[ih] more bounce management, was update about bogus list unsubcribe requests

[Miles Fidelman]

On 8/26/16 6:03 PM, John Levine wrote:

>> - list functions & behaviors (or we wouldn't be dealing with the impact
>> of DMARC on lists)
> That's definitely a losing battle.  I've used list managers that do
> things like flattening HTML parts into text, and downcoding images to
> make them smaller.  ARC is a crock but it will likely be good enough.
>
>> - the administrative realities of running mail systems and lists - such
>> as trying to track down things like who's sending bogus unsubscribe
>> requests, or who's been using their "spam" button as a delete key
>> (things that get very hard when, for example, someone is forwarding list
>> mail to a new address, but has left their old address on the list - I
>> can't tell you how many times I've been asked to remove someone from a
>> list, and simply can't find an address that remotely resembles the one
>> they send the removal request from)
> Commercial mailers solved these problems long ago, and discussion list
> managers have lagged beind.  Commercial mailers send every recipient a
> separate copy of the message, usually with a unique bounce address and
> some tokens in the List-Unsubscribe or elsewhere that identify the
> recipient.  For webmail spam buttons, if you have static IPs it is not
> hard to register your IP range with Hotmail and AOL, or your DKIM
> signing domains with Yahoo and automatically get copies of the mail
> that's reported as spam so you can turn them into unsubs.  I've done
> that, even with the redaction some of them do it's not hard to
> automate.
>
> As far as I can tell, too many discussion lists are stuck in a 1996
> mindset where separate copies are "inefficient" or "bandwidth hogs"
> which is ridiculous if you consider what a tiny slice of the
> Internet's bandwith mail takes, even including all the spam.  For
> lists that have other lists subscribed, my advice is don't do that, or
> if you do, the sublist better handle its own bounces.

It's not just bandwidth.  It's also a matter of CPU cycles - at both 
ends (particularly when you factor in antivirus and antispam processing).

I can't tell you what a pain it is to turn on VERP processing because 
the large hosts will tell you that someone marked a message as spam, but 
won't tell you who.  The only way to deal with that is to send each 
recipient their own message, then go through and match message id's in 
the log file.  What a waste.

Miles

-- 
In theory, there is no difference between theory and practice.
In practice, there is.  .... Yogi Berra