Katriel,
Note that ARPKI <http://dl.acm.org/citation.cfm?id=2660298> (among
other academic CT variants) similarly has no need for gossip; roughly,
monitors sign that they have seen a cert and then clients just verify
the signatures. Multisignatures sound like an elegant way to achieve
that; are they more efficient than /n/ individual signatures for
smallish cothorities e.g. size <10?
Note that CT does not require all Monitors to be third parties;
self-Monitoring is a viable
option and avoids the problem of detecting a misbehaving (3rd party)
Monitor. In that
context, having multiple Monitors sign something about a cert may not
work out well.
The Audit function notifies Monitors (and browser vendors, indirectly?)
about log
misbehavior, not clients. I say this because clients (browsers) get log
metadata
from browser vendors, and thus most will not need to consume data from
Auditors.
Monitors need audit info to decide whether the logs they are watching
are misbehaving,
and browser vendors need to know the same, to properly manage the log
metadata
distributed to users of their products.
I've never quite understood why CT mandates gossip instead of having
monitors sign that they have seen an STH. Perhaps this is a good
opportunity to clarify that point?
Monitors watch logs for entries of interest to the Subjects that serve.
(This is true
even for self-Monitors.) So, it's not clear on whose behalf a Monitor
would sign info
about an STH.
Steve
_______________________________________________
Trans mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/trans
