Being able to reject chance as an explanation is critical to identifying design. The way to do this is to compare the structure of an artifact against some pattern that can help us rule out chance as an explanation.
Sometimes designers can anticipate the knowledge of observers in order to craft designs which can be recognized as designs. They can structure it according to a pre-existing pattern that the supposed observer has in their inventory or some pattern that the designer thinks can be recognized by an observer.
Here I described this exercise:
“Take the coins and dice and arrange them in a way that is evidently designed.” That was my instruction to groups of college science students who voluntarily attended my extra-curricular ID classes sponsored by Campus Crusade for Christ at James Madison University (even Jason Rosenhouse dropped in a few times). Many of the students were biology and science students hoping to learn truths that are forbidden topics in their regular classes…
They would each have two boxes, and each box contained dice and coins. They were instructed to randomly shake one box and then put designs in the other box. While they did their work, I and another volunteer would leave the room or turn our backs. After the students were done building their designs, I and the volunteer would inspect each box, and tell the students which boxes we felt contained a design, and the students would tell us if we passed or failed to recognize their designs. We never failed!
Bill Dembski worked hard to rigorously define criteria for patterns which observers can use to recognize designs constructed by designers willing to communicate evidence of design. It is worth noting, such a method will not help us find patterns to detect all possible designs, but only a small space of designs, especially those designs that designers want to be discovered as designs:
Masters of stealth intent on concealing their actions may successfully evade the explanatory filter. But masters of self-promotion intent on making sure their intellectual property gets properly attributed find in the explanatory filter a ready friend.
It seems Bill is asserting the Designer of life has gone to great lengths to make his designs recognizable as designs.
But a necessary ingredient to being able to detect design is to establish the pattern is one that isn’t concocted after the fact. Ideally the pattern can ascertain some degree of intentionality.
What is an after-the-fact specification? Suppose an archer randomly shot arrows in the field, then you came along and wherever an arrow landed, you paint a bullseyes around it and declare, “wow, I demonstrated the intentionality and skill of the archer because the arrows all landed on bullseyes. The improbability of this happening is astronomical.” The painting of the bullseyes after the arrows had landed was an after-the-fact specification. The intentionality by the archer was non-existent.
What Darwinists have done is argue that all the impressive probability arguments made by ID proponents are merely painting bullseyes around arrows or making after-the-fact probability arguments. This is the line of argument if often made by Michael Shermer and indirectly by neutral evolutionists. One internet Darwinist said:
Creationists and “Intelligent Design” theorists claim that the odds of life having evolved as it has on earth is so great that it could not possibly be random. Yes, the odds are astronomical, but only if you were trying to PREDICT IN ADVANCE how life would evolve.
I countered by showing the fallacies in that criticism here.
Another form of this after-the-fact objection is to say, “one configuration of fair coins is no more probable than any other, so you can’t detect design”. In an internet debate, I said:
For example, consider if we saw 500 fair coins all heads, do we actually have to consider human subjectivity when looking at the pattern and concluding it is designed? No. Why? We can make an alternative mathematical argument that says if coins are all heads they are sufficiently inconsistent with the Binomial Distribution for randomly tossed coins, hence we can reject the chance hypothesis.
and one of my detractors countered:
if you have 500 flips of a fair coin that all come up heads, given your qualification (“fair coin”), that is outcome is perfectly consistent with fair coins,
I demonstrated the fallacy of my detractor’s claim. How I did so and the history of those debates can be found by following the links starting here. It led to many exchanges culminating in A Statistics Question for Nick Matzke.
However I should amend something I said previously:
Do we actually have to consider human subjectivity when looking at the pattern and concluding it is designed? No. Why?
What I meant to ask was, “do we have to consider level of human subjectivity that is above what is common in the sciences?” Someone determined enough could insist the way I rejected chance as an explanation for 500 fair coins heads was subjective, but to argue that, he would have to pretty much argue all of science is subjective!
Specification depends on the knowledge of subjects. Is specification subjective? Yes….
No Free Lunch
Remark 2.5.9, page 66
Critics of ID have exploited that statement (pretty much in the way I truncated it too). Thus, I prefer to demonstrate specification can be found without assuming any more subjectivity than is already present in the practice of science. By doing this, the issue of subjectivity doesn’t become an issue.
The 500 fair coin heads example was not using any more subjectivity than is already present in entire disciplines of accepted and proven math, statistics and science. In fact, most would consider the procedures I used to reject the chance hypothesis was fully objective by the standards of science. Saying otherwise made my detractors look foolish, so much so a label was coined to describe their behavior as DDS.
The challenge for ID proponents is building specifications for non-trivial design patterns, especially patterns that entail functionality. The patterns I explored at UD have been very trivial. I’ve endeavored to find increasingly more complicated design patterns that can objectively rule out chance. But the trick is being able to formally and convincingly make the case that the specifications are not after-the-fact.
Unfortunately, I feel I must now make a retraction. I used the word “postdictive” to describe after-the-fact specifications. And after some investigation last night, the word has so many meanings that I think it best the word be deprecated in ID discussions. I also mistakenly attributed the word to Bill’s writings, and I now doubt he ever used the word. Wikipedia had one entry on Postdiction
According to critics of paranormal beliefs, postdiction (or post-shadowing, retroactive clairvoyance, or prediction after the fact) is an effect of hindsight bias that explains claimed predictions of significant events, such as plane crashes and natural disasters. In religious contexts it is frequently referred to by the Latin term vaticinium ex eventu, or foretelling after the event.
and that’s not exactly the sense I was using it, though close. Still other defintions don’t capture that sense. Only a few authors use it the way I have used it over the last 10 years, and I think after talking to Eric Anderson, it’s time I stop using it. Apologies to all my faithful readers for leading you astray for 10 years.
Instead of “not postdictive”, Bill uses “detachable” to describe patterns that are not-after-the-fact or not a bullseye painted around an arrow. Bill uses the archer analogy to describe the concept of detachable, but he also defines it formally:
DEFINITION Given an event E, a pattern D (which may or may not delimit E), and a requisite precondition Σ = (H,P,I, Φ = (φ , λ )), we say D is detachable from E relative to Σ if and only if the following conditions are satisfied:
CINDE P(E|H & J) = P(E|H) for any information J generated by I, TRACT φ(D|I) < λ. Design Inference, page 145
Over the years, the rigorous math used by Bill was sometimes so large and complex, it made it easy for critics to make subtle misrepresentations and knockdown arguments Bill never made. They’d construct elaborate mathematical theatrics counting on the fact that sometimes Bill’s math is so difficult, few would know what was being said by either side anyway, and what counted was how well one could at least look like they were right.
Jason Rosenhouse used the phrase, “math mongering”. Complex math was abundant in a debate which I thought could be argued in far simpler terms with far more accessible math and definitions.
Instead, debates lasting years emerged that amounted little more to arguing over definitions and procedures, and less time was spent actually demonstrating specifications were not after-the-fact. The homochirality argument took almost two years of internet debate before the critics finally capitulated that the pattern of homochirality is not after the fact. It made me enthusiastic that we can also prove other well accepted ID patterns are not after-the-fact either.
Intuitively we know they aren’t after-the-fact specifications, but demonstrating it rigorously is not easy. We can either go the route as outlined by Bill’s definition, or try other means, or just remain content with our intuitions. I’ve opted for other means that won’t be as all-encompassing as Bill’s, but simpler even though they work only with specialized cases. But if you can use Bill’s high-powered math, more power to you!
I myself won’t even attempt to make design inferences using CSI V2. CSI V2 is a great all encompassing procedure to cover almost every possible specification under the sun, but I’ve not been successful using it. That doesn’t mean readers of this essay can’t succeed where I failed, however.
The reason I wrote this essay was connect my usage of the phrase “non postdictive” with ID literature. In the process, I decided I have to make an effort to use a new word or phrase instead of “non postdictive.”
Unfortunately, what phrase can we use to describe a specification as not being after the fact? I used to say, non postdictive specification. I could say, not-after-the-fact specification. I could say not-painting-a-bullseye-after-the-arrow-was-shot specification. Or I could use the word detachable (which the uninitiated will have no clue of). If I reverted to non-postdicitive specification, that may just lead to confusion. So far “not-after-the-fact specification” is the best I could come up with.
1,636 total views, 1 views today