On 13/02/2021 10:28, The Natural Philosopher wrote:
On 12/02/2021 22:18, Mathew Newton wrote:
On Friday, 12 February 2021 at 14:32:41 UTC, The Natural Philosopher
wrote:

Dropping mail silently contravenes the protocol. But it is exactly what
big mail servers and spam filters do

How would you expect/prefer them to behave?

Bounce, or inform recipient

Trouble is the spam will be forged sender so you end up creating
unwanted backscatter. I drop all such bounce messages that do not
originate from one of my domains on the floor. There are a lot less now
than in the past so most big setups most be adopting a rule along the
lines of if it is obvious bulk UCE spam accept and then junk it.

The old Demon philosophy of accept everything and leave it to the end
user to sort it out is long gone now. Even Demon was forced to add
antispam measures to their email service after the great Swenfest.

Given that large mail providers can't perform filtering during receipt
of mail (i.e. whilst the SMTP dialogue is still underway)
They could if they wanted to

if a message is subsequently found to be spam it can't be sent back to
the purported sender as this is likely to fake and quarantining
everything doesn't really work for anyone (mail platform or recipient)
given the volumes involved.

They tend to score it as it is on its way to the recipients mailbox.
Increasingly email services no longer offer a catchall mailbox and you
have to define aliases for anything you want to be accepted.

In days of yore I used to get a lot of spam to partial Turnpike msgids
beginning Ewok & seq and before that Snews client msgids snz123456

I run a mail server and silently drop (as in don't deliver to the
recipient or bounce) the 'spammiest of spam' and have never knowingly
had a single false positive. I would expect large mail platforms to
perform just as well. False positives for 'possibly spam' are few and
far between these days too (and usually down to the poorly configured
sender domains/software).



--
Regards,
Martin Brown