Many of the web sites that discuss backscatter encourage mail administrators to not further contribute to the problem of backscatter. I would be remiss if I did not include a section on it.
While that is the primary problem of backscatter, Al Iverson of Spam Resource has some more suggestions for not contributing to the problem of backscatter, some of which I have paraphrased and reprinted below.
Don't use Challenge/Response, and don't allow your users to, either. While Challenge/Response spam filters fix your spam problem, it irritates and annoys everyone else. SpamCop says that this is a selfish method of spam filtering, and I agree. It offloads the spam filtering onto the senders (ie, I send you a message and you expect me to filter it... hey, filter your own mail!). But even worse, if a spammer spoofs an email address and you send a challenge back to the sender in the MAIL FROM, you are sending piles of challenge notifications to people who never sent mail. In other words, you become a source of backscatter. Don't do this.
Configure your virus scanner to silently strip or discard viruses/worms instead of sending a notification back to the sender. Viruses and worms that sent via email do not have valid sender addresses, they are spoofed. If your mail filter catches the virus and sends a notification back to the sender, it goes to the wrong person, an innocent third-party. This is backscatter, and your server just succeeded in annoying somebody. It's the same as the Challenge/Response problem.
Don't run autoresponders, out-of-office notifications, etc. The idea here is that autoresponders annoy people and your mail could end up being blocked. The issue here is the same as the above, spammer sends a mail to your autoresponder which bounces a message back to the "sender" who is thinking "Who is this guy? I don't care if he is out-of-office." I have mixed feelings on this one. While on the one hand I can see the anti-autoresponder's point, on the other hand I find OOF notifications very useful. Whenever I send someone at my company an email and I get an OOF notification, I am glad to know that their response will be delayed. Similarly, when I have been on vacation and people sent me mail, they were wondering where I was because I hadn't responded. So, in the business world I believe that there is a need to use an autoresponse/OOF notification. Microsoft Exchange has the ability to separate autoresponders to internal users vs external users. At the very least, I think that you should let people know you are OOF if they are within your organization. You might be able to get away with not setting anything for senders outside of your org. Or maybe you only send auto-responses to senders who pass a DKIM or SPF check.
So, even if we all can't stop backscatter, let's not make it worse by contributing to it. Don't send mail automatically if can't deliver it or without verifying that whoever sent the message actually sent it.
"You might be able to get away with not setting anything for senders outside of your org."
In my experience, it's the opposite. Awareness of vacation time can be made available via other methods (FREE/BUSY, internal company calendaring, etc). It's untimely customer/client communication that companies are trying to avoid with out-of-office messages. And those companies would need a very compelling business case for turning the feature off.
C/R, vacation, and similar pains are easy to handle. If receivers decide to use them anyway they MUST follow RFC 3834 and maybe 5230 (for SIEVE), i.e. send their crap to the envelope sender address. That avoids issues with mailing lists (=> deactivation or forced unsubscription, but no SpamCop report). If those users are no hopeless cases they could limit their "efforts" to white listed (= known) addresses and/or SPF PASS.
Another way to reduce backscatter at the border could be "MINGER" specified in an Internet Draft: Roughly a very cheap way for border MTAs to figure out which mailboxes exist and are ready (anything else can be directly rejected at the border, instead of bouncing it later causing backscatter).
Around the internet world, specifically dealing with email and MTAs, there are people who are familiar