Blockchain applications to decentralise finance (DeFi) are quickly developing and are gaining more and more traction in practice as well as in policy and academic debates. One of the most far-reaching applications of this sort are Decentralised Autonomous Organisations (DAOs) whose aim is to decentralise entrepreneurial decision-making through smart contracts, creating a new form of business organisation.
Many only know DAOs because of the hack that happened in 2016 to ‘The DAO project’, a smart contract running on Ethereum that should have functioned as a venture capitalist. If the DAO had not failed because of the initial hack, the decision on which projects to undertake should have been made with decentralised voting by token holders. Since then, the technology has evolved considerably and sizably, and running projects already exist.
This post aims at informing the discussion on DAOs and business organisation theory. We discuss the theory of business organisation applied to DAOs, present preliminary anecdotal evidence on different levels of governance decentralisation of two successful projects and conclude briefly with a contrast of these theoretical and anecdotal findings.
DAOs and the theory of business organisation
As any blockchain application, DAOs emphasise decentralised decision-making. This is at odds with the basic building blocks of the mainstream theories of business organisation. According to these, centralised decision-making is necessary to establish an entity as a bundle of property rights over assets, whose members commit capital to the common venture and have the entity shielded by the legal personality granted by law. The ownership of the members (or shareholders) and the (centralised) control of the managers are separated by design. Such separation generates agency conflicts.
However, if we stretch the idea behind DAOs to its ultimate potential, most of the theory of business organisation should be rewritten. The blockchain (allegedly) assures that property rights over assets are clearly allocated, since smart contracts are, in principle, always complete. Moreover, smart contracts prevent personal creditors from forcing the liquidation of the venture, effectively shielding the entity without legal personality. Most relevant for this post is the decentralised decision-making through ‘governance tokens’ protocols. This (allegedly) eliminates the separation of ownership and control now that those who commit the capital are the ones making the decisions directly, obviating agency problems. Thus, for DAOs corporate law would be neither essential nor enabling, but simply trivial. The question remains whether DAOs are as decentralised in practice as their theoretical foundations suggest.
Preliminary anecdotal evidence
Even though all DAOs make use of smart contracts on the blockchain, a myriad of different procedures can be coded so to allocate control to capital providers. Relatedly, the level of decentralisation of voting can vary depending on the initial distribution model of governance tokens. By distributing these tokens, decision-making power is allocated. Some DAOs allow members with a certain amount of governance tokens to put forward proposals, whereas other DAOs envisage a multi-tier token system whereby only holders of a specific type of token can advance proposals on which the others vote. Further, developers may have veto rights on the implementation of proposals, and so on.
A DAO often launches in a centralised manner, granting a sufficient amount of governance tokens to the developers so to safeguard its core functionality and implement necessary upgrades. After this starting phase, the governance tokens are allocated amid the other (non-developing) capital providers in a manner established in the protocol.
Previous empirical studies showed that wealth in DeFi tends to concentrate because of capital accumulation incentives but also, to a larger degree, because of the initial distribution of governance tokens. Any DAO distributes its governance tokens differently, depending on its specific goals and desired governance structure. Governance tokens grant voting power; thus, the distribution of these tokens determines the level of decentralisation.
We consider the governance of two of the most successful DAOs: Compound and Uniswap. Compound is an algorithmic money market protocol on Ethereum worth 1.8 million Ether (ETH), representing around 5 billion dollars. Uniswap is a protocol for token exchange on Ethereum worth 775 thousands ETH, representing around 2 billion dollars. These projects are sizeable and successful enough to preliminary assess their ‘decentralised governance’. We specifically consider the level of such decentralisation.
Compound and Uniswap both reserved some governance tokens to the founders and initial developers (respectively 26% and 22%) and distributed the rest, but they did it differently.
Compound distributed a portion of its governance tokens to founders and investors, who could subsequently freely delegate the associated votes to other participants or vote the tokens themselves.
Uniswap also applied a so-called retrospective distribution. Not only developers and investors, but also past traders had the chance to be involved in governance decisions.
The subsequent distribution of the governance tokens showed that the method adopted by Uniswap allowed for more decentralised governance. However, both have not yet succeeded in achieving remarkable levels of decentralisation. It has been calculated that the top 5 addresses holding Compound tokens have 42.18% of the voting power, whereas the top 5 addresses holding Uniswap tokens have 12.05% of the voting power (55.69% considering the top 100). The estimation is conservative, as it is complex to identify token holders in the blockchain. Therefore, the level of concentration may be even higher.
This exemplifies how DAOs’ governance structure largely depends on the initial distribution and the protocol-specific voting mechanisms. Though the eventual level of decentralisation is not entirely clear for every case, many DAOs make use of an initial centralised allocation of tokens (see, for instance, Index Coop, Alchemy and Habitat).
Crucially, the characteristics of the ‘decentralised and automated’ governance of DAOs are set centrally by the founders and the team of core developers and may well be resistant to future changes. In this context, many of the typical aspects of corporate law and governance arise: one can think of resolving conflicts between minority and majority token holders, duties of developers vis-à-vis token holders, the possible restrictions of some governance structures, and so forth. In other words, there are considerable islands of conscious power in the ocean of unconscious (and automated) co-operation and corporate law will prove all but trivial in addressing the challenges ahead.
Edoardo Martino is an Assistant Professor of Law at the Amsterdam Center for Law & Economics (ACLE) and a Research Associate at the European Banking Institute.
Simone Spijkerman is a Research Intern at the Amsterdam Center for Law & Economics (ACLE).