Hello everyone. I had an idea for an encryption system based on irrational numbers a while ago and wrote down a rough specification. It is basically like this:

1. Decide on an irrational number to define the encryption, e.g. pie, e or an irrational n root of a number. It must be possible to construct an algorhythm that can compute the digits of this number with no limit.

2. The input file can be any sequence of bytes, i.e. this system is designed to work with a 256 character set.

3. Use the algorhythm (the key) to produce a series of hex digits twice as long as the sequence of bytes in the file. Then perturb each byte in the file with the corresponding 2 hex digits generated by the key. You now have the encrypted file.

4. Decrypting the file requires the algorhythm used and the parameters that were fed to it. The bytes that contain this information are slotted into the encrypted file in one byte chunks in a series of random locations. The random number sequence used is a one time pad shared between the two parties by non - computer network means.

So, what I was wondering is: does this system sound secure or insecure. Would the use of an indefinetely long number sequence generated by deterministic means be an strength or a weakness? Any comments appriciated.


You could say you have an irrational number, but what you really have is a pseudo-random sequence of bytes, that you're using to xor to the bytes of your plaintext. This is called a synchronous stream cipher.

And if you're capable of sending beforehand a one time pad that tells where the algorithm's parameters are stored in the text, why not just send the algorithm's parameters as the one time pad, instead of putting them in the text?