This sounds almost scientific, but it isn’t. It’s an arbitrary technique. First, he speaks of permutations when he means combinations, and assumes that all var

This sounds almost scientific, but it isn’t. It’s an arbitrary technique. First, he speaks of permutations when he means combinations, and assumes that all variables are binary. Then he decides that eight test cases is “getting ridiculous.” Huh? It depends on how they are performed. No number is ridiculous a priori. What he should be doing is thinking through the technical risk and his coverage goals. It may be that all pairs coverage is worth doing, or random testing, or exhaustive testing. Or testing based on sensitivity analysis. Or testing based on a decision table. I don’t have a problem with simple heuristics, but the author does nothing to justify his personal preference.