Quote:

care to vulgarise (i'm not sure is real word in english but i give it a try, so it mean explainning for stupid people :D) this ?

One-time pads are long strings of random bits - easy to make if you are careful. You keep one copy for yourself, and give one to the person you want to message. When you want to send the message you XOR the bits together: you flip every bit in your message where there is a '1' in the key, otherwise leaving the bit unchanged. For example, if your message looks like this: 10101010101010101010 and you have the key 01111000110001011011 the result is 11010010011011110001 You send this to your recipient. They do the same process: for each bit, flip it if there is a 1 in the key at that place. Now they will have a readable message. You both immediately discard your keys. Why is this safe? It is "information-theoretically secure": the encrypted message does not provide any information about the original message. The right key can turn any given string of bits into any other given string (just put in ones and zeros in the right places). From the encrypted message you could "decrypt" it into any message! This means that the only thing you know about the message is its length (and that could be padded with extra space). The problem is that you need to distribute the keys safely, prevent them from being copied, never re-use them, destroy them after use, and use truly random keys. In reality this is often tricky. Lots of cases in real life where one of these failed and enemies got an advantage. http://en.wikipedia.org/wiki/One-time_pad Here is an OTP adventure idea I came up with: "Greetings, Firewall agents. Two weeks ago, this person, "Agent Omega" of [Fill in suitable conspiracy] did an emergency farcast from Earth. As you know, normally egocasting involves setting up a secure quantum channel to exchange cryptographic keys and then using them to encrypt the ego. However, emergency farcasters cannot do this since they have to blast out the message before there is time to set up a channel. Instead they rely on an embedded one time pad. While everybody can receive a copy of the broadcasted ego, only people with a copy of the pad can decrypt it. We have a copy of Agent Omega's broadcast, but it is entirely useless to us. However, we figured out the black resleeving facility in the belt where the message was received, decoded and Agent Omega was resleeved after the broadcast. Your mission, should you choose to accept it, is to infiltrate this facility and get a copy of the current one-time pad for Agent Omega's emergency farcaster. This will ensure that next time he emergency farcasts we will get a copy of him. This message will self-destruct in five seconds."

Quote:

i'm not really sure i bind my head over the public key encruption so

This is how it works: I make a pair of keys, [Kprivate, Kpublic], using some fancy mathematics. Someone who knows Kprivate can figure out Kpublic, but knowing Kpublic will not tell you anything about what Kprivate is. You can use these keys to encrypt things, and someone with the *other* key can decrypt the message. Now, I give you Kpublic. In fact, I can tell everybody about it. If you encrypt a message with it and send it to me, I can decrypt it. But nobody else can decrypt it, since they haven't got my secret key Kprivate. So the whole world can send me messages only I can decrypt. Even better, I can sign messages by writing something like "This message is signed by Arenamontanus", encrypt it using Kprivate, attach it at the end of my normal messages and send them to anybody. The recipient can use Kpublic to decrypt the signature and see that it indeed does say that it is from me - since I am the only person with Kprivate. This is how you can authenticate your identity on the Internet or the Mesh. This is a very powerful scheme, and it can be used for a lot more devious things (like making messages that can only be decoded when 3 out of 5 people combine their keys, play poker or vote over telephone lines with no risk of cheating etc.). However, the security depends on nobody finding out my private key and that the mathematics behind the particular encryption method is correct. Quantum computers might be able to figure out Kprivate from Kpublic - this is why so many people want to build one. Eclipse Phase use: when joining, Firewall agents get a few public and private keys allowing them to authenticate themselves to Firewall and vice versa. They can check that the invitation to a meeting really is from Firewall, and Firewall can check that the person joining really is who they claim to be (assuming nobody has been hacked or gone traitor ;-) of course). The Eye is running using various shared keys: you need to have a Firewall public key and a recognized member private key to get into it. The same is true for a lot of other networks, but Firewall takes security a bit more seriously. Just imagine the danger if someone were to steal a file containing a lot of private keys... http://en.wikipedia.org/wiki/Public_key (Next obvious big topic is what quantum encryption is, but let's leave that for later)

Rhyx wrote:

A fun and safe method that my mentor taught me about that theoretically works really well

Yes, book cyphers are pretty nice. However, they can be cracked even for obscure books with enough cleverness. (Simon Singh's codebreaking challenge had a book cipher as one of the steps, and it was cracked by some people from my almost-alma mater, http://codebook.org/codebook_solution.html ) It is all about making cracking the code harder than it is worth for the enemy.