What is a good DAO (and how do we know)?
The following is a transcript of my talk at DAWO24 in Winterthur, Switzerland, on 4 July 2024.
These days, I mostly introduce myself as an ethnographer of on-chain communities. People often misinterpret that as someone who studies ‘vibes’ or group sentiment inside Discord servers. Researching ‘community’ is instead about seeking to understand the sphere of activity that exists outside the market and the state, which, from my and others’ observations, is marginalized by contemporary institutions. Community technology initiatives are often dismissed as weak ‘alternatives’ to the mainstream, but for those involved, they are expressions of local meaning, knowledge, agency, cooperative action, and shared practice. While my training was in political science, I taught myself ethnography because, flawed as it is, it remains the only method for understanding group interaction and coordination, including when it breaks down or is undermined.
When I first encountered Ethereum in 2016, the notion of technology-enabled ‘sovereignties’ was not new to me (see ‘Our Media’). What has been surprising is the speed at which it has become a viable prospect for a different institutional approach.
DAOs are not just bank accounts for subreddits, as Aaron Wright once said (although they can certainly be that too). DAOs are a technology that can assist groups in coordinating better internally and with each other. As a result, the domain we call civil society, community, or the third sector, can potentially develop capabilities that go beyond what firms do today. My talk today draws on various strands of my ethnographic research, projecting them into a possible future through a mash-up Entity that includes a DAO.
The Entity
The Entity I am going to describe does not exist but has elements of some real and observable projects. It involves:
- An automated bureaucracy (the DAO component).
- A system to organize knowledge and to collectively agree on the actions it hopes to achieve with that knowledge.
- A contributions system.
- A system for mutual support across entities that share a common interest/concern.
Why do we need another imaginary DAO?
My intention today is twofold:
1. To locate DAOs further along an evolutionary timeline of knowledge technologies in general.
2. To consider how we can know whether a DAO (or the entity associated with it) has remained on course.
This rough table suggests evolution or transformation in knowledge-sharing technologies for individuals and organizations. The change described starts with private knowledge, which becomes increasingly shareable through technological advancements like writing, printing, and social media. The left column represents self-expression, where it’s up to the individual what they choose to share (within cultural and institutional boundaries). The right column represents activities conducted with others, where organizational rules make group efforts and relationships legible to the state and each other.
I use North, Wallis, and Weingast’s (2009) definition of organizations, which provides people with:
- A means to coordinate behavior.
- Where the organization is more than the sum of the actions of individuals.
- Involves internal rules, norms, and shared beliefs that influence how people behave within its boundaries (p15–16).
The individual column has dominated research on knowledge technologies to date, including work on platform governance and the political consequences of social media. Organizational or group knowledge has SAAS. Nonetheless, in a legally-constituted organization (including an unincorporated association), knowledge is generally only accessible to other organizations in a limited form through annual reports, marketing, and standards bodies, agreements, or industry associations where they choose to share with other organizations.
Firms have succeeded by internalizing costs, including knowledge management, so they have what they need at hand, thereby reducing transaction costs and developing or accruing capabilities. Communities, including associations, can struggle at this.
DAOs promise to level the playing field on some dimensions by automating processes, providing even the most loosely formed communities of interest, such as subcultures, with the ability to vote or manage a treasury. But they also have vulnerabilities, some of which are not to do with the code but relate to knowledge problems at the community level.
The Future DAO & Entity
The DAO that lives in this possible future was established by a group of people seeking to regenerate an area of land. Those involved set out to give rights to nature and wanted to create what they called a self-defending ecological institution (here I am explicitly drawing on Austin Wade Smith’s work). People contribute to regenerating and protecting this land through routine, ritualistic, or creative actions. These activities are recorded in a knowledge system that also holds information collected from monitoring equipment and processes.
The DAO is not designed to achieve a predefined impact or reach an externally set target but to aid efforts to keep an ecology intact and thriving, which requires that certain things be done and others avoided. Nature might have a deciding vote or economic power so that this land can never be used for pure extraction again. You are rewarded for contributing, and with that comes stewardship rights.
By this indeterminate point in the future, there are other similar autonomous ecological entities, and they have the ability to share knowledge and resources easily, to collectively govern areas of common concern. This DAO is not pure fiction — its code base is being created now.
A couple of months ago, I observed Regen Network create a pilot for such a DAO at Wilbur Hot Springs in California. My role there was to write a speculative ethnography — a fiction based on observations of that place and the components of the system being built (you can read that here). From that experience, I take the animating purpose of the Entity and its associated DAO — the ambition to harness local knowledge and steer towards ecological benefit.
The Robot Bureaucrat
Standard definitions of DAOs begin with smart contracts and their use in solving coordination problems. I tend to agree with the following definition of DAOs, which emphasizes programmability:
DAOs are limited-programmability robots where the limited programmability power is widely dispersed among users who can only adjust the program in accordance with hard-coded meta-programming rules — Delphi Digital
There is a DAO-adjacent entity, which I will refer to as the Entity. From a political science perspective, the software component of a DAO is a robot bureaucrat tasked with implementing policies and procedures set by the political principals. It operates within a defined scope to produce services and coordinate activities among its members (see my extended explanation here). We know from experience now that DAO coordination problems extend beyond what smart contracts can achieve.
A functional bureaucracy is one that is not prone to political corruption. As per Fukuyama’s framework for assessing bureaucracies, a DAO must balance capacity and autonomy:
Capacity refers to the DAO’s effectiveness in performing its designated tasks and services, including technical competence and the ability to execute functions reliably.
Autonomy indicates the degree to which the software component of the DAO can operate independently, without political intervention and avoiding corruption.
These two goals bring us to knowledge problems and questions of governance, including membership boundaries, who the political principals are, and their ability to change what occurs in this DAO when needed. For this, our Entity needs technical capabilities too.
Knowledge Organisation Infrastructure
DAO-adjacent entities have knowledge needs. In the case of this future Entity, it needs to conduct planning, keep track of monitoring information, and retain knowledge offered by those who know about this land and its needs. Perhaps there is an Indigenous group involved who need to document stories and practices to activate and manage Indigenous cultural IP.
The Entity establishes a knowledge system that enables them not only to keep information but to express collective decisions over how that knowledge is used. They can govern their knowledge and decide who else it is shared with and on what terms. A project I am leading with Michael Zargham at Metagov is designed to achieve this. The KOI-Pond project is an early-stage attempt to create knowledge networks (see this article by Zargham and team at Blockscience). A KOI stands for Knowledge Organization Infrastructure, and it’s a graph structure within which knowledge objects can be described using reference identifiers.
The KOI enables people to express intents about how knowledge objects they are associated with will be used. This differs from privacy terms, which only say what you opt into and out of regarding how your data enters (or not) a system. The KOI-Pond provides agency and governability, which we hope will result in a governed context that provides automation technologies with parameters and boundaries.
An LLM can plug into the controlled context of the KOI. One aim of Metagov’s KOI-Pond is to ascertain whether the LLM produces outputs better informed by the organization’s rules, practices, and norms (as per this article). The KOI is therefore a means to activate the collective intelligence of the Entity’s knowledge, a memory system that requires maintenance, classification, and sorting at a collective level. The Entity can also create specific microservices that use specific knowledge objects for particular purposes.
For our future Entity, the outcomes of using a KOI include better onboarding as people can easily access knowledge from before their arrival. It will be easier to rotate or hand over leadership because knowledge is shared, and there is a continuous collective agreement in place over how that knowledge is to be used. The Entity can also share that knowledge with other like-minded Entities, increasing the capabilities of all. The KOI, therefore, brings us closer to organizations that are ‘shareable’ in the sense of a mesh of organizational intelligence that enhances capabilities among networks of self-governing organizations.
The KOI also stores information about who is associated with a particular knowledge object, be it a team or an individual.
Contribution System
Attaching identities — individuals or teams — to knowledge objects is useful for sorting and expressing intents. If we add a contribution system, it also opens new economic capabilities. Contribution systems are not simply ‘work for a DAO’ tools. A contribution system is a general class of solution that Jason Potts and I argue is an under-recognized component of the emerging on-chain economy. They are an approach to addressing how an ensemble of people can cooperate under a rule system to achieve a common joint task that benefits from the participation of many, and who each seek to benefit from the joint production.
A simple example of a contribution system is academic citations (see Kealey &Ricketts 2014 who define a contribution good is defined as a good whose benefits are non-rival over contributors but that cannot be accessed by non-contributors).
Blockchains are contribution systems. For instance, Ethereum achieves security by rewarding validating nodes and penalizing them for bad behavior. Our work on contribution systems commenced from my ethnography of SourceCred, which attempted to reward people for their contributions to the SourceCred product. The catch was that it was a permissionless organization — anyone could enter or leave at their will. SourceCred was a sophisticated piece of software. It did not simply record people’s actions but assigned a score based on how those actions were received — built on, liked, referred to etc. At times, the community made decisions to reward what doesn’t easily fit into our notions of work, including care work and community maintenance.
They also learned that membership boundaries are important. They lacked a sophisticated knowledge infrastructure (although there were attempts to organize their archives towards such a system), which meant that onboarding and passing on leadership roles were difficult. Newcomers would repeat lessons that had already been learned by others.
Our future DAO’s contribution system enables it to know who has contributed what. Importantly, by computing over contributions, it can help know the combined expressions of what those in the group deem important (intersubjective value). It does not just keep track of actions; it tracks when an action has been used, built on, or appreciated. In this Entity, people can come and go and (to a degree) offer what they think is useful. They can decide to weight actions related to some outcomes as more important than others. Without a legible contribution system, this could lead to extractive behavior — intentional or otherwise — or conflicting priorities. The contribution system means that efforts that produce useful outcomes are seen and rewarded.
There is a commons process occurring, whereby contributors are creating rules and ascertaining the quality of contributions according to those rules. Property may be created as a spillover — in the form of a token, access rights, etc. Property rights are agreements concerning the definition of an object, including its administrative description and enforcement. In this context, property rights do not originate from state enforcement but from the commons. These rights might be represented by tokens, and encompass access rights or serve as collateral, options, or resources for exchange with other entities.
Connections Between DAOs
There are other components to this DAO that are only just beginning to be developed. Importantly, it requires a means to connect to other DAOs. Last year, I commenced an ethnography of DAO DAO DAOs, which were attempting to create systems of mutual support across DAOs, initially through token swaps. That particular ethnography stalled because token swaps were not adequate on their own. At Edge Esmerlda recently, Jake Hartnell, a founder of DAO DAO, explained to me his recent work on a protocol for re-staking across DAOs, which is something our future DAO may instead utilize to strengthen ties with other like-minded DAOs and enable them to define what they may need to co-govern.
A parallel project called Coordi-nations/Network Nations out of BlockchainGov has been discussing this in relation to the concept of supersidiarity. Where subsidiarity entails a decentralization, it is done through a coordinating entity that devolves power to localized entities. Supersidiarity is where localized entities have the ability to coordinate and govern on points of common concern, but it is done as a bottom-up process (or middle-out).
The Entity’s Use of These Systems
Let’s now return to the future DAO and its associated Entity. It has developed an economy so that what is valued locally is rewarded accordingly, and that value is determined through the system itself, which is able to compute across all the contributions and recognize needs/gaps as they arise so that the group can steer towards these. The administrative burden is less than it sounds due to the automation capabilities of this knowledge and contribution system. The software component of the DAO administers tasks at the direction of the stewards (those who contribute and whose ongoing monitoring and expression of needs and actions are used to steer, in addition to the high-level deliberation that takes place around ceremonies/planning). There is no appointed leader; rather, leadership is something that anyone can step into as required because this automated support is in place. It is also possible for it to share knowledge with other DAOs and provide information to external institutions as required.
How Do We Know if This DAO is Effective?
In our current time, we measure DAOs by their TVL, values alignment within the community, resilience against governance attacks, and ability to withstand the weight of nation-state regulatory requirements. In this future time, we will focus on the extent to which collective knowledge is able to be activated as intended/expressed by those who govern over it. To know if this DAO is on course, we would ask the following questions:
- Does it encourage high-worth contributions?
- Do contributor rewards outweigh opportunistic speculation?
- What keeps people here (creativity, care, play, etc.), and is this also rewarded?
- Is routine and ritualistic work, such as infrastructure maintenance, sustainable and rewarded?
- Are there effective membership boundaries in place that minimize extractive behavior?
- Do the onboarding/offboarding processes, including through the use of LLMs, assist it in managing these membership boundaries?
- Can leadership be easily rotated?
- Does the DAO’s software bureaucrat minimize the administrative burden on the contributors?
- Can the DAO create networks with other DAOs over shared concerns?
By this point in the future, Entities like this can share knowledge. We now have a more precise picture of what that means:
- Increased capabilities resulting from local knowledge management and commons-style self-organizing.
- Coordination of knowledge and resources with other DAOs for mutual support over shared areas of concern.
The infrastructural components I have described could enable an organization’s knowledge to be shareable with other organizations, enabling new forms of institutions as they coordinate among themselves. That in turn may produce new practices, norms, creativity, and rituals that we cannot yet begin to imagine.
About me
I am a Professor at RMIT University and an Australian Research Council-funded Future Fellow working on the project “Cooperation Through Code”. My research is examining permissionless systems (including public blockchains) using ethnographic methods. I am also a Research Director in Metagov.
