4 ms·

Judging by the headline, it sounded like they tried to implement a one-time pad, but had only heard of them by rough description.

jacques_chester 14y ago
That was my thought too. In theory, OTPs are nigh uncrackable. In practice, they suck. 1. The XOR text needs to be as long as the plaintext. 2. The XOR text needs to be truly random. 3. You have to distribute the XOR test somehow. Remember all those spy novels where they burn the codebooks? Yeah.
- derleth 14y ago
  > In theory, OTPs are nigh uncrackable. One-time pads are mathematically proven to be uncrackable if used correctly. This is because, for any given cryptext, every possible plaintext of the same length is equally probable. Using a one-time pad correctly means only using a given pad once, the 'pad' being the key material. If you ever reuse the same key, you are no longer using a one-time pad and the above proof no longer applies. And if your pad is 13 bytes which you keep reusing over and over again, you're a bleeding idiot.
- cbsmith 14y ago
  It very much depends on the use case. If I was a CIA director having an affair with a biographer, an OTP would seem quite helpful. During our liaisons, we could exchange thumb drives with several GB of randomly generated data from any of a number of reasonably random entropy sources (hardware PRNG works pretty well by itself, but you can throw in EGD and other sources for greater confidence) with some software whitening and the occasional bit of chaff plaintext for good measure. At each exchange destroy any previous pads (and yes, one would have to go to some lengths to ensure destruction without interception, but tossing them in to an incinerator would probably work pretty well). Now I have a means to communicate over the internet while apart from my paramour without any concerns about the content of the messages being decoded while in transit (compromises at the source or destination are obviously still in play). Not every crypto problem involves establishing a secure connection party selected ad-hoc without a higher-bandwidth secure channel.
billsix 14y ago
One time pad requires the key to be as big as the plaintext. This looks like a weak variant of the Vigenere cipher.

jacques_chester 14y ago

That was my thought too. In theory, OTPs are nigh uncrackable. In practice, they suck. 1. The XOR text needs to be as long as the plaintext. 2. The XOR text needs to be truly random. 3. You have to distribute the XOR test somehow. Remember all those spy novels where they burn the codebooks? Yeah.

derleth 14y ago
> In theory, OTPs are nigh uncrackable. One-time pads are mathematically proven to be uncrackable if used correctly. This is because, for any given cryptext, every possible plaintext of the same length is equally probable. Using a one-time pad correctly means only using a given pad once, the 'pad' being the key material. If you ever reuse the same key, you are no longer using a one-time pad and the above proof no longer applies. And if your pad is 13 bytes which you keep reusing over and over again, you're a bleeding idiot.
cbsmith 14y ago
It very much depends on the use case. If I was a CIA director having an affair with a biographer, an OTP would seem quite helpful. During our liaisons, we could exchange thumb drives with several GB of randomly generated data from any of a number of reasonably random entropy sources (hardware PRNG works pretty well by itself, but you can throw in EGD and other sources for greater confidence) with some software whitening and the occasional bit of chaff plaintext for good measure. At each exchange destroy any previous pads (and yes, one would have to go to some lengths to ensure destruction without interception, but tossing them in to an incinerator would probably work pretty well). Now I have a means to communicate over the internet while apart from my paramour without any concerns about the content of the messages being decoded while in transit (compromises at the source or destination are obviously still in play). Not every crypto problem involves establishing a secure connection party selected ad-hoc without a higher-bandwidth secure channel.

billsix 14y ago

One time pad requires the key to be as big as the plaintext. This looks like a weak variant of the Vigenere cipher.