[RC5] Do not forget about the cheaters :)

Richard Farmbrough rawkiw at cix.co.uk
Thu Jan 8 08:19:57 EST 2004


This is a nice tweak, clearly the number of bits will have to make 
"collisions" reasonably likely,  for some reasonable value of reasonable.
(I think the overhead is the pre -calcualtion by the server so this works 
very well.)
In turn this means that you could send some random bits, and only do a 
check on a random basis on the answers returned.
Early abort would not work against this in the long run as a cheater would 
have a lower average number of returns per block.
(Conversely both your system and my system make it near impossible to 
"safely" cheat on a single block, although mine suffers from the "kind 
executioner" 
(http://www.psych.upenn.edu/sacsec/abir/_private/Pamla/surprise-exam-abs  
tract.pdf) paradox according to the number of dummy keys.)

-----Original Message-----
From:	Decio Luiz Gazzoni Filho [SMTP:decio at revistapcs.com.br]
 Decrypt the
RSA-provided cyphertext with this key, and extract [...]x bits. x is a 
parameter that allows you a tradeoff between
overhead (mainly communication) and probability of catching a cheater. Now
you ask the client to return *all* instances of cyphertext matching the 
bits[...]
this is very resilient to an early-abort strategy [...]

---
This mail has been checked for viruses, by Richard, but you should ensure 
your system check all incoming mail itself.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.532 / Virus Database: 326 - Release Date: 27/10/2003




More information about the rc5 mailing list