"The real work in an attack, at least an attack against a well-designed cipher, is modifying the attack technique so that it works. Knudsen's papers are an excellent example of this; he is a master at making an attack work where others have failed. Differentials work where characteristics don't. Truncated differentials work where normal differentials don't. Even this year's exciting find, impossible differentials, are simply another way at looking at a differential attack. A cryptanalyst with a "menu" would have never found any of those attacks, and would have broken far fewer ciphers."