On 5/12/15 6:58 PM, Stephen J. Turnbull wrote:
> Douglas Otis writes:
>
>  > DMARC being unable to assert the domain
>
> I'm not sure what you mean by "assert the domain".  AFAICS no new
> protocol is needed to validate Sender -- SPF and DKIM allow that
> already, and it's not obvious to me where the big threat is from a
> misaligned or spoofed Sender.  (A BCP might say that Sender should be
> aligned with the SPF domain if available, and otherwise with a valid
> DKIM signer otherwise).  I suppose some receivers already use this
> information in their reputational models.
>
>  > Many have not realized double signing is wide open to abuse
>
> Please present your threat analysis.  As far as I can see, double
> signing is no more vulnerable than the current practice for mailing
> lists when relaying mail from p=none sites.  It would increase the
> attack surface for the kind of abuse that caused some major sites to
> publish p=reject last April, but it's something that can be turned on
> incrementally as a matter of local policy (just as DMARC itself was),
> and it can be turned off as fast as you can propagate the config
> change to your SMTP server farm (unlike p=reject itself, which suffers
> from DNS caching lag).
>
> I wouldn't call that "wide open".
>
> Steve
>
>

_______________________________________________
dmarc mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dmarc

Reply via email to