Paper 2021/897

A Rational Protocol Treatment of 51% Attacks

Christian Badertscher, Yun Lu, and Vassilis Zikas


Game-theoretic analyses of cryptocurrencies and---more generally---blockchain-based decentralized ledgers offer insight on their economic robustness and behavior when even their underpinning cryptographic assumptions fail. In this work we utilize the recently proposed blockchain adaptation of the rational protocol design (RPD) framework [EUROCRYPT '18] to analyze 51% double-spending attacks against Nakamoto-style proof-of-work based cryptocurrencies. We first observe a property of the originally proposed utility class that yields an unnatural conclusion against such attacks, and show how to devise a utility that avoids this pitfall and makes predictions that match the observable behavior---i.e., that renders attacking a dominant strategy in settings where an attack was indeed observed in reality. We then propose a generic remedy to the underlying protocol parameters that provably deter adversaries controlling a majority of the system's resources from attacks on blockchain consistency, including the 51% double-spending attack. This can be used as guidance to patch systems that have suffered such attacks, e.g., Ethereum Classic and Bitcoin Cash, and serves as a demonstration of the power of game-theoretic analyses.

Available format(s)
Publication info
A major revision of an IACR publication in CRYPTO 2021
blockchains and distributed ledgersrational cryptography
Contact author(s)
yunlu mail @ gmail com
vassilis zikas @ gmail com
christian badertscher @ iohk io
2021-07-01: revised
2021-07-01: received
See all versions
Short URL
Creative Commons Attribution


      author = {Christian Badertscher and Yun Lu and Vassilis Zikas},
      title = {A Rational Protocol Treatment of 51% Attacks},
      howpublished = {Cryptology ePrint Archive, Paper 2021/897},
      year = {2021},
      note = {\url{}},
      url = {}
Note: In order to protect the privacy of readers, does not use cookies or embedded third party content.