Trying to stay within Barry's guidelines, my original comment was about
this list, but with an eye to how it can be generalized.

In any specific case, the message stream from a  list like our own should
be identifiable by seeing expected values for:

- MailFrom address = list bounce address
- DKIM signature with DKIM PASS (to retain authentication after secondary
forwarding)
- List ID
- To address = list's posting address
- SPF PASS or SPF NONE at first hop after list forwarding
- ARC data that identifies the forwarding point and affirms the author
identity

I do not suggest that a general mail stream can be parsed to detect a
previously unknown list.  Nor do not suggest that an automated detection
process can be safely used to grant the privileged status needed by lists.
I do think ARC will need additional tokens to describe tools other than SPF
and DKIM which a list may use to authenticate a post's author.

Options for our specific list:

We have alternatives to the current munging behavior, and this group is
highly motivated to see munging go away.   So something should be done.

Baptiste's proposal is clearly the easiest to implement:   Admins inform
the group that IETF is going to stop munging on a specific date.  After
that date, subscribers are switched to digest mode if the MLM or the user
detects problems.   Admins switch them back when the user reports that the
list traffic has an exception in place by the subscriber's evaluator.  This
approach may also be sufficient for this list, as I suspect that most of
our evaluators will already have an exemption for IETF.

Emanuel's approach requires software development by every evaluator or
their software providers.   The Intranet has a decreasing number of
software providers and hosting services,, but this scope seems problematic
even for the limited scale of this list.  The larger problem is that the
software implementation still needs human guidance about which messages
should be un-munged.   Malicious actors will add un-munging signals if they
think it will enhance their attacks and permit impersonation.

So I keep coming back to the most difficult proposal, which is my own.   An
evaluator needs to be able to identify specific list traffic using
information provided by the list, probably through a disclosure statement
or service agreement document.  To facilitate automation, the disclosure
could eventually be standardized as an XML document, perhaps supplemented
by free-form content for human consumption.  We need an initial list, our
own, to work out that standard.  Ultimately, my approach still requires a
transaction where a human decides to trust the list and restore it to
"privileged" status, typically in response to a support ticket from a
subscriber to his own email administration.

I admit that my approach also has scaling problems, because each list must
be considered individually.  For the moment, the scope is one list, not all
lists.

Being a trustworthy list is also very important to me.  Baptiste's approach
does not require the list to make any promises, so it does not require the
list to undertake any security improvements.   I want to see this list
implement 100% sender authentication even more badly than I want munging
removed.  I want that level of authentication to become the norm when the
concept scales upward.   Disclosure statements only work when they are not
announcing vulnerabilities.

Doug Foster







On Wed, Jul 19, 2023 at 2:20 AM Murray S. Kucherawy <[email protected]>
wrote:

> On Tue, Jul 18, 2023 at 4:27 AM Douglas Foster <
> [email protected]> wrote:
>
>> 1) For evaluators that enforce DMARC against lists, are they willing to
>> consider any concessions to list traffic?   If so, do they favor an
>> exemption process where the list avoids munging, or an unmunging solution
>> implemented at their inbound gateway?
>>
>
> How do you determine that an evaluator is enforcing DMARC "against
> lists"?  How can one definitively identify list traffic?
>
> Also, I imagine that if munging is to be considered reversible, either the
> munging itself, or a way to describe it, needs to be standardized in some
> form.
>
> -MSK
>
_______________________________________________
dmarc mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dmarc

Reply via email to