When attempting cryptanalysis against a cipher not equivalent to a one time pad, there are ways to program a computer to detect a plausible ciphertext, e.g., if the cryptanalyst knows the language of the plaintext and its character encoding, then the computer can be programmed to analyse for such characteristics. With sufficient ciphertext, brute force will then obtain the one most likely plaintext.Quote:

Originally Posted bykryptkat

But with a one time pad, there is no one most likely plaintext, regardless of the amount of ciphertext available. With knowledge of the language of the plaintext and its character encoding, the cryptanalyst can only obtain all such possible plaintexts of the corresponding length that have those characteristics, and all are equally likely.

Yes, you do need some external knowledge. But the amount of external knowledge required to work out the plaintext encrypted with a one time pad is the same amount of external knowledge required to work out the plaintext without the ciphertext. Since the attacker would know the plaintext even without cryptanalysis, one can then reason that the cipher must be unbreakable: to "break" it, one must know the plaintext, yet that is the very thing that cryptanalysis sets out to obtain from the ciphertext.Quote:

Originally Posted bykryptkat

Frankly, there is no point in continuing this. Accept that you are wrong. If you refuse, then do as I say: write a paper and publish it. Prove the experts wrong. Even if they do not accept you now, maybe a later generation of experts will, and you will leave your mark in history, just like Claude Shannon.