Can DAOs be truly decentralised?

Written as part of the Devcon Scholars Program (Legal) for Devcon SEA. Thanks to the generosity of the Ethereum Foundation, I was invited to partake in the Devcon Scholars Program as a Legal Scholar, and attended Devcon 7 in Bangkok in November 2024 as part of that program.

“Decentralised autonomous organisations” — i.e., DAOs — are a new kind of decentralised collective in which individuals can organise and govern themselves through pre-defined and self-enforcing rules, made possible by technologies such as blockchain systems.¹ The promise of DAOs is that their decision-making and governance processes are decentralised without the involvement of centralised management structures or ownership and/or control by a single person or a group of people.

“Decentralisation”, however, is difficult to define. Work and research on measures of decentralisation are still emerging, including various classification systems and systemisations of knowledge.² Alexandra Sims notes that decentralisation could mean, amongst other things, permissionless participation or decentralised decision-making.³ Sims suggests that fully decentralised decision-making would allow DAO members to put forward as many proposals on anything they want.⁴

As a working definition, Henrik Axelsen, Johannes Jensen, and Omri Ross propose a definition of “sufficient decentralisation” that is a verifiable state where: the DAO’s design is resistant to collusion and based on long-term equilibrium, and the DAO’s governance processes have unrestricted and transparent access.⁵ They propose evaluation of sufficient decentralisation across five dimensions: (1) token weighted voting and incentives, (2) infrastructure, (3) governance, (4) escalation, and (5) reputation.⁶

Gabriel Shapiro considers “decentralisation” to mean that: “any residual human discretion (i.e., intrinsic modalities of power) are systematically dispersed over a large, agile, and potentially anonymous group of incentive-aligned persons, preferably with permissionless access of third parties to acquiring that power on fair terms”.⁷

The core of Shapiro’s characterisation turns on the dispersion of discretion amongst DAO members, and the acquisition of both membership and the means to partaking in that power are (preferably) permissionless. Note that being permissionless does not mean being open. Rather, it means that once an individual meets certain criteria — for example, by holding a relevant token — that individual can then be admitted as a member of the DAO (and, subsequently, take part in the power of that dispersed discretion) without the need for an existing member of the DAO to grant permission to the individual.⁸

I like this definition because it captures the spirit of what sets DAOs apart from modern day corporations and limited liability companies: the “decentralisation” aspect is that which allows DAO members to take part in steering the collective. By contrast, traditional companies recognised by law are typically governed, variously, by a combination of a board of directors and shareholders who help to guide the day-to-day matters of the company and hold decisive votes on significant decisions, as set out in a company constitution, articles of association, and/or a shareholders’ agreement. A regular employee will have no direct say or impact on the decisions and direction of the company. Entry into the company — whether as an employee, director, or shareholder — is permissioned (noting, however, that the acquisition of shares in a publicly traded company can be permissionless where anyone with sufficient funds can buy shares at the share price). Governance power is strictly held by directors and shareholders, and its dispersal is often governed by contractual terms.

As Sims notes, however, due to the complexities of operating DAOs and the near-impossibility of fully decentralised decision-making, many DAOs are more accurately described as “DINOs” — i.e., “DAOs in name only” or “decentralised in name only” — because elements of centralisation often persist in DAOs.⁹ And while the traditional company structure described above is antithetical to the ethos of DAOs, there may be lessons to be learned. Indeed, it may be that DAOs need some measure of centralisation in order to work.

Some common challenges faced by DAOs are as follows:

1. De facto centralisation — DAOs may begin, necessarily, with a small group that controls governance with the promise to decentralise later,¹⁰ or they may trend that way as the majority of members may not participate in voting, leading to de facto control by the group of founders or holders of significant amounts of tokens or others who choose to be actively involved.
2. Rigidity of smart contract governance — Decision-making through smart contracts on the blockchain can make it difficult to adapt how decision-making occurs, which has sometimes led to the creation of sub-DAOs where committees comprised of the DAO’s members govern specific tasks, thus avoiding the need to involve all members of the DAO in each decision. This is arguably a form of centralisation introduced into the governance process.
3. Poorer quality in governance — Decentralised governance may not produce governance outcomes as sophisticated or in as timely a manner as may be produced by a board of directors or an executive team with professional expertise.
4. Governance attacks — A single actor or a group of actors may move to take control of the DAO and its assets through the DAO’s own governance procedures in a way that does not align with the original mission of the DAO.

DAOs also face legal issues. Legally, they are not a recognised entity. Consequently, they do not receive legal protections in the same way that an incorporated company would. However, jurisdictions may nevertheless seek to analogise the structure of a particular DAO to an existing organisational structure, such as an unincorporated society, association, or partnership.¹¹ Sims notes that the law has traditionally “separated the roles of capital, decision-making and labour”, whereas DAOs often combine those roles for their members, which may expose members to personal liability where they partake in decision-making and disincentivise their participation.¹² This feeds back into the issue, referred to above, of members not choosing to take part in decision-making processes.

To mitigate the effects of some of the problems discussed above, such as poorer quality in governance, defending against governance attacks, and legal entity issues, Shapiro has proposed the creation of an entity called a “BORG” — from “cybernetic organisation” — which he defines as: “a traditional legal entity incorporated with a legal requirement to use decentralized or autonomous technologies (such as smart contracts or AI) to augment all or a portion of the entity’s governance and activities”.¹³ These can take the form of or DAO-adjacent entities or tech-augmented companies.¹⁴ Importantly, however, BORGs are not DAOs because: some of its rules will be off-chain and legal; a BORG’s rules may change without consent from its members or stakeholders because some of those rules will be governed by statute; a BORG may have its own dissolution or extinction as a goal in, for example, a merger or acquisition situation; and BORGs lack decentralisation because they may be controlled by a small group of members akin to a board of directors.¹⁵

To address trust issues, Shapiro suggests a range of on-chain and off-chain techniques that can help. I am specifically interested in the off-chain techniques that he puts forward, including using legal wrappers, using qualified code deference to clarify the embedding of some legal rules, and having an “emergency supervisor” who can step in where a BORG may be breaking some of its own rules.¹⁶

These techniques draw BORGs, and any DAOs to which they are adjacent, closer to the structure and protections of existing law. This isn’t a bad thing, and it can be seen as a recognition that there are concepts in traditional corporate law and structures that have been developed to address the same issues that arise for any individuals seeking to organise themselves as a group. As much as the ethos of DAOs and decentralisation has been to go against traditional centralisation, it isn’t necessary to throw everything out all at once — in fact, it may prove more effective to go down a hybrid, BORG-style route by beginning with traditional structures and shifting feature by feature towards decentralisation until an optimal, stable blend is achieved.

Whether true decentralisation is possible is unclear, but this kind of design approach may in turn influence the development of the underlying technology. Rather than taking what exists and trying to shoehorn a vision into it, a co-creative approach can point out what needs to be built in order to deal with the specific challenges that arise.