Be that as it may, we can be skeptical of Dembski's design
argument and still find much of value in his paper. For
example, suppose the following:
1) Dembski provides a definition of CSI that allows quantification.
2) He gives a really good argument that CSI is a good measure
of biological information.
3) He is able to develop the Law of Conservation of Information
and show by numerous examples that the Law is obeyed.
This is a phenomenal achievement even if he never says a word
about design.
Below is a lengthy excerpt from Dembski's paper starting at the
second paragraph. I am going to leave the quote intact for
ease of reading and then follow with my comments
====Begin Bill D===========================================
Let us then begin with information. What is information?
The fundamental intuition underlying information is not, as
is commonly thought, the transmission of signals across a
communication channel, but rather, the ruling out of
possibilities. To be sure, when signals are transmitted
across a communication channel, invariably a set of
possibilities is ruled out, namely, those signals which
were not transmitted. But to acquire information remains
fundamentally a matter of ruling out possibilities, whether
these possibilities comprise signals across a communication
channel or take some other form. As Robert Stalnaker
(1984, p. 85) puts it, ÒTo understand the information
conveyed in a communication is to know what possibilities
would be excluded by its truth.Ó Information in the first
instance presupposes not some medium of communication, but
contingency. For there to be information, there must be a
multiplicity of distinct possibilities any one of which might
happen. When one of these possibilities does happen and the
others are ruled out, information becomes actualized. Indeed,
information in its most general sense can be defined as the
actualization of one possibility to the exclusion of others.
Complex Information
This definition of information is highly abstract
and by itself of little use to biology and science
more generally. To render information a useful concept
for science we need to do two things: first, provide
a means for measuring information; second, introduce
a crucial distinctionÑthe distinction between specified
and unspecified information. First, let us consider how
to measure information. In measuring information it is
not enough to count the number of possibilities that
were ruled out, and offer this number as the relevant
measure of information. The problem is that this simple
enumeration of excluded possibilities tells us nothing
about how the possibilities under consideration were
individuated in the first place. Consider, for instance,
the following individuation of poker hands:
(i) A royal flush.
(ii) Everything else.
To learn that something other than a royal flush was dealt
(i.e., possibility (ii)) is clearly to acquire less
information than to learn that a royal flush was dealt
(i.e., possibility (i)). Yet if our measure of information
is simply an enumeration of excluded possibilities, then the
same numerical value must be assigned in both instances since
in both instances a single possibility is excluded. It follows,
therefore, that how we measure information needs to be independent
of whatever procedure is used to individuate the possibilities
under consideration. And the way to do this is not simply to
count possibilities, but to assign probabilities to these
possibilities. For a thoroughly shuffled deck of cards, the
probability of being dealt a royal flush (i.e., possibility (i))
is approximately .000002 whereas the probability of being dealt
anything other than a royal flush (i.e., possibility (ii)) is
approximately .999998. Probabilities by themselves, however,
are not information measures. Although probabilities properly
distinguish possibilities according to the information they
contain, nonetheless probabilities remain an inconvenient way
measuring information. There are two reasons for this. First,
the scaling and directionality of the numbers assigned by
probabilities needs to be recalibrated. We are clearly acquiring
more information when we learn someone was dealt a royal flush
than when we learn someone wasnÕt dealt a royal flush. And yet
the probability of being dealt a royal flush (i.e., .000002) is
minuscule compared to the probability of being dealt something
other than a royal flush (i.e., .999998). Smaller probabilities
signify more information, not less.
====end===================================================
Now I'm going to go back through and insert my comments.
Quotes from Dembski have a "#" at the beginning of each
line. Yes, I know this makes for a rather long post but
I think its helpful to read Dembski first without the
chops, especially since this material has not been
previously posted.
#Let us then begin with information. What is information?
#The fundamental intuition underlying information is not, as
#is commonly thought, the transmission of signals across a
#communication channel, but rather, the ruling out of
#possibilities.
Here I strongly disagree. The fundamental intuition underlying
information is, IMHO, descriptive length. How much information
is there in some object X? How long does it take you to describe
it to someone?
For clarification, I think Dembski is discussing here information
in the sense of classical information theory. Here the channel
is very important. It's also important not to get caught up in
thinking about channels in terms of physical cables or wires.
The basic idea is that information normally carries with it
some idea of transmission. There is some source of information
and there is some receiver of information.
Algorithmic information theory AIT (roughly descriptive length),
while similar in some ways to classical (Shannon), is also
different in many ways. Most importantly, there is no need
to discuss ensembles of messages, probability distributions,
channels etc. These ideas can be used in conjunction with
AIT, but they aren't necessary. The algorithmic information
content (AIC) is independent of the process used to construct
a message (stochastic, deterministic etc.).
For this reason, AIC is an intrinsic measure, i.e. it is
determined from the "structure" of a message irrespective
of how the message came about.
#To be sure, when signals are transmitted
#across a communication channel, invariably a set of
#possibilities is ruled out, namely, those signals which
#were not transmitted. But to acquire information remains
#fundamentally a matter of ruling out possibilities, whether
#these possibilities comprise signals across a communication
#channel or take some other form. As Robert Stalnaker
#(1984, p. 85) puts it, "To understand the information
#conveyed in a communication is to know what possibilities
#would be excluded by its truth."
This is a really nice quote and all that (from a book
entitled <Inquiry>) but has little if anything to do
with a technical or scientific understanding of information.
Let's recall my item (2) above. Consider the information
passed from DNA to protein. What on earth do understanding
and truth have to do with anything? What does a protein
understand about excluded possibilities or truth? Well, I
suppose one could say this is a metaphor with excluded
possibilities being nonfunctional proteins and truth
being functional proteins. The protein may not understand
any of this but we, as observers, do. So what? The message
wasn't sent to us!, we're just eavesdropping. More
importantly, it still requires information to specify a
non-functional protein and the genetic information
processing system has the capacity of processing non-functional
messages just as an engineering communication channel can
transmit pornography and meaningless junk.
#Information in the first
#instance presupposes not some medium of communication, but
#contingency. For there to be information, there must be a
#multiplicity of distinct possibilities any one of which might
#happen. When one of these possibilities does happen and the
#others are ruled out, information becomes actualized. Indeed,
#information in its most general sense can be defined as the
#actualization of one possibility to the exclusion of others.
#
#Complex Information
#
#This definition of information is highly abstract
#and by itself of little use to biology and science
#more generally. To render information a useful concept
#for science we need to do two things: first, provide
#a means for measuring information;
Amen to this. But let's not forget that information measures
such as Shannon entropy and AIC can be measured. What's
wrong with these? Well, I'm guessing that what's wrong
with them is that they cannot distinguish specified from
unspecified information. That's ok by me. It may be possible
to think of some other way of distinguishing them, provided
we know what they look like ;-).
#second, introduce
#a crucial distinctionÑthe distinction between specified
#and unspecified information. First, let us consider how
#to measure information. In measuring information it is
#not enough to count the number of possibilities that
#were ruled out, and offer this number as the relevant
#measure of information. The problem is that this simple
#enumeration of excluded possibilities tells us nothing
#about how the possibilities under consideration were
#individuated in the first place. Consider, for instance,
#the following individuation of poker hands:
#
#(i) A royal flush.
#
#(ii) Everything else.
#
#To learn that something other than a royal flush was dealt
#(i.e., possibility (ii)) is clearly to acquire less
#information than to learn that a royal flush was dealt
#(i.e., possibility (i)).
Sorry, but this is not clear to me at all. Suppose that
you are forced to draw five cards and are told that if
you draw anything other than a royal flush you are going
to be executed tomorrow. Just before you draw you notice
a little twinkle in the eye of your potential executioner.
Ah, he's playing a ghastly joke you think to yourself.
If I draw the royal flush he will execute me anyway. Or
perhaps he does not intend to execute me at all and just
wants to see me squirm. Or perhaps he will execute me no
matter what I draw. So, no matter what I draw, I won't
know the truth of the matter until tomorrow. If I'm
brought out before the firing squad, then I'll know. Well,
maybe not. Perhaps the guns will have blanks in them ...
I am trying to illustrate primarily two points with this
story. First, the information Dembski is talking about
with this example is not intrinsic to the cards, in fact
it really has little to do with the cards. The information
has to do with some type of external agreement as to what
the cards mean. Second, I wanted to illustrate the type
of convolutions one can get into if you try to attach truth
and meaning to the concept of information.
#Yet if our measure of information
#is simply an enumeration of excluded possibilities, then the
#same numerical value must be assigned in both instances since
#in both instances a single possibility is excluded. It follows,
#therefore, that how we measure information needs to be independent
#of whatever procedure is used to individuate the possibilities
#under consideration.
It should also be independent of subjective qualities like truth,
value or meaning.
#And the way to do this is not simply to
#count possibilities, but to assign probabilities to these
#possibilities. For a thoroughly shuffled deck of cards, the
#probability of being dealt a royal flush (i.e., possibility (i))
#is approximately .000002 whereas the probability of being dealt
#anything other than a royal flush (i.e., possibility (ii)) is
#approximately .999998. Probabilities by themselves, however,
#are not information measures. Although probabilities properly
#distinguish possibilities according to the information they
#contain, nonetheless probabilities remain an inconvenient way
#measuring information. There are two reasons for this. First,
#the scaling and directionality of the numbers assigned by
#probabilities needs to be recalibrated. We are clearly acquiring
#more information when we learn someone was dealt a royal flush
#than when we learn someone wasnÕt dealt a royal flush. And yet
#the probability of being dealt a royal flush (i.e., .000002) is
#minuscule compared to the probability of being dealt something
#other than a royal flush (i.e., .999998). Smaller probabilities
#signify more information, not less.
This confused me. I had assumed that the "something else" was
some specific something else. Anyway, lets just note that
Dembski's something else, with probability .999998, also
contains things like flushes, straights, straight flushes,
four aces, four kings. And, if jokers are wild, .... ;-).
Brian Harper
Associate Professor
Applied Mechanics
The Ohio State University
"Should I refuse a good dinner simply because I
do not understand the process of digestion?"
-- Oliver Heaviside