Exploring Randomness in Blockchains

Nowadays blockchain systems are widely used in many different fields, not only as a kind of payment method, i.e. cryptocurrency, but also as the infrastructure of decentralized applications. A smart contract, which is a program running on the blockchain, e.g. Ethereum, enables decentralized applications without the need for any trusted third party. To deploy a smart contract every miner of the blockchain needs to perform the same function of the smart contract such that they reach a consensus on its final state. Therefore, current blockchain systems are deterministic and non-probabilistic, disallowing any randomness in smart contracts, which is a significant limitation for the applicability of blockchains. A wide range of real-world applications depend on random functions, most obvious examples are games and lottery applications. Various methods have been proposed To address the random number generation problem, such as using a trusted oracle or the block hash. All of those have different disadvantages and advantages. Noticing the lack of concrete guidance for the inclusion of randomness in smart contracts on blockchains, we investigate the state-of-the-art random number generation methods and compare them in several critical aspects including availability, unpredictability, unbiasability, verifiability, scalability, execution time and cost.