[RC5] Splitting buff-in.des?

James Mastros root at jennifer-unix.dyn.ml.org
Fri Jan 16 18:46:52 EST 1998


On Fri, 16 Jan 1998, George D. Nincehelser wrote:
> Anssi Saari wrote:
> > Since I somewhat stupidly downloaded 1000 DES blocks, which would take my
> > Pentium 40 days or so to check, I'm wondering if it would be possible to
> > just "shave off" some blocks from my buff-in.des and give them to some
> > other computer to check? Otherwise, with bad luck, the contest might be
> > delayed if the right block is sitting on my hard disk...
> 
> Look on the bright side.  More than likely, your 1000 blocks DO NOT 
> contain the correct key.  Therefore, you have most likely saved the group
> the time of checking them.
> 
> Ain't statistics wonderful? ;-)

This is probably the most common error on this list: just because it is
unlikely that any individual block contains the key, that dosn't mean that
we should plan around only that probability.  (Think about it: if that 
were
true, we could have the correct answer in a moment: there is no correct
key.)  Instead of doing that, we need to handle each block with its true
value: 5^56/10,000 = 8 millionths of a cent.  That may not seem like much,
but it is infinatly greater then zero (infinity*0=8/1,000,000, not
0+infinity=.000 000 8).  It is far safer to assume that EVERY block has 
The
One True Key.

> In all seriousness, I don't think it really matters.
> 
> On a related note, however, I have one machine that isn't always on the
> network, thus I gave it a very large in-buffer (also 1000 blocks).  It 
> won't
> be networkable again for a while, so the out-buffer is also going to grow 
> for
> a while.
> 
> On the off-chance that I hit the key with this machine, what happens? Will
> anything alert me that a potential winner is sitting in my out-buffer
> (spurring me to get re-connected sooner and flush the buffer)?  Or will
> nothing be known until the actual flushing of the blocks to the server?

Nothing will show to you (unless you use a perproxy)...  This was a 
concous
decision on the part of the inner circle: if the client said somthing, 
then
the unscruplous could get the full prize by turning in The Key themselves.
(The personal proxy does show a slight difference in the logs: or at least
it did at the end of the rc5-64 contest (he didn't see it untill he looked
for it, however)).  I don't know the reasoning behind this.

> I know it probably doesn't matter statistically, but I'm just currious.
Yes, it does matter.  Those 1000 blocks are worth eight thousandths of a 
cent.

	-=- James Mastros

--
To unsubcribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest



More information about the rc5 mailing list