Murray S. Kucherawy wrote:
>> I have trouble with the 3rd separation sentence and the potential
>> ignorance it presents by breaking the original responsible party.
>>
>> What is the actual question does it separate?
>>
>>   An association between the purported author and the signer?
>>   Is an authorization question?
>>   Does it absolve the responsibility of the original domain signer?

> The sentence is meant to make explicit the fact that the author 
> of a message and the signer of a message are not necessarily the 
> same thing.  So I guess then the first of your three examples is 
> the right one.

Which implies the 2nd and 3rd?

But consider, if the 5322.From is a required DKIM signature binding 
requirement, including for all collective signatures, there will 
always an association that can not be removed.

So I can only see that this is an attempt to remove a possible 
recognized association, remove any signing restrictions by ignoring 
any possible unauthorized signing restriction and by absolving any 
previous single and/or collective signer responsibility.

>> I don't think the raw DKIM-base document should be making any
>> conclusion about that it intends to separate or absolve by moving the
>> responsibility to that of the signer.
> 
> But the signer (d=) is the only provable entity on a signed 
> message.  This was what was said in the update draft as 
> well (RFC5671).

I don't follow. 5322.FROM is a required binding. While d= is the 
domain required for verify, the 5322.From is bound to it as a provable 
entity as well, i.e - you can't change it.

>> By having it, it implies that those using the DKIM-BASE
>> implementation can effectively 100% ignore the original 
>> responsible domain own signature without technical and even 
>> possibly legal repercussions.
> 
> I think the problem is that terms like "original responsible 
> domain" are undefined given that there are no assurances 
> of the validity of any other part of the message.  If you mean 
> the From: field domain, that domain may or may not match "d=" 
> even if there's a plurality of signatures.

Does "Previous Single or Collective Responsible signing domains" help 
better express the concern?

>> I don't think a reference to POLICY needs to be made, but only focus
>> on the idea that the LAST SIGNER is the responsible party.
> 
> I don't think that's necessarily a correct assertion.  If a 
> message has four valid signatures on it, then four parties 
> have accepted some responsibility for the message.  The From: 
> domain doesn't need to match the "d=" on any of them.

I can understand the historical perspective why the 3rd sentence is 
there as a statement to remove any non-authorized 3rd party signing by 
DKIM-CORE.

But please note this is not a ADSP policy concern, but also 
REPUTATION, including MLM issues since we are trying to give it a 
"DKIM Permission Slip" to not only absolve all previous single or 
collective responsible signing domains, but also break any bound 
"entity" used to couple policy and trust concepts from previous 
responsible resigner(s), and it can do so as a whole or selectively. 
It is the final arbiter.

So even with future unknown reputation ideas, or with MLM with its 
List-ID proposals, or anything else down the pike, the concern is that 
it equally applies and in even more possibly undesirable ways - 
because 4871bis says not about these "entities."

The 3rd sentence is an old idea that still based on the idea that 
POLICY would never emerge as a standard. It was a mentality that 
perpetuated the long time conflicts we had here.  I believe we need to 
be honest about these conflicts if we wish to get any workable widely 
endorsed usage of this stuff.  One doc that implies "ignore it" while 
we are still talking about DKIM Protective Wrapper technology, since 
to be counterproductive for a long time.

So I all am just suggestion that we take the opportunity now with 
4871bis to add compromising semantics that inform readers that SIGNING 
is not always desirable or the right thing to do and they should be 
aware of any current or future augmented DKIM related technology or 
functional specifications.

-- 
Hector Santos, CTO
http://www.santronics.com
http://santronics.blogspot.com


_______________________________________________
NOTE WELL: This list operates according to 
http://mipassoc.org/dkim/ietf-list-rules.html

Reply via email to