Whether executing smart contracts or trading crypto-currencies, the digital assets that blockchains protect exist only in computer code. When stolen, it is possible for hackers to evade detection by rolling back the blockchain to a previous version of the code that existed before the hack. Basically, if more than half of the computers working as nodes to service the network tell a lie, the lie will become the truth. This is exactly what happened with the Ethereum blockchain when an attacker tried to steal about $50m of the digital currency, Ether.4 Two other blockchains based on Ethereum, Krypton and Shift, suffered what are commonly referred to as 51% attacks in August 2016.5,6 The attack works when hackers are able to compromise over half the nodes participating in the distributed ledger, in which case, they can prevent new transactions from gaining confirmations and halt transactions between some or all users. They also can reverse transactions that were completed while they were in control of the network, meaning they could double-spend coins if attacking a crypto-currency blockchain. Blockchains (like all distributed systems) are not so much resistant to bad actors as they are ‘anti-fragile’ – mean ing, they respond to attacks and grow stronger. However, this requires a large network of users. If a blockchain is not a robust network with a widely distributed grid of nodes, it becomes more difficult to ensure the immutability of the ledger.