[hinkley]

Let's say I have a machine that runs one cron job. It looks for work, writes a few files if it finds some. Over a three day holiday weekend, no work is queued so the machine hardly does anything. And it's on a vlan so there's little to no broadcast traffic hitting it.

Tuesday morning I ssh into the box. How much high quality entropy do you think I got from the nearly deterministic network and disc traffic on this machine? Maybe a few bits.

What I'm hoping comes of this is that we can find a better set of inputs for several/random, and that we can use more conservative estimates of entropy from the sources that are already used. Also I hope we find a way to treat intel's hardware RNG as a low quality input (currently it's a no quality input.)