If I am uncertain about the desirability of a message, I send it to Quarantine to decide whether the message and its sender are acceptable or not.
If acceptable, I deliver the message and update my filtering rules (relax a content filter or whitelist a sender). The sender does not need to be notified because his message is delivered (although after a delay.)
If unacceptable, I will usually blacklist the sender. I have no desire to help bad actors, so I provide them with no information. Telling them that I blocked them for characteristic X will only help them try their next attack with characteristic Y.
In the less common case that a sender is acceptable but this particular message is nuisance advertising, I will silently discard the message. Telling him that I rejected one message is likely to create confusion and may suspend delivery of messages that I do want from that sender.
The technical obstacles to bounce are also significant:
- The volume of unwanted inbound messages is likely greater than your normal volume of outbound messages. Why would you want to more than double your outbound traffic volume?
- Unwanted messages are a distributed denial of service attack on your network. If you respond to those messages, you amplify the attack on yourself.
- The bounce address may be invalid, causing you to attack the bounce address, a problem called backscatter.
Similarly, I do not use SmarterMail's recipient verification option. My inbound gateway accepts everything, then does a recipient check after the SMTP session is closed, using a database query to offload that workload from my main mail server. Messages with no valid recipients are discarded quickly. Invalid recipient messages are currently 50% of my accepted messages. That is an improvement -- the percentage was 65%, until I blocked some unwanted senders and informed some business partners about terminated employees. Most of the invalid recipient traffic is from directory harvesters who are trying to guess their way into my network and possibly market my address list as well. I do not help them guess. I do run periodic reports, and where applicable, senders with high failure rates get blocked at HELO.
In a perfect world, I might want to notify a few business partners if they send to an invalid recipient, and do so while the SMTP session is open. Implementing that design requires a different product as my inbound gateway mail system, and a lot of special code, all of which seems low priority. I cannot imagine a reason to notify for any other reason than this obscure one.