2 ms·
From a practical point having sources of randomness/entropy isn't very useful. A CSPRNG will do a great job generating random numbers. It will create secure Tls
by matt_wulfeck 11y ago
From a practical point having sources of randomness/entropy isn't very useful. A CSPRNG will do a great job generating random numbers. It will create secure Tls connections and a ssh_key that's never been seen before. However it's Achilles heel is that it MUST be seeded with random numbers. That's hard to do in a predictable state, such as a headless vm.
If the seeds no good it can't establish a secure Tls connection to fetch random data (ala Ubuntu pollinate service).
The next thing to push crypto security forward is easy, efficient, and trustworthy ways to generate this seed.
- tptacek 11y agoYeah, you're missing the point a little. If you want to generate and standardize a random curve (like, in this paper, the "random" version of Curve25519, which doesn't rely on specific magic parameters carefully chosen for speed), you have a big problem: even though your CSPRNG might be perfectly fine, nobody is going to believe that you actually used it, because you can't prove that you did. Hence: lotteries.