[Cryptography] Certificates and PKI

Nico Williams nico at cryptonector.com
Tue Dec 23 15:27:38 EST 2014


On Tue, Dec 23, 2014 at 05:47:05PM +0000, Ben Laurie wrote:
> I do agree that this is an important difference. However, I don't see
> why we should trust registries/registrars any more than we trust CAs?

You already have to trust registries/registrars to a large degree.
Trusting them to properly safeguard their DNSSEC private keys isn't
asking that much more of them, gives them less power than unconstrained
CAs[*], doesn't prevent CT for DNSSEC (so we can also keep registrars
honest, so that the additional extent to which we must trust them isn't
intolerably more).

In the long run it also means one fewer sort of entity to trust: the
unconstrained CAs, as hopefully they will go the way of the dodo.  In
the medium term we might even see registrars and CAs merge, producing
that same long term result sooner.

No real new problems arise as to authenticating domain owners either,
just as to enrollment.

[*] Unconstrained because unconstrainable, because name constraints were
    never deployed in time and as critical.

Nico
-- 


More information about the cryptography mailing list