The design of their solution, which is based on ideas that have been floating around on the Net for years, uses two simple ideas from economics. The first is that people usually benefit from allowing trade rather than blocking it. How does this relate to email? While it may seem a bit weird to think of email as trade (rarely is money attached), email is always about some form of exchange and in its best form contains information that is in some way valuable to the recipient. The exchange of this information is likely to be valuable to the sender as well; after all, he took the initiative to compose and send it.

So email allows people to "trade" information. However, as with any trade, if favorable terms of exchange are unknown to the parties involved, further communication may be required to discover what, if any, favorable terms exist. We normally think of this important process as negotiation or simply opportunity discovery.

As "information good", email has a property that makes discovering favorable terms of exchange a bit of a challenge: additional communication is the subject of the trade itself. The impact of this property is that email is difficult to manage through regular means. The most common type of spam control (various forms of filtering) can halt the process of discovery because it may block some communication completely. In addition, since filters alone lack a way for senders to pay receivers (to make a side payment in economics terms), such filters block exchange that could be beneficial to both parties.

Negotiations do have cost, but what if these costs could be made relatively small in comparison to those required by direct involvement of the recipient? Furthermore, what if senders who are initially rejected are given a means to buy a moment of the target recipient's time at a price set by the recipient?

If an agreement can be reached, then all parties are better off; legitimate senders have gotten their message out while the recipient has been compensated for the value of her costs for receiving, reading and processing the email. (Note that if any further communications are expected, the recipient can whitelist the sender, causing the process of negotiation to be skipped. Of course, whitelist status can be revised - the recipient can always remove someone.)

Let's assume that this negotiation can be carried out between the sender and recipient using software they control and which acts in their behalf. The trick is to ensure that the recipient's "negotiator" screens out the undesirable messages but allows the desirable messages pass through unimpeded. If not using a traditional filter, how should such a negotiator be configured, and how does it actually accomplish the screening?

Effective screening hinges on the second key idea from economics, that of information asymmetry. Critically, a person writing a message knows more about the message than the person receiving it, especially before the receiver has read it. This presents an information asymmetry problem, where one party has knowledge the other does not. With email, spammers know they are sending spam, and your long-lost high school buddy knows she is your long-lost high school buddy. Yet you can't know which one wrote the email until you've read it.

Getting the right information exchanged involves a shift in focus: a shift from the information available in a message to the information known to the sender. It is one application of the line of research by Akerlof, Spence, and Stiglitz that received the 2001 Nobel Prize. The solution is to get the party with the information to make a commitment based on that private knowledge.

Through the use of an appropriate economic mechanism, the sender may make such a commitment (an economic mechanism, as studied in the field of mechanism design, is a set of rules, roles, and processes for exchange). The commitment is essentially a warranty in the form of a monetary pledge or posted bond. No spammer is willing to pledge that his email is not spam when he knows that it is and his bond is at stake. In contrast, your long-lost high school buddy knows perfectly well that her message is not spam and so she can happily pledge that it is not spam.

The warranty (technically a bond since it is posted in advance) has the same effect as it does for cars, computers and other consumer goods. You have much more reason to believe that the car you are buying is a lemon if the seller offers it "as is" than if the seller offers a solid warranty. In the case of the mechanism for email, however, the recipient (buyer) gets to choose the size of the bond; it might be 1 cent, $1, or any amount he chooses.

Even a with a very small bond, the numbers work in favor of your long-lost high school buddy but heavily against the spammer sending millions of messages. Marketers are therefore forced to become more focused and less wasteful in their communications. The bond creates the motivation for the marketplace to improve the quality of information going between the sender and recipient. Since it will be possible (and perhaps likely) that people will setup mailboxes for the sole purpose of collecting the warranties of unwary marketers, those marketers that stay in the game will have a real financial reason to be careful in regard to whom they contact.

Using sender warranties is different from other sender-pays proposals in that there is no fixed per-email delivery charge and no electronic stamp. Also, there is no requirement to classify types of mail. Rather, the sender risks the bond amount, and only takes this risk for email when his identity is absent from the recipient's pre-approved whitelist. Email from whitelisted senders may pass through to the recipient without a challenge and therefore without a cost. Should a sender receive a challenge, they can always opt not to post the bond (and not to send the email).

A well integrated software design can make most of the setup and management activities nearly if not entirely transparent to the average user. Although an infrastructure is required to support sender warranties, it's complexity need not be in the face of the user (people buy securities via the web and yet the behind-the-scenes complexity is astounding.) With a simple mouse-click, the recipient can claim or release the bond.

Aside from just blocking email spam, the mechanism itself is general and can apply just as easily to instant message spam or unwanted SMS messages (SMS is a popular medium in Europe). It can also be applied to telemarketers, who interrupt you at home.

There are some nice secondary effects. Once an infrastructure is established, it becomes available for other uses. The same payment system used for warranties could be used for buying information goods or physical products. The potential to receive payment from strangers initiating a first contact removes the fear of publishing one's email address. In fact, it is motivation for mailbox owners to share their contact information on websites and directories, causing a desirable reduction in market friction (figuring out how to reach some one will become easy). Ultimately, the mechanism restores control of your time, attention, and contact information to you.

The other documents are:

• a FAQ, which provides a less technical introduction to the ABM
• an Overview of the ABM protocol that provides a step-by-step walkthrough of the concept protocol
• an academic treatment entitled Information Asymmetry and Thwarting Spam that provides a rigorous formal analysis.