Douglas Otis <dotis(_at_)mail-abuse(_dot_)org> wrote:
On 9/23/11 3:58 AM, Richard Kulawiec wrote:
Opt-out is spamming. (Or conversely, any mailing list operated
without a proper opt-in procedure is a vehicle for spam.)
I can't resist the opportunity to disagree with Doug...
On 23 Sep 2011 01:39:22 -0000 "John Levine" <johnl(_at_)taugh(_dot_)com> wrote:
["BOBOTEK, ALEX" <ab3778(_at_)att(_dot_)com> wrote:]
A notable example of 'opt-out' that comes to mind is not in the world
of email, but the 'do not call' list used for telephony.
True. The do-not-spam list is one of those ideas that keeps
coming around. Phones and e-mail are not really comparable here
because there is a fixed well-known set of phone numbers, while
there isn't a fixed set of e-mail addresses. I can easily list all
the numbers I don't want people to call, but I can only describe the
set of email addresses not to spam by using pattern matching.
Opt-out, in fact, is entirely possible; but it needs to be a
distributed service, with database and decisions at or very-near
(And I make no claim there is a supportable economic paradigm for
it: but after all, this is a RESEARCH group; this is a legitimate
research topic nonetheless.)
In the SMTP world, only the MDA _can_ know what mailbox an email
will be delivered to -- thus it's plain that the MDA is the ideal
(if not only) place to implement a workable opt-out mechanism.
Subscription to the opt-out service by the recipient has to be a
private transaction between the recipient (or his agent) and the
operator of the MDA. As such, the details of the subscriptions are
necessarily private (and any attempt to end-run that guarantees
the information to be out-of-date).
IMHO, the subscription -- to give value to the subscriber --
must include _whether_ to return an error, as well as whether to
bit-bucket or quarantine the emails covered by opt-out.
Likewise IMHO, the subscription must allow different classes of
opt-out conditions. Ideally, some of these might be set by "honest"
mass-mailers; others will necessarily imply filtering algorithms
by the MDA. Probably, the subscription would include whitelisting
for "known-good" senders, mainained by the subscriber (but note
that whitelists which include only "sender email address" have many
Anyone want to take this as a research topic?
(Yes, this _is_ remarkably close to what many ISPs already do...)
John Leslie <john(_at_)jlc(_dot_)net>
Asrg mailing list