's theory is generalized to apply universally. At best
Pitman's formula applies to the specialized case of numerical
sequences. And he has yet to offer the rigor that Dembski has done
with his work published in a Cambridge University Press peer reviewed
work.
In effect Pitman is playing to an intentionally false strawman theory
wihich is more or less a wasted effort as far as I'm concerned. He is
attempting to beat the devil at his own game; not a wise tactic on
even a good day.
> This
>is much more than Pagano has ever done, and Brave Sir Tony is being a
>real shit for dumping on him without presenting any alternative or even
>a real critique. But I think Sean also lends aid and comfort to the
>false atheist ID straw man, as is Pagano. Sean shows that CSI is vacuous
>by presenting a formula that does nothing useful, while Pagano shows it
>by presenting nothing at all.
I calls 'em the way I sees 'em. My job here is not to be brave or
ingenous but to show the atheists for what they are. And I've left
Harshman running with his tail between his legs more than once. And
little 'ol me against the intelligencia elitist.
Regards,
T Pagano
hersheyhv
> On Jul 16, 11:58 pm, Seanpit <seanpitnos...@naturalselection.
>
>
>
> 0catch.com> wrote:
> > On Jul 16, 1:07 pm, Vend <ven...@virgilio.it> wrote:
>
> > > > > > You haven't provided enough information to calculate CSI. You've only
> > > > > > given the size and specific characters per position. You haven't given
> > > > > > any other "target" or reference string for comparison. The
> > > > > > calculation of CSI, like symmetry, requires a comparison to be made
> > > > > > between strings Q and Z.
>
> > > > > > CSI = X^n - (X^hd)
>
> > > > > > Where:
> > > > > > n = the size of the sequence
> > > > > > hd = the Hamming Distance between strings Q and Z
>
> > > Say I have this sequence: 11121123344432992877777777551111121 and I
> > > want to know whether it's designed. How should I choose the reference
> > > string? How should I interpet the resulting CSI value?
>
> > As I said before, CSI, by itself, does not tell you if the string's
> > pattern was or was not designed. CSI only tells you if a pattern was
> > the likely result of some sort of bias. Once the bias is detected, one
> > can then try to hypothesize about the origin of the bias. However,
> > this type of hypothesis requires additional information about the
> > material which carries the pattern as well as some history with the
> > material in question as it relates to various forces of nature -
> > intelligent and non-intelligent.
>
> It seems to me that your CSI isn't capable of detecting bias in a
> sequence.
> It's just a way of comparing two sequences.
And CSI is not a feature of *a* sequence at all. It is a variable
whose size can change (and change dramatically) whenever you change
one of the two sequences you are looking at (whether the teleologic?
target sequence or the reference sequence).
[snip]
hersheyhv
0catch.com> wrote:
> On Jul 16, 3:18 pm, Vend <ven...@virgilio.it> wrote:
>
> > > > Say I have this sequence: 11121123344432992877777777551111121 and I
> > > > want to know whether it's designed. How should I choose the reference
> > > > string? How should I interpet the resulting CSI value?
>
> > > As I said before, CSI, by itself, does not tell you if the string's
> > > pattern was or was not designed. CSI only tells you if a pattern was
> > > the likely result of some sort of bias. Once the bias is detected, one
> > > can then try to hypothesize about the origin of the bias. However,
> > > this type of hypothesis requires additional information about the
> > > material which carries the pattern as well as some history with the
> > > material in question as it relates to various forces of nature -
> > > intelligent and non-intelligent.
>
> > It seems to me that your CSI isn't capable of detecting bias in a
> > sequence. It's just a way of comparing two sequences.
Specifically, a high value of what you call CSI is an indication of
sequence similarity. The greater the similarity of two sequences, the
higher the value of what you call CSI.
> Yes - a comparison where if the two sequences were produced
> independently by truly random processes,
And, of course, the easiest explanation for two sequences having
similarity is that they have a common origin or source. What *you*
have to demonstrate is that two identical or similar sequences cannot
have a common source and were, in fact, produced *independently by a
truly random process*. Evolution assumes that sequence similarity is
evidence of common origin, not independent truly random process.
> they would not have maximum
> CSI. Rather, they would have the maximum CSI divided by the number of
> possible characters per position. Therefore, extremely low and
> extremely high CSI values are strong indicators of bias.
>
> > > > If I did my calculations well, this means that with your definition of
> > > > CSI as the length increases it becomes more and more difficult to
> > > > distinguish two completely unrelated random sequences from two
> > > > completely correlated (one the copy of the other) ones.
> > > > This makes me wonder what your CSI is good for.
>
> > > As the length of a sequence increases, the CSI for two independently
> > > produced random strings will approach: (1/X) (X^n - X^hd). Compare
> > > this to the CSI for two completely correlated sequences which equals
> > > X^n.
>
> > Sorry but I don't understand where you get that (1/X) from.
> > Moreover, if we are talking about random strings, then hd is a random
> > variable (binomially distributed) thus CSI is also a random variable.
> > I was considering expectations.
>
> Actually not. The Hamming Distance of two randomly produced strings
> will tend toward a match between the strings that is 1/X or the
> maximum possible distance divided by the number of possible characters
> per position.
So what you are presenting here is the idea that *some* mechanism
involves a random walk in single aa steps starting with two random
sequences that are the same size. Alhough, of course, in *real*
proteins, the match will be greater than 1/X because some aa's are
much more common in proteins than others. But what does such a model
of random walks in total sequence space for a given sized protein have
to do with reality? And wouldn't that be what you rail against as my
"strawman" version of your model?
> For example, if you flip a coin 10 times and then do it again and
> compare the two sequences, how many of the positions will match, on
> average? The answer is 5 or the maximum Hamming Distance of 10
> divided by the number of possible characters per position (i.e., 2),
> which equals 5 matches. If the experiment produced 10 matches or 0
> matches, that would be out of line with a prediction of truly random
> flips of the coin. Over time, if the flips are truly random, the
> number of matches will fit this formula to a greater and greater
> degree.
>
> Sean Pitmanwww.DetectingDesign.com
Seanpit
> > > > There would be no beneficial mutations unless mutations were in fact
> > > > able to produce novel beneficial molecules or molecular systems.
>
> > > A beneficial mutation is a mutation that increases the fitness of the
> > > oranism bearing it.
>
> > A beneficial mutation is a mutation that produces a beneficial
> > molecule or physical system that wasn't there before.
>
> Actually, it doesn't have to be so.
> A beneficial mutation can also
> change some structure that existed
> previously or even remove a structure.
A functional change in a structure that previously existed is the same
thing as creating a new structural with a new functional element.
Also, a benefit that arises from removal of a structure is still the
creation of a new physical system.
> > > What is a beneficial molecule or molecular system?
>
> > Any system added to an organism that provides a survival/reproductive
> > advantage - regardless of how such a system was obtained - i.e., via
> > mutation, horizontal or vertical transfer, insertion by deliberate
> > human or non-human design, etc.
>
> Thus anything obtained through a beneficial mutation.
Vertical and horizontal transfer, genetic recombination, etc, don't
necessarily involve mutations.
Sean Pitman
www.DetectingDesign.com
hersheyhv
Biased (not random) is what one observes under the assumption that
sequences evolve by descent with modification rather than by each
protein evolving *independently* by a random walk from a random
sequence in total sequence space. Moreover, descent with modification
should give a particular type of pattern of biased sequences. And
*structure* even more so. What evolutionary mechanism involves each
and every new protein arising by a random walk from some random non-
existant proteins?
> > Again, that's hardly interesting.
>
> I guess what's "interesting" is in the eye of the beholder . . .
>
> > We knew that.
>
> The question is, how did you know that? You had to have some basis
> for "knowing" that the origin was most likely biased . . .
>
> > Similar sequences do not arise by chance, and nested
> > hierachies of similar sequences do not arise by chance.
>
> How do you know? The fact is that they could have arisen by purely
> random "chance" as you put it. It is possible. It just isn't
> statistically likely.
Yep. And that is why we get the nested pattern of descent with
modification so consistently.
> > They demand
> > explanation. But we know the explanation: descent with modification and
> > branching. Rarely, the explanation for some similarity is not common
> > descent but common function, as with a few important residues in the
> > lysozymes of some folivores. Don't see where this gets you, though.
>
> There is a difference between detecting bias and detecting the actual
> origin of that bias. We agree on the bias part. What we don't agree
> on is the origin of the bias.
So why did your designer *independently* for each protein make all
these neutral differences (in a protein with a particular function)
that only make sense as a pattern of descent with modification?
> The real problem comes with the fact that those systems with
> increasing "irreducible complexity", as a measure of the minimum
> structural threshold requirements needed to produce the functions in
> question, show less and less CSI (as I defined the term) compared to
> anything else in the gene pool - as a fraction of the maximum possible
> CSI. It is the significant decrease in CSI relative to each increase
> in IC that gets the ToE into real trouble. Increasing the CSI at
> higher levels of minimums structural threshold requirements is a real
> challenge for the proposed mechanism of random mutation and function-
> based selection that rapidly reaches the point of incredulity.
Only if one stupidly assumes a random *functionless* walk single
nucleotide by single nucleotide from a sequence position that is very
far away for a protein of that size. Of course, given enough time
(about 100 million years to be exact) any truly neutral site (and
there are many) in a protein is likely to have undergone neutral drift
(with sites that allow less variance in aa's being slower to change).
For really distant homologous proteins without a lot of 'crucial'
aa's, only the structure and function remains.
Of course, if you actually had *evidence* that life on earth has only
existed for 6000 years....
> Sean Pitmanwww.DetectingDesign.com
hersheyhv
> On Jul 16, 3:40 pm, John Harshman <jharshman.diespam...@pacbell.net>
> wrote:
>
> < snip >
>
> > > As it turns out, there is a very clear pattern of limitation when it
> > > comes to the proposed mechanism of random mutation and function-based
> > > (i.e., natural) selection that falls well shy of the known and
> > > theoretical limits of what intelligent deliberate forces of nature can
> > > achieve with the same material.
>
> > This is your claim. I have very little interest in discussing it with
> > you. If you want to argue, let's argue about common descent, which is my
> > field.
>
> And, it is your assumption that you can adequately hypothesize common
> descent over extended periods of time by simply looking at the pattern
> in question - without any additional knowledge as to how the material
> you are considering actually relates to various forces of nature over
> time.
>
> Your problem is that you assume common descent over common design
Yes indeed. You would have to claim that the "common design" appeared
to have been done with the explicit purpose of fooling us into
thinking common descent if you were to claim "common design".
Remember that most of the evidence for the nested *pattern* of
sequence relationships presents largely a consistent pattern
regardless of which protein and function (or morphological feature)
you look at. And the changes we are looking at in sequence data are
*specifically* changes that have little or no functional effect, so
they are basically unnecessary changes from a *design* or *functional*
perspective (and there is clear experimental evidence that this is the
case). They are the consequence of the nature of neutral drift over
long time frames.
So if the differences are not due to *functional* *design*, perhaps
they are due to *decorative* design or artistic license. But then,
why create basically the *same* pattern for all the different proteins
examined? The only reasonable explanation, if you insist on
intentional or intelligent *design* for the observed pattern is that
the designer wanted to fool us into thinking the design was due to
common descent. Bad designer! Bad, bad designer! What a nasty
practical joke.
> even
> though you do not claim any basis in mechanism for your assumption.
> You simply know that "common descent" is the answer based pretty much
> on your philosophical position that "No intelligent designer would
> have done it that way".
Well, no well-intentioned sane intelligent designer would have done it
that way. A trickster designer might have.
> How do you know this? Without reference to a
> viable mechanism for your common descent when it comes to its
> functional elements beyond extremely low levels, what real scientific
> basis beyond some philosophical assumptions do you have?
More bullshit that assumes that the size of the end product determines
the number of mutational steps needed to produce that end product. I
am still waiting for you to produce *that* particular algorithm.
>
> Sean Pitmanwww.DetectingDesign.com
Gary Bohn
news:apagano-qevm93df8mtc0...@4ax.com:
> On Sun, 15 Jul 2007 20:37:38 -0700, "R. Baldwin"
> <res0...@nozirevBACKWARDS.net> wrote:
>
>>Sean Pitman has been good enough to provide his mathematical formula
>>for Complex Specified Information (CSI). Here is the Pitman CSI
>>formula:
>>
>>Given a string A and a reference string B
>>
>> CSI = X^n - (X^hd)
>>
>>Where:
>>
>> n = the number of symbols in string A
>> X = the number of symbols in the symbol set
>> hd = the Hamming Distance between A and B (i.e., the number of
>> symbols
>>that do not match)
>>
>
> CSI is an acronym for Complex Specified Information which is made up
> of two separate measurable attributes----at least as far as Dembski's
> theory is concerned. Both complexity and specified information have
> defined criteria which must individually be met before any inference
> can be made and the inference is NOT deductive.
>
> Atheists in this forum have consistently raised a straw ID theory
> which asserts that CSI is
> (1) a single measure,
> (2) can be arrived at mechanically (that is, algorithmically), and
> (3) produces a result which is used to conclusively deduce whether or
> not an intelligent agent is required.
>
> All three are false. And it's hard to tell if Ptiman is on to
> something with his derived value which he also labels "CSI" or whether
> he is lending aid and comfort to the false atheist ID straw man.
>
> Probability theory alone doesn't provide any tools for determining
> degree of randomness of a bit string. Chaitin, Kolmogorov, and
> Solomonoff offered a theory whereby they argued that the
> compressibility of some bit string given some reference class was such
> a measure. The less compressible a string then the more random it
> was. In other words, the collection of algorithmically compressible
> (and therefore non random) sequences has small probability among the
> totality of sequences, so that observing such a sequence is reason to
> look for explanations other than chance. I'm guessing that Pitman is
> attempting to use this outcome as a means of deriving some single
> measure. Dembski does no such thing.
>
> Lastly, creationists and other like-minded people should avoid
> engaging ID challenges as currently formulated. They are largely
> sterile (that is, fruitless) because they are divorced from any real
> world problem situation. Atheists can't help themselves in this
> regard. Abiogenesis and neoDarwinism can only flourish when divorced
> from real world biology and biochemistry. Forcing the atheist to
> produce a real world problem will undoubtedly end the challenges
> completely.
>
> Regards,
> T Pagano
>
>
Real world problem like what: a recreation of God in the lab?
--
Gary Bohn
Science rationally modifies a theory to fit evidence, creationism
emotionally modifies evidence to fit a specific interpretation of the
bible.
“Colorless green ideas sleep furiously.” — Noam Chomsky, 1957
Seanpit
wrote:
> >>If all you're saying is that homologous sequences resemble each other
> >>more than expected by chance, that's nice but hardly interesting.
>
> > It is actually very interesting. It is the basis for your notion of
> > common ancestry.
>
> It's a useless statement of some small fraction of the basis of my
> notion of common ancestry.
Yet, without sequence homology in varying degrees you'd have no basis
at all.
> > I also use this evidence as the basis for my belief
> > in the common origin of such biosystems. The only thing we differ on
> > is the nature of the origin; not the fact that that origin was common.
>
> And this is your major obfuscation here, intentional or not. You are
> confusing the origin of variation with the origin of the nested
> hierarchy that contains that variation. This notion of "common origin"
> is not useful, largely because of that conflation.
You believe that the origin of the variation was random mutation and
natural selection. This mechanism, acting over time, produced the
nested hierarchical pattern. That is in fact your position. No reason
to hide it.
While the mechanism of random mutation and natural selection most
certainly can produce a nested hierarchical pattern, that isn't the
only feature seen in living things. The pattern itself, by itself,
isn't a problem. The problem comes with trying to explain the
functional aspects of the systems within the pattern. Lower-level
systems aren't a problem. The systems beyond these very low levels
quickly do become a problem for your proposed mechanism.
Oh, I know, you don't care about the "mechanism". You argue that
common descent from a common evolutionary ancestor is evident without
any need for knowledge about the mechanism. You assume that a nested
pattern would only be the result of common descent. Yet, this
assumption is not based on evaluation of the potential or limitations
of deliberate vs. non-deliberate processes. A nested pattern could
easily be the result of common design. It is well within the range of
possibility. You reject this possibility, not on the basis of
scientific demonstration of the limits of what intelligence can
achieve, but upon your notion that an intelligent agent simply
wouldn't have done it that way. That, my friend, isn't a conclusion
based in science, but in subjective conjecture - a feeling of how you
think an intelligent designer would or should create. That's very
very weak "science".
> >>I'm
> >>not even sure of you definition of "biased"; it seems to be something
> >>similar to "unlikely by chance".
>
> > Biased means not random.
>
> Not as I understand it. The roll of a loaded die is biased, but it's
> still random. The random distribution happens to differ from that of a
> fair die.
Random actually means non-predictable with better than even odds of
success given the assumption that certain numbers are likely to fall
in a certain distribution pattern. Anything that predictably deviates
from this assumption is the result of some sort of non-random bias.
There are all kinds of potential biases. A biased distribution,
however subtle, is more predictable than a non-biased distribution. A
loaded die could be loaded in any number of different ways. It could
be loaded with a computer chip that biases the next roll based on the
outcome of the previous roll in any number of ways. It could be
biased to produce a 2 after a 6 or even a specific series like pi or
the square root of 2.
> >>Again, that's hardly interesting.
>
> > I guess what's "interesting" is in the eye of the beholder . . .
> >
> >>We knew that.
>
> > The question is, how did you know that? You had to have some basis
> > for "knowing" that the origin was most likely biased . . .
>
> Your use of language is bizarre. The basis is simple enough, and it's
> simple probability, under any distribution you may prefer. No need to go
> into anything like CSI.
Detecting bias isn't always as simple as you make it appear. If you
didn't know about pi, the pattern produced by this algorithm would
indeed seem quite "random" to you since its distribution follows that
produced by other seemingly random sources of sequences - i.e.,
radioactive decay, etc. In fact, this little subtly is the reason by
no sequence taken by itself, not even an infinite sequence, can ever
be proven to be the result of a random process - even though it can be
proven that the vast majority of potential sequences are in fact
random or have maximum Kolmogorov Complexity.
> >>Similar sequences do not arise by chance, and nested
> >>hierachies of similar sequences do not arise by chance.
>
> > How do you know? The fact is that they could have arisen by purely
> > random "chance" as you put it. It is possible. It just isn't
> > statistically likely.
>
> It's unlikely enough that we can afford to ignore the possibility. On
> this we are all agreed, so I can't imagine why you belabor the point.
The question is at what point can you be confident in your hypothesis
of a non-random origin? Some cases are easy because the distribution
pattern is so obviously biased. However, other cases, like pi, are
not so obvious and may in fact be extremely difficult to discover even
though they were in fact produced by a very simple algorithm.
> >>They demand
> >>explanation. But we know the explanation: descent with modification and
> >>branching. Rarely, the explanation for some similarity is not common
> >>descent but common function, as with a few important residues in the
> >>lysozymes of some folivores. Don't see where this gets you, though.
>
> > There is a difference between detecting bias and detecting the actual
> > origin of that bias. We agree on the bias part. What we don't agree
> > on is the origin of the bias.
>
> > The real problem comes with the fact that those systems with
> > increasing "irreducible complexity", as a measure of the minimum
> > structural threshold requirements needed to produce the functions in
> > question, show less and less CSI (as I defined the term) compared to
> > anything else in the gene pool - as a fraction of the maximum possible
> > CSI. It is the significant decrease in CSI relative to each increase
> > in IC that gets the ToE into real trouble. Increasing the CSI at
> > higher levels of minimums structural threshold requirements is a real
> > challenge for the proposed mechanism of random mutation and function-
> > based selection that rapidly reaches the point of incredulity.
>
> Once again you conflate common descent, which is what I'm interested in,
> with the mechanism of change, which is what you are apparently
> interested in. I have very little interest in arguing about whether
> mutation and selection are sufficient to account for adaptation. I'm
> talking about the cause of nested hierarchy, which has nothing to do
> with selection except as selection is one underlying cause of fixation.
Again, the cause of a nested pattern is very much related to the
proposed mechanism. Different mechanisms of change may or may not
predictably produce such a pattern. Also, different mechanisms may
produce such a pattern besides common descent.
Sean Pitman
www.DetectingDesign.com
Seanpit
wrote:
> It is pointless, to anyone with a fair knowledge of math. The metric itself
> is so badly flawed that it was not necessary to go into what makes one
> reference string better than any other.
The selection of the reference string is like a selection of UTM. As
the string gets longer and longer, the initial selection matters less
and less until it doesn't really matter at all.
Your supposed "problems" with my definition of CSI seem to me to be
based largely on a misapprehension of what a definition of CSI should
be able to accomplish. For example, you seem to think that CSI should
be able to define some intuitive notion of what a random number should
look like. Yet, this is impossible without some sort of reference.
KC is based on a reference and so is my definition of CSI.
A production of a truly random string will produce maximum KC
regardless of the reference UTM as the string grows longer and
longer. And, this same string will produce maximum CSI * (1/X). In
this way, very low or very high CSI values (like 0 or the maximum CSI
value) are strong indicators of non-random bias.
As far as I'm concerned, CSI is only about detecting bias given a
specific or even a large number of reference algorithms. In my view
this is in fact what SETI scientists propose with their search for a
certain type of signal containing certain types of patterns. I mean
really, SETI scientists themselves say that if they found signals
tagged with the first 100 digits of pi, prime numbers, or of the
Fibonacci series repeated over and over that they would themselves
conclude ID. How so? Because, obviously, such a match would have
extremely high CSI according to my definition between a mathematical
algorithm of known non-random production and a radiosignal - clearly
demonstrating the non-random nature of the radiosignal in a manner
that is beyond anything that any known non-deliberate process even
comes close to being able to achieve with the medium of radiosignals.
Sean Pitman
www.DetectingDesign.com
Victor Eijkhout
> basic formula (i.e., the circumference of a circle
> divided by its diameter).
That "formula" is not very useful. It requires an axiomatization of
geometry so that you can define a euclidean plane. You need to rule out
hyperbolic geometries and such.
> But, if you don't like pi, then use any
> other simple algorithm - like the square root of 2.
So what do you think sqrt(2) is? It's the limit of a sequence. Again, an
infinite construction. That "V with a side arm" is just a notation.
Victor "not sure why I bother; it's clear how much you understand of
math"
--
Victor Eijkhout -- eijkhout at tacc utexas edu
rupert....@gmail.com
0catch.com> wrote:
> On Jul 16, 3:18 pm, Vend <ven...@virgilio.it> wrote:
>
> > > > Say I have this sequence: 11121123344432992877777777551111121 and I
> > > > want to know whether it's designed. How should I choose the reference
> > > > string? How should I interpet the resulting CSI value?
>
> > > As I said before, CSI, by itself, does not tell you if the string's
> > > pattern was or was not designed. CSI only tells you if a pattern was
> > > the likely result of some sort of bias. Once the bias is detected, one
> > > can then try to hypothesize about the origin of the bias. However,
> > > this type of hypothesis requires additional information about the
> > > material which carries the pattern as well as some history with the
> > > material in question as it relates to various forces of nature -
> > > intelligent and non-intelligent.
>
> > It seems to me that your CSI isn't capable of detecting bias in a
> > sequence. It's just a way of comparing two sequences.
>
> CSI. Rather, they would have the maximum CSI divided by the number of
> possible characters per position. Therefore, extremely low and
> extremely high CSI values are strong indicators of bias.
>
> > > > If I did my calculations well, this means that with your definition of
> > > > CSI as the length increases it becomes more and more difficult to
> > > > distinguish two completely unrelated random sequences from two
> > > > completely correlated (one the copy of the other) ones.
> > > > This makes me wonder what your CSI is good for.
>
> > > As the length of a sequence increases, the CSI for two independently
> > > produced random strings will approach: (1/X) (X^n - X^hd). Compare
> > > this to the CSI for two completely correlated sequences which equals
> > > X^n.
>
> > Sorry but I don't understand where you get that (1/X) from.
> > Moreover, if we are talking about random strings, then hd is a random
> > variable (binomially distributed) thus CSI is also a random variable.
> > I was considering expectations.
>
> Actually not. The Hamming Distance of two randomly produced strings
> will tend toward a match between the strings that is 1/X or the
> maximum possible distance divided by the number of possible characters
> per position.
>
> compare the two sequences, how many of the positions will match, on
> average? The answer is 5 or the maximum Hamming Distance of 10
> divided by the number of possible characters per position (i.e., 2),
> which equals 5 matches.
That's only true for a fair coin. If your coin comes up heads nine
times as often as tails (which is still random), you would expect 8
matches, even though there are still only two results.
P(HH)= 81/100; P(TT)=1/100; P(match)= 82/100; Expected matches = 10 *
P(match) = 8.2
The expected number of matches if you generate your sequence with 2D6
is 1.1 (1460/1296), not 0.9 (10/11)
John Harshman
> On Jul 16, 4:31 pm, John Harshman <jharshman.diespam...@pacbell.net>
> wrote:
>
>
>>>>If all you're saying is that homologous sequences resemble each other
>>>>more than expected by chance, that's nice but hardly interesting.
>>
>>>It is actually very interesting. It is the basis for your notion of
>>>common ancestry.
>>
>>It's a useless statement of some small fraction of the basis of my
>>notion of common ancestry.
>
> Yet, without sequence homology in varying degrees you'd have no basis
> at all.
Presumably you mean "sequence similarity". Sequence homology is an
inference from similarity.
>>>I also use this evidence as the basis for my belief
>>>in the common origin of such biosystems. The only thing we differ on
>>>is the nature of the origin; not the fact that that origin was common.
>>
>>And this is your major obfuscation here, intentional or not. You are
>>confusing the origin of variation with the origin of the nested
>>hierarchy that contains that variation. This notion of "common origin"
>>is not useful, largely because of that conflation.
>
> You believe that the origin of the variation was random mutation and
> natural selection. This mechanism, acting over time, produced the
> nested hierarchical pattern. That is in fact your position. No reason
> to hide it.
This is not true. The hierarchical pattern results from fixation of
changes on a branching tree. The processes that caused the fixations is
irrelevant to the existence and explanation of the hierarchy.
> While the mechanism of random mutation and natural selection most
> certainly can produce a nested hierarchical pattern, that isn't the
> only feature seen in living things. The pattern itself, by itself,
> isn't a problem. The problem comes with trying to explain the
> functional aspects of the systems within the pattern. Lower-level
> systems aren't a problem. The systems beyond these very low levels
> quickly do become a problem for your proposed mechanism.
>
> Oh, I know, you don't care about the "mechanism". You argue that
> common descent from a common evolutionary ancestor is evident without
> any need for knowledge about the mechanism. You assume that a nested
> pattern would only be the result of common descent. Yet, this
> assumption is not based on evaluation of the potential or limitations
> of deliberate vs. non-deliberate processes. A nested pattern could
> easily be the result of common design.
How? Since any pattern whatsoever, including a complete lack of pattern,
could be the result of "common design", your statement says nothing.
Present a reason why we would *expect* a nested hierarchy from common
design or abandon it as an explanation for the pattern, because an
explanation does indeed require reasons, more than "well it could have".
> It is well within the range of
> possibility. You reject this possibility, not on the basis of
> scientific demonstration of the limits of what intelligence can
> achieve, but upon your notion that an intelligent agent simply
> wouldn't have done it that way.
Again, I reject it not for that reason, but because we have no reason to
expect any particular pattern from "common design". An explanation that
explains anything you could conceivably have observed is no explanation
at all. You need to deal with this.
> That, my friend, isn't a conclusion
> based in science, but in subjective conjecture - a feeling of how you
> think an intelligent designer would or should create. That's very
> very weak "science".
I suppose it would be, if indeed that was my reasoning. Note that I'm
not saying here than an intelligent designer didn't create the diversity
of life. I'm only saying that if he did, he used common descent as part
of the process. If we can't tell this, then scientific investigation of
the history of life is impossible. Is that your position?
>>>>I'm
>>>>not even sure of you definition of "biased"; it seems to be something
>>>>similar to "unlikely by chance".
>>
>>>Biased means not random.
>>
>>Not as I understand it. The roll of a loaded die is biased, but it's
>>still random. The random distribution happens to differ from that of a
>>fair die.
>
> Random actually means non-predictable with better than even odds of
> success given the assumption that certain numbers are likely to fall
> in a certain distribution pattern.
That is not the definition of "random" that I am familiar with.
I see no point to this digression.
>>>>Similar sequences do not arise by chance, and nested
>>>>hierachies of similar sequences do not arise by chance.
>>
>>>How do you know? The fact is that they could have arisen by purely
>>>random "chance" as you put it. It is possible. It just isn't
>>>statistically likely.
>>
>>It's unlikely enough that we can afford to ignore the possibility. On
>>this we are all agreed, so I can't imagine why you belabor the point.
>
> The question is at what point can you be confident in your hypothesis
> of a non-random origin? Some cases are easy because the distribution
> pattern is so obviously biased. However, other cases, like pi, are
> not so obvious and may in fact be extremely difficult to discover even
> though they were in fact produced by a very simple algorithm.
Conventionally, P<.05 is taken as the threshhold of confidence. And
there are tests that allow one to reject particular models of randomness.
You are misinformed on both counts. But feel free to elaborate, with
examples.
R. Baldwin
news:1184616155.2...@m37g2000prh.googlegroups.com...
> On Jul 15, 8:37 pm, "R. Baldwin" <res0k...@nozirevBACKWARDS.net>
> wrote:
>> Sean Pitman has been good enough to provide his mathematical formula for
>> Complex Specified Information (CSI). Here is the Pitman CSI formula:
>>
>> Given a string A and a reference string B
>>
>> CSI = X^n - (X^hd)
>>
>> Where:
>>
>> n = the number of symbols in string A
>> X = the number of symbols in the symbol set
>> hd = the Hamming Distance between A and B (i.e., the number of
>> symbols
>> that do not match)
>>
>> 2007:news:1184513686.4...@g12g2000prg.googlegroups.com
>>
>> So.. how well does this work?
>>
>>
>> Case 1:
>> A = 0000000000
>> B = 3141592653
>> X = 10
>> n = 10
>> hd = 10
>>
>> CSI = 10^10 - 10^10
>> CSI = 0
>> 10 digits of zero have zero CSI with respect to the first 10 digits of
>> pi.
>> This seems reasonably intuitive.
>>
> each other even though each is the product of independent processes or
> origins - like a mathematical algorithm vs. a radiosignal from outer
> space. A radiosignal that matches a mathematical algorithm perfectly,
> over time, becomes less and less likely to be the result of a truly
> random process - regardless of the original algorithm(s) chosen for
> comparison purposes.
Since all finite strings can be the product of mathematical algorithms, all
conceivable radio signals match the output of some algorithm or other. Where
does that leave you?
>
>> Case 5:
>> A = 010110111
>> B = 101001000
>> X = 2
>> n = 10
>> hd = 10
>>
>> CSI = 2^10 - 2^10
>> CSI = 0
>>
>> Here's another problem. When A is the exact complement of B, it has zero
>> CSI., even though a trivial computation changes A into B.
>
> It is only a trivial computation given the proper algorithm - which
> itself is a "match" and therefore has maximum CSI.
While it is true that you need the proper algorithm, complement is a
computation that occurs readily and naturally in LTI systems all over the
place. If you wanted an algorithm for deciding whether pattern A stemmed
from pattern B, it would be a desirable trait that it also test for the
complement of B.
>
>> Case 6:
>> A = 1245340028
>> B = 8200435421
>> X = 10
>> n = 10
>> hd = 8
>>
>> CSI = 10^10 - 10^8
>> CSI = 9,900,000,000
>>
>> Two new problems: First, when A is B in reverse order, whether CSI turns
>> out
>> high or low is hit or miss. This is not a very intuitive result, having
>> CSI
>> depend so drastically on whether you got the reference pattern forwards
>> or
>> backwards. Second, the CSI for a string with a large Hamming Distance is
>> not
>> much smaller than the CSI for a string that matches the reference pattern
>> exactly.
>
> Again, this is an algorithm issue. Pick the correct algorithm, and
> the match becomes clear. Don't pick the correct algorithm and the
> match is not so clear. The apparent match is only clear to you
> because you already have access to the algorithm, in your mind, that
> allows you to quickly recognize that A is simply the reverse of B.
The same argument I made for complement applies here. The computation
argument is more appropriately applied to whether you picked a good
reference to begin with.
>
>> Looking at the case for X = 10, n = 10:
>> hd = 0 CSI = 9,999,999,999 10 left-hand 9's, 0 right hand
>> 0's
>> hd = 1 CSI = 9,999,999,990 9 left-hand 9's, 1 right hand
>> 0's
>> hd = 2 CSI = 9,999,999,900 8 left hand 9's, 2 right hand
>> 0's
>> hd = 3 CSI = 9,999,999,000 7 left hand 9's, 3 right hand
>> 0's
>> hd = 4 CSI = 9,999,990,000 6 left hand 9's, 4 right hand
>> 0's
>> hd = 5 CSI = 9,999,900,000 5 left hand 9's, 5 right hand
>> 0's
>> hd = 6 CSI = 9,999,000,000 4 left hand 9's, 6 right hand
>> 0's
>> hd = 7 CSI = 9,990,000,000 3 left hand 9;s, 7 right hand
>> 0's
>> hd = 8 CSI = 9,900,000,000 2 left hand 9's, 8 right hand
>> 0's
>> hd = 9 CSI = 9,000,000,000 1 left hand 9, 9 right hand 0's
>> hd = 10 CSI = 0 0 left hand 9's, 10
>> right
>> hand 0's
>>
>> Now lets try X = 2, n = 4:
>> hd = 0 CSI = 15 = 1111B 4 left hand 1's, 0 right hand 0's
>> hd = 1 CSI = 14 = 1110B 3 left hand 1's, 1 right hand 0
>> hd = 2 CSI = 12 = 1100B 2 left hand 1's, 2 right hand 0's
>> hd = 3 CSI = 8 = 1000B 1 left hand 1, 3 right hand 0's
>> hd = 0 CSI = 0 = 0000B 0 left hand 1, 4 right hand 0's
>>
>> From this we see:
>> 1. CSI is independent of the contents of a string, except with regard to
>> its
>> Hamming Distance from some reference pattern.
>
> Yep . . .
>
>> 2. CSI for a given string depends heavily on the choice of reference
>> string,
>> since that determines the Hamming Distance.
>
> Yep . . .
>
>> 3. CSI may be seen as simply the Hamming Distance expressed as a variety
>> of
>> unary numeral system. For CSI in radix X, the number of zeros on the
>> right
>> hand of the number is equal to the Hamming Distance (that is, of course,
>> an
>> artifact of formulae having the form X^a - X^b).
>> 4. CSI does not have very good dynamic range.
>
> I don't see the relevance . . .
Your CSI value IS the Hamming Distance!
This is a unary numeral system:
0 = 0
1 = a
2 = aa
3 = aaa
4 = aaaa
5 = aaaaa
N = n repetitions of a
where "a" is any symbol at all.
Take a look at the pattern you get. In radix X, your CSI value IS the
Hamming Distance in a Unary numedral system, with "0" as the symbol, padded
on the left with X-1.
You might as well just use the Hamming Distance by itself and be done with
it.
>
>> 5. CSI does not have very good granularity. There are only n+1 possible
>> values it can take, if it is defined.
>
> Don't see your point . . .
A good metric would allow more fine tuning of values.
>
>> 5. CSI does not provide an intuitive understanding of the "complexity" of
>> a
>> string.
>
> The only reason a string has "intuitive" complexity or randomness is
> when you don't know any simple algorithm for producing it. The fact
> is that there could always be such an algorithm. The fact that you
> haven't yet found it doesn't mean it isn't there. So, the best you
> can say is that a particular string looks "intuitively" random - to
> YOU. It may not look so random to someone else who knows how it was
> actually produced.
Well, this is an argument that I entirely agree with - and an argument
against the utility of ANY CSI algorithm based on an assumption that some
patterns are specifications and others are not - for the very reason that
EVERY possible finite string meets the criteria for being a specification.
>
>> Case 7:
>> A = 3141592653
>> B = 31415926535
>> X = 10
>> n = 10
>> hd = undefined
>>
>> CSI = undefined
>>
>> Even though Sean was aware of modified Hamming Distance as one possible
>> metric to compare the match between two strings of different length, he
>> chose not to use it. He specifically selected, for reasons known only to
>> himself, Hamming Distanct. That means CSI can only exist for a string if
>> it
>> is compared to a reference of exactly the same length. But it does lead
>> to
>> unintuitive results here.
>
> Modified Hamming Distance would work just fine - as I've already
> pointed out to you before. Just use the size of larger sequence space
> in the calculation.
Great. Why don't you use it, then? Why don't you have another go, and try a
new formula?
>
>> Given all this, it is rather difficult to see how the Pitman CSI formula
>> will tell us anything useful about a string. It is highly unlikely to
>> make a
>> design predictor.
>
> It isn't a design predictor, by itself. It is a bias detector and can
> be used to predict the future nature of the bias of a growing string.
> A hypothesis concerning the origin of bias requires additional
> knowledge that goes beyond CSI.
The formula does not accomplish that purpose.
>
>> But, kudos to Sean for at least giving it a try. That's
>> more than most ID-ists have done.
>
> Not really . . . The concept has been there all along. Despite your
> assertion to the contrary, bias can be detected without any knowledge
> of the source of the string. This bias can then be used to predict
> the future appearance of the string as it continues to grow.
> Determining the actual source of the bias requires additional
> information.
>
Bias exists in EVERY string. ALL SORTS of predictions can be made about ANY
string. Without background knowledge, it is not possible to determine which
biases are relavent.