mnot’s blog

Design depends largely on constraints.” — Charles Eames

Sunday, 15 February 2004


Economic Approaches to Spam

SPF is getting a lot of attention, but it’s got some pretty fundamental limitations, as well as some shorter-term practical problems. What else is there?

An approach that makes sense is to acknowledge that spam is in the eye of the beholder; rather than finding technical ways to identify unsolicited mail, just charge spammers when you don’t want their mail; economics takes care of the rest.

The Economist discusses this type of approach in its latest edition (subscription required). In particular, in (finally!) unveils the approach that Bala has been working on for a while;

Balachander Krishnamurthy, a boffin at AT&T Labs, the research arm of the phone company, proposes a system in which ISPs would establish a consortium—similar to banks creating Visa—that would act as a clearing house. ISPs would then give all of their subscribers credit limits. From then on, every time a recipient declares an e-mail “unwanted” to the clearing house, the sender is charged, say, $1. Once his credit limit—$200, for instance—is reached, the ISP shuts down his account.

The way that he explains it, it’s pretty simple; you’re issued cryptographic stamps that are used, one per message, by putting it into a header. When people decide that what you’ve sent is spam, they’ll cash a stamp, and your account will be charged. There is some overhead to the crypto, but it’s not nearly as much as other proposals which rely on the overhead to discourage spammers (this always seemed like a short-term approach). See the preliminary paper (PDF) for more.

I like this, in that I don’t have to pay for e-mail unless what I send is considered spam, and it still allows e-mail marketing, as long as the practitioners are willing to pay if people don’t like it. The only hitch is the adoption curve (which plagues just about any serious technique for battling UCE), but Bala’s employer gives him an advantage, as one of the world’s bigger ISPs.

This will be interesting to watch.