[Redcloak]

Someone has already suggested a similar approach, though a bit more high tech. What you're describing is a series of "one time pads"; keys that are used only once and then tossed. These are great unless someone compromises the book of code sheets. A researcher recently suggested using a continuous stream of data as the source for the one time pads. Users would simply pick a particular moment in time to start collecting data from the stream to generate the keys. When they're done, they discard the data. The next time the want to encrypt data, they go back to the stream and collect more data for a fresh set of keys. So long at the agreed upon time to start collecting data is kept secure, the given message is secure; even though the data stream is being broadcast to the general public. (Except, of course, from brute force attacks. But all codes are eventually suceptible to brute force attacks.)

[Wurlitzer]

"But all codes are eventually suceptible to brute force attacks.) "

You may be right Redcloak but to my knowledge, the method I described has not been broken without one of the parties giving up the book.

[Britton J Wingfield]

I believe that the NSA uses white noise from a radio telescope as a one-time key. A 650-meg CD of random electromagnetic crap from the sun makes for a hell of a secure key.

[altair]

But all codes are eventually suceptible to brute force attacks.

You mean ciphers not codes. The point of cryptography is to make the expense of breaking the cipher greater than the amount of time the data has to be kept secret. For example the message "The suicide bombing starts in two hours Abdul" doesn't need to be kept secret as would a message planning something six months in the future.