TCS Daily

Worse Than Spam

By Arnold Kling - June 12, 2003 12:00 AM

"...public policy that applies governance models from traditional telecommunications to the Internet [is] one of the few ways to do real damage by making the Internet brittle and backward-looking rather than resilient and full of opportunity. It is the duty of those who set policy to understand why the Internet works and the limits and dangers of trying to 'do good'."
- Bob Frankston

Once upon a time, Americans trusted in self-reliance and believed in limitations on government. Today, it seems, many Americans trust in government and believe in limitations on self-reliance. This modern attitude collides with the architecture of the Internet, which was not designed to be a nanny state.

Perhaps the most dramatic illustration of the way that Internet architecture supports self-reliance rather than regulation is the persistent problem of unsolicited commercial email, commonly known as spam. There are solutions to spam based on self-reliance that are reasonably effective. Other solutions, which threaten the architecture of Internet email, are much worse than the problem.

The Economics of Spam

Spam earns a profit for senders. A seller of a product or service will pay a spammer to send an email advertisement to millions of users. Do the sellers obtain enough orders to recover the prices charged by spammers? I would think not, but evidently the spammers have no trouble finding businesses willing to pay to send spam, so there must be a lot of people who believe that spamming works.

Spam imposes a cost on those of us who do not like to get it. We spend time sorting through our email and deleting spam. Spam also uses up bandwidth and storage space at Internet Service Providers (ISPs). Without spam, they could operate with less capacity, and they could pass along the reduced costs to consumers.

Spammers, and the companies that use spammers, do not pay for the costs that they impose on ISPs and on those of us who do not wish to receive spam. Understandably, this upsets many people, some of whom cry out for government to "do something." For example, Christopher Caldwell wrote for the Weekly Standard that
"there is no chance that the Internet will return to its old level of user-friendliness until lawmakers recognize that the decision to leave it unregulated was a serious, ideologically driven mistake."

Caldwell is by no means alone. Anti-spam legislation has been passed in many states, and a number of bills are pending in Congress. All such bills are misguided. The cost of creating the mechanisms on the Internet to track down spammers in order to tax or regulate spam is greater than the cost of spam itself.

Engineering, not Ideology

As Frankston points out, the Internet's email protocol, called Simple Mail Transfer Protocol (SMTP), is designed to be fast but stupid. I like to use Ed Krol's metaphor of the Pony Express. The horseman galloping to the next station does not have time to examine the return address or peek inside the envelopes and look for spam. His job is just to bring the mail closer to its destination. Similarly, the routers on the Internet are not designed to open up the packets and check for spam. Their job is to move the packets to their destination.

This separation between header information (which tells where the email is going) from the content of the email is fundamental to the Internet. It is an engineering decision, not an ideological one. Trying to re-engineer email without that distinction would be like trying to re-engineer the financial system to operate using Roman Numerals.

Sentries for Senders

Because it is impractical to inspect email packets while they travel, spam can only be controlled at the endpoints. You either have to stop it at the spammer's computer, at the recipient's mail server (such as AOL or Earthlink), or at the recipient's computer.

In order to stop spam at the spammer's computer, each provider of Internet service would have to create a sentry system of some sort for its email senders. For example, Yahoo! recently announced that its email senders will be subject to a "verification" system. When you send an email from your Yahoo! account, you may be asked to view an image that contains a code and type in the code. The idea is to force you to be personally involved in the process of sending email. As an email service provider Yahoo! has every right to try this and other means to stop its users from sending spam. I hope that Yahoo!, AOL, and other large email providers come up with means to prevent the use of their services to initiate spam without causing too much inconvenience to their non-spamming customers.

Basic Email Efficiency

I would conjecture that spam is not the major cause of people wasting time handling their email. The biggest cause of poor email productivity is that people leave email in the "in" box too long.

I define basic email efficiency as quickly sorting email into folders and deleting unwanted mail. People who follow these practices are able to use email far more productively than people who fail to follow such practices. Spam causes a decline in productivity for both types of users, but the impact is relatively small. People who fail to practice basic email efficiency are wrong to blame their poor productivity on spam. Even without spam, their "in" boxes will tend to become crowded and difficult to use.

Spam Filtering

Most email programs have features that allow you to sort mail into folders using keywords. This helps to implement basic email efficiency. It also provides anyone with the ability to create spam filters. Recently, Mark Hurst explained how he employs spam filters. He writes, "I spend less than two minutes per day dealing with my incoming spam... Spam isn't a problem for me, and it shouldn't be a problem for you."

I use a system that is slightly more sophisticated than Mark's. I use an open source program called Popfile to pre-process email before putting it through my email filters. Popfile employs a statistical algorithm, which tends to be more accurate than rule-based algorithms for sorting mail. I am a strong believer in statistical methods, and I am very happy with Popfile. Here is how I deal with spam.


I wish that spam would go away. I am against spam. But all solutions to the spam problem are costly. The least costly solution is self-reliance.

The way I see it, becoming self-reliant takes two steps. The first step is to adopt basic email efficiency. This requires discipline, organization, and a willingness to change bad habits. That can be quite difficult.

The second step is to configure mail filters. This requires a bit of technical sophistication (particularly to use Popfile as I do), but not much. Moreover, we can expect each new generation of email programs to make filtering more powerful and easier to implement.

The alternative of regulation requires fundamental changes to the architecture of the Internet. The Internet is designed for self-reliance, not for centralized regulation. While Christopher Caldwell and many legislators wish that government regulation of the virtual world could rise to the level found in the physical world, there are those of us who would prefer the opposite.

TCS Daily Archives