From: Sam Quigley <poodge@econ.Berkeley.EDU>
Date: Sat, 19 Aug 95 21:33:12 PDT
To: cypherpunks@toad.com
Subject: Re: Economic Model for Key Cracking
In-Reply-To: <199508190800.BAA26793@ix9.ix.netcom.com>
Message-ID: <199508200433.VAA07070@quesnay.Berkeley.EDU>
MIME-Version: 1.0
Content-Type: text/plain


>>>>> "Bill" == Bill Stewart <stewarts@ix.netcom.com> writes:

    > Alternatively, you could do a model where everybody gets paid,
    > but only after the answer is found, which discourages scammers
    > (since they don't get paid if they lie about searching the range
    > that has the real key.)  If a Bad Guy lies about the key not
    > being in his range, people do have an incentive to look for it
    > if the first pass fails, and have an incentive to finger him if
    > they do find the key on a later pass.  

But the scammer who has done his or her math homework will recognize
that it's reasonably unlikely that his or her assigned range contains
the true key (depending, of course, on the size of the assigned
range...).  Thus, on the average, if everyone gets paid, a scammer
will make money by not actually searching the range assigned (perhaps
actually using the CPU time on a different key search).

I think it'd also be important to reward those who throw more compute
power at the task more than those who just sic a desktop on the job.
This is handled elegantly by the first method -- the person who
dedicates some giant supercomputer to the task is that much more
likely to find the key and be rewarded...

-sq