Other Directory Sites: SeekWonder | Directory Owners Forum
The following archive was created by hippie-mail 7.98617-22 on Sat Apr 10 1999 - 01:18:05
mgraffam@idsi.net
Sat, 23 Jan 1999 14:43:19 -0500 (EST)
On Sat, 23 Jan 1999, Alex Alten wrote:
> If it didn't work then some cryptographic researcher would find out, publish
> his results on the web, Intel stock would drop by $10, they would vigorously
> deny it for a couple of weeks, then finally they would fix it in the next
> production run, and run an ad campaign "Intel RNG Inside". At some point
> it would be accepted by everyone, including hardcore coderpunk's mothers, as
> a good RNG and the stock would recover by $30.
How exactly would you suggest that a research find out about the internals
of the CPU?
We can run a battery of tests on the output of SHA-1 and a good hardware
generator until we are blue in the face, and they will look basically
the same.
It's no easy matter to find deviations from randomness just by staring
at the output. A researcher would likely not be able to look at the
internals of the CPU.
I'm not a big fan of crypto in hardware. Not in today's climate. I love
the idea of tamperproofing the algorithms, and burying the keys in there
too. Good idea. The problem is that as of today, there are powerful forces
at work that would love to compromise such "solutions" .. and being
basically tamperproof, we have no way of knowing what the hell is in them,
nor if they do what they claim too.
This is precisely why we all demand published algorithms.. so we can
review it. This is why we demand public source, so we can check for bugs.
I can't think of any particularly compelling reason to give this sort of
demand up just because we are talking CPUs and hardware.
I don't trust Snake Oil Inc's claims about their security, and just
because Intel is a multibillion dollar corporation doesn't mean their
word is any better.
I don't want words. I want science.
What scares me is that this will set a precedent and that everyone will
start using weak RNGs. 256 bit keys can look mighty weak when you know the
RNG's deviations from randomness.
It's good to know that I have plenty of Pentium and 486 boards ready to
run Linux and PGP for me. :)
Michael J. Graffam (mgraffam@idsi.net)
"If you die, you win.."
John Landry, statistics professor discussing life insurance
and probability
The following archive was created by hippie-mail 7.98617-22 on Sat Apr 10 1999 - 01:18:05