Security of Blockchains against Incentive-Driven Attacks

In traditional cryptographic analysis, an application or protocol is deemed "secure" if it realizes its task against any adversarial strategy, however mischievous. While this approach yields strong security guarantees, in many cases it's overpessimistic since it neglects the incentives that lead participants to deviate from their prescribed behavior, resulting in solutions that defend against unlikely attacks.

In this talk we present the "Rational Protocol Design" (RPD) framework, which formally incorporates adversarial incentives into a composable cryptographic protocol design. We showcase the benefits of the framework by applying it—in combination with recent developments on the composable analysis of blockchains—to analyze blockchain protocols like Bitcoin. Our treatments captures, in a cryptographic framework, how knowledge of the attacker's incentives can be used to circumvent known impossibility results and/or derive a fallback rational-security notion.