Monday, February 27, 2006

Detecting Design

On the KCFS forum, I decided to put the Design apparatus to the test. A creationist was explaining that "We do have processes and criteria in place that allow us to determine the design is real," but was being evasive about whether he could detect design in a string of bits. After I had linked to this example of Billy Dembski using the Da Vinci Code as an example of Design Detection, I posed this challenge:

Billy Dembski was able to identify design in one of these four strings of numbers. I will show you how if you will guess which was designed. At least one is a completely random string. I may or may not have designed the remaining two.

a) 1332211185
b) 2271802943
c) 4991636425
d) 4349155943
To my surprise and amusement, the creationist replied
D- final answer- 434 915 xxxx is an area code and exchange in Virginia.
Which it is. It was not, however, designed. It was generated by the R software package's random number generator, as was item "b." The first item was Dembski's number, a rearrangement of the Fibonacci sequence. The third item was a rearrangement of the square numbers (sticking with the theme).

There was amusement aplenty about the confident assertion that he had identified design. But then, after I explained the score, rather than defending the "processes and criteria in place that allow us to determine the design is real," our creationist tried to turn the challenge back:
Why should I believe you? Can you prove any of this?
So much for those processes and criteria.

The challenge was simple, the failure absolute. Next time a creationist tries to insist that there are all these great tools for detecting design, pull this test out and see how they do. The Online Encyclopedia of Integer Sequences will give you lots of raw material, as will RAND's tables of random digits. Or just choose a random or designed starting place in pi or e. Your creationist won't know the difference.

Why? Because there are a million random digits in the RAND table, so any sequence you find in pi will probably occur in the random table, as will any arrangement of any of the 100,000 integer sequences. And we're playing by creationist rules, since Billy Dembski himself used it as an example. Some sequences of numbers are "too random."