Robert A. Rosenberg wrote:
At 11:36 -0700 on 07/17/2007, Ned Freed wrote about Re: acceptings
particularly a message (3):
A significant number of list managers that receive an
>> authoritative indication that an entire group of
>> addresses they just tried to send to are bad are going to
unsubscribe them all right away rather than wait for
>> a later probe to confirm that the addresses are bogus.
If they are going to want to unsubscribe all the addresses, IMO, they
should first sanity check the list by either doing an immediate probe to
that list of addresses (and unsubscribe based on the results, thus not
unsubscribing the good addresses), or doing individual resends of the
message but this time sending one copy per address (in effect doing the
same thing as the probe and/or acting as if each address was the only
one at that domain/MX).
I believe what Ned as implying was that this "probe" was already done by
virtue of the SMTP attempt to deliver the distribution and its recorded
HARD FAILURE to deliver the mail to the address and/or domain provides
dynamic, immediate and SMTP level feedback to the MLS (Mail List Server)
or any one other client.
The next time the MLS connects to its smart host (or embedded SMTP
server) to begin a distribution, it will get some 45x/55x for some of
the RCPT TO addresses based on the SMTP recorded history for the
address/domain. At that point, the MLS will automatically unsubscribe
or de-activate the address.
This works because most, if not all, good SMTP servers today will record
HARD FAILURES for relays and will have keep a history of the bad
addresses. The next time an attempt is made to relay mail to the
address, it will issue a negative response. Why accept an RCPT TO when
the SMTP server knows the address is bad and can't be routed?
For our MLS system, it makes the account in-active rather than
unsubscribe. That gives someone a chance to make any individual
corrections and reactivate the account.
--
Sincerely
Hector Santos, CTO
http://www.santronics.com
http://santronics.blogspot.com