Re: Re: Unbreakable crypto

[thor]

One time pads are the only information theoretically secure cipher. Of course, the devil's in the details. The key has to be truly random. There is the issue of key exchange. But, once you meet these requirements, there is no way to break a one time pad. All possible decryptions with the same length as the message are equally likely, so how do you know that you know that a given 6-letter string decrypts to "rob_au" vs "chester"?

[Roy Johnson]

Because "chester" is 7 letters?

[thor]

If you want to be a stickler for the "details", I suppose you're right. But what you didn't see was there was a NUL at the end of "rob_au", thus making it seven characters. ;)

[Roy Johnson]

And "6-letter string" was the encoded form of "7-letter string"? ;-P

[hardburn]

All possible decryptions with the same length as the message are equally likely

This is one property of OTPs that I think is often overlooked for practical applications. This could be used for plausible deniability reasons. For instance, say you took the source code to a DeCSS program and a letter to your Grandma, which are both exactly the same length. You then generate a random string of characters, also of the same length. There exist two seperate OTP keys for that character stream, one of which will decrypt the stream into the DeCSS source and the other into the letter. Generating these keys is as simple as XOR'ing the stream against the document. If you get caught, you can give the authorities the key for the letter, and they thus have no evidiance. You can then walk away and give your freinds the key for the source code.

[Anonymous Monk]

If you get caught, you can give the authorities the key for the letter, and they thus have no evidiance.

Until they knock on your door and politely ask you to hand over all your computer equipment.

[hardburn]

That's what thermite is for.