i hope you all will forgive the slightly pedantic interjection of an outsider, but ark_templar brings up an interesting point about arbitrary transformations when he asks, "could you explain the concept of an encryption key whats its aim if the algo of encryption is unknown then I think (thats what I wanted to test) that the thing is pretty much unbreakable." this is a topic in "nonlinear algebra" which interests me.

in a sense, encryption is just an arbitrary mapping from a string of data (a vector) to another string of data. decryption is accomplished by guessing an inverse mapping which maps the encrypted data back to the original data. it does not require one to know exactly what algorithm was used to encrypt the data in the first place. in fact most reasonably simple keyless algorithms have the same inverse mapping, up to a variation in a couple of parameters. a good encryption algorithm has very few inverse algorithms regardless of the simplicity of the encryption itself. this is the purpose of the key. the key is an arbitrary set of steps in your encryption whose sole purpose is to reduce the number of inverses to your encryption. modifying a keyless encryption algorithm is, at best, equivalent to changing a couple of bits of the key of a keyed algorithm, and thus we can much more efficiently increase the difficulty of cracking our encrypted message by incorporating arbitrary, randomly generated keys into our algorithm than by trying to write more or different steps into it.

and tim axe, i may be mistaken, but i think that while the allies cracked the nazis' encryption due partly to stealing the machines off of sinking submarines, the allies codes remained intact through the end of the war. the fact that the wind-talkers code remained tight through the war is more a testament to human bigotry than mathematical brilliance, methinks. the polish resistance under nazi occupations used, as radio call signs, surnames such as brzeszczesciekiewiec (sp?) to assure themselves that they could not be infiltrated by nazi trojans, and to great avail. never underestimate a brilliant mind's unwillingness to extend his intellectual powers outside his own cultural bounds. to that end, a great read for ark_templar would be stephenson's cryptonomicon, which covers in great detail the world war two history of cryptography and its mathematics, be it only a novel.

and on a more extraneous note, does anybody know how to set up encrypted swap partitions on a linux box? i figure it's kind of silly to encrypt a filesystem if the key is going to sit in memory and then perhaps be written to the swap partition, whose persistence is god knows how long.

forgive the length of my wind.