On May 10, 2015 11:55:22 AM EDT, "Stephen J. Turnbull" <[email protected]> 
wrote:
>Scott Kitterman writes:
>
> > Yahoo, for example, already consider the impact of this and other
> > breakage to be less than the benefit of p=reject.
>
>True, but the benefit of p=reject is huge: Yahoo! claims malicious
>mailflows of more than a million messages per minute disappeared like
>magic when they published p=reject.  Such a flow surely stressed their
>systems, and I'm sure they consider the potential for high losses due
>to contact-list-based phishing to be important, if only for the damage
>to their reputation that would ensue.
>
> > I expect their willingness to invest engineering resources in
> > further reducing a level of breakage they've already determined is
> > acceptable will be limited.
>
>Of course it's limited.  But the only cap I can be sure of is the
>difference between benefit of p=reject (see above) and cost (1 day of
>manager-admin meetings to decide to do it, and 15 seconds of admin
>time to change the DNS record, ie, basically zero).  I believe that
>difference to be orders of magnitude larger than the cost of
>implementing a dozen delegation protocols, and therefore irrelevant.
>
>What matters is the benefit that the p=reject domains perceive to
>improving service to their mailbox users.  I have no information about
>that.  Of course I suspect that the value to them of such improvements
>is close to nil, but until they actually say that, I'm going to hope.
>
>If you have better information about how much they value such
>improvements, I'd love to hear about it.  But I rather doubt you do;
>the techs surely don't have the authority to say they will do it, and
>at best a guess of what they need to offer management to get
>permission, and management won't discuss the value of a bird in the
>bush (not if they have half a brain amongst them, anyway).

I don't have any particular insight. I do think that the group ought to be 
paying close attention to those who do.  As I attempted to communicate in my 
assessment framework proposal, whatever solution the group coalesces around has 
to be workable for entities from the largest to the smallest. 

I think we have three outcomes:

1.  We develop something that's broadly deployable

2.  Single actor approaches such as from rewriting dominate the solution space

3.  Mediated mail communication becomes rare

Scott K


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

Reply via email to