The DAO

“The DAO” is still for many people a particular initiative that happened on the Ethereum network in 2016. It was mentioned previously in the context of the hard fork which occurred in the aftermath of its failure, to nullify the damage it did.

This early attempt at a Decentralized Autonomous Organization almost destroyed the entire Ethereum commons where it was constructed, and in the end split it asunder.

The DAO aimed to create a decentralized venture capital fund, similar to Coaese’s concept of production organized through a nexus of (smart) contracts. It is unfortunate that we never got to see whether the DAO would overcome the transaction costs associated with this method of organizing production, whether it would make good or bad decisions, and whether decentralization of its “directors” would help or hinder.

The DAO was phenomenally successful as a crowdfunding effort, holding 14% of all ETH in existence ¹, worth more than $100 million. This is particularly impressive for such a novel approach which had never been tried before, and is testament to the degree of excitement and buzz that must have permeated the Ethereum community at its launch.

Before the DAO could achieve anything of consequence it was “hacked”. Someone exploited a series of vulnerabilities in its smart contracts to “steal” ETH valued at around $50 million. The DAO had been configured with a 28 day waiting period before the funds could be withdrawn, and this gave the Ethereum community time to consider how it would respond.

Some Ethereum founders and developers were likely exposed to the DAO’s losses personally, giving them an incentive to make an exception and set the network’s rules aside to nullify it. To have such a large proportion of all ETH be stolen also would not bode well for the price of the asset in a scenario where the attacker dumped even a small portion of their stolen ETH on the market. The only entity that stood to benefit directly from the enforcement of the rules in this case was the hacker.

This open access book chapter ² by Quinn DuPont provides a detailed history and ethnography of the DAO and its aftermath. It draws a stark contrast between the way the DAO’s governance was believed to function by participants and how it actually functioned in practice when under stress.

The Ethereum Foundation released new node software which defaulted to a hard fork upgrade that would undo the DAO. This was adopted by most but not all of the Ethereum ecosystem, with 15% of PoW miners refusing the hard fork and the survival of this chain giving other constituencies (developers, users) a choice to also reject the fork. The chain which persisted with the consensus rules as they were defined became known as Ethereum Classic (ETC) - it lost the right to call itself Ethereum because that trademark was controlled by the Ethereum Foundation. The implications of this for the Ethereum commons have already been considered.

The hard fork was effectively a bailout, and the nature of the crypto commons is such that this kind of rollback is always possible if the stronger constituencies within a network are negatively affected. This can act as a kind of defence mechanism too, because an external attacker who wishes to destroy the network cannot be assured that its constituents will not “fork around” them and their attack. This likely helps to discourage attacks which are very costly.

One of the lessons to be learned from the DAO is to be wary of complexity when dealing with blockchain-based assets. The “immutable” nature of these systems (when it holds) means that mistakes can result in catastrophic losses. If your autonomous organization is built on flawed foundations it can crumble in an instant. Greater complexity means it is harder to be sure that such flaws are not present, and there are great incentives for people to find them if they do exist.

In hindsight, it seems hard to believe that people were willing to entrust so much money to a brand new initiative that was deploying its very first iteration in the wild. Even without the fatal flaws, one wonders how well such an ambitious first attempt at a DAO would have made use of the resources which had been allocated to it. DAOs became less popular for a time after The DAO episode, but in mid-2019 we are witnessing a rapid proliferation of this form. This time around, even the DAOs that have been online for months or years are not being entrusted with more than a few million dollars, and we are yet to see compelling evidence that they will make efficient use of the resources that are allocated to them.

It has recently been announced that a new attempt to build a DAO with the same objectives is forthcoming and will perhaps give us the opportunity to see how the concept fares when it doesn’t get exploited at launch.

References