Phrasenation
Zachriel
First a note of thanks to Dr. Sean Pitman. In the midst of attempting to
defend his contrarian position, Dr. Pitman has kept his good humor, his
wits, and has always attempted to make clear his various meanings (though
not always with complete success). He has acted as a spur to the creative
impulses, and I must say that I know much more now about the evolution of
words and the combinatorics of letters than I ever did before (for whatever
*that* might be worth).
-------------------------
In the Beginning
This project started as a word-game challenge by Dr. Sean Pitman. He said to
start with a word, and changing just a letter at a time, try and evolve
longer and longer words. That was too easy, so I introduced a poem, "O Sean
Pitman" that used step-wise mutation and manual selection to create a
multi-line poem in iambic pentameter. This poem demonstrated that there was
a selectable path from the single-letter word "O" to the entire poem.
Then he indicated that this wasn't sufficient, but that the word-game had to
be played without human intervention, and that I had to calculate every
single possible insertion. Without human intervention doesn't sound like
much of a word-game, but I introduced Word Mutagenation on talk.origins,
which showed that evolution could not only evolve words through random
mutation and mechanical selection, but could pretty much exhaust the entire
dictionary. In addition, Dr. Pitman's original and specific numerical
predictions were contradicted by the data. You can find the original
discussion, including links to the original introduction of Word
Mutagenation on talk.origins, here:
http://www.zachriel.com/mutagenation/
Today, we introduce Phrasenation.
http://www.zachriel.com/phrasenation/
-------------------------
The Meaning of Words
As a general principle, we could define a word found in the dictionary as
more meaningful than a jumble of letters; so "king" is more meaningful than
"kxjz". Two words forming a valid phrase are, of course, more meaningful
than a single word; so "the king" is more meaningful than simply "king".
Consequently, a single word with a leading or trailing space is more
meaningful than just a single word as it implies a connection to another
word; so " king" is slightly more meaningful than simply "king". The
addition of special symbols can also add meaning; so "the king!" is more
meaningful, and of a somewhat different meaning, than "the king".
Some Valid Phrasenations
"king"
" king"
"the king"
"the king!"
So how do we determine, for the purposes of Phrasenation, what constitutes a
valid word or phrase? Well, we will take a miniscule sample of English
literature, and compare our phrase to that sample! If the sample is found,
then it will be considered a valid phrase. One proviso: All words must be
complete. No half words. (As a technical matter, we will count certain
symbols, such as "!" and "?" as separate words.)
-------------------------
Our Phrase Book
For our Phrase Book, we will use the Tragedy of Hamlet, Prince of Denmark by
William Shakespeare. I'm sure most everyone can agree that just about
anything the Bard said is in some sense meaningful. After all, he
practically invented the English language single-handedly! Keep in mind that
this will exclude the vast majority of valid phrases, including even most of
Shakespeare. However, you can add phrases to the Phrase Book, if you choose.
Phrases NOT Included
"O, happy dagger"
"a tale told by an idiot"
"the quality of mercy"
"O Sean Pitman"
Finally, all phrasenations are ranked by numbers of characters. Longer is
"better".
-----
Hamletations
When generating a phrasagen (a mutant phrase), we will use random
mutation and various types of recombination. Starting from just two words,
"the" and "question", some not-so-valid phrasagens might look like these:
Some Not-So-Valid Phrasenations
Point quextion
Snip ues
Remainder qtion
Exchange thestion
Insertion quethstion
Complex quthon
Phrasenation allows one to adjust the relative frequency of each type of
mutation. Once having generated a phrasagen, we must compare it to our
Phrase Book in order to determine its meaningfulness. If it is not found, we
will ruthlessly eliminate it. And indeed, we will eliminate invalid
phrasenations in "full proportions".
-------------------------
Indexing the Bard
To make this a practical matter, we must index every valid phrase in our
Phrase Book. But it isn't enough to index just the first word in a phrase.
We must index every single word. Consider that each of these are valid
phrases:
"to be, or not to be"
"be, or not"
"or not to be"
"not to be"
But what if the first words are the same? Well, then we will compare the
word that follows, and if necessary each succeeding word until we find a
word which is different. For instance, "to be" is not a unique phrase, as it
could be found as "to be, or not" or as "to be- that is the question". In
fact, the phrase "to be" shows up 34 times as first words, including "to be
your valentine", but "to be," (with a trailing comma) only once, in "to be,
or not to be".
The index includes 36,176 words (including symbols).
-------------------------
A Most Interesting Result
When testing the index, the Phrasenator outputted every indexed word
followed by a specified number of words. Then the Phrasenator counted the
number of unique phrases. For one-word phrases, 4,801 were unique phrases.
But what about other numbers of words? For a large number of words, the
answer is surely 36,176, but what is a "large" number? Somewhat
surprisingly, if you select any four words in series, the vast majority
(97.63%) will constitute a unique phrase! And for that small percentage
which are not unique, those phrases are nearly all purposefully repeated,
such as "a pit of clay for to be made" from the singing Clown's refrain,
A pickaxe and a spade, a spade,
For and a shrouding sheet;
O, a pit of clay for to be made
For such a guest is meet.
-------------------------
Definitions
L = Length of phrase or collection of phrases
N = Number of valid phrases of length L. N is generally difficult to define.
In Phrasenation, we explicitly define it as the contents of The Tragedy of
Hamlet, Prince of Denmark, by William Shakespeare.
M = Total number of Mutations, the number of Phrasagens whether valid or
not.
C = Number of allowable characters (letters or symbols).
C^L = Sequence space. Number of possible combinations of letters in a string
of length L.
(C^L)/N = Pitman's Number, the number of possible combinations of characters
divided by the number of valid phrases. "Zillions and zillions".
N/(C^L) = Pitman's Ratio, the probability of evolving a valid sequence
through Pitman's random walk.
With "O Sean Pitman", we introduced our modest project with simple
concatenation and point mutation. Then, in accordance with Dr. Pitman's
wishes, we added Insertion. Now we add Exchange and Complex Recombination.
Each of these categories are approximately related to powers of L.
L^0 Concatenation: There are only two ways to join two strings end-to-end.
The number of ways to choose is not dependent on L.
L^1 Point Mutation: There are L+2 possible locations for each such mutation,
including adding one at either end. The number of ways to choose is
proportional to L.
L^2 Exchange: Exchanging same-length snippets between strings. We must
choose the beginning and end of snippet to be exchanged.
L^3 Insertion: Inserting snippets from one string into another string. We
must choose the beginning and ending of each snippet, plus the insertion
point.
L^4 Complex: Exchanging snippets of possibly different lengths between
strings. We must choose the beginning and ending of the snippet, and the
beginning and ending of the section to be replaced.
-------------------------
Pitman's Assertions
Dr. Pitman's assertions relate to the behavior of Phrasenation as L
increases.
Pitman's Vastness Assertion
Pitman's Ratio approaches zero
N/(C^L) --> 0
Pitman's Principle of Zillions
Pitman's Number approaches Zillions and Zillions
M --> (C^L)/N
As L increases, Dr. Pitman claims that the ratio of valid phrases to the
totality of sequence space, approaches zero. Valid sequences get lost in the
vastness of sequence space. He concludes that it is impossible to evolve
sequences beyond the " lowest level of complexity". However, Dr. Pitman has
failed to provide a method of calculating N, much less a map of how valid
sequences are distributed. Generally, any collection of valid phrases and
sentences have some validity. Language can ramble somewhat and still be
valid. We could start by talking about Pitman's handwaving and suddenly, for
no particular reason, change the subject to ghosts and the murder of kings.
Imagine that!
-------------------------
Results Matter
However, even using a tiny sliver of the English language--just one play by
one playwright--it can be shown that phrases of substantial length can be
easily evolved. Our experiments have shown that words and phrases appear to
have some underlying connection related to their own evolution. As such,
words and phrases make ideal subjects for evolutionary algorithms.
Beware a War of Words, Sean Pitman,
Ere you err.
http://www.zachriel.com/mutagenation/Beware.asp
-------------------------
Specific Pitman Extrapolations
Originally, Dr. Pitman claimed that it would take a Pitman Number of
mutations to discover new words or phrases, indeed, it would take zillions
and zillions of years. He was only off by a few gazillion, and he revised
his predictions accordingly.
http://tinyurl.com/3wuub
Pitman's revised assertions:
L=16, M=5e8
L=32, M=2e17
Phrasenation on a sliver of the English language
(one play by one playwright)
Size of Book = 2500
Seed = "to", "be"
L= 8, M=2e4
L=16, M=2e6
L=32, M=2e7
L=48, M=6e7
There is sufficient data, available to anyone who cares to look at the open
source code, to flatly contradict Dr. Pitman's original predictions based on
the Pitman Ratio, N/C^L--and his revised predictions, as well.
-------------------------
Zachriel's Phrase Mutation and Evolution Experiment
And it takes less than "zillions of years"!
http://www.zachriel.com/phrasenation/
Tired of letting Shakespeare have all the fun?
Coming Soon! Phrasomatic!!
Zachriel
"Zachriel" <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com>
wrote in message news:DbqdnUHqBcS...@adelphia.com...
<snipped>
Minor update (so it doesn't keep changing pictures when using the sliders).
>
A few results for small Book size. I let the Phrasenator run until it timed
out.
Book Size = 25
Timeout = 5 seconds (set in the variables screen)
Seed = "to", "be"
M = Total mutations
L = Length of longest phrase
---------
100% Point Mutation
M = 698K, L = 12
100% Exchange
M = 54K, Nothing
100% Insertion
M = 109K, L = 3
100% Complex
M = 50K, L = 3
Note that without Point Mutations there is no material for recombination to
work with.
---------
50% Point Mutation, 50% Exchange
M = 813K, L = 15
50% Point Mutation, 50% Insertion
M = 294K, L = 20
50% Point Mutation, 50% Complex
M = 250K, L = 23
40% Point, 20% Exchange, 20% Insertion, 20% Complex
M = 455K, L = 23
A mixture of Point Mutation and recombination is best. A mixture of
different types of recombinations even better.
---------
Default
40% Point, 20% Exchange, 20% Insertion, 20% Complex
Mutation Rate = 1
M = 455K, L = 23
Mutation Rate = 2
M = 1317K, L = 17
Mutation Rate = 3
M = 1685K, L = 14
Mutation Rate = 4
M = 3609K, L = 14
Mutation Rate = 5
M = 4108K, L = 13
Mutation Rate = 10
M = 5140K, L = 12
Having multiple mutations before selection results in generally slower
evolution. Basically, information is lost.
Bill Rogers
<snip>
> Specific Pitman Extrapolations
>
> Originally, Dr. Pitman claimed that it would take a Pitman Number of
> mutations to discover new words or phrases, indeed, it would take zillions
> and zillions of years. He was only off by a few gazillion, and he revised
> his predictions accordingly.
>
> http://tinyurl.com/3wuub
> Pitman's revised assertions:
> L=16, M=5e8
> L=32, M=2e17
>
> Phrasenation on a sliver of the English language
> (one play by one playwright)
>
> Size of Book = 2500
> Seed = "to", "be"
> L= 8, M=2e4
> L=16, M=2e6
> L=32, M=2e7
> L=48, M=6e7
>
> There is sufficient data, available to anyone who cares to look at the open
> source code, to flatly contradict Dr. Pitman's original predictions based on
> the Pitman Ratio, N/C^L--and his revised predictions, as well.
Thank you, Zach.
Once again Pitman's predictions look sort of OK for low L and get
worse and worse as L increases. It even seems to be something of a
rule that as L doubles, the Pitman Prediction/Zach Observation ratio
increases by a log of logs. Quite a dramatic trend. Doubtless he will
claim that he expected it and that once L gets big enough (defined as
safely out of range of your computing power) the trend will
miraculously reverse itself and C^L/N will begin to accurately predict
the number of mutants that must be screened to find a new phrase of
length L.
I suppose there is an off chance that results like this might lead him
to reconsider whether the model behind C^L/N is actually correct.
Bill
Alas, poor Pitman, I knew him, Zachriel,
A fellow of infinite jest, of most exponential fancy. He hath told us
of the gaps a zillion times. And now, how disproved by Phrasenation
they are. My mutants recombine at it. Here failed those powers of ten
that he has multiplied I know not how oft. Where be your L's now, your
marbles, your odds, your merry multiplications that were apt to set
the howlers on a roar. Not one now, to mock your own mis-estimation?
Zachriel
"Zachriel" <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com>
wrote in message news:DbqdnUHqBcS...@adelphia.com...
Book Size = 10,000
Default Mutation = 40,20,20,20%
Seed Words = "to", "be"
3 62 "toe"
4 73
5 100
6 1,400
7 11,045
8 19,423 "meet it "
9 48,187
10 67,742
11 692,346
12 844,180
14 1,403,390 ". o, woe is me"
15 112,948,722
16 3,701,287
17 3,780,502
18 10,361,067
19 10,537,393
21 112,948,968
22 16,714,376 "think not so, my lord "
23 18,349,145
24 27,734,360
25 31,112,510
30 34,340,269 " have you not my lord, i have "
31 113,003,041
32 74,085,679
33 80,527,518
34 114,281,622
35 96,952,395
36 97,549,310
39 114,946,466
41 133,615,397
47 141,753,360
65 220,962,305
66 231,402,806
67 236,898,314
"now, 'tis not to come', if it be not to come, it will be now; if it"
Compare to Dr. Pitman's (revised) predicted values
http://tinyurl.com/3wuub
> L = 4: 20 mutations on average
> L = 8: 20,000 mutations
> L = 16: > 5e8 mutations
> L = 32: ~ 2.5e17 mutations
> L = 64: ~ 6e34 mutations
Compare to the totality of sequence space, C^L, C=30.
30^16 = 4e23
30^32 = 2e47
30^64 = 3e94
30^67 = 9e98
Please note that 236,898,314 <>
927,094,631,000,000,000,000,000,000,000,000,
000,000,000,000,000,000,000,000,000,000,000,
000,000,000,000,000,000,000,000,000,000,000.
As we have a finite-sized Phrase Book, knowing there are 36,176 phrase
indexes in our Phrase Book, and assuming each phrase can vary from 1 to 100
words, there are 3e6 possible phrases (with a small portion of duplicates).
Please note that solving the Pitman Number for N does not give us the
predicted value.
236,898,314 <> 9e98 / 3e6.
Nope, not even close.
---------------
Book Size = 2500
Default Mutation = 40,20,20,20%
Seed Words = "to", "be"
2 5
3 20
4 274
5 700
6 1,381
" do so"
7 6,097
8 25,745
11 52,860
12 335,823
13 679,316
14 960,567 "but there is, "
15 1,010,058
17 1,986,231
18 2,155,279
19 2,899,586
20 3,176,405
23 3,367,821
24 3,480,790
25 5,652,823
30 7,366,917
31 7,607,293
35 18,872,866 " nay, i know not. is it the king? o"
37 22,770,593
38 35,578,958
41 38,139,477
47 43,580,235
48 63,390,237
50 88,277,829
51 130,697,741
52 136,837,881 ", we know what we are, but know not what we may be. "
30^52 = 6e76
A similar calculation clearly shows that random mutation and mechanical
selection for length exceeds any notion of a mere random walk.
Zachriel
"Zachriel" <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com>
wrote in message news:zuOdnYVeZYb...@adelphia.com...
>
> "Zachriel"
> <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com> wrote in
> message news:DbqdnUHqBcS...@adelphia.com...
> <snipped>
>>
>> Today, we introduce Phrasenation.
>> http://www.zachriel.com/phrasenation/
>>
>
> Book Size = 10,000
> Default Mutation = 40,20,20,20%
> Seed Words = "to", "be"
>
> 3 62 "toe"
> 4 73
> 5 100
> 6 1,400
> 7 11,045
> 8 19,423 "meet it "
> 9 48,187
> 10 67,742
> 11 692,346
> 12 844,180
> 14 1,403,390 ". o, woe is me"
> 15 112,948,722
> 16 3,701,287
Hmm. 15 = 112,948,722.
There's a bug when stopping and restarting the program when calculating
these values. I'll look into it. . . . Oops found it already.
In sub "readDataArrays"
Was
.Cells(1, "h") = maxSoFar
(which probably made it zero)
Should be
maxSoFar = .Cells(1, "h")
(which sets it to the value when last saved)
Oh, well. It doesn't affect the other values. What happened it that when
restarting the Phrasenator, it updated the first-time it found a phrase of
length 15--since restarting. This would not effect the largest values which
were calculated without interruption last night. Once it had found a longer
one on the previous run, it should not have updated that value, but it did.
This should fix it. I'll post it, but won't be able to completely debug it
till later. Certainly the Phrasenator found some phrase of length 15 after
no more than 112,948,722 mutations, certainly sooner, and length 67 after no
more than 236 million.
--
RobinGoodfellow
Hey Zachriel,
Here's some fun preliminary analysis to go with your Phrasenator. If
there are a total of W (not necessarily distinct) words in Hamlet, given
the way you define a phrase, the total number of valid phrases N of any
length is given by:
N = number of one word phrases +
number of contiguous two-word phrases +
number of contiguous three-word phrases + ... +
number of contiguous W word phrases
= W + (W-1) + (W-2) + ... + 1 = W(W+1)/2
I don't know what W is (you say that there are 36,176 words, including
symbols, but certain symbols - e.g. spaces - repeat multiple times). To
be generous to Sean, let's say there are 250,000 non-symbol words, plus
a symbol for every word. That makes for 500,000, or 5e5 words. Using
our formula for N, and generously approximating in Sean's favor, we get
5e5*5e5 ~ 2.5e10, or some 25 billion valid phrases. Mind you, this is
for the length equal to the number of characters in Hamlet: if the
average number of chracters per word is c, Sean's "sequence space" size
would be 26^(cW). The smaller c the better for Sean, so saying c=3
yield a sequence space sizeof 26^(1.5e6) - I don't want to think about
how large that is. If you limit your length to, say, 67 characters, N
would be much, much smaller - but, because we all like Sean, we'll keep
it at 2.5e10.
So, let us now compare your results with Sean's predictions: (L=word
length, M=number of mutations, P=Pitman Number=C^L/N)
For L=3, M=62, P~0 (26^3 << 2.5e10)
L=4, M=73, P~0
L=5, M=100, P~0
So far, Sean's holding up marvelously: his prediction actually
understimates the number of mutations needed. Of course, that is
because N so grossly overestimates the number 3, 4, and 5-letter
phrases. Let's try larger values of L.
For L=10, M=67742, P~5,467
L=11, M=692346, P~146,810
L=12, M=844180, P~3,817,200
So far, so good for Sean, though it seems like his predictions are
beginning to overtake the actual values. But maybe there's still
something to his arithmetic. Let's continue
For L=22, M~1.7e7, P~5.3e20
L=23, M~1.8e7, P~1.4e22
L=24, M~2.7e7, P~3.6e23
Hmmm ... the actual number of mutations is an embarrassing number of
orders of magnitudes below what is given by Seans formula (it's lower
than the cube root of his prediction), despite the fact that we are
still grossly overestimating N. But, as Dr. Pitman has asserted,
phrases will start to grow apart in "sequences space" as the sequence
length gets bigger, and his formula will once again become a useful
predictor? So, let's keep going
For L=36, M~9.8e7, P~3.5e40
For L=39, M~1.1e9, P~6.1e44
For L=41, M~1.3e9, P~4.1e47
All I can say is ouch, ouch, ouch. The only thing that the actual
number of required mutations and Sean's formula have in common is that
they are both monotonically increasing. Which also be said a whole slew
of functions, such as log(L).
But, as Bill pointed out, I'm sure Sean would claim that his
approximation will start working again far beyond the bounds of modern
computational ability. Or that you "stacked the deck" in favor of
evolution one way or another with your phrasenator. Or that you are
evolving all your phrases to fit various "templates", so it doesn't
count as real evolution. With so many ways to hand-wave around actual
observations, the fun never needs to stop!
I'll try to post some more analysis of your data a bit later, as time
allows. In the meantime, let me just say, wonderful job! Not only does
your Phrasenator technically impressive and provides strong empirical
support for your position, but your web-page design is just damn cool!
Reading through was really enjoyable.
Cheers,
Leonid.
PS: All my calculations were very back-of-the envelope, so I disclaim
all warranty, explicit or implied, on my math. No purchase necessary,
void where prohibited, human calculators need not apply.
Zachriel
"Zachriel" <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com>
wrote in message news:QdOdnR2fC43...@adelphia.com...
>
> "Zachriel"
> <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com> wrote in
> message news:zuOdnYVeZYb...@adelphia.com...
>>
>> "Zachriel"
>> <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com> wrote
>> in message news:DbqdnUHqBcS...@adelphia.com...
>> <snipped>
>>>
>>> Today, we introduce Phrasenation.
>>> http://www.zachriel.com/phrasenation/
>>>
>>
>> Book Size = 10,000
>> Default Mutation = 40,20,20,20%
>> Seed Words = "to", "be"
>>
>
> Hmm. 15 = 112,948,722.
> There's a bug when stopping and restarting the program when calculating
> these values. I'll look into it. . . . Oops found it already.
>
3 5 bet
4 107 too
5 395 woo
8 455 too too
9 2,215 too too
10 16,023 let that
11 93,159 it to the
12 615,557 not to the
13 1,258,860 and there i
14 1,600,017 and that they
15 2,210,268 in my heart of
16 4,282,888 on him by them
17 5,981,931 sworn, my lord,
18 6,654,419 hear and see the
22 9,606,379 do this? ay, my lord,
23 25,033,567 do this? ay, my lord,
25 26,891,190 my lord, my lord, i have
30 27,541,821 in my heart that i shall live
31 32,544,665 in my heart that i shall live
32 84,436,722 my father. o, where, my lord? in
34 89,277,623 my father. o, where, my lord? in
36 90,644,474 my father. o, where, my lord? in my
37 145,220,165 , that follows not. what follows then
38 164,282,752 this? ay, my lord, i would know that
39 175,125,116 father; but you must know, your father
46 179,948,114 you. you are welcome. nay, good my lord, this
48 223,186,675 to you. you are welcome. nay, good my lord, this
49 287,345,293 come', if it be not to come, it will be now; if
51 302,104,946 ? my lord, the king your father. the king my father
56 354,398,047 a thing- a thing, my lord? of nothing. bring me to him.
57 378,358,775 a thing- a thing, my lord? of nothing. bring me to him.
69 433,409,166 be not to come, it will be now; if it be not now, yet it
will come:
Got up to 69 letters this time, but it took a bit longer. The program
records the largest value so far. So, for instance, there is a gap between
57 and 69. It discovered a 57-letter sequence. The next longest word it
discovered was 69 letters. Later on it discovered sequences of length 58
through 68, but these numbers are not recorded. The sudden jump is
noteworthy.
Zachriel
"RobinGoodfellow" <lmuc...@yahoo.com> wrote in message
news:cm6t5q$2mo$1...@news01.cit.cornell.edu...
> Zachriel wrote:
<snip>
>> Zachriel's Phrase Mutation and Evolution Experiment
>> And it takes less than "zillions of years"!
>> http://www.zachriel.com/phrasenation/
>>
>> Tired of letting Shakespeare have all the fun?
>> Coming Soon! Phrasomatic!!
>>
>>
>
> Hey Zachriel,
>
> Here's some fun preliminary analysis to go with your Phrasenator. If
> there are a total of W (not necessarily distinct) words in Hamlet, given
> the way you define a phrase, the total number of valid phrases N of any
> length is given by:
>
> N = number of one word phrases +
> number of contiguous two-word phrases +
> number of contiguous three-word phrases + ... +
> number of contiguous W word phrases
> = W + (W-1) + (W-2) + ... + 1 = W(W+1)/2
>
> I don't know what W is (you say that there are 36,176 words, including
> symbols, but certain symbols - e.g. spaces - repeat multiple times).
Spaces and apostrophes aren't counted. Spaces are used as separators only.
Apostrophes are counted a letters, such as in 'twas and 'twere. However,
this resulted in counting quotes as apostrophes, so we are overcounting the
number of unique words somewhat.
> To
> be generous to Sean, let's say there are 250,000 non-symbol words, plus
> a symbol for every word.
There are 4801 distinct words, including one-letter symbols that are counted
as words, and words that begin or end with quote marks. There are 36176
words altogether with an average length of around 4.3 letters each. The
incredible results of the Hamletator!
But let's be generous to Sean. By the way, nearly all the stage direction
has been removed leaving only the poetry (except the description of the play
within the play).
Something else to point out is that Hamlet represents just a sliver of
Shakespeare, and all of Shakespeare represents just the tiniest portion of
valid English phrasing. If the Phrasenator finds the phrase "four score and
seven years" or "can you spare a dime" or "they misunderestimate me", then
it would be rejected. Indeed, the vast majority of English is simply thrown
away.
I considered a larger Phrase Book, but indexing takes quite a while. Of
course, I could include the index for download. But no matter how big the
Phrase Book, it would always be just a tiny portion of valid phrases. And
the Phrasenator works quite well just like it is.
> But, as Bill pointed out, I'm sure Sean would claim that his
> approximation will start working again far beyond the bounds of modern
> computational ability. Or that you "stacked the deck" in favor of
> evolution one way or another with your phrasenator. Or that you are
> evolving all your phrases to fit various "templates", so it doesn't
> count as real evolution.
The Phrase Book is just used to determine whether a given phrase is
meaningful, or not. The process of creating phrasagens is completely random.
But these are certainly valid phrases:
37 o, i could tell you- but let it be.
39 i think you did not love your father;
50 nay, that follows not. what follows then, my lord
52 my lord, the king your father. the king my father?
They are even meaningful and evocative.
> With so many ways to hand-wave around actual
> observations, the fun never needs to stop!
Tired of letting Shakespeare have all the fun?
Coming Soon! Phrasomatic!!
> I'll try to post some more analysis of your data a bit later, as time
> allows. In the meantime, let me just say, wonderful job! Not only does
> your Phrasenator technically impressive and provides strong empirical
> support for your position, but your web-page design is just damn cool!
> Reading through was really enjoyable.
>
> Cheers,
> Leonid.
>
> PS: All my calculations were very back-of-the envelope, so I disclaim
> all warranty, explicit or implied, on my math. No purchase necessary,
> void where prohibited, human calculators need not apply.
>
--
Zachriel
"Zachriel" <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com>
wrote in message news:DbqdnUHqBcS...@adelphia.com...
<snip>
>
> open
> source code, to flatly contradict Dr. Pitman's original predictions based
> on
> the Pitman Ratio, N/C^L--and his revised predictions, as well.
>
You can access the source code by pressing alt-F11 when the spreadsheet (not
the control panel) is active, or by tools/macro/visual basic editor. Or you
can interrupt the run of the program by pressing the escape key which will
drop you into the debug mode of the editor. You can examine the code and
variables, then resume the program when you want. The heart of the code is
in the Phrasenate module which contains the mutation algorithms. Lots of
calls to the Random(A,B) function, which returns a random integer between A
and B. The Hamlet module was used only to prepare the Phrase Book; removing
stage directions, counting words, testing the index, etc.
I got the transcript of Hamlet here:
http://darkwing.uoregon.edu/~rbear/shake/hamlet.html
Don't forget to unzip the file to its own folder if you want to enjoy the
graphics included with the software package. If you can't read the labels in
the control panel due to the images, just click on a label and the picture
will temporarily disappear. If you don't like the graphics in the control
panel at all, then just delete the images from the folder.
--
John Wilkins
<"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com> wrote:
This whole thing is fascinating. Can you put it up on a coherent webpage
when you are finished, say alongside the Weasel programs Ian Musgrave
has set up?
--
John S. Wilkins jo...@wilkins.id.au
web: www.wilkins.id.au blog: evolvethought.blogspot.com
God cheats
Zachriel
"Zachriel" <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com>
wrote in message news:DbqdnUHqBcS...@adelphia.com...
<snipped>
Minor updates:
Maximum size of book increased to 25000 for those who want to hurt
themselves.
Default Mutations, 40,30,20,10.
Default Seed Word, "mad".
Doesn't update the book when flipping between worksheets.
(But the program still updates when changing values on the Phrasagens
worksheet. Keep in mind that if you change these values manually, and if the
Book is very large, it might take a while to update the control panel.)
Zachriel
"Zachriel" <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com>
wrote in message news:DbqdnUHqBcS...@adelphia.com...
-------------------------------------------------
Zachriel's Extravagant Extrapolation
--------------------------------------------------
Zachriel can handwave with the best of them, or Zachriel's extravagant
extrapolation . . .
We noted before with Word Mutagenation that maximum growth size of words was
related to the Book Size. In other words, you can't grow elephants in Petri
dishes. If you want large phrases, you need to allow for enough room for
them to grow. These are typical phrase sizes that are easily achievable
(with the limited number of trials I have attempted so far).
Book size - Max phrase length
100 - 15
1,000 - 30
10,000 -60
In the spirit of Dr. Sean Pitman, and using these numbers as the basis for
Zachriel's extravagant extrapolation, these are my, ahem, projections:
Book size - Max phrase length
100,000 -120
1,000,000 - 240
10,000,000 - 480
100,000,000 - 960
So with about a Book size of about a hundred million, we can project the
literary output of a poem the length of "O Sean Pitman". With a population
similar to the number of bacteria in the average human intestines, we can
expect the entire poetic play, "The Tragedy of Hamlet, Prince of Denmark".
Now what would happen if we were to increase the size of the Phrase Book?
You can either take these predictions on their face, or you might rather
consider the possibility that Zachriel has been Phrasenating for a little
bit too long.
Zachriel
"John Wilkins" <john...@wilkins.id.au> wrote in message
news:1gmo145.1t65cnbv6yltzN%john...@wilkins.id.au...<snip>
>> 69 433,409,166 be not to come, it will be now; if it be not now, yet
>> it
>> will come:
>>
>> Got up to 69 letters this time, but it took a bit longer. The program
>> records the largest value so far. So, for instance, there is a gap
>> between
>> 57 and 69. It discovered a 57-letter sequence. The next longest word it
>> discovered was 69 letters. Later on it discovered sequences of length 58
>> through 68, but these numbers are not recorded. The sudden jump is
>> noteworthy.
>
> This whole thing is fascinating. Can you put it up on a coherent webpage
> when you are finished, say alongside the Weasel programs Ian Musgrave
> has set up?
I have been collecting some interesting data. What would you envision?
-----
John Wilkins
> "John Wilkins" <john...@wilkins.id.au> wrote in message
> news:1gmo145.1t65cnbv6yltzN%john...@wilkins.id.au...
> > Zachriel
> > <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com> wrote:
> <snip>
> >> 69 433,409,166 be not to come, it will be now; if it be not now, yet
> >> it
> >> will come:
> >>
> >> Got up to 69 letters this time, but it took a bit longer. The program
> >> records the largest value so far. So, for instance, there is a gap
> >> between
> >> 57 and 69. It discovered a 57-letter sequence. The next longest word it
> >> discovered was 69 letters. Later on it discovered sequences of length 58
> >> through 68, but these numbers are not recorded. The sudden jump is
> >> noteworthy.
> >
> > This whole thing is fascinating. Can you put it up on a coherent webpage
> > when you are finished, say alongside the Weasel programs Ian Musgrave
> > has set up?
>
> I have been collecting some interesting data. What would you envision?
>
A step-by-step introduction into the major claims (mentioning but not
focussing on Sean's objections to evolution) for the uninformed
intelligent laity. With plenty of wry humour, which given Shakespeare as
a target won't be hard. Talk to Ian.
Sean Pitman
the truly important problem for the ToE. That is far more than I can
say for most in this forum. Unfortunately, though you continually
misstate my position both here and on your website, my challenge has
always been to evolve a 1000-character sequence by random mutation and
function based selection where each step is more functionally
beneficial than the step before (I never claimed the limit to
evolution was at the level of single words). You claimed you could
easily do this and presented your long "O Sean Pitman" poem as
evidence. Of course, you haven't come close to anything of this size
and your current attempt, though admirable as far as effort, falls way
. . . way short. There is just one tiny little problem with your
program that turns out to be the same lethal problem that Dawkins had
with his program.
Evidently, you need to go back and re-read the following post I wrote
in response to similar assertions of yours:
It basically destroys the premise behind your "Phrasenator" program.
Your program is based on the notion that every part of a sequence in a
collection of phrases like Hamlet is meaningfully beneficial as long
as no partial words are present. This notion is simply ridiculous. I
means that sequences like, "and in the" and ", no, not" and "is let
the" are defined as meaningfully beneficial in your program.
Basically, absolutely any addition to a string will be defined as
beneficial as long as it is a complete word found as part of this
particular sequence in Hamlet. It need not represent an intact
thought much less an internally relevant thought. Well, using those
parameters, all of Hamlet would be rapidly evolved using your
parameters since it would take an average of less than 20 million
mutations to find a new beneficial addition regardless of the size of
the "evolving" sequence.
Formula used: (N/C^L)s * (N/L)l where s = the small sequence and l =
the larger evolving sequence.
This is really not different than an expanded form of Dawkins's
template based "Methinks it is like a weasel" evolution program. The
only difference is that you are using intact single words instead of
intact single letters. However, like Dawkins, you do not consider the
presence or lack of collective meaning in the steppingstone sequences
as you head toward an ideal sequence. You simply define every single
match to the model as beneficial. Weather or not it makes collective
sense is simply not considered. And, that is fatal to your position.
You see, I showed you how single letter changes couldn't get you
across the gap once you went beyond very short sequences. You even
recognized this fact yourself. After a meaningfully beneficial
sequence reaches a couple dozen or so characters in size, you are
forced to change, add, or subtract multiple characters at the same
time in order to get very far off your island. So, you resorted to
adding or subtracting whole words to a sequence instead of single
letters. Well, that does help you for a while. However, as your
sequence gets a bit longer, you start running into walls were the
addition of just one word will not get you very far as far as
beneficial meaning or function - just like it was with the
single-letter-gap problem. Now, you must change, add, or subtract
multiple words at the same time in order to get very far off your
island.
So, you see, with increasing sequence length, the total number of
letters that must be changed at the same time to get to a new island
of beneficial function - *increases*. You just can't get off your
staring island with 1 letter change, or 2 or 3 or 4 . . . or 10 or 20
or 30 . . . with increasing L. It is this growing requirement that
creates the exponentially expanding neutral gap. You see, a gap of
just 10 characters is equal to about 300 trillion random walk steps
and a gap of 20 is equal to 80,000 trillion trillion random walk
steps.
Bigdakine
>From: seanpi...@naturalselection.0catch.com (Sean Pitman)
>Date: 11/4/04 4:43 PM Hawaiian Standard Time
>Message-id: <80d0c26f.04110...@posting.google.com>
>
>"Zachriel" <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com>
>wrote in message news:<DbqdnUHqBcS...@adelphia.com>...
>
>Great try Zach - really! At least you are actually trying to consider
>the truly important problem for the ToE. That is far more than I can
>say for most in this forum. Unfortunately, though you continually
>misstate my position both here and on your website, my challenge has
>always been to evolve a 1000-character sequence by random mutation and
>function based selection where each step is more functionally
>beneficial than the step before (I never claimed the limit to
>evolution was at the level of single words)
Why does it have to be "more beneficially functional"?
Thats a crock and you know it.
I haven't seen a goal post move this fast since Northwestern snapped a losing
streak 20 years a go.
<rest snipped>
Stuart
Dr. Stuart A. Weinstein
Ewa Beach Institute of Tectonics
"To err is human, but to really foul things up requires a creationist"
"Creationists aren't impervious to Logic: They're oblivious to it."
Zachriel
"Sean Pitman" <seanpi...@naturalselection.0catch.com> wrote in message
news:80d0c26f.04110...@posting.google.com...
> "Zachriel"
> <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com> wrote in
> message news:<DbqdnUHqBcS...@adelphia.com>...
>
> Great try Zach - really!
Hi Sean. I noted you made the same post twice. I'll answer in the other
thread, "Zach's 'Phrasenation' evolution program".
(It might be best not to do that. Some people may answer in this thread,
some in the other. A new thread is appropriate though. I'm kinda keeping
this one for program notes, bug reports, etc. No biggie.)
Sean Pitman
> >Great try Zach - really! At least you are actually trying to consider
> >the truly important problem for the ToE. That is far more than I can
> >say for most in this forum. Unfortunately, though you continually
> >misstate my position both here and on your website, my challenge has
> >always been to evolve a 1000-character sequence by random mutation and
> >function based selection where each step is more functionally
> >beneficial than the step before (I never claimed the limit to
> >evolution was at the level of single words)
>
> Why does it have to be "more beneficially functional"?
What is the point of natural selection as part of mechanism of
evolution? To give functionally biased direction to random mutations
over time - right? Well, if two genetic sequences are the same as far
as beneficial function is concerned, where is the ability of natural
selection to distinguish between them? Without a detectable increase
in beneficial function, the best that can happen is neutral drift or
random walk. It is this random walk that takes exponentially
increasing amounts of time with wider and wider neutral gaps.
> Thats a crock and you know it.
It's not a crock since this is exactly how evolution is supposed to
work.
> I haven't seen a goal post move this fast since Northwestern snapped a losing
> streak 20 years a go.
Perhaps the goalposts are moving in your own mind? They certainly
haven't moved on my account. My position has always been that the
expanding neutral gap problem is lethal to the ToE.
> Stuart
> Dr. Stuart A. Weinstein
> Ewa Beach Institute of Tectonics
> "To err is human, but to really foul things up requires a creationist"
> "Creationists aren't impervious to Logic: They're oblivious to it."
It depends on the creationist and/or evolutionist. Both can be quite
hardheaded and oblivious to logic and both can be quite reasonable and
downright logical. Each must decided for his or her self.
Zachriel
"Zachriel" <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com>
wrote in message news:9sadnUaBMYl...@adelphia.com...
I haven't done a complete analysis yet, but with Book Size = 25,000, and
after a couple of nights of phrasenating, I'm up to L = 83.
Anecdotally, I've found that for a given population, and after a long enough
time, the upper ranks of the Phrasenator fill up with variations on a basic
theme. This seems to be the limiting factor. So with the Book Size = 25,000,
I have a lot of variations of "the body is with the king, but the king is
not with the body". In other runs, it fixes on different sections of Hamlet.
But there's still plenty of variety in the lower ranks. The smallest phrases
still extant have L=20. It's a similar situation for smaller Book Sizes.
The basic conjecture seems to hold. The larger the Book Size, the larger the
phrases can grow. Realizing this, Dr. Pitman has apparently abandoned this
line-of-attack and has decided that the entire concept is faulty--even
though Phrasenator works exactly like the Word Mutagenator, which is modeled
on his original word-game. Indeed, both programs share the same basic code.
Oh well.
Of course, there is the latest generation, up-and-coming Phrasomatic!
Bigdakine
>Message-id: <80d0c26f.04110...@posting.google.com>
>
>bigd...@aol.comGetaGrip (Bigdakine) wrote in message
>news:<20041104220826...@mb-m05.aol.com>...
>
>> >Great try Zach - really! At least you are actually trying to consider
>> >the truly important problem for the ToE. That is far more than I can
>> >say for most in this forum. Unfortunately, though you continually
>> >misstate my position both here and on your website, my challenge has
>> >always been to evolve a 1000-character sequence by random mutation and
>> >function based selection where each step is more functionally
>> >beneficial than the step before (I never claimed the limit to
>> >evolution was at the level of single words)
>>
>> Why does it have to be "more beneficially functional"?
>
>What is the point of natural selection as part of mechanism of
>evolution?
To give functionally biased direction to random mutations
>over time - right?
You know Sean, I would figure that somebody who has been involved with this
issue would learn how to state the problem in English.
If I didn't know you already " To give functionally biased direction to random
mutations over time" would make no sense to me.
Natural selection does one thing, and one
thing only. Change allele frequencies as a result of the effects of
differential reproductive propensity.
That one thing however, does have important consequences.
Well, if two genetic sequences are the same as far
>as beneficial function is concerned, where is the ability of natural
>selection to distinguish between them?
Why does it have to distinguish between them? If I have a set of mutations that
produce similar increases in reproductive propensity, then that set will be
favored over of the set of mutations which favor lower reproductive propensity.
Within the set that maximizes reproductive propensity, selection may not be
able to distinguish among its members.
This is one component of what we call "diversity".
Much like your arguments regarding protein functionality, you have mapped that
fallacy to the level of genes.
Without a detectable increase
>in beneficial function, the best that can happen is neutral drift or
>random walk. It is this random walk that takes exponentially
>increasing amounts of time with wider and wider neutral gaps.
Which is what the phrasenator just debunked.
>
>> Thats a crock and you know it.
>
>It's not a crock since this is exactly how evolution is supposed to
>work.
Um no. Lets not confuse evolution with Sean's version of it.
>
>> I haven't seen a goal post move this fast since Northwestern snapped a
>losing
>> streak 20 years a go.
>
>Perhaps the goalposts are moving in your own mind? They certainly
>haven't moved on my account. My position has always been that the
>expanding neutral gap problem is lethal to the ToE.
I referring to your new added criteria, tha each mutation must be increasingly
beneficial. Second, another objection is what about mutations that are
"differently beneficial".
Sean Pitman
> >> Why does it have to be "more beneficially functional"?
> >
> >What is the point of natural selection as part of mechanism of
> >evolution? To give functionally biased direction to random
> > mutations over time - right?
>
> You know Sean, I would figure that somebody who has been
> involved with this issue would learn how to state the
> problem in English. If I didn't know you already " To give
> functionally biased direction to random mutations over
> time" would make no sense to me.
>
> Natural selection does one thing, and one
> thing only. Change allele frequencies as a result of the effects of
> differential reproductive propensity.
>
> That one thing however, does have important consequences.
Yes, it biases the direction of random mutations over time in a
population. In other words, if a mutation results in a more
reproductively fit overall phenotype, the offspring of that individual
will do better than their peers and eventually overtake the colony.
Obviously, this increases the frequency of that mutation. On the
other hand, if a mutation reduces the reproductive fitness of an
individual, its offspring will die out of the population over time.
This clearly reduces the frequency of that mutation over time.
However, if a mutation has no phenotypic effect on reproductive
fitness, its allelic frequency will drift randomly over time. It may
one day become fixed or it may be eliminated from the gene pool
altogether. In this way mutations can be added or subtracted to a
sequence over time in a biased, directed, non-random way even though
the mutations themselves occur randomly.
> > Well, if two genetic sequences are the same as far
> >as beneficial function is concerned, where is the ability of natural
> >selection to distinguish between them?
>
> Why does it have to distinguish between them?
If it cannot detect a difference in functional benefit between
different sequences, it cannot guide the random process of random
mutation. At this point evolution truly becomes a completely random
process - along the lines of a Boeing 747 arising from an explosion in
a junkyard type thing. Without the ability of natural selection being
able to distinguish, in a biased way, between different genetic
sequences, evolution is dead. That is why the neutral gap problem is
so serious.
> If I have a set of mutations that
> produce similar increases in reproductive propensity, then that set will be
> favored over of the set of mutations which favor lower reproductive propensity.
That's fine. You have "more" reproductively fit sequences compared to
what came before. Zach's program does not do this. It does not
analyze the evolved sequences for increased beneficial meaning. All
it does is analyze the evolving sequences as far as a sequence match
to a pre-established sequence (i.e., Hamlet in this case).
Have you ever read about Dawkins's "Methinks it is like a weasel"
evolution program? Do you know what is wrong with that program as far
as a model of biological evolution? Of course the argument is that
Dawkins never intended it to be taken as a model of biological
evolution, but many evolutionists have done just that. Many people
actually think that Dawkins's program models real biological evolution
for every type of cellular function. Though Dawkins's program does
model certain types of biological evolution, like certain aspects of
the immune system and the like, it does not model many other types of
functional systems that do not have access to such a sequence
template.
You see, Dawkins's program was template based. It selected those
sequences that best matched a pre-established sequence without regard
to anything else as far as increased functional advantage is
concerned. In real biological evolution Nature cannot read sequences,
it can only read changes in beneficial function. There is no template
model sequence to compare individual sequence changes with. All
mutational changes are therefore subject to selection pressure only as
far as they affect phenotypic function. The interesting thing is that
many mutations do not change phenotypic function. They are neutral
with regard to function and are therefore immune from selection
pressure. They best that these mutations can do is drift randomly
along in the gene pool. Such mutations may add up over time by sheer
random luck to run across something that is actually beneficial, but
such a random walk takes a whole lot longer than a more directly
guided process.
Zach's program does exactly the same thing that Dawkins's program did.
It selects based on sequence comparison, not on any sort of analysis
of improved beneficial meaning or "function". The only minor
difference between the two programs is that Zach's program only
selects for intact word matches while Dawkins's program selected for
single letter matches. Zach has just changed the number of characters
that must match to be selectable is all. Fundamentally though, the
process is still one of sequence comparison without regard to change
in meaning. Using this process, all of Hamlet could be quickly
evolved since the average gap between selectable sequences of
increasing length is less than 20 million mutations irrespective of
sequence length.
> Within the set that maximizes reproductive propensity, selection may not be
> able to distinguish among its members.
Selection will be able to distinguish between this set and that which
came before - or else this set of improved sequences would not be
preferentially selected over other potential sequences. Zach's
program doesn't do this.
> This is one component of what we call "diversity".
You don't seem to understand that Zach's program doesn't work as
you've described evolution as working here.
> Much like your arguments regarding protein functionality, you have mapped that
> fallacy to the level of genes.
And where do you think proteins come from?
> > Without a detectable increase
> > in beneficial function, the best that can happen is neutral drift or
> > random walk. It is this random walk that takes exponentially
> > increasing amounts of time with wider and wider neutral gaps.
>
> Which is what the phrasenator just debunked.
Not so. You just don't understand that Zach's program does not detect
changes in meaning, only in sequence matching to a pre-established
sequence without regard to meaning or "function". It is the same
Dawkins' fallacy - thinly disguised.
> >> Thats a crock and you know it.
> >
> >It's not a crock since this is exactly how evolution is supposed to
> >work.
>
> Um no. Lets not confuse evolution with Sean's version of it.
Um yes. I'm sure that at least some of the less biased minds that
follow these threads can see the fallacy behind your misunderstanding
here. My "version" of evolution is no different than the standard
"version" of evolution.
> >> I haven't seen a goal post move this fast since Northwestern snapped a
> >> losing streak 20 years a go.
> >
> >Perhaps the goalposts are moving in your own mind? They certainly
> >haven't moved on my account. My position has always been that the
> >expanding neutral gap problem is lethal to the ToE.
>
> I referring to your new added criteria, that each mutation must be increasingly
> beneficial.
Random mutations must be increasingly beneficial if the neutral gap
problem is to be avoided. This was never an added criterion. It is a
fundamental criterion of the ToE.
> Second, another objection is what about mutations that are
> "differently beneficial".
Then they will all be positively selectable as long as they are more
beneficial than what came before. Again, Zach's program doesn't ask
this question. It only asks if the differences are a better match to
a pre-established sequence without regard to increased beneficial
meaning/function.
> "To err is human, but to really foul things up requires a creationist"
If you have an exceptionally bright creationist - maybe . . .
> "Creationists aren't impervious to Logic: They're oblivious to it."
"Logic" = Anything that agrees with Stuart! ; )
> Stuart
> Dr. Stuart A. Weinstein
> Ewa Beach Institute of Tectonics
Zachriel
avoidance in the thread "Pitman's Eyebrows".
"Sean Pitman" <seanpi...@naturalselection.0catch.com> wrote in message
news:80d0c26f.04110...@posting.google.com...
> bigd...@aol.comGetaGrip (Bigdakine) wrote in message
> news:<20041106155556...@mb-m18.aol.com>...
>
<snip>
>
>> > Well, if two genetic sequences are the same as far
>> >as beneficial function is concerned, where is the ability of natural
>> >selection to distinguish between them?
>>
>> Why does it have to distinguish between them?
>
> If it cannot detect a difference in functional benefit between
> different sequences, it cannot guide the random process of random
> mutation. At this point evolution truly becomes a completely random
> process - along the lines of a Boeing 747 arising from an explosion in
> a junkyard type thing. Without the ability of natural selection being
> able to distinguish, in a biased way, between different genetic
> sequences, evolution is dead. That is why the neutral gap problem is
> so serious.
>
>> If I have a set of mutations that
>> produce similar increases in reproductive propensity, then that set will
>> be
>> favored over of the set of mutations which favor lower reproductive
>> propensity.
>
> That's fine. You have "more" reproductively fit sequences compared to
> what came before. Zach's program does not do this.
Sure it does. All sequences must be valid phrases, and we define more fit
sequences to be the longer ones. This is per your own instructions.
http://tinyurl.com/55bw5
> It does not
> analyze the evolved sequences for increased beneficial meaning.
Sure it does. We define beneficial as being valid phrases in Elizabethan
English, with longer phrases being more "beneficial". You had wanted to
evolve long phrases. If you had wanted to evolve phrases with high scrabble
score, we could have done that too--just as we did in Word Mutagenator. We
use a very inefficient method of checking our phrases for validity. We
compare them to a list of phrases. We could instead use a list of famous
quotes, a play from Shakespeare, the Bible, or even in a book of every
phrase ever uttered. We could use a grammar checker instead. It really
doesn't matter. Even with the very inefficient method we are using here, it
is quite obvious that we can evolve long phrases.
> All
> it does is analyze the evolving sequences as far as a sequence match
> to a pre-established sequence (i.e., Hamlet in this case).
This is no different than the Word Mutagenator, or your own original
word-game. You said take a string of letters found in a standard and agreed
list of such strings of letters. Let's call the string of letters a "word",
and the agreed list a "dictionary". Mutate this "word" and compare the
mutant to the "dictionary". Do this repeatedly trying to find longer and
longer such "words".
All the Phrasenator does is use a different "dictionary" containing longer
sequences of words which we will call "phrases". The algorithm is the same.
And you are in complete avoidance of this rather obvious fact.
<snip>
> Zach's program does exactly the same thing that Dawkins's program did.
> It selects based on sequence comparison, not on any sort of analysis
> of improved beneficial meaning or "function".
Sure it does. Phrasenator checks to make sure every phrase is a valid phrase
in Elizabethan English. It does this in a very inefficient manner by
comparing it to a list of known phrases. It would be more efficient if it
had a longer list. It could conceivably have a complete list, just as we
previously used a relatively complete dictionary.
Longer phrases are considered "better" per your original instructions.
> The only minor
> difference between the two programs is that Zach's program only
> selects for intact word matches while Dawkins's program selected for
> single letter matches.
The difference is that all the living members of the population in
Phrasenator are valid phrases.
> Zach has just changed the number of characters
> that must match to be selectable is all. Fundamentally though, the
> process is still one of sequence comparison without regard to change
> in meaning. Using this process, all of Hamlet could be quickly
> evolved since the average gap between selectable sequences of
> increasing length is less than 20 million mutations irrespective of
> sequence length.
<snip>
Wrong again, Dr. Pitman. It would take far more than 20 million mutations to
evolve arbitrarily long sequences.
By the way, you never answered this question. Are any of these meaningful
phrases?
"a cold God soon berates" (23)
"short zany theatre began" (25)
"what fat agent may stare far" (28)
"watery heart so thin beat apart" (31)
"what hot and fat breast bears far" (33)
"hate began here if a heart beat apart" (37)
How about these?
"my salad days" (13)
"wear my heart on my sleeve" (26)
"laugh yourself into stitches" (28)
"I have been in such a pickle" (28)
Now provide a rule. If you can't, then your word-game analogy is fatally
flawed and should be abandoned.
--
Bigdakine
>From: bigd...@aol.comGetaGrip (Bigdakine)
>Date: 11/6/04 10:45 AM Hawaiian Standard Time
>Message-id: <20041106155556...@mb-m18.aol.com>
Actually, I meant "variability".
Bigdakine
>Message-id: <80d0c26f.04110...@posting.google.com>
>
>bigd...@aol.comGetaGrip (Bigdakine) wrote in message
>news:<20041106155556...@mb-m18.aol.com>...
>
>> >> Why does it have to be "more beneficially functional"?
>> >
>> >What is the point of natural selection as part of mechanism of
>> >evolution? To give functionally biased direction to random
>> > mutations over time - right?
>>
>> You know Sean, I would figure that somebody who has been
>> involved with this issue would learn how to state the
>> problem in English. If I didn't know you already " To give
>> functionally biased direction to random mutations over
>> time" would make no sense to me.
>>
>> Natural selection does one thing, and one
>> thing only. Change allele frequencies as a result of the effects of
>> differential reproductive propensity.
>>
>> That one thing however, does have important consequences.
>
>Yes, it biases the direction of random mutations over time in a
>population.
That is difficult to parse. Random mutations are random with respect to
selection. Your statement above seems to imply that natural selection
influences the process of mutation. It does no such thing, and I'm sure you
know that.
In other words, if a mutation results in a more
>reproductively fit overall phenotype, the offspring of that individual
>will do better than their peers and eventually overtake the colony.
All other things being equal.
>Obviously, this increases the frequency of that mutation. On the
>other hand, if a mutation reduces the reproductive fitness of an
>individual, its offspring will die out of the population over time.
>This clearly reduces the frequency of that mutation over time.
All other things being equal.
>However, if a mutation has no phenotypic effect on reproductive
>fitness, its allelic frequency will drift randomly over time. It may
>one day become fixed or it may be eliminated from the gene pool
>altogether. In this way mutations can be added or subtracted to a
>sequence over time in a biased, directed, non-random way even though
Drift is neither biased or directed. Thats why its called *drift*.
>the mutations themselves occur randomly.
>
>> > Well, if two genetic sequences are the same as far
>> >as beneficial function is concerned, where is the ability of natural
>> >selection to distinguish between them?
>>
>> Why does it have to distinguish between them?
>
>If it cannot detect a difference in functional benefit between
>different sequences, it cannot guide the random process of random
>mutation.
Among the suite of elite mutations, selection might not be able to distinguish
among them. On the other hand selection will be certainly make distinctions
between that elite set and the rest.
Seriously, do you suppose that there is only one fit genetic makeup?
At this point evolution truly becomes a completely random
>process - along the lines of a Boeing 747 arising from an explosion in
>a junkyard type thing.
Boy, creationists just can't resist a worn out and debunked false analogy.
Without the ability of natural selection being
>able to distinguish, in a biased way, between different genetic
>sequences, evolution is dead. That is why the neutral gap problem is
>so serious.
In fact if you have a population of different mutations confering similar
differential reproductive success, evolution has much more flexibility in
proceeding as conditions change. I ahve the impression you seem to think that
evolution proceeds like vectors placed end to end. That is a poor
characterization.
>
>> If I have a set of mutations that
>> produce similar increases in reproductive propensity, then that set will be
>> favored over of the set of mutations which favor lower reproductive
>propensity.
>
>That's fine. You have "more" reproductively fit sequences compared to
>what came before.
No, that is not quite accurate enough. What I have is suite of different
reproductively fit sequences. Thats a far different thing, than claiming more
of one type of reproductively fit sequence.
Zach's program does not do this. It does not
>analyze the evolved sequences for increased beneficial meaning. All
>it does is analyze the evolving sequences as far as a sequence match
>to a pre-established sequence (i.e., Hamlet in this case).
My understanding of Zach's algorithm is that it works with a population of
sequences. I could be wrong. In any event, a population is not constrained to
have only one fit genotype.
>
>Have you ever read about Dawkins's "Methinks it is like a weasel"
>evolution program? Do you know what is wrong with that program as far
>as a model of biological evolution?
Among the things it lacks is that it does not deal with a populations of
expressions. Zach's methods do.
Of course the argument is that
>Dawkins never intended it to be taken as a model of biological
>evolution, but many evolutionists have done just that.
No, from my experiences, creationists assume it is intended to be taken as a
model of biological evolution, so they can have a straw man to play with. Much
better to play with that than things like multiply nested heirarchies and
embryonic teeth in baleen whales.
Many people
>actually think that Dawkins's program models real biological evolution
>for every type of cellular function.
Dawkins simple program serves to illustrate the benefits of cumulative
selection over pure random mutation. It does that quite well.
Though Dawkins's program does
>model certain types of biological evolution, like certain aspects of
>the immune system and the like, it does not model many other types of
>functional systems that do not have access to such a sequence
>template.
>
>You see, Dawkins's program was template based. It selected those
>sequences that best matched a pre-established sequence without regard
>to anything else as far as increased functional advantage is
>concerned. In real biological evolution Nature cannot read sequences,
>it can only read changes in beneficial function. There is no template
>model sequence to compare individual sequence changes with.
Again Dawkin's simulation was only designed to show the benefits of cumulative
selection. Templates in the real world are much more abstract and dynamic than
in Dawkin's simple model. Essentially Dawkins makes a simplifying assumption
(among many) that a current phenotype is the template "Me THinks Its Weasel".
You can then ask the question how many generations does it take to evolve to
this template starting from a previous, much different template..
I think one can make the argument, that for so called living fossils, the idea
that the template is the current phenotype, is approximately true.
All
>mutational changes are therefore subject to selection pressure only as
>far as they affect phenotypic function. The interesting thing is that
>many mutations do not change phenotypic function. They are neutral
>with regard to function and are therefore immune from selection
>pressure. They best that these mutations can do is drift randomly
>along in the gene pool. Such mutations may add up over time by sheer
>random luck to run across something that is actually beneficial, but
>such a random walk takes a whole lot longer than a more directly
>guided process.
Which is the point of Dawkins's demo.
>
>Zach's program does exactly the same thing that Dawkins's program did.
> It selects based on sequence comparison, not on any sort of analysis
>of improved beneficial meaning or "function".
Zach's main purpose here isn't to simulate evolutionary biology. His main
purpose is to challenge unsupported assertions made by yourself.
Second it doesn't work with generations of a single individual like Dakwins's
program. It works with a population.
Third, it doesn't work with a fixed template either. The goal of the
phrasenator is to produce phrases, and the longer the better. The purpose of
the phrase book is to decide what is beneficial. Not to provide a template. If
you disallow that, you might as well say your claims can't be tested. To call
"longer phrases the better" a template is to equivocate on the meaning of
"template". Zach is not looking for a particular long phrase, just one that is
long.
<snip>
Stuart
Dr. Stuart A. Weinstein
Ewa Beach Institute of Tectonics
"To err is human, but to really foul things up requires a creationist"
Zachriel
Let's check the current task list:
1. finish experimentation with Phrasenator
2. detail creationist arguments and how Phrasenator refutes same
3. defend Phrasenator as required on talk.origins
4. make progress on next-generation Phrasomatic!
5. have a life
Good.
Hmm. I might want to work on prioritizing that list a bit.
;-)
--
Zachriel
"Zachriel" <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com>
<snip>
>
I have just uploaded the new Phrasenation: Bible Version. The Phrase Book
consists of Genesis from the King James Bible. Also, updated the standard
Shakespearean Phrasenation, including changes resulting in significantly
improved performance. Genesis is somewhat longer than Hamlet, so please be
patient while the Phrase Book is indexed.
WARNING: Phrasenator: Bible Version is rated 'R' for nudity, violence, adult
situations, and unspeakable abominations.
--
Zachriel
Zachriel
"Zachriel" <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com>
wrote in message news:krqdnVYSHMU...@adelphia.com...
> <reposted, first didn't show, lost in the ether>
>
> "Zachriel"
> <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com>
> wrote in message news:DbqdnUHqBcS...@adelphia.com...
> <snip>
>>
>> Today, we introduce Phrasenation.
>> http://www.zachriel.com/phrasenation/
>>
>
> I have just uploaded the new Phrasenation: Bible Version. The Phrase Book
> consists of Genesis from the King James Bible. Also, updated the standard
> Shakespearean Phrasenation, including changes resulting in significantly
> improved performance. Genesis is somewhat longer than Hamlet, so please
> be
> patient while the Phrase Book is indexed.
>
> WARNING: Phrasenator: Bible Version is rated 'R' for nudity, violence,
> adult
> situations, and unspeakable abominations.
>
>
Please bear with us. We are having heavy traffic at Zachriel.com. Rest
assured that our engineers are busily manning the pumps.
Mitch Alsup
>
> WARNING: Phrasenator: Bible Version is rated 'R' for nudity, violence, adult
> situations, and unspeakable abominations.
How is this different from any other reasonable translation of the Bible?
Zachriel
"Mitch Alsup" <Mitch...@aol.com> wrote in message
news:e90782f7.04111...@posting.google.com...
It's the King James. The warning was for those who may not be familiar with
the contents of Genesis, which includes stories of incest, rape and murder.
Zachriel
"Zachriel" <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com>
wrote in message news:7qydnSZaIs4...@adelphia.com...
For reference, this is the text I used:
http://www.jambe.co.nz/bible/01.htm
However, a much better version for Bible research can be found at Crosswalk.
You can compare versions quite easily and includes related reference
material.
http://bible.crosswalk.com/
You can use any text you want for Phrasenation. The project was to create a
genetic algorithm that could be easily viewed and understood without any
specialized knowledge.
Zachriel
"Zachriel" <"http://www.zachriel.com/mutagenation/contact.asp"@giganews.com>
wrote in message news:DbqdnUHqBcS...@adelphia.com...
<snipped>
New update. Several changes to improve the system performance. Also, if you
modify the Phrase Book, the Load Phrase Book button will now activate, so
you can reload it into memory.
Zachriel
our desk somewhat and are starting to work on a new writeup of the
Phrasenator. (That's the "royal we".) As suggested by John Wilkins, we need
to make the argument from "scratch", not assuming the reader is already a
participant in the Pitmanian discussion of the past (or is that Pitmania?).
Hopefully, this will be a useful resource for the casual, but interested,
reader.
Here's a partial list of topics I'd like to cover.
* The Creationist Argument
* Description of Genetic Algorithms
* Types of Mutations-Recombinations
* Population dynamics
* Randomness and Selection
* Creationist Fallacy of Big Numbers
* How Mutagenation Works
* What Mutagenation is not (a proof of biological evolution)
These are not necessarily in order or a complete list. Any suggestions would
be welcome. I would especially appreciate links to the Creationist arguments
concerning information and complexity arguments.
--------
Re: Phrasomatic
Still working on that. Still "coming soon". Of course, it depends on what
the meaning of "soon" is. Are we talking glacial time? Or geologic time?
If you remember from Mutagenation, I found a text dictionary of about 70K+
words. Now I need a new dictionary, but it needs to include the word syntax;
noun, verb, adjective, adverb, etc. Maybe the pronunciation, too. I've
tried going through the dictionary manually, but that really is glacial.
I've found some noun-lists and verb-lists, but hardly complete. If anyone
knows where to find such a dictionary in some sort of list-format, let me
know.
Actually, the Phrasomatic is working, sort of, and has been quite a lot of
fun. I'm not ready to release it though. Soon, though.
--