An acquaintance wondered why the people who run the systems that receive mail get to make all the rules about what gets delivered. After all, he noted:

The sender pays for bandwidth and agrees to abide by the bandwidth provider’s rules.

It is useful to think of the Internet as a collection of tubes, all leading from the periphery to the middle, where the middle is approximately “the peering point.” The sender has paid for the tubes leading from himself to the middle, but he most definitely has not paid for any of the other tubes through which his mail might flow from the middle to the recipients, nor does he have any contractual or financial relationship with any of the providers of those other tubes. Furthermore, due to the asymmetric way that e-mail works, the aggregate amount that recipients spend to process his mail (deliver, filter, reject, whatever) dwarfs the amount he spent to send it. Bulk mail senders deeply do not want to believe these facts, but they remain true whether they believe them or not.

Recipient networks subsidize mail delivery because it’s what their users want. If the users complain, which they do with great vigor and frequency whenever [a sender with bad practices] turns on the faucet, well, what do you expect?

There have been a variety of attempts for senders to share the costs of the recipients’ tubes, of which the only one that is even arguably successful is Goodmail.

You could imagine a system where recipients are required to deliver mail, but then you would have to add a system of cost settlements along the lines of those used by phone companies or postal systems, which would cost more than the entire existing system does, if you could even build it which you probably couldn’t. (See my white paper for why.) Furthermore, you would have a whole new legal mess as bad guys gamed the new system, which they certainly would because real money would be involved, and in practice you’d probably have to set rules for entry into the e-mail system analogous to those involved in being a telco or a post office, to get the necessary financial accountability from the players, and maybe not even then. (Who’d trust a network in Nigeria?)
Really, you don’t want to go there.

He responded:

All that being written, from a philosophical perspective, I continue to grapple with the issue that any group can make a decision that permanently prevents a particular individual or organization to participate in a specific activity without any form of public process. It just does not sit well with me.

I had two somewhat unrelated responses:

A) Every form of communication ever invented has been abused by bad guys if the bandwidth isn’t rationed either by physical characteristics, by restrictions on who can use it, or by price. I gather that there were short-lived flat rate telegraph tariffs in the late 1800s which were quickly changed when people started sending sleazy junk messages, and free TDD-to-voice gateways intended for deaf people in North America have an incredible problem with 419’ers scamming through the gateways.

The predecessors to Internet mail were either metered per message, like MCI Mail and Compuserve, or had enforceable AUPs with every user, like the Arpanet. Now we’re in a situation I think is pretty much unprecedented, a communication technology with flat rate pricing, no meaningful physical restriction on sending, and no limits on who can use it. This means that mail systems have little clear precedent for anti-abuse rules, with superficially plausible precedents like paper mail and telephones turning out not to apply.

So the model we use is roughly that of the company mailroom. Ages ago I wrote a note about rusty pickup trucks driving up to the company mail room and dropping off giant stacks of cheaply printed circulars, pointing out that it shouldn’t be a big surprise if the mailroom staff didn’t give those the highest priority when sorting and delivering.

If you think about it, it’s quite remarkable how well the mail system still works. You can set up a brand new mail server and so long as you get the technical details right (proper DNS and rDNS, IP in a range appropriate for servers, etc.) for the most part so long as you don’t send a vast amount of mail, people will still just accept it.

B) Recipients’ opinions of senders really do evolve. Spamhaus drops listings all the time, and there are plenty of senders who used to be listed, cleaned up their act, and now we’ve forgotten about them because their mail doesn’t annoy anyone. But the way you get people to improve their opinion of you is to behave yourself for long enough that they forget that they used to hate you. Mailers say “we used to be naughty but now we’re nice” all the time, and the most reasonable response is “we don’t believe you, and we don’t care enough to argue.” We’ve been burned too many times.

I fully understand that it is a very difficult business problem to migrate from spamming to legitimate mail, since there’s a revenue hole from the time that the spamming stops to the time that one can start successfully sending legit mail, but that’s not our problem.