A Massive Analysis of Ethereum Smart Contracts Empirical Study and Code Metrics

Abstract

In this work, we perform a comprehensive empirical study of smart contracts deployed on the ethereum blockchain. The objective of the analysis is to provide empirical results on smart contracts features, smart contract transactions within the blockchain, the role of the development community, and the source code characteristics. We collected a set of more than 10000 smart contracts source codes and a dataset of meta-data regarding their interaction with the blockchain from etherscan.io. We examined the collected data computing different statistics on naming policies, smart contract ether balance, number of smart contract transactions, functions, and other quantities characterizing the use and purpose of smart contracts. We found that the number of transactions and the balances follow power-law distributions and the software code metrics display, on average, values lower than corresponding metrics in standard software but have high variances. Focusing the attention on the 20 smart contracts with the topmost number of transactions, we found that most of them represent financial smart contracts and some of them have peculiar software development stories behind them. The results show that blockchain software is rapidly changing and evolving and it is no longer devoted only to cryptovalues applications but to general purpose computation.