Re: WSJ/gmail/ML, was a permission to...

On 05/03/2014 08:29 PM, Alessandro Vesely wrote:
> On Fri 18/Apr/2014 14:37:21 +0200 John Levine wrote:
>> I also note that this hack, with or without Ale's changes, does
>> nothing to solve the send from gmail and WSJ article problems.
> Those two problems can be solved in different ways.  Gmail could use a
> third party's submission server just like they use its pop/imap one.
> WSJ could write "WSJ.com" in the From: and the purported issuer in the
> Subject: (and possibly also Reply-To:) instead of their currently
> doing the other way around.

In your proposal one spam/phishing fighting technique (DMARC) requires 
(at least) three techniques to solve its problems (which in turn may 
require another 3^2 techniques to solve their problems, which ... 3^n 
techniques ... et cetera).

>
> We need to refine the spec specifically for mailing lists.  I'd love
> to read something like:
>
>     A mailing list MUST NOT tamper with the From: field.  Those who do
>     that are not mailing lists, and can work their own way out of
>     DMARC.
>
> Is it practical to mandate the same also for To:, Cc:, Date:?  Are
> there lists which alter them?
>
> Also:
>
>     In order to get weak signatures, a mailing list needs to let its
>     posters' domain admins know which posters post to which addresses.
>     It is advisable to ask for posters' permission to do so.
>
> That can be done manually for the time being.  Imagine lots of
> ML-admins writing to the relevant postmasters asking to apply
> low-profile signatures for specific MAIL FROM/RCPT TO pairs.  A
> postmaster can verify that those users really post to those addresses,
> and believe that it is a mailing list since its owner says so.  A
> couple of scripts would do.  Automation can later be improved.

This really doesn't scale. Maybe you have the few TBTI ESP's in mind, of 
which there are only a few. But each domain that starts using p=reject 
will introduce a new set of relations (ML-admins-postmaster), which in 
turn will grow exponentially.

>
> The same manual requests can be done for whitelisting, let's see the
> pros and cons.
>
> Either way, we have to do it at our expenses, since it's us who want
> mailing lists.  Or we may hope that users' rebuttal will discourage
> receivers from honoring DMARC policies, but that sounds like letting
> email problems grow.

I agree there is a need to solve the current problems, but let's come up 
with one solution, not many.

/rolf

_______________________________________________
ietf-822 mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/ietf-822