[ NNSquad ] Re: Catch-22 for sending email

[Rich Kulawiec <rsk@gsp.org>]

First, let me point that Comcast is completely, utterly, and absolutely
incompetent when it comes to controlling spam from their network.  While
they're doing idiotic things like this to you, they're simultaneously allowing
other systems on their network to spew spam at ferocious rates indefinitely.

(I've gotten spam from one address on their network for five YEARS.  It's now
in the firewall, of course -- no point in letting the SMTP connection through,
no point in logging it, no point in reporting it.  Just drop the packets.)

So your "error" is expecting anything remotely resembling competence
from them. ;-)  I'm completely willing to believe that they labeled your
IP address as a spam source based on one message because that's exactly
the sort of boneheads they are.  (No offense to any Minbari.)

Second, as to not identifying the message more closely, that's SOP when
dealing with outsiders (to avoid list-washing, as has been pointed
out) but shouldn't be SOP when dealing with one's own customers.
After all, if you really were spammers, then "helping you listwash"
based on a single message wouldn't do you any good -- for listwashing to
be effective, it has to be done on a reasonably large scale and it has to
specifically target known complainers, their domains, their networks, etc.
(My "postmaster" address here, for instance, rarely gets spam even though
it's exempted from all anti-spam measures.  It's been quite thoroughly
listwashed by every competent spammer out there.  So on those infrequent
occasions when I get something, I know I'm either dealing with a new
player or an idiot.)  Their rationale is bogus.

Third, here's my guess: at least one of your recipients is at AOL.  AOL has
a feedback loop which, *when used properly*, is really quite useful.
It enables those responsible for IP address ranges to receive notification
when email originating from those ranges is marked by AOL users as spam.
That notification doesn't identify the AOL user in question, although
various methods (e.g., VERP) can be used to pre-tag messages headed for
AOL so that it can be worked out post-mortem.  It *does* preserve the
"Subject" header, though -- which is how Comcast knew it had something
to do with a party.

The thing is...apparently the "mark as spam" button is insufficiently
distinguished from the "delete" button in the GUI.  As a result, AOL
users will hit that button when they shouldn't.  I see it periodically
with a mailing list I run -- it's about 60 people, I've known them all
personally for at least 20 years, and every now and then, I get an AOL
feedback loop report subsequent to a message on that list.

And this is where knowing how to use that feedback loop properly comes
in: any real spam run launched from one's network, say from a compromised
Windows system, will generate a flood of feedback loop reports and will
make itself highly visible in short order.  But a single feedback loop
report from a known outbound mail server means nothing...unless, of
course, you're Comcast, you're staffed entirely by idiots, and you want
to continue telling the trade press that you "take the spam problem very
seriously" with a straight face while your network contends for the #1
spot on the worldwide list of spam sources for years on end.  In that
case, it provides a convenient pretext for taking ineffective, misdirected
action that achieves nothing useful.

---Rsk