On Fri, 21 Mar 2003 09:29:34 -0500, "Damien Morton"
<dm-temp-310102(_at_)nyc(_dot_)rr(_dot_)com> writes:
Could a daily, globally published, random number somehow come into this
computation. Incorporating this number into the hash computation somehow
would then validate a stamp for 24 hours (or some other period).
No. Stamps are intended to be good for months/years. The reason for
that design decision is to reduce resouce use, to avoid having to
remember *every* past used stamp to detect double-spending.
Unused random nonces (the first type of stamp) may be flushed after a
year or two and no longer accepted.
On the other hand, how does one avoid having to remember every past
hash-stamp, forever? Thus, the only reason I include a HashNonce in
the protocol is to allow me to invalidate past (say, 2 years old)
hash-stamps. If I could, I'd take them out of the protocol, they
require additional semantics, and they're one main reason that
prevents this protocol from being implemented soley by user-agents.
On reflection, the correct design decision may be to remove the
HashNonce and remember all hash-collision stamps forever. At 20 bytes
each. (hash table implementation), the cost to store used Hash-Stamps
is pennies/million. Accidently losing that table allows stamps to be
double-spent, but if that occurs only rarely, its not an issue.
Scott
_______________________________________________
Asrg mailing list
Asrg(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/asrg