Shooting spammers first requires that one finds them, and this proposition is complicated by their being physically situated in Brungaria, Itchistan, Lower Slobbovia, etc. So we settle for electronic countermeasures, which fall short, partly because fools are so ingenious. Spam filters are too little, too late. We have to get them before their submission even arrives at their server, much less ours. So I hereby create a new acronym: NOOI (National Opt-Out Infrastructure).

In this, there are two classes of e-mail recipients, those subscribing to the services of the infrastructure and everyone else, which presently means you. Subscribers only accept mail from other subscribers, and the mail between these parties follows a collection of very strict rules, which the recipient can enable and disable to their tastes.

Some of these rules are easy, such as a ‘reply-to’ domain that agrees with the sender’s IP address. The more subtle ones require that the sender not have any embedded scripts; does not use animated GIFs, does not use the word ‘Free’, and limits their message to what could reasonably fit on one screen. The sender is alerted by their own e-mail software that the messages they are posting are rejected when they fail to meet criteria: these are expressed in ‘Average Probability of Delivery’, which means that out of e-mail accounts generally only Y out of X will accept this particular message. If the APD is low (say below 10%), then the sender goes through the more complicated scan of SPD (Specific Probability of Delivery) where the list of intended addressees is checked. The sender is informed that 90% of the receivers reject the message because of false headers, 60% reject due to the presence of the phase ‘Russian Women’, and 99.8% reject due to the presence of script triggered by an embedded HTML ‘Submit’ button.

There are several strategies the sender can employ to get the probability of delivery raised. First is to offer something that people are actually interested in. This presumes the presence of enough intelligence on the part of the sender to be concerned with their audience’s needs and interests.

The remaining 90% that are still clueless use the subterfuges: ‘F R E Ee’, ‘Bonus bingo’, ‘Belarus and Tajikistan’, ‘Die mands’, etc. The hustle is a bit cuter. However, as readers start right-clicking on the header and selecting the ‘Reject’ option, the rapidly rising resistance to reception effectively votes the sender ‘off the island’. It still takes awhile to send an economically significant amount of spam, and the protest votes cut it off at the pass.

The ‘body filters’ are looking for ‘Platinum Card’, ‘Newt Gingrich’, ‘Trial Subscription’, etc. and discarding anything that is either too long to keep the readers attention, contains words and phases that the reader is already tired of receiving, or seems to contain a lot of tortured spelling and grammatical constructions.

Certain senders are, needless to say, exempted, such as immediate family members and working associates. Another layer of senders is given the benefit of the doubt, since their original transmissions were acceptable and their subsequent messages have proven trustworthy. As correspondence continues, the receiver can unblock messages with various kinds of attachments, various questionable words and phrases, and so on.

Another test is the longevity of the sender’s account. Those that are less than two weeks old can be held until a statistically significant number of other subscribers get messages they like. ‘Held’ messages may stay that way for months, if the sender’s address raises more than a trivial number of validation issues. Identical sender addresses that seem to simultaneously condense out of multiple disparate providers are blocked by a recipient community that doesn’t find this amusing.

The sender is now having to navigate a minefield. This raises the cost to the point where getting an actual delivery would be cheaper by the use of snail mail.

The rejection mechanism is the same as publishing in general: the sender may not know why it was rejected, but they keep trying until some strategy is successful. Such strategies might include humor, information content, desirable products and services, or simply a reputation among recipients of having properly targeted proposals. Effective feedback forces the spammer to constructively differentiate on the product axis, distill their audience list on the prospect axis, and creatively package their message on the marketing axis. The ‘cheap shot’ business disappears, the only people that stand a chance are those that know what the consumer community is interested in right now. Since that is mercurial, e-mail marketers have to stay on their toes.

What makes this work is the aggregate voice of the receivers. You can block messages with particular rules, but you don’t know what other people are blocking and vice versa. This knowledge needs to be pooled. The filter mask then migrates ‘upstream’ to the points of origin. A lot of computing is involved in making this work. However, as we have noticed recently, computing costs are trivial compared to the time wasted by people holding their fingers in the dike.