Download PDF version of this article PDF

Sender Pays?

Eric Allman, the Curmudgeon author of “The Economics of Spam” (ACM Queue 1(9), December/January 2003-2004), has it just about right regarding why we are deluged in spam: it costs no more for a sender to put out a million messages than one. Given recent statistics (spam allegedly passed 50 percent of all e-mail traffic some time ago), I suspect that the time isn’t too far off before some of the “sender pays” approaches start to happen.

Really, though, there’s nothing new about “sender pays” in the e-mail field. In many corporate environments, the network infrastructure is provided on a chargeback basis using, among other things, traffic in terms of packets or messages as a metric. In timesharing systems, I/O activity is one of the things users are charged for, along with connect time and the like.

Back in my old CompuServe days (I was a CompuServe user from the early 1980s until a few years after AOL bought them out), e-mail messages within the system were a nickel per 1,000 characters, and 15 cents per 1,000 characters to the Internet when they first made that available. Eventually, user demand and dropping costs for equipment and bandwidth forced CompuServe just to roll it all into the monthly charge, but for a long time it still charged extra for monthly e-mail sending over some reasonable amount. With charges and policies like that, very little spam was sent from CompuServe even after it started to appear in earnest in the early 1990s—until AOL took over.

Now that the Internet is fundamentally commercial, couldn’t it be thought of as essentially the RSCS (remote spooling communications subsystem) portion of a planet-scale VM system? And be billed accordingly, with perhaps some breakpoints to allow ordinary usage at a reasonable fee while discouraging damaging quantities of activity?

Would I hate it? Sure! Once you get used to having a service or resource essentially for free (pay a low monthly fee and use it as much as you want), it’s going to hurt, emotionally at least, to have to pay more. But the phone company has done it for years—and as the cable companies move in to the .comm’s business with more Internet service, they’re probably going to find out they have to do the same thing.

Mike Brady, California


I enjoyed Eric Allman’s article “The Economics of Spam” (ACM Queue 1(9), December/January 2004). It’s a very clear explanation of why spam has become a plague that’s difficult to stop.

At first sight, if you analyze the spam problem from an economic point of view, you are tempted to try to find an economic solution to it. Any kind of economic solution will imply adding a cost that will become a tax for everybody using the Internet.

I think we should follow a path different from filters to finish spam. The first and most basic idea should be to add some kind of screening for e-mail, the same as when using caller ID on the phone.

The e-mail system should include some way of knowing who’s e-mailing you before you download the messages to your computer. I know this means redefining and modifying the way the current e-mail system works, but I believe you should be able to redefine the Internet e-mail system to confront the people who take advantage of its open and good-will approach. Proposals such as the Tripoli e-mail environment are a next step in the Internet e-mail system in the 21st century. I think we need to include security aspects such as digital certificates as part of the e-mail system in order to assure identity and confidentiality in e-mails. When talking about these topics, we have these eternal questions:

When? As if there will be a future moment with no more complex technology or lesser traffic on the Internet.

How? Well, that’s why we are IT professionals, to find a way to migrate between both systems, with both systems running at the same time. And we cannot forget about making it as simple as possible for the normal user or for granny to e-mail her grandchildren.

Vladimir Sanchez, Santiago, Chile



The February issue of ACM Queue contains an error in Alexander Nareyek’s “AI in Computing Games” (see page 62). The game Counter-Strike is produced by Valve, not Microsoft.

We edit letters for content, style, and length.


Originally published in Queue vol. 2, no. 3
see this item in the ACM Digital Library


© ACM, Inc. All Rights Reserved.