By Darcy Allen
***
One sign that the DAO landscape is maturing is the continuing acceptance of delegation to subgroups to avoid unending DAO votes. My aim in this post is to explore why and how DAOs should delegate their powers to committees, teams, councils, pods, and so on.
My argument is that the reason why a DAO delegates decisions matters. Is it to access specialist knowledge? To make sensitive decisions more private? To reveal solutions to hard problems? These reasons matter because they shape the governance relationships between those subgroups and the DAO. Importantly, DAO tools must cater to experimentation and evolution in these subgroup relationships.
Why delegate?
The ideal of all token holder voting on every DAO decision is wearing off. Asking (and expecting) governance token holders to vote on every decision is rarely reasonable nor is it effective at achieving DAO aims.
As DAOs mature they have started to delegate decisions to smaller subgroups. Those subgroups are linked back or accountable to the overarching DAO, perhaps through recurring reviews, elections, and veto powers. They might have preferential access, or control over come capital in a multisig.
DAO subgroups move us towards polycentricity. That is, where there are many overlapping centres of decision making within an overarching set of rules.
DAO polycentricity emerges in many forms: teams, working groups, subDAOs, pods, optimistic pods. I use the word emerges intentionally: polycentricity is best understood as an emergent phenomena rather than a designed one.
Subgroups exist for different reasons. Some reasons are obvious: speeding up governance (proposal > snapshot > multig can be slow) and voter apathy.
But there are also more fundamental reasons for DAO subgroups. These reasons relate to revealing or concealing types of knowledge.
Some subgroups reveal specialist knowledge. For instance, where decisions are highly technical or involve expert knowledge, a smaller and tighter subgroup might be an effective governance structure.
Many web3 communities are reasonably sceptical of this expertise rationale for good reason (for centuries we’ve seen expertise deployed to expand arbitrary powers, but that’s for a separate article).
Other subgroups exist to conceal knowledge — that is, for privacy.
By default DAOs are open, and they should strive to be. But issues, such as strategic diversification of the treasury, might be best made in more closed groups to prevent risks. Frontrunning, for instance.
Again, there are complexities here around the input of the community, which further underscores the need for feedback and accountability in subgroup governance.
Finally subgroups can be a tool to reveal local knowledge through experimentation.
Subgroups can be an effective tool where decisions are made under uncertainty, but where local information might help to resolve that uncertainty.
Subgroups can be diverse and can overlap on similar issues (e.g. what grants to fund). They can be competitive. Such autonomous subgroups can make decisions that might not have been approved by the entire DAO, but this nonetheless reveals (good and bad) local information.
DAO delegation questions
There are many open questions around what DAO subgroups should look like.
Some decisions, of course, should remain with the DAO. But which ones? (this is for a separate post, but it should take into account the tradeoff between decision costs and agency costs).
Let’s assume some subgroup delegation. How many delegated groups should exist? What is the relationship between those delegated groups? If different subgroups compete on similar decisions, how is performance rewarded and resources distributed? What are the mechanisms of accountability to the DAO, and what is the timeframe of that accountability?
The answers to these questions ultimately hinge on the reasons for the delegation. Specialisation? Privacy? Experimentation?
Take the example of experimental competing grant programs to reveal local information about what the treasury should fund. It seems unreasonable here for a rapid veto power from token holders. This governance relationship, by contrast, might be to arrange through periodic elections of grant givers over longer time frames.
A specialist decision group, on the other hand, may have a different governance relationship, with less autonomy.
The transition: progressive polycentricity
How can DAOs move towards more polycentric decision making?
One strategy is to build-in delegated on-chain subgroups at launch.
I’m sceptical of this approach. Maybe DAOs will get this process right, learning from existing frameworks and models.
But likely this will lead to subgroups with unclear or overlapping responsibilities. DAO members might (rightly) grow distrust of the groups and their lack of transparency.
Importantly, early-stage DAOs simply don’t have the relevant knowledge about what those subgroups should or might look like.
Other DAOs will add in their groups later — what I call progressive polycentricity.
This is what we’re seeing now across many projects and proposals.
In the progressive polycentricity approach it is clearer what groups should exist, and even the (initial) people who might occupy each group.
Time will have revealed what decisions the community cares about controlling directly. It will have revealed some of the tradeoffs in governing those groups.
But this approach must also develop those groups in a (perhaps messy) public process.
Either way, there’s a risk that many DAOs create too many subgroups as stakeholders clammer for power. Many DAOs could easily move from all token holder voting to a plethora of working groups and committees that don’t bring the benefits of knowledge revelation or expertise, but rather opacity and bureaucracy.
Delegation tools
Delegating DAO decision making is experimental. It’s an experimental process.
This experimentation — discovering the number, relationship, purpose, structure of delegated groups — will lean heavily on DAO tooling.
There are several DAO tooling projects working on these problems, aiding in the frameworks and design of pods and working groups.
Further development of composable DAO tooling — those tools that facilitate new mechanisms of transparency and accountability of subgroups — is critical to establish more dynamic, polycentric and ultimately more effective DAO organisation.