On Sun, 23 Mar 2003 11:23:52 -0500, Kee Hinckley
<nazgul(_at_)somewhere(_dot_)com> writes:
At 12:47 AM -0600 3/23/03, Scott A Crosby wrote:
> What is the minimumm speed CPU that you will allow to send mail? A
25 MHz 486 is at least 200 and perhaps 500 or 1000 times slower than
a 3 GHz P4. A 25 MHz 486 needs several minutes to generate 60
About 7 hours to compute a partial collision stamp. But, if the sender
already has a valid stamp (previously sent to them by the recipient),
they may use it with no computations. Recipients can compute
nonce-stamps at zero cost. Thus the 600ghz*seconds of CPU would only
be spent on new senders.
So you're saying that it's okay that people on old machines have to
wait seven hours before they can send a message to someone they
haven't received a message from?
Correct. Unless they can use out-of-band means to obtain a nonce-stamp
in other ways such as a phone call, solving a CAPTCHA on a website, a
auto-reply robot, that sort of thing.
As the back of the envelope calculation I did earlier showed, its
about $1 for 100,000 ghz*seconds. 10ghz*seconds wouldn't seem to be
enough to stop a spammer, yet would already cost a 25ghz system at
least 7 minutes.
Of course, I suppose it doesn't matter in some sense. Because the
odds of anyone writing the hashcash software for their machines is
pretty close to nil. They'll just always end up in the junk bin.
Perhaps.. If there was an easy solution, its likely someone would be
using it by now. And with alternative techniques to obtain a stamp (as
above, it may be incrementally deployable.)
Is anyone collecting system proposals, with their plusses, minuses,
and practicality listed?
Scott
_______________________________________________
Asrg mailing list
Asrg(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/asrg