If you flip either 0% of bits or 100% of bits, the crypto is easily breakable. (In the 0% case, the ciphertext is bit-by-bit identical to the plaintext!) For numbers very close to 0% or very close to 100%, it's not a lot harder. About 50% seems the optimal number of bits to flip, provided you don't do it by a predictable pattern.
The base64 encoding would happen after encrypting, and the base64 decoding would happen before decrypting. So although it does make the ciphertext longer, it doesn't affect the required key length, which is still equal to the length of the plaintext.
My algorithm has the advantage that you could not only reasonably commit a very short key to memory, but you could also commit to memory the script necessary to decrypt it!
In reply to Re^4: crypto with core modules only
by tobyink
in thread crypto with core modules only
by morgon
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |