Another Look at ALGORAND (1905.04463v2)

Abstract: ALGORAND is a celebrated public ledger technology. In this paper, we identify several design flaws of the ALGORAND protocol. In particular, we show that the claimed (proved) fork-free property is not true and several assumptions in ALGORAND are not realistic in practice. The ALGORAND wiki page https://golden.com/wiki/Algorand claims that "the probability of a fork in the protocol is estimated at 1/1,000,000,000 and therefore blocks can be considered final upon validation". However, our first attack in this paper shows that a malicious adversary who controls less than 1/3 of the users (or money units) could fork the ALGORAND chain very easily. Our second attack shows that a malicious adversary could use a bribery attack to fork the ALGORAND chain very easily also. Furthermore, we show that the celebrated Byzantine Agreement component in ALGORAND is not necessary. The Byzantine Agreement is the most expensive part and one of the most innovative parts in the ALGORAND protocol. It is used to avoid forks in ALGORAND. We show that a simple majority vote could be used to achieve the same property that Byzantine Agreement achieves in ALGORAND under the same network assumption.