While the topic of this paper is Nigerian scammers, the model that it builds is much more general. At heart, it's a quantitative cost-benefit model of filtering attacks, vs launching unsuccessful attacks. This model can explain quite a lot of things. For example, it can model the viability frivolous lawsuits, including IP extortion (patent trolling). From that, you could extract how expensive filing (or losing) a lawsuit needs to be in order to put the breaks on patent trolling as an industry.