God and the Explanatory Filter
Posted January 20, 2003
Dembski's Explanatory Filter (EF) is intended to distinguish regularity, chance and design. Dembski claims that the filter does not register false positives for design, though it is susceptible to false negatives. When we apply the EF to God we get some interesting results.
The filter has three stages, first a check for regularity, second a check for chance and third a check for design.
1 First Stage - Check for Regularity
If something is due to regularity then it is the outcome of the working of the laws of the universe. Since God is the creator of the universe it is not possible for God to be due to regularity. I am sure that all believers would agree that God is not a product of the laws of the universe.
2 Second Stage - Check for Chance
For something to be due to chance the probability of it happening by chance must be greater than Dembski's Universal Probability Bound (UPB). This is 1 in 2 ^ 500, equivalent to 1 in 10 ^ 150, or 10 ^-150. This equates to 500 bits of information. Evaluating the probability of God is not easy. There are two possibilities: God is simple or God is complex.
2.1 God is Simple
If God is simple, then God does not contain a lot of information. Low information corresponds to high probability, so a simple God would be more probable than the UPB. In this case the EF would assign God to chance. I do not think that there are many believers who would agree that the existence of God is due to chance alone. This incorrect result might indicate a problem with the assumption that God is simple or else it might show a problem with the Explanatory Filter.
2.2 God is Complex
God is complex, but how complex relative to the UPB? This question is dependent on how much information God contains. God is omniscient, knowing everything; a very large amount of information indeed. However we can put a lower limit on the information contained in God. Given that God wrote the Bible, the minimum amount of information is the amount contained in the Bible. God knows more than is in the Bible; knowing less is not possible.
Looking at one of my Bibles, I find that it has about 30 characters per line, 56 lines in a column and two columns per page. The Old Testament contains 840 pages and the New Testament 240 pages. A total of 1080 pages. This is 30 x 56 x 2 x 1080 = 3628800 characters. For simplicity let us take the number of possible characters as 30, 26 letters plus space and some punctuation. Therefore by the standard probability argument the likelihood of the Bible having arisen by chance is 1 in 30 ^ 3628800. Remember that this is an upper bound, God is less probable than this because he contains more information than is in the Bible.
Working out the numbers, 1 in 30 ^ 3628800 is a probability of 1.8 x 10 ^ -3628942. This is less than the UPB of 10 ^ -150 with a good margin for error. Hence the EF does not assign God to chance at this stage.
This is a better result. God is not due to chance, which is in agreement with the opinion of all believers. I will proceed on the assumption that God is complex.
3 Third Stage - Check for Design
3.1 Look for a Specification
The third stage starts by looking for an independent specification. God certainly has a specification, scripture. Given that God is specified then God is again not a result of chance: the EF assigns low probability non-specified events to chance. Again this is a good result indicating that God is indeed specified. Had God not been specified then the EF would have indicated chance which we have already rejected in 2.1.
3.2 Design is Detected
At the third stage the EF says that if something is both complex and specified then it is due to design. God is both complex, by 2.2, and specified, by 3.1. Hence the EF says that God is due to design. Remember also that Dembski claims that the EF does not show false positives for design so this result is supposed to be reliable. Of course the EF says nothing about the nature of the designer, it merely asserts the presence of intelligent design.
I think all believers would agree that God did not arise from either regularity or chance. However by Dembski's definition of design only regularity, chance and design are allowed -- Dembski defines design as everything which is neither regularity nor chance. With this definition of design it is inevitable that the Explanatory Filter decides that God is designed.
Given the wider aims of the ID movement it is amusing to see that Dembski's Explanatory Filter appears to give support to the atheist argument that God was designed by humans. I am more inclined to think that either the EF is flawed or that this is an example of a false positive for the EF.
Originally published at TalkOrigins.
rossum is the nom-de-net of Martin Ross.