Thought Provoker
Posts: 530
Joined: April 2007
|
Let me start by asking for forgiveness for the length of this post. It will take a lot to show the difference between smoke and mirrors BS and what I think it an honest attempt.
Unlike most people I think there is some room for giving the benefit of a doubt to a select few ID proponents. Specifically, Mike Gene.
Mike and I disagree on a lot of things. But I think he is sincerely trying to approach the ID question honestly. I think of him as an ID scientist even though Mike doesn't claim ID is science.
My opinion of Dembski probably matches those of you who detest the ID Movement.
Here is your chance to reward sincere effort, even if you think the effort is valueless. Listen and understand what I am presenting. For those of you who only want to blindly defend the status quo, please don't bother those willing to think for themselves. I wish you all the luck in fighting the Culture War against those who are your mirror opposites. You all deserve each other.
Mike Gene's book, The Design Matrix finally came out a month ago. I have read it. It is not bad. It is the best pro-ID book I have seen by far (wasn't that hard). Mike actually presents his arguments. They are understandable. Discussing it isn't like nailing jello to the wall. Mike builds to his conclusion in the last chapter (Chapter 10) which is a proposed methodology for inferring design. Please consider getting Mike's book and reading it for yourselves. It is available for $16.47 at Amazon
"The Design Matrix is a method by which you can score a particular feature according to four different criteria to assess and quantify the strength of a design inference."
The four criteria are…
1. Analogy - How similar is the phenomenon to something known to be designed?
2. Discontinuity - How irreducibly complex is the phenomenon?
3. Rationality - How purposeful (i.e. functional) is the phenomenon?
4. Foresight - How much front loading is involved in the phenomenon?
By necessity, this is going to be an oversimplification of Mike's process. I very much encourage people to get The Design Matrix to read how Mike explains that, while subjective, this method has similarities of scoring in the Olympics along with medical and other scientific circumstances.
Mike chooses to use a system of scoring -5, -4, …, -1, 0, +1, …, +4, +5 for each criteria. In addition to matching the duck versus rabbit theme prevalent throughout the book (-5 = "looks like a duck", +5 = "looks like a rabbit") it also allows for the balanced position being "0".
Mike's first example is to test the book, itself, for a design inference.
"Since the book is perfectly analogous with other books that are known to be designed, we can give it a Analogy score of 5. As for Discontinuity, I can safely say that there is no hurricane, volcano, beam of energy, or any other non-teleological force that can substitute for me as author. As such, the writings found within are fundamentally discontinuous with anything known to be caused by non-teleological processes found in nature. The book thus deserves a Discontinuity score of 5. When we turn to the criterion of Rationality … I'll humble myself and take a mere score of 4. … I'll give myself a Foresight score of 3."
Mike's explanation for the Foresight score is longer than I wish to type. I guess you will just have to get the book if you want to know. ;)
The resultant score for inferring the book itself is designed is 4.25, near a "strong" indicator of design.
Mike provides 6 other examples of the Design Matrix method being applied to real world situations.
"[A] pseudogene, a string of nucleotides that has no function [has a] Design Matrix score of -4.5".
"I would give the genetic code an overall Design Matrix score of 3."
In the book, Mike goes into detail as to how he arrived at these numbers.
Right away you will note Mike's approach is significantly different from Mr-I-don't-do-pathetic-details.
In 2005 Dembski wrote a paper called Specification: The Pattern That Signifies Intelligence
While Dembski has filled it with a lot of bombastic smoke and slight of hand, there is enough meat to actually understand it which I have taken the time to do. I believe there are several critiques demonstrating the weaknesses of this paper.
However, until Mike Gene's proposal, it was the only game in town. Asking IDists to give it up was like... well... asking them to repudiate God.
We have been having an interesting discussion over a Telic Thoughts comparing Dembski's method to Mike's.
When I couldn't shame any Dembski supporter into providing an example like Mike did, I supplied my own and used both methods to attempt to infer design. Here is the updated version (the one in TT had some minor errors)....
------------------------------------------------------------
Let’s play YAHTZEE!
For those unfamiliar with the game. Yahtzee is a dice game where you roll five dice in an attempt to get certain patterns. Each player gets three chances to get a pattern they need.
In the real Yahtzee game there are 13 different patterns to choose from. For this exercise, we will simplify it to four patterns. The four patterns are…
1. Yahtzee = all five dice are the some (i.e. five of a kind)
worth 50 points
2. Straight = five dice in a sequence (i.e. 1 thru 5 or 2 thru 6)
worth 40 points
3. Full House = pair and three of a kind (e.g. Two 4s and Three 6s)
worth 25 points
4. Chance = none of the above
worth the total of the dice
As in the real Yahtzee game players must score one of the patterns each turn. A score of “0” must be used if the pattern does not match. In our simple game, the maximum total score is 145 points (50 + 40 + 25 + 30). The player with the highest total score wins.
My little sister and I used to play Yahtzee alot. Being the brat she was, she would cheat if she thought she could get away with it. In this hypothetical example, I will be using the two design inference methodologies to detect cheating by my sister.
From page 21 of Dembski's paper…
X = –log2[M·N· ?S(T)·P(T|H)]
Where…
X = "context-dependent specified complexity"
M = Number of throws
N = Number of dice
T = Pattern
?S(T) = Number of T-like patterns in the Semiotic agent’s lexicon.
P(T|H) = Probability of pattern T happening according to H
The Semiotic agent’s basic lexicon is…
“Yahtzee” (five dice all the same)
“Straight” (five dice in sequence)
“Full House” (a pair and a three of a kind)
“One”
“Two”
“Three”
“Four”
“Five”
“Six”
The lexicon provides for different patterns having different probabilities yet be able to fully describe any single throw. For example the “One”, “Two”, “Three”, “Four”, “Five” throw could be simplified to “Straight” whereas “Two”, “Five”, “One”, “Six”, “Four” can’t be reduced to a simpler pattern. Both throws have equal probably but one is more specified than the other.
First Turn
The practical application is a situation where my sister and I are playing Yahtzee in the livingroom. We set up the game and I go to the kitchen to get a drink. My sister rolls the dices and yells out that she got “Four”, “Five”, “One”, “Six”, “Four”, I say “fine”. Mathematically ?S(T) is 7779 (3+6^5) since the lexicon we are using doesn’t allow for a pattern reduction in this situation.
X = –log2[M·N· ?S(T)·P(T|H)]
M = 1 (first throw)
N = 5 (five dice)
?S(T) = 7779
P(T|H) = 1/7776 (five dice have 7776 permutations)
X = –log2[ 1 • 5 • 7779 • 1 / 7776 ] = slightly negative, definitely less than 1
My sister shouts, “I’m keeping the fours” and rolls the three dice then says “I got another four, a six and a two”. I say “fine”.
X = –log2[M·N· ?S(T)·P(T|H)]
M = 2 (second throw)
N = 4 (average four dice per throw)
?S(T) = 7779
P(T|H) = 1/216 (three dice have 216 permutations)
X = –log2[ 2 • 4 • 7779 • 1 / 216 ] = negative, less than 1
My sister shouts, “I’m rolling the last two dice” and rolls the pair of dice then says “I got a pair of fives, full house!”. I say “you got lucky”.
X = –log2[M·N· ?S(T)·P(T|H)]
M = 3 (third throw)
N = 3.33 (average 3.333 dice per throw)
?S(T) = 9
P(T|H) = 1/36 (two dice have 36 permutations)
X = –log2[ 3 • 3.333 • 9 • 1 / 36 ] = negative, less than 1
According to Dembski’s methodology I should not infer that my sister cheated.
What about the Design Matrix?
Analogy – this situation isn’t similar to how my little sister cheats. She usually isn’t that subtle, however I am in the kitchen. We will score it a -2.
Discontinuity – The full house evolved over three throws. Score it a -4
Rationality – Big help to my sister’s score but not the best, Score it a +3
Foresight – Full House isn’t that hard to get, doesn’t overly predict a future need, score it a 0
Average score = -0.75 Looks like a lucky duck
Both methods infer that my sister didn’t cheat.
Second Turn
After this I take my turn and notice I forgot to put ice in my drink and return to the kitchen. My sister rolls again and yells out that she got a “Three”, “Three”, “Three”, “Three”, “Three”. I respond with “I think you cheated”. This time the lexicon allowed for pattern reduction to something the Semiotic agent (me) could recognize as a “Yahtzee”. In this case ?S(T) is 9.
X = –log2[M·N· ?S(T)·P(T|H)]
M = 4 (fourth throw)
N = 3.75 (average 3.75 dice per throw)
?S(T) = 9
P(T|H) = 6/7776 (only six of the 7776 permutations are a Yahtzee)
X = –log2[ 4 • 3.75 • 9 • 6 / 7776 ] = 3.26 = greater than 1
Dembski’s method suggests Design (cheating) should be inferred
What about the Design Matrix?
Analogy – this situation is exactly how my little sister usually cheats. We will score it a +5.
Discontinuity – First roll means scoring it a +5
Rationality – The ultimate help to my sister’s score, another +5 score
Foresight – A straight is actually harder to get than a Yahtzee, but a Yahtzee is still hard and, therefore, useful for the future. Score it a +3
Average score = +4.5 Looks like a wascally rabbit (who cheats)
Both methods infer that my sister cheated.
Third Turn
Being the nice brother I am, I let my sister get away with it and we continue playing. To my utter amazement, on her very next turn she throws five ones.
X = –log2[M·N· ?S(T)·P(T|H)]
M = 5 (fifth throw)
N = 4 (average 4 dice per throw)
?S(T) = 9
P(T|H) = 6/7776 (only six of the 7776 permutations are a Yahtzee)
X = –log2[ 5 • 4 • 9 • 6 / 7776 ] = 2.85 = greater than 1
Dembski’s method suggests Design (cheating) should be inferred
What about the Design Matrix?
Analogy – my little sister has never cheated this well before, but maybe she has practiced. We will score it a -4.
Discontinuity – First roll means scoring it a +5
Rationality – This doesn't help my sister since she already has a Yahtzee, a -5 score.
Foresight – She needs a straight, this is of no future help, another -5 score
Average score = -2.25 making it a duck
Dembski’s method infers my sister cheated.
The Design Matrix infers she did not.
What do you think?
As Fox News would say “We report, you decide”
-----------------------------------------------------------
Other than mixing "Design" with "Context-sensitive Specified Complexity" I feel I have appropriately applied Dembski's method. Please ask for clarification on either method and I will try to explain it the best I can. Like I said, I have taken the effort to understand Dembski's paper.
I don't want to get too far into trying to defend Mike's method. My point here is to point out that Mike is making an honest effort at communicating a real idea as compared to the snake oil salesman approach of Dembski.
Here is the link to the Telic Thought thread where we are discussing it. You might want to take a look at it. I think some people might be squirming over the comments.
|
