On 11/05/2026 16:48, Tero Kivinen wrote:
Alessandro Vesely writes:
In that sense, ARC may have attempted to address an authorization problem using only authentication primitives.

Actually ARC provides the authentication information that can be used to create authorization in the final recipients email filters.


The authorization data that can be derived from authentication cannot cover indirect mail flows.


Lets say ietf list would add ARC header before the mailing list email
is modified.


An ARC Set must be added /after/ any modifications, otherwise it will not validate. The AAR is derived from any trusted Authentication-Results header fields present at that time:

   Because there is only one AAR allowed per ARC Set, the AAR MUST
   contain the combined authres-payload with all of the authentication
   results from within the participating ADMD, regardless of how many
   Authentication-Results header fields are attached to the message.
           https://datatracker.ietf.org/doc/html/rfc8617#section-4.1.1


Your message had DKIM signature, but it of course got broken by the mailing list. The mail2.ietf.org already checked that as it added header saying:>
Authentication-Results: mail2.ietf.org (amavisd-new); dkim=pass (1152-bit key)
        header.d=tana.it

If it would have added also ARC header saying:

ARC-Authentication-Results: i=1; mail2.ietf.org; dkim=pass ...


The information conveyed is the same in both cases. Having it signed is a matter of authentication, not authorization. Indeed, a malicious forwarder can insert arbitrary data into an AAR header field.


then when the email ends at my final filters, I could have configured that to say that for "List-Id: <*.ietf.org>", I do trust ARC Authentication results from *.ietf.org, and those results should be considered while decideding whether to filter the email or pass it to my mailbox.

Currently I do this by rule that says

* ^List-Id:.*(\.ietf\.org)
* ^Precedence:.*(list|bulk)
$DEFAULT


The fact that you have this rule suggests that the authorization decision is being derived from sources other than ARC Sets —presumably your knowledge of the IETF and of the mailing lists to which you subscribe. In addition, this is only one side of the authorization relationship: ietf.org has no way to know which recipients are willing to accept list-modified messages despite a resulting DMARC failure, and therefore must still munge From: header fields for some domains.


If forwarding semantics actually require explicit authorization relationships between participants or mail streams, then that would also explain why ARC struggled to produce stable evaluation semantics across independent operators.

I think ARC header generation is very useful and provides information to the final recipients filters that really can help them.


However useful, it does not by itself solve the mailing-list problem, which was ARC's primary motivation. Whitelisting subscribed mailing lists is a time-honored practice, but it is inherently one-sided. An authorization layer would also provide mailing lists with knowledge of which recipients are willing to accept list-modified messages despite DMARC failure, allowing From: munging to be avoided for those recipients.


Best
Ale
--




_______________________________________________
dmarc mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to