🚀 go-pugleaf

In article <11496@lanl.ARPA> jlg@a.UUCP (Jim Giles) objects to my
assertion that the RSA encryption-providing company will make it easy
for someone to come up with a randomly-selected key; I suggested that
they would start from a user-selected key and go forward until we find
a suitable prime.  He suggests:

>    Step 1 - have customer hit space bar N times (to get an N bit number).
>	     The output will be an N bit binary number which consists
>	     of the least significant bit of the microsecond clock at
>	     each space bar detection.  If the leading bit is 0 complement
>	     the whole number (we are looking for an N bit number, so
>	     the leading bit must be one).
>...
>
>This doesn't seem to provide much foothold for the codebreakers to
>use.  Maybe factoring would be easier than trying to reconstruct the
>generation sequence used.

This works fine for getting your initial secret primes; I don't know
if microseconds are good enough, but am willing to buy it.  But it has the
disadvantage that after getting the primes they must stay on the computer
as long as this public key will be used.  This leaves them unprotected
for a long time: public keys will typically be longer-lived than private
ones, since you need to publish them in public places, and this can take
time as well.

Your suggestion would be similar to building a communication package to
transfer your files for you that happens to have your login password sitting
there in clear.  It doesn't give your password to everybody in the world,
but it's exposed to anybody on your system.

Another poster says that detemining whether a number is prime is just
as hard as factoring it.  This is false.  See Knuth vol 2, for example; the
statistical method he gives has been tightened up to certainty in recent
years.
-- 
	Jim Gillogly
	{hplabs, ihnp4}!sdcrdcf!randvax!jim
	jim@rand-unix.arpa