I really can't find a nice summary definition of what the are meaning by "Detachable specification", but supposedly this is why weather is complex but not designed.
However, I could be wrong about my interpretation of it - anyone have more insigh?
It's not a term I had heard before, so I had to look it up. It refers to "detaching" an event from its surroundings. I don't know if it's a made up Dembski term, or if philosophers use the term. That's one of Dembski's problems. His first book, "The Design Inference" was actually his philosophy PhD dissertation. Philosophers love new terms and bandying them about. Clarity of speech is not exactly rewarded in those circles. So, what is it?
Let's go back to those poker hands. Suppose I deal 5H 5C 3D JH QS. That hand is no more nor less likely to come up than any other hand. So would you say that it is a totally unremarkable hand, or an incredibly unlikely hand? In most circumstances, you would say that it is unremarkable, and you would keep the pair of fives and hope for the best. But suppose this was the third time in this game that you had drawn exactly the same hand? Suddenly that unremarkable hand becomes a remarkable hand. Those previous hands "specify" a hand that should be considered remarkable. The knowledge of the previous hands allows you to "detach" those apparently random cards from their surroundings, and recognize them as something significant. And if you recall that each of the two prior times, you drew your third five, but the other guy had a flush, you would probably figure out the significance of the hands, and try to find out how he is stacking the deck. In other words, you would say that the cards fell that way by design.
The whole subject relates to quantifying complexity. How do you say that the DNA sequence that specifies a bacterial flagellum is significant? In general, for anything you want to quantify complexity for, you have to decide how many ways there are to rearrange the parts in order to meet a given specification, and how many total ways there are to rearrange the parts. Then you have to apply a formula I don't remember.
That's why it's proper to treat the parts of the flagellum as a discrete combinatorial object. Remember? That's how you quantify its complexity.
But, now, suppose that you have an existing object, with no a priori specification. You just have a thing. The DNA looks like just long strings of letters. Every string is like every other string. The only way you can make any judgement about the complexity of DNA, or parts of an organism is to first find a specification that allows you to detach it from its surroundings. I think, and here I'm a bit fuzzy, Dembski uses functionality as his "detachable specification". In order to analyze the complexity of a flagellum, you would have to make an estimate of all possible ways to arrange those parts in such a way that they would perform the flagellum function, which I think he would say is motility.
So that is what he is trying to do. He's trying to get an accurate measure of complexity. Is he doing it well? I doubt it. First, he is rather biased. Second, it is almost impossible to make such estimates in a meaningful way. In poker, we know the ways in which cards are dealt, so we can make exact measurements of the combinations in which the cards can be dealt. In biology, we don't know that. Exactly how many parts are there in a flagellum anyway? So, while his estimates are probably wrong and might even be meaningless, it's a legitimate attempt. It isn't some weird Dembski thing that no scientist has ever heard of. It's just not clear that his application is correct.
And about the weather. Why isn't it as complex as life? Well, I hesitate to say that it isn't, but it seems to me that in most cases, all the weather pretty much looks like weather. If you could find a meaningful pattern that allowed you to "detach" random looking weather from the background noise, then you could determine that some weather was more complex than other weather, and also compare it to living systems.