Their method of measuring "distances" is very complex.

	Two five letter words typically require the measurement of over 6 million distances
	The results are very sensitive to small changes in the method.
	If one uses the mathematically natural square root in a distance, their results would be worse by a factor of 92
	Dropping obsolete forms of 15 and 16 decrease probability by a factor of 5
	When Haralick redid the method (but used what I believe is flawed data), the probability dropped from 1 in 62,500 to 1 in 400