How to configure proof-of-work functions to stop spam

Abstract

Spam email is a growing problem for today's Internet infrastructure. Besides many filtering techniques, proof-of-work functions have been proposed to fight spam email. Proof-of-work functions are moderately hard cryptographic functions which allow a computer to proof that a certain amount of resources has been spent. Since spammers have limited resources calculating proof-of-work functions can reduce the amount of email they can send out. In this paper, we put the costs of calculating proof-of-work functions in relation to the potential profit. This relation must be known in order to parameterize these functions such that a spammer makes no profit. We investigate in detail the monetary costs of different categories of proof-of-work functions. This allows us to determine how much hardware resources have to be spent per email message in order to make sending spam email unprofitable. The main result of our work is that proof-of-work functions must be a lot harder to calculate than usually assumed by other authors.