[RC5] Re: An idea to prevent cheating

Joe Zbiciak j-zbiciak1 at ti.com
Tue Sep 1 14:23:40 EDT 1998


'Ryan Schmidt' said previously:

| OK, I don't know if I understood your idea right, so correct me if I'm
| wrong, but wouldn't that require that the client check each and every block?

No.  I think the idea is that each block gets its own checksum computed
(independently of all other blocks cracked by the client) so that d.net
can spot-check returned blocks to ensure that nobody is cheating and
that nobody has a broken client.

I don't know about the numerical properties of DES, but I would think
XOR'ing all of the possible encryptions of some plaintext over a range
of the keyspace may be reasonable, and it's fairly quick.  Any
cryptographers out there care to speculate on how reasonable such a
checksum would be for this application?

At the very least, it would allow d.net to identify potentially broken
clients, by spot-checking 1 in every 10,000 - 100,000 blocks or
something, and seeing what turns up.  A higher checking rate could be
achieved by comparing the checksums on dupes to the previously returned
checksums, although the storage requirement could be fairly large
unless you were creative.

Yesterday, we completed ~11,000,000 blocks.  Checking 1 in 10,000
blocks, d.net would only need to check 1100 blocks -- not an onerous
task.  With my two PCs and one workstation, I provided 576 blocks
yesterday.  Get 5 or 6 strong processing machines on the problem, and
you've got some error detection built into the d.net infrastructure.

Regards,

--Joe

-- 
 +------ Joseph Zbiciak -----+
 | - - j-zbiciak1 at ti.com - - | "Can I ask you a really stupid question?"
 |-Texas Instruments, Dallas-|
 | - #include <disclaim.h> - | "Yes, and history will bear me out on that." 
 +---------------------------+

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



More information about the rc5 mailing list