Probability as a SciCre Polygraph

(Originally posted by Wesley R. Elsberry on 1992/09/16 to the Evolution Echo.)

H. Morris quotes the following (with ellipses removed) from a source text:

The catfish range in length from 11 to 24 cm., with a mean of 18 cm. Preservation is excellent. In some specimens, even the skin and other soft parts, including the adipose fin, are well preserved strongly suggests that the catfish could have been transported to their site of fossilization.

As pointed out in a recent message, the quoted text is congruent with the actual passage, except that it leaves out precisely the sequence of characters that inverts the meaning of the text. Henry Morris claims this is an "error", as opposed to a deliberate misquotation.

With leading spaces, double spaces, ellipses, etc. (all non-necessary characters) removed, there is a sequence of 291 characters. The character set from which this quote can be formed is the set of capital letters (26 characters), lowercase letters (another 26 characters), the space character (1 character), the period (1 character), the comma (1 character), and the numerals (10 characters). This is a set with cardinality 26+26+1+1+1+10 = 65.

The missing sequence is:

the catfish were a resident population. It is highly improbable that

This sequence is 69 characters in length (remember the space), formed from the same base character set (you can't swap characters sets in the middle).

Using SciCre Probability, what is the "pure" chance that this neat excision of a contradictory quote would be observed (that is, that it was a simple error instead of a deliberate misquote)?

Let's see, the original quote is 291+69 = 360 characters. We need to add a null character to the original set, so that deletion can be modelled, so the character set now includes 66 characters. Hitting the "deletion" character for replacement of each of the members of the offending 69 length sequence while hitting the identical character for replacement on each of the rest of the quote gives a 1/66^360 chance of coming up with the resulting SciCre Quote by simple error. This greatly exceeds not only the Golay number (10^450), but also is far less likely an outcome than Gish's estimates of the "pure chance" derivation of a hundred amino acid protein. (This still holds for a "minimal" character set of cardinality 34 (remember the null and the space are needed) derived from the characters actually used in the 360 character sequence in question, since 34^360 is still quite huge.)

By Gish's SciCre Probability, we have "proved" that Henry Morris is an out-and-out baldfaced liar.

Let's not hear any chatter about n-grams and Shannon. Recall that I said "using SciCre probability", and definitely do not claim that the above will withstand the slightest scrutiny that uses real-world probability.

Which is, after all, the point.