A special National Cipher Challenge for extraordinary times › Forums › Bureau of Security and Signals Intelligence Forum › A cipher is only as strong as its key
29th April 2020 at 1:58 pm #47980
A common issue of sorts in cryptography is that a cipher is only as strong as its key, a cipher can completely change the structure of the language so that it is completely non-analysable but if it only has 26 possible keys then you can just try them all and see which one gives English.
This gave me an idea and I am wondering whether anyone has either heard of such a cipher or as any ideas about how it could be implemented (mathematics, hashing etc.). The idea is to have the human who is encrypting their text first generate a group of very similar messages but only on of them actually holds the information you wish to pass on while all the rest read like they could be the correct message but are incorrect. Then the cipher would take these messages and somehow combine them all into a message where depending on what key you use it decrypts into one of these messages. This would then mean that guessing at the key would be impossible and so under certain circumstances it would be an unbreakable cipher while still having a short key! Thoughts?29th April 2020 at 2:02 pm #47981HarryKeymaster
I think you are working towards the idea of the One Time Pad. If you can generate a truly random key stream and use it just once (by, say, adding it letter by letter to the text a la the Vigenere cipher) then no-one can tell what the original message was without access to the key-stream. Any feasible decryption could be obtained by choosing a suitable key stream. Using other possible messages as the stream has the weakness that the underlying structure in the text makes it feasible to attack the cipher, and there would have to be some way for the intended receiver to determine which decryption is correct, which is itself a weakness. Either that can be determined from the cipher text, which leaves it open to detection without you knowing it has been broken, or it is a separate part of the key, and that needs distributing.
Harry29th April 2020 at 2:08 pm #47982
However Harry, the OTP requires a key the length of the entire message whereas the idea of this cipher is to have a very short key which just distinguishes which of the previously given possible messages is the correct whereas in a one time pad any possible message the length of the ciphertext could have been the plaintext, so this has less possible messages than a One Time Pad but each of possible messages could feasibly be the correct one. For instance in war time they could send the cipher with the original messages squashed into it saying that their ship is in 20 different places and without knowing the key you can’t tell which of those 20 locations is the correct one.29th April 2020 at 2:11 pm #47983HarryKeymaster
Oh, now I see. You are basically looking at using a form of codebook which is distributed. I think the problem is that other intelligence would help the enemy to distinguish between the likely locations and plan for defence. Better that they don’t even know the message is about a ship!29th April 2020 at 5:19 pm #47991
I saw this a few weeks back:
- You must be logged in to reply to this topic.