On Jun 6, 2006, at 5:46 PM, John Levine wrote:
The question then becomes, while exploring schemes to have response
codes per-recipient, what can be done to the protocol to make it
usable by a B2C sender.
The protocol works just fine for B2C senders now. They send each
recipient a separate message.
In a world where ISPs charge B2C senders by bandwidth instead of
per-message (fantasy land, or not) bandwidth stops being cheap.
ISPs do charge by bandwidth, but it's an insignificant part of the
total cost except in small remote countries with satellite links. I
don't know of anyone who charges by the message other than the exotic
stuff Goodmail is doing.
B2C senders are rolling in profits and worried sick about the
potential impending desubsidization of the delivery of their
commercial speech.
The B2C senders I know are worried about deliverability and phishing.
Steve Atkins works with them more closely than I do, so perhaps he
could opine here.
"What he said."
Bandwidth is a barely noticable cost to ESPs, and not one that drives
their
business model beyond certain easy tweaks to reduce it. Sending
one message per recipient is current best practice, and is used by
most all
the reputable ESPs I'm aware of. VERP is one reason, but the ability
to do proper customization is the driving reason.
"Templating" is nowhere near powerful enough for a lot of email as
sent today - there's complex database work used to customize emails
for segmented recipient lists, and also there are custom tokens and
often customized content, per recipient. It might be adequate for some
fraction of an ESPs customers, but a huge overhead at an ESP is
maintaining the software systems used to allow customers to create
the mails - can you imagine them wanting to maintain two entirely
separate content creation and delivery engines, solely so they can
save a bit of bandwidth on the mail sent with one of them?
At most ESPs where I'm aware of their internal issues the dominating
factor is database disk I/O bandwidth to build those messages and do
the huge amount of list management needed. External bandwidth or
SMTP CPU is down in the noise as far as operational issues are
concerned.
There would be zero interest at any ESP, and I think at any reasonably
sized B2C internal mailer, in anything that came out of this thread.
B2C senders are mostly concerned about deliverability. To the extent
that some are more concerned about deliverability than the profitability
of certain products or customers. That should give you some idea as to
how important that is to them. Because of that they're extremely
concerned
about reputation - far more than you are. Their sole interest in SPF is
to allow domain-based whitelisting at AOL. A major reason for their
interest in DKIM is that it allows a list-owner (such as IBM) to move
their
business from one ESP to another and to maintain their reputation as
tied to "ibm.com" rather than the mail being based on the reputation
of the ESP they happen to be sending through.
Phishing isn't a general concern of anyone outside the financial
services industries at the moment. I suspect that'll change, but I'm
hoping that the financial services folks, and the ESPs that provide
them with email services, will have got somewhere with solving
the problems before then.
Cheers,
Steve
_______________________________________________
Asrg mailing list
Asrg(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/asrg