TontineCoin: Survivor-based Proof-of-Stake

Proof-of-Stake cryptocurrencies avoid many of the computational and environmental costs associated with Proof-of-Work protocols. However, they must address the nothing-at-stake problem, where a validator might attempt to sign off on competing blocks, with the hopes of earning coins regardless of which block becomes accepted as part of the blockchain. Cryptocurrencies such as Tendermint resolve this challenge by requiring validators to bond coins, which can be seized from a validator that is caught signing two competing blocks. Nevertheless, as the number of validators increases, it becomes increasingly infeasible to effectively monitor all validators, and to reach consensus. In this work, we incentivize proper block monitoring by allowing validators to form tontines. In the real world, tontines are financial agreements where payouts to each member increase as the number of members decreases. In our system, a tontine is a group of validators that monitor each other's behavior, "murdering" any cheating tontine members to seize their stake. As the number of validators in a tontine is smaller than the number of validators in the currency as a whole, members can effectively police each other. We propose two methods whereby a Tendermint-like currency can be extended to allow for the creation of tontines: a pure PoS model, and a hybrid Proof-of-Stake/Proof-of-Work model. We describe snitch mechanisms for both the inter- and intra-tontine setting, argue our incentive mechanisms increase monitoring, and describe how it handles a variety of possible attacks. We extend our model to act as a validator delegated cryptocurrency, with the users having an incentive to partially participate. We show that these strategies may benefit validators as well as speed up the block formation process. Moreover, we describe a prototype implementation of TontineCoin, and perform various experiments that support our theoretical analysis