-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

In message <[email protected]
il.com>, Douglas Foster <[email protected]> writes

>    The presumed success of DKIM2, as a solution for the mailing list 
>    problem, is the incentive for terminating ARC.  

I don't think so .. the incentive for producing a document is that the
experiment has shown that ARC is not in practice useful at scale and it
is unwise to fail to record this, lest further people spend time and
effort discovering its lack of utility for themselves.

Arguably I should stop there because most of the rest of your email
should have been sent to the DKIM working group instead of here.

>Therforee, it is 
>    relevant and appropriate to define that path.
>
>    * At minimum, all p=reject domains for a specific mailing list must 
>      choose to implement DKIM2

I think you must mean that senders of mail from domains with p=reject
must implement DKIM2.

>    * The mailing list detects that all p=reject domains which are 
>      eligible for posting have switched to DKIM2.  This becomes its 
>      incentive to implement DKIM2 as well.

no .. the incentive occurs earlier in time than that; because by using
DKIM2 the mailing list can let mailbox providers know that it is a
mailing list and hence it is appropriate to apply different counting
rules for detecting DKIM replay

>    * The mailing list implements DKIM2, including definition of its 
>      transforms.

the last clause is unnecessary

>    * Recipient domains that currently enforce p=reject must understand 
>      DKIM2 and trust the transformations made by the mailing list.

broadly, yes

>    * The mailing list collects data to confirm that all subscribers 
>      are now willing to accept its transforms.

I think you will find that very few mailing lists operate as democracies

>    * The mailing list is now able to stop munging the From address, as 
>      long as:
>       * New subscribers must agree to accept the DKIM2 transforms.

nonsense

>       * New subscribers with p=reject must implement DKIM2 before 
>         making posts.
>       * Existing subscribers must implement DKIM2 before switching 
>         from p=none to p=reject, or they will lose the right to make 
>         posts.

they will not lose the right, it's just that their email may not reach
everyone

>    Analysis
>    Multiple parts of this process assume actions that are inconsistent
>     with our experience to date/
>    * ARC can be implemented at the server level by the server 
>      organization.   DKIM2 must be implemented by the Mailing List 
>      Management software, and the implementation is more complex.

our experience to date is that very few "server organisations" (whatever
they are) have implemented ARC -- I don't see why that will change

>    * Implementing DKIM2 as an alternative to munging requires the list 
>      to put restrictions on new subscribers, something they have 
>      seemed unwilling to do so far.
>    * DKIM2 requires recipient email filters to maintain a list of 
>      trusted mailing lists, so that those DKIM2 transforms can be 
>      accepted.   This also has been difficult to achieve to date.

I expect that mailbox providers will apply heuristics as the type (and
scale) of recipes that they will deem appropriate. Note that the
exploded flag allows identification of mailing lists, which current
systems do not do  (yes, I know about List-* header field, but I have
done enough analysis of what is seen on incoming email at scale to know
that they are of little practical use at that scale)

>    * Mailing lists need to collect data about the security posture of 
>      their recipients, something that requires some effort and has not 
>      had visible uptake so far.

I don't think any of the mailing lists I use have ever asked me about my
security posture. I think you may mean something else.

>    Hence, I conclude that DKIM2 will not solve the mailing list 
>    problem.   Evaluating whether it will solve another important 
>    problem would require the same level of analysis, not blind faith.

The alterative to this type of speculation is interworking experiments.
Will you be joining in ?

>    DKIM2 does not provide a way to identify mailing list traffic that 
>    is a malicious impersonation of a non-DKIM2 domain.   ARC did 
>    address that problem, by rewarding mailing lists that enforced 
>    authentication and documented that action.

at $DAYJOB$ it has rewarded no-one because the list of trusted ARC
signers is extremely short (and shrinking as we identify flows that are
only trusted because we believed that an ARC signer was doing something
sensible -- and they were not).

>    DKIM2 does not help ESPs like Constant Contact or Otter.AI.   These 
>    organizations must continue to mung the From address.

Constant Contact (who can speak for themselves of course) only use a
special domain when their customer does not own a domain or is unable
(with the current level of tooling) to set up a DKIM key of their own.  

Note that that means those senders will not have SPF records either to
ARC would tell you nothing at all.

>    DKIM2 may solve the general problem of malicious impersonation, but 
>    only if all legitimate participants implement it.

It is certainly true that not implementing DKIM2 will mean that no
benefits of DKIM2 will accrue to you.

- -- 
richard                       writing to inform and not as company policy

"Assembly of Japanese bicycle require great peace of mind" quoted in ZAMM

-----BEGIN PGP SIGNATURE-----
Version: PGPsdk version 1.7.1

iQA/AwUBagZE+GHfC/FfW545EQJmuACdEBsCYn7lSd31cPmaFFlc9I3ZuDsAoM2C
anUz+sAVJ7wb3awwhVodAC70
=Dlha
-----END PGP SIGNATURE-----

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

Reply via email to