[ietf-dkim] a bit of philosophy on working group productivity andscope

[Scott Kitterman]

Dave Crocker wrote:
>>> But first we need to do *anything at all* that is useful.
> 
> ....
> 
>>> As of today, there is no standardized transit-time message authentication
>>> technique.  If we can produce a standard that permits validating ANY
>>> identity
>>> with a signed message, we will have created a stable base for all sorts
>>> of enhancements.
>>
>> Perhaps, but a stable base for future enhancements that will actually have
>> some utility is not, I would think, something useful.
> 
> 
> You seem to have missed the "but first" paragraph.
> 
I think not.  I think I'm trying to say that what you defined as 
minimally useful isn't sufficently useful to be worth the effort of a 
working group.  I think the minimum needs to be higher.
> 
> 
>> Unless the output of this putative group would at least enable a receiver to
>> reject a 'bad' message or have more confidence in a 'good' message there is
>> no incentive for either senders or receivers to deploy.
> 
> 
> for some definitions of good messages and bad message.
> 
Yes.  Up to the receiver to decide that.  I won't even try to come up 
with a universal definition.
> 
> 
>> It would seem to me that there is a necessary tie between the identity being
>> signed, some e-mail identity that end uses actually see, and some type of
>> sender policy declaration that would allow receivers to have some idea how
>> to interpret the presence, absence, and validity of signatures.  
> 
> 
> Quite a bit of useful filtering is done today that does not require the end-user 
> to participate directly and does not involve knowing the sender's "policies" and 
> does not require using the rfc2822.from field.
>
Yes.  All of which does not require MASS or DKIM.

What I am attempting to say is that I do not believe there is any 
sigificant value in signing some new, invisible e-mail identity.  For it 
to have value, it needs to relate to a current, visible identity.  It 
also needs to be tied to some sort of sender policy because otherwise 
there is no reliable way to know what the presence or absence of a 
particular signature is supposed to mean other than in the case of a 
valid signature for an identity that is identical to the body From:.

If all I want is a cryptgraphically valid signature, there are other 
ways to get it.

I thought your thread was about what is the minimum we can accomplish 
that will be worthwhile.  I think that's about as low as the bar goes.

Scott Kitterman