Random numbers

[Kent Borg]

Sandy Harris wrote:
> http://www.av8n.com/turbid/paper/turbid.htm
> If you have serious randomness needs and your board does
> not have an RNG built-in, I'd say Turbid [using a sound card] was the obvious
> thing to use.

Sampling the real world as a source of randomness is a good idea to 
consider.

But if you have serious randomness needs, stop to carefully analyze your 
randomness needs first.

Do you need random numbers particularly quickly? (/dev/urandom is pretty 
slow, there are faster ways.)

Do you need particularly high quality random numbers? If so, what is 
your measure of "quality"?

Are you up against an active foe who might try to degrade the 
randomness? (The crook who wants to know what keno number to play next? 
The NSA the might try to decode the diplomatic cable with weeks to work 
on it? Or the secure phone call with seconds to work on it? The grad 
student who wants to fake the data?)

How you create random numbers has a lot to do with what they are for, 
and what larger system they are part of.


-kb