Eviction and specification

So I have finally been evicted from an Internet discussion. I knew I was bound to get evicted from Uncommon Descent at some stage - although I had no idea what the reason would be. You have to be extremely respectful and cautious to remain on that blog if you are an ID opponent. I am impressed by those who manage to hang in there, such as Chris Hyland.

Anyone who is interested in the Design Inference should read ttp://www.designinference.com/documents/2005.06.Specification.pdf. Dembski makes it quite clear that it supersedes all previous explanations of the design inference. Look for the following issues:

* Faulty explanation of classical hypothesis testing. In fact all hypothesis testing requires a clearly defined alternative hypothesis. Without that it is impossible, for example, to justify the difference between one-tailed and two-tailed testing or confidence intervals.

* Sudden leaps from simplicity defined in terms of complexity theory in well-defined spaces such as bit strings to simplicity defined in terms of number of concepts to simplicity defined in terms of number of words. (And then you get the problem that you can make up a single word for any outcome)

* Subtle change from specification as "all events simpler than the observed outcome" to "all events simpler and less probable than the observed outcome". This seems to be just slipped in without explanation or comment.

* Total lack of justification for definition of specification.

There are a host of other problems - but these will do for a start.

SETI and lasers

I have been participating in a discussion on Uncommon Descent about a possible new approach to SETI (where they would look for extraordinarily bright nanosecond laser pulses) and whether it is an example of Dembski's explanatory filter. Unfortunately Dave Scott asked me to leave the blog before I had got to the meat of what I wanted to say so I will continue it here.

I am struggling to relate this example to Dembski's most recent paper on specification http://www.designinference.com/documents/2005.06.Specification.pdf - which he says supersedes previous descriptions of specification such as No Free Lunch. If the design inference applies then presumably the phenomenon - nanosecond laser pulse - would need to show specified complexity. In this paper Dembski defines "specified" in terms of the simplicity with which the phenomenon can be described. Essentially the specification of an outcome is the outcome in question and all other outcomes that are at least as simple and less probable. There is no justification in the paper for this definition of specification, but that is not my concern. I am just trying to understand how on earth you apply this concept to a laser pulse.

In Dembski's examples he mainly concentrates on outcomes which are in a well-defined domain e.g. the Royal Flush is in the domain of poker hands. So you can vaguely get the idea that there are some poker hands that harder to describe and others that are as simple or simpler because they conform to a pattern. I think there are problems even in the case of poker hand - but how simple is the description of a nanosecond laser pulse and what is the domain we are operating over? All possible observations? All possible optical observations? All possible laser observations? There seems to be no logical basis for the specification.

One could argue that nanosecond laser pulse is very simple because it only requires three words to describe it. Dembski appears to use roughly this definition of simplicity in the paper. But any phenomenon can be described in one word by simply making up a new word that refers to just that phenomenon. So simplicity must be something to do with a description that uses prior concepts. The word LASER is of course a recent one made up specifically to refer to a new phenomenon. So really any description should at least include a full description of what a laser is which appears to make this rather a complicated phenomenon.

I believe this example shows that the paper fails to provide a useful definition of specification. I have a second and deeper concern. The alternative to the explanatory filter is a straightforward comparison of hypotheses. Which explains the observed phenomenon better? A natural cause or one including design. In the aborted discussion I had got as far as establishing that even a committed IDist would not deduce design for just any observation that had no known source. So there had to be something peculiar about this one that makes it a suitable candidate. The answer, from Dave Scott at least, was that he could not conceive of a natural process that would create such a pulse. However, we have come across many natural processes in the universe that were inconceivable to virtually everyone before they were discovered - the whole of quantum mechanics is a good candidate. So this really is a non-answer. The thing that really convinces us that such a phenomenon is very likely to come from a designed source is the fact that we as humans have created such a thing. A designed solution would have been equally inconceivable before the invention of the laser. It is the plausability of a designed solution, knowing the detail of how it can be done and what kind of intelligence is required to do it - which makes this a convincing test for alien civilisations. And this is not an example of the explanatory filter.