Why have Bitcoin, with its distributed consistent ledger, and now Ethereum with its support for fully fledged “smart contracts,” captured the imagination of so many people, both within and beyond the tech industry? The promise to replace obscure stores of information and arcane contract rules – with their inefficient, ambiguous, and primitive human interpretations – with publicly visible decentralized ledgers reflects the growing technological zeitgeist in their guarantee that all participants would know and be able to foresee the consequences of both their own actions and the actions of all others. The precise specification of contracts as code, with clauses automatically executed depending on certain sets of events and permissible user actions, represents for some a true state of utopia.
Regardless of one’s views on the potential for distributed ledgers, one of the most notable innovations that smart contracts have enabled thus far is the idea of a DAO (Decentralized Autonomous Organization), which is a specific type of investment contract, by which members individually contribute value that then gets collectively invested under some governance model. In truly transparent fashion, the details of this governance model, including who can vote and how many votes are required for a successful proposal, are all encoded in a smart contract that is published (and thus globally visible) on the distributed ledger.
Today, this vision met a serious stumbling block: a “bug” in the contract of the first majorly successful DAO (which broke records by raising 11 million ether, the equivalent of 150 million USD, in its first two weeks of operation) allowed third parties to start draining its funds, and to eventually make off with 4% of all ether. The immediate response of the Ethereum and DAO community was to suspend activity – seemingly an anathema for a ledger designed to provide high resiliency and availability – and propose two potential solutions: a “soft-fork” that would impose additional rules on miners in order to exclude all future transactions that try to use the stolen ether, or, more drastically (and running directly contrary to the immutability of the ledger), a “hard-fork” that would roll back the transactions in which the attack took place, in addition to the many legitimate transactions that took place concurrently. Interestingly, a variant of the bug that enabled the hack was known to and dismissed by the creators of the DAO (and the wider Ethereum community).
While some may be surprised by this series of events, Maurice Wilkes, designer of the EDSAC, one of the first computers, reflected that “[…] the realization came over me with full force that a good part of the remainder of my life was going to be spent in finding errors in my own programs.” It is not the case that because a program is precisely defined it is easy to foresee what it will do once executed on its own under the control of users. In fact, Rice’s theorem explicitly states that it is not possible in general to show that the result of programs, and thus smart contracts, will have any specific non-trivial property.
This forms the basis on which modern verification techniques operate: they try to define subsets of programs for which it is possible to prove some properties (e.g., through typing), or attempt to prove properties in a post-hoc way (e.g., through verification), but under the understanding that they may fail in general. There is thus no scientific basis on which one can assert generally that smart contracts can easily provide clarity into and foresight of their consequences.
The unfolding story of the DAO and its consequences for the Ethereum community offers two interesting insights. First, as a sign that the field is maturing, there is an explicit call for understanding the computational space of safe contracts, and contracts with foreseeable consequences. Second, it suggests the need for smart contracts protecting significant assets to include external, possibly social, mechanisms in order to unlock significant value transfers. The willingness of exchanges to suspend trading and of the Ethereum developers to suggest a hard-fork is a last-resort example of such a social mechanism. Thus, politics – the discipline of collective management – reasserts itself as having primacy over human affairs.
This is absolutely right. Back in the nineties, many people got into cryptography research in the belief that new technical mechanisms from digital cash through online elections to censorship-resistant systems and peer-to-peer sharing could change the world. Indeed they did, but rarely in the ways the pioneers expected. The music majors lost their grip on the industry, but power went to Apple, Amazon and Google rather than to musicians. Annoying bank branches were replaced by even more annoying bank websites. And the losers from globalisation are starting to find their voice, in one country after another, in ways that might be profoundly damaging. Technology cannot replace politics; but when most politicians don’t understand technology, where is the leadership to come from?