Re: [Asrg] pre-rfc thought balloon: ESMTP DATAFIRST

[Daniel Feenberg <feenberg at nber.org>]

On Mon, 5 Jun 2006, Jim MacLeod wrote:

...
> If I read it correctly, the problem sounds like "Messages to lots of
> recipients have to be sent to each recipient individually if the server
> won't accept the message to some recipients."  Is the solution that, by

I don't think many MTAs send such messages, since 99.44% of them would be
to forged addresses. Instead they just discard mail whose content appears 
to be spammish. In some cases that judgement is wrong, and inefficiency or 
ill-feeling can result. So operators of MTAs using content analyzers would 
presumably be interested in a mechanism for rejecting mail during the SMTP 
command phase on a recipient by recipient basis. Sending the data first 
seems to be an easy way to enable this, although it has been pointed out 
that the information could be returned after the data phase.

Presumably only MTAs with content based spam scanners, and among those 
only ones with individual databases, would offer the DATAFIRST option to 
connecting MTAs. And among connecting MTAs, only those who cared if 
messages were delivered would act on the offer. It isn't obvious that 
anyone would be harmed. There is plenty of experience to suggest that 
individual databases work much better than system-wide databases.

Nor does this option imply any additional wasted bandwidth over the 
current widely practiced method of accepting the message and then 
discarding it.


> sending the DATA first, the receiving server could separately 250/5xx for
> each recipient, based on a complex decision involving the content?
>
> I wonder what's next, an ESMTP extension replacing DATA with a URI that
> could be retrieved ala ETRN?

That suggestion is frequently made, and this seems to be a way to avoid 
such things.

Daniel Feenberg
NBER



> -Jim
>
> On 6/5/06, David Nicol <davidnicol at gmail.com > wrote:
> > This e-mail is a presentation of an idea for comments[...]
> > The new ESMTP extension is tentatively called "DATAFIRST"


_______________________________________________
Asrg mailing list
Asrg at ietf.org
https://www1.ietf.org/mailman/listinfo/asrg