Thursday, December 14, 2006

Property rights on your attention

I enjoyed watching the video of this Google Tech Talk, "An Economic Response To Unsolicited Communication" by Marshall Van Alstyne about e-mail spam.

I liked Marshall's framework for the spam problem. He talked about spam as an externality, like pollution, and proposed a solution based on the Coase theorem that attempts to give people "property rights over their attention."

From his paper:
We propose an "Attention Bond," allowing recipients to define a price that senders must risk to deliver the initial message.

Requiring attention bonds creates an attention market ... to price this scarce resource. In this market, screening mechanisms shift the burden of message classification from recipients to senders, who know message content ... In certain limited cases, this leads to greater welfare than use of even "perfect" filters.
I was mostly interested in the theory discussed in the talk, but Marshall did propose an application for trying to eliminate spam. The basic idea is a whitelist system where senders not on your whitelist have to post a micropayment bond ($.01 - $.05) for you to receive the message. If you determine the message is spam, you seize the bond.

While I enjoyed the talk, the proposed solution has problems. The biggest I see is that there is a quiet assumption that e-mail senders can be identified.

Yes, if you implement a strong identification system over e-mail, you can implement all kinds of promising anti-spam solutions. However, as security guru Bruce Schneier said, "These solutions generally involve re-engineering the Internet, something that is not done lightly."

Marshall addressed other criticisms near the end of his talk, including how the system would deal with honeypots and botnets, but, I think, also may have oversimplified the challenges there.

For example, Marshall claimed that marketers would be careful who they send e-mail to, so someone who sets up a honeybot to seize "attention bonds" would not get much business. But, I suspect enterprising people would not just set up one honeybot, but billions of them, each of which has a forged identity behind it made to look as attractive as possible to marketers. True, we may not have much sympathy for e-mail marketers, but this may threaten to ruin e-mail marketing completely, which would create opposition from the business community to this system.

Slashdot has a post on Marshall Van Alstyne's work, including some snarky comments ([1] [2] [3]) in the discussion.

See also Bruce Schneier's Crypto-Gram post, "The Economics of Spam".

Thomas Claburn at InformationWeek also has an interesting article on Marshall's work with good comments from others in the field.

No comments: