The United States operates under a system of representative democracy. Constituents vote for Congresspeople, who in turn will vote on issues of governance. But when a citizen votes for a representative, are they voting for that rep because the rep best aligns with their individual views, or because they most align with that rep’s views?

It’s a subtle difference, but one that I think is important. There’s no real way to answer it within the constraints of our current system. However, I think it’s one area DAOs could be effective in exploring.

Enter: dynamic delegation

A model of dynamic delegation would operate somewhere between a direct and representative democracy.

Model #1

Individuals within the DAO would be divided into constituencies. Each constituency votes on a representative. Here’s how voting structures would work:

1. For each governance proposal, everyone has a vote they can cast
2. If a certain percentage of a constituency casts a vote, then the constituency’s vote gets cast in favor of whichever way those people voted
3. If the percentage of constituents that vote falls under that threshold, the delegate’s vote represents the entire constituency
4. Each constituency gets one vote in the larger governing forum

It might be useful to illustrate this with an example. Say each constituency has 20 people, and one of those twenty is the constituency’s representative. When a proposal comes up for a vote, everyone in the constituency is able to cast a vote. Yet just because someone can vote doesn’t mean they do. That’s where dynamic delegation comes in.

Say 13 people in the constituency vote (9 vote “yes” for the proposal and 4 vote “no” for the proposal). If the threshold for point #2 (above) was that at least 51% of the constituents voted, then the individual votes would be counted. In this case, because the majority of those who voted chose to vote for the proposal, the constituency’s vote would be cast for the proposal.

Now consider a scenario where only 9 people voted (4 “yes” and 5 “no”). If the threshold was still 51%, then the constituency’s vote would be delegated to their representative. The representative chooses how the constituency votes — even if that representative chooses to vote “yes” for the proposal.

Model #2

Another way to structure a model of dynamic delegation would be to always count votes cast in the constituency, and allow the delegate to vote for those in the constituency who chose to abstain from voting.

Say constituencies are composed of 100 people. Each person in the constituency gets a vote that can count toward the DAO’s final tally. If 30 people vote, each of their votes counts. The constituency’s delegate then votes for 70 people who didn’t vote. That way, if someone cares about a proposal enough to cast a vote, their vote counts. If they’re indifferent (i.e., choose not to cast a vote), they defer to their representative.

The idea in both models is that people will vote on issues they deem important. For those more important issues, democracy is more direct — the way the constituency votes is determined primarily by the constituents. On less important issues, where constituents are relatively indifferent, the delegate has more sway over the vote.

Core to both models

These are just baseline models. There are many components worth testing, including:

• How to determine a constituency (my hypothesis is that all constituencies should at least be of equal size). Should constituencies have aligned views? Different views? I imagine a good starting point would be to randomize the assignment of individuals.
• Should all constituencies be the same size? Allowing people to choose a constituency based on the delegate might be the best answer for how to determine a constituency. In model #2, it probably doesn’t matter given that each person has a vote. In model #1, if constituencies were different sizes, votes per constituency could be weighted by size of constituency.
• Frequency of delegate election. Should it be every quarter? Every year? Only if a delegate appears to consistently vote in their constituency’s minority?
• The threshold for delegation to the representative. 51% was illustrative in the above example, but it’s unclear if that’s actually the ideal threshold

DAOs should be laboratories for governance

The way I view it, DAOs as reformed or refurbished companies isn’t really that exciting. We’ve had hundreds of years to test company structures. In many cases, there’s some value in having a hierarchical structure with levels of decision-makers. And if DAOs end up doing that, too, well then it’s not really that revolutionary. At least in my view.

DAOs are incredible tools for experimentation. That they buck centralization inherently promotes more democratic systems of governance. Governance is one area where it’s difficult to test or change existing structures and the fact that DAOs are new provides opportunity for such experimentation.