Abstract

Transaction systems, such as Bitcoin and Ethereum, have positioned blockchain as a mainstream technology. Furthermore, it has been argued that blockchain has a strong potential to drastically increase the global economy’s efficiency. However, blockchain’s vulnerability to cyber-attacks requires improvements in order for blockchain to realize its full potential. The latter is not only the opinion of cyber security experts, but also a lesson left by the cyber-attack on Bitcoin’s largest platform (Mt Gox) in 2014. The latter attack generated losses for 450 million dollars. Moreover, as blockchain-based applications become more popular, the frequency and magnitude of cyber-attacks could increase. Blockchain technology uses the most demanding kind of consensus in distributed computing systems to render maximal fault tolerance to cyber-attacks: a Byzantine agreement. The latter concept assumes that the majority of the system’s members will truthfully pass on any information they receive. Therefore, blockchain ignores the impact of cyber-attacks on well-intentioned blockchain participants’ incentives to honestly report new blocks. This paper shows that the behavioral assumptions embedded in the maximal fault tolerance analysis of blockchain-based applications are not the unique equilibrium outcome. Multiple equilibria exist because honest blockchain participants experience a trade-off between counteracting cyber attackers’ actions and correcting any noise present in the signal they receive. Moreover, even in the absence of cyber attackers, there are games where well-intentioned blockchain participants’ honest reports do not exist in equilibrium. In addition, there exist conditions under which an equilibrium, where no participant provides honest reports, renders maximal ex ante equilibrium welfare. Finally, this paper discusses when the standard fault tolerance analysis is compatible with the maximization of a system’s ex ante equilibrium welfare.