[mailto:owner-ietf-mailsig(_at_)mail(_dot_)imc(_dot_)org] On Behalf Of Dave
Crocker
And if dkim is NOT intended for user-level?
The group has first explain why not and then explain why we do not
expect user-level agents to attempt to make use of the information
provided.
One approach is to pretend that DKIM is not in competition with 'user
level' signature schemes such as S/MIME and PGP. I do not think that
this is a sensible or sustainable strategy. The criticizms that have
been made wrt the legacy signature schemes in the context of edge
protection are equally valid when mua verification is concerned.
I believe that we have a much crisper, more logically coherent argument
if we simply say 'S/MIME has had ten years and the sustained support of
all the primary providers of MUA and has not managed to establish itself
as a ubiquitous signature solution, it is time to try a different
approach that does not attempt to simulataneously solve signature and
encryption'.
Although the short term objective is not to replace S/MIME as a
signature mechanism it is inevitable that DKIM will do so if it is
successful and is capable of supporting the same range of security
features.
The limited security assurances provided by S/MIME are the result of the
limitations of DNS as a key distribution mechanism. Defining the
interfaces to existing PKI work such as PKIX and XKMS is all that is
required to provide the type of end-to-end security assurances and
persistent signature guarantees that S/MIME is intended to provide.
The likelihood that DKIM will replace the S/MIME signature component
does not however mean that S/MIME supporters should see it as a threat,
in fact quite the contrary since deployment of DKIM in the fashion
described will provide the missinbg infrastructure pieces required for
ubiquitous adoption of S/MIME and PGP encryption.
There is a potential win-win situation here, if we are prepared to look
at the bigger picture.