Playing Games with Cybercriminals

The talk begins with a concern that conventional policing of cybercrime, built on arrests, marketplace takedowns and botnet dismantling, may only shift activity elsewhere through displacement. It asks whether softer, economics-based levers could make criminal markets less efficient instead. The speaker, a sociologist working with experimental economists, uses the classic lemons problem of information asymmetry as a starting point.

An online experiment placed 392 paid participants into 56 small markets, each with four sellers and three buyers trading over 20 rounds. Sellers chose output, advertised quality, which could be misrepresented, and prices, while buyers saw average ratings and seller history. Against a control, three interventions were tested separately and together: a Sybil-style attack where a share of purchases never arrived, a defamation attack that randomised a share of ratings, and a combination. The Sybil-style attack cut seller profits sharply, especially in later rounds, and increased buyer inactivity, while the combined version performed about the same. The defamation attack showed little measurable effect.

The findings are presented as an early first step rather than settled answers. Limits include a narrow design, the difficulty of moving the approach from lab to field, since interception has no simple online equivalent, and the cost of building long-term undercover identities. Audience questions raised targeting high-value sellers, the closed trust networks used by some ransomware groups, and displacement. The practical takeaway is that softer economic interventions may be worth testing, with law enforcement and industry input on implementation.