Back in 1997, I posed a challenge to Dembski at the NTSE
conference to explain the information contained in a solution
to the Traveling Salesman Problem found via a genetic
algorithm. I specified a tour of 100 cities as one which
exceeded Dembski's own complexity threshold for CSI. Dembski
did not argue then that the task was too simple or that the
solution-state reached did not represent CSI. What Dembski
did argue is that the information of the solution state was
"infused" somehow by the intelligence that went into the
operating system, programming language, and the genetic
algorithm program itself. I have answered this objection with
an analysis of the specific problem I posed then. The
statement that evolutionary algorithms are "too undemanding"
ignores the fact that some applications of evolutionary
algorithms do demand quite a lot, where the solution state
exceeds Dembski's own CSI threshold. The problem I specified
in 1997 is one of those.
Dembski's most recent approach has its own problems. Dembski
says in TDI that one may put either some event/object through
his Explanatory Filter, or one may put the event that produced
the event/object of interest through his explanatory filter.
Dembski illustrates this in his new post via looking at
Dawkins' WEASEL program and its output of "METHINKS IT IS LIKE
A WEASEL". The string "METHINKS IT IS LIKE A WEASEL" is the
event/object in question, and the specific run of Dawkins'
WEASEL program is the event which produced it. But Dembski
runs into a problem here, which is that the object itself is
specified and also has measurable complexity. Thus, Dembski
turns to the run of WEASEL to analyze it. And Dembski's
analysis is merely that WEASEL's end product is determined,
thus its complexity is 0. Dembski's error here is that he is
still only evaluating the end object, and does not properly
account for all the complexity inherent in the running of
WEASEL, for each run of WEASEL has a different starting point
and intermediate states along the way to generating its
output.
Other evolutionary algorithms, though, do not even share
WEASEL's reliance upon a "distant ideal target". The
complexity of the solution reached is not tainted by the
pre-existing representation of that solution within the
evolutionary algorithm. To assert that those instances of
evolutionary algorithms cannot produce "actual" CSI because
they necessarily produce CSI is question-begging at its most
basic.
Dembski wants us to use the evidence of biological phenomena
to conclude that life was designed, or that certain features
of living systems were designed. (See his First Things
article from October 1998.) In those cases, we do not have
definitive evidence that shows what kind of process produced
the systems in question. Thus, what gets fed to Dembski's
Explanatory Filter is by necessity the produced object alone.
The level of specified complexity inherent in that object is
our guide to whether one must conclude regularity, chance, or
design. We cannot feed the process that produced the object
to Dembski's Explanatory Filter without presupposing the
answer, and thus begging the question.
Feeding an object into Dembski's Explanatory Filter determines
whether the object has the attribute of high probability,
intermediate or low probability without specification, or
"complexity-specification". Dembski's "reiterations" post now
implies that while feeding an object into his Explanatory
Filter may find "complexity-specification" in that object,
feeding the event that produced the object may find only
regularity, and thus only "apparent CSI".
But Dembski previously claimed that his
"complexity-specification" was a completely reliable indicator
of the action of an intelligent agent back in the "First
Things" article. His "reiterations" post stance completely
obviates that claim. If the determination of "actual CSI" or
"apparent CSI" requires the evidence of what sort of process
produced the object in question, then finding that an object
itself has CSI is necessarily ambiguous and uninformative on
the issue of whether it was produced by an intelligent agent
or an unintelligent natural process.
My review of TDI showed that natural selection shared the same
triad of attributes that Dembski claimed for intelligent
agents alone. It appears that Dembski must concur with me,
given his recent post and its distinction between "actual CSI"
and "apparent CSI".
Wesley