HOUSE_OVERSIGHT_015906.jpg

216 Are the Androids Dreaming Yet?

One perfect way to encode a message is to use a one-time pad. On a sheet of paper I write a completely random set of numbers or letters – since we are going to translate numbers to letters it does not matter which. I make a copy and give it to a person I later want to send a coded message. Because I will only use these two paired sheets once it helps to make a few of them – a pad in fact. By convention, we refer to a single sheet or a whole book as a one-time pad code. Here is the one-time pad I created earlier. It is just a random sequence of letters and spaces.

kaleygnqaloiuebldlan dlkawoqyevbax gmlsosuebal

To code a message, I substitute numbers for letters as with the progressive cypher earlier again using modulo arithmetic to wrap around if I reach the letter ‘z’. I have applied my one-time pad to the hello reader message below to get ‘sfacngfvbpta’.

hello reader
sfacngfvbpta

This code is unbreakable – almost! Notice there are very few clues for anyone wanting to decode it without holding a copy of the pad. Spaces do not necessarily indicate breaks between words, and letter patterns are absent. It has only one flaw. The total number of characters and spaces could have some meaning. This is a problem because if I routinely communicated bombing targets and my message was “Bomb Bath”. You could figure out the sender was not going to bomb Bristol if the message were shorter than 11 letters and spaces. To avoid this problem, messages are extended with nonsense at beginning and end to make sure no information can be gleaned from the length. The convention is to code messages to the full length of the pad. You must never reuse a pad. Each time you code a message, rip off that page rather like a calendar. Destroy it and use the next page for the next message. At the other end, the recipient uses his copy of the pad to run the process in reverse. Decode the message by swapping each letter according to the modulo method, rip the page from the pad, and burn it. Because each key is only used once you can’t use any sort of statistical method to work out the message, making the one-time pad perfectly secure. Claude Shannon proved this in 1945 while working for Bell Corporation but, due to wartime secrecy, his proof was not published until 1948.

HOUSE_OVERSIGHT_015906