On Mon, 26 Apr 2004, Hallam-Baker, Phillip wrote:
This is the key.
WE ARE AUTHENTICATING THE GOOD EMAIL.
Then I don't see the point, as you define good email as that with a
Caller-ID record. Surely S/MIME or even server to server TLS
would do that better?
It would do the job 'better' in the same sense that a fully armoured
tank would do the job of getting to work safely 'better'.
S/MIME is a great protocol for the wrong market. It was designed for
ultra security concerned geeks, not the masses. Microsoft will not
even send out its security alerts using S/MIME because messages appear
in obsolete mail clients as attachments.
TLS is good for privacy, as authentication it is only good if the mail
servers use decent certs, which almost nobody bothers to do because
nothing checks. Moving to use good certs would be a good idea, but
will not be possible until the market is developed which will take
longer than is acceptable.
Hence the value of MARID, it is the cheapest, easiest to deploy means
of authentication from edge outgoing gateway to edge incomming gateway.
In the context of spam we are doing crowd control, we do not need
fort knox security, we do not even need airport security. Stadium
anti-gatecrasher security is fine.
In the context of anti-phishing it is a different matter, that is
a case where my bank clients would very much like to be able to use
S/MIME and are looking for deployment possibilities. But that is
in the context of securing geniune communications from banks to
re-establish confidence.
In terms of fraud prevention it is actually much more important to
suppress 'ordinary' spam. Drain the swamp. But also because large
numbers of 'ordinary' spams are actually a different type of phishing
fraud tactic, one that impersonates an honest merchant rather than
a well known brand.
Phill