> A stream of n bits from /dev/random should not be repeatable with any
> frequency greater than 1 in 2^n, but it definately should be uniformly
> distributed. Otherwise the numbers aren't really random, and therefore
> they aren't useful for any purpose I can think of -- certainly not monte
> carlo integration or anything else that uses "random" numbers, and not for
> cryptographic applications.
ok, ok, this is all going nowhere. as has been pointed out to me
`multiple times, it is *not possible* to have 'unguessable' numbers that
*aren't* uniformly distributed. because if they're not 'statistically
random' you can guess them. right? we all know that.
my point was more subtle: it is your responsibility to verify that the
uses you put /dev/random to are appropriate. monte-carlo integration and
n-body simulations are *not* kernel issues and thus *not* appropriate for
linux-kernel. attacks on tcp sequence numbers *are* kernel issues, but no
one has ever claimed that statistical uniformity was relevant here.
--s
@ @
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-oOO-(_)-OOo-=-=-=-=-=
C. Scott Ananian: cananian@lcs.mit.edu / Declare the Truth boldly and
Laboratory for Computer Science/Crypto / without hindrance.
Massachusetts Institute of Technology /META-PARRESIAS AKOLUTOS:Acts 28:31
-.-. .-.. .. ..-. ..-. --- .-. -.. ... -.-. --- - - .- -. .- -. .. .- -.
PGP key available via finger and from http://www.pdos.lcs.mit.edu/~cananian
KGB Rule Psix Albanian FSF atomic [Hello to all my fans in domestic surveillance]
Suharto PLO blowfish Cocaine quiche Ortega Waihopai Hawk Clinton Uzi
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/