Fighting spam: Theory and practice

There is now no doubt that spam is on the top of your minds. The response to last week’s column regarding the bad ideas being proposed to fix the spam problem was overwhelming – greater than I’ve ever had.

There is now no doubt that spam is on the top of your minds. The response to last week’s column regarding the bad ideas being proposed to fix the spam problem was overwhelming – greater than I’ve ever had.

Reader Arthur Byrne pointed me to a well-written paper, “Information Asymmetry and Thwarting Spam“, which discusses the mechanism that Bill Gates touted as “payment at risk.” The basic idea is a message recipient could set a price to be paid by the sender if the recipient rejected a message as unwanted. The paper is worth reading and does make it sound reasonable.

But Byrne commented: “The technical formatting of the required infrastructure and [other details] are a non-trivial set of problems.” He also was kind enough to go into these problems in some depth. As usual, the devil is in the details and the simple solution becomes really, really complicated, which makes the chances of it working next to zero.

Reader Benjamin Vogel wrote: “Let’s not also forget the legal issues if we need to pay for e-mail. If I send an e-mail and I am charged for that service, I can reasonably expect a level of service. If my e-mail is bounced, delayed or somehow lost, I could hold the ISP responsible for that lost data. In a day and age where viruses are rampant and denial-of-service attacks are common, we would be hard-pressed to find any ISP that would guarantee any level of service for their e-mail transfer. Thus, many ISPs would not even offer e-mail as a service. But no e-mail means no spam, right?”

Reader David Neill wrote: “You’re all wet, and so is Bill Gates (nice to be in the same league with that guy on something, isn’t it?). Charging for e-mail is the only way spam is ever going to come under control, because you make it advantageous for anyone with any significant volume of received mail to collect from those who originated it, with charges maybe being forwarded through those who relay mail.”

Neill adds: “The billing infrastructure can be written in Perl in less than a day. It probably can be a one-liner. Have a contest. How hard can it be?”

How optimistic can you be? Neill, if you ever had anything to do with billing systems you would understand what a nightmare they are. Tracking charges is not the only thing they have to do. There are all the supporting operations, such as reconciling accounts, authorizing payments, checking credit, chasing late payments, meeting auditing and tax requirements . . . in short, it is anything but simple.

That illustrates the core point from last week wonderfully: Your idea of a billing solution is simple, neat and wrong. The reason is that you vastly oversimplified the problem and real-world problems of any importance are, unfortunately, rarely simple.

Moreover, this is the reality of most IT problems. Just consider networking a few thousand desktops. In theory, it’s simple enough: You just run cables from the PCs and servers to hubs and routers. In practice there are the hundreds or thousands of decisions that are required to actually make everything work – where to locate sockets, how to route cables, how to deal with solid concrete walls, cross factory floors, connect buildings separated by public highways, and on, and on and on.

Yogi Berra put it succinctly: “In theory, theory and practice are the same. In practice, they are not.” Whether we’re talking about wiring PCs or fighting spam, the practical gap between theory and practice is where simple, neat solutions are proved to be wrong.

Shameless Plug Department: If you want to get a handle on enterprise messaging and spam attend the Network World Messaging and Spam Technology Tour hosted by yours truly. The dates are March 23 in Arlington, Va.; March 25 in Framingham, Mass.; March 30 in Schaumburg, Ill.; and April 1 in San Jose.