Comparing the proposed values with present quorum values is unfair as current values have proven to be impractical. The proposal suggests reverting back to a historical quorum that was practical for the DAO – given the total delegated voting power. As @possumlabs mentions, “it’s important to acknowledge that the suggested changes are not a quorum reduction by definition. The suggested minima of 100m & 150m for non-/constitutional are understood as fail-save values.”
The data supports this as you can see from historical quorum averages. If delegated voting power goes up, then so will quorum, and it can in fact far-exceed the 4.5% threshold that is used today. So in essence, we are resetting the system with a more reasonable configuration.
Adaptive quorum is something we explored. In particular, we looked at an adaptive quorum biasing-inspired model as well as a custom model where the quorum oscillates between a set range. You can find dashboards built by Claude on the models here and here. The main issue with these models is that they change the voting strategies for participants compared to the existing system – which warrants further analysis and exploration. We may release a research doc at some point on this work.
In summary, for the time being, we did not find enough upside to implementing such models to justify their added complexity and untested edge cases. In any case, a DVP-based quorum model provides strong flexibility for future experimentation. For example, it would be possible to add an adaptive quorum or other systems on top as ‘future extensions’.
Any incentive program tied to delegating tokens should only be rewarded if the voting power is actively used. If that is the case, then we should see active voting power remain high, above quorum. So even if quorum does go up because more tokens are delegated, as long as those new tokens are actively used, we will continue to meet quorum.
In regards to reducing security, this is not strictly true. For example, let’s consider the scenario where an adversary attempts a hostile take over by acquiring enough ARB to overwhelm the DAO and pass a malicious proposal.
We make the following assumptions:
- Honest voters: ~223M voting power, assuming 100% of the current delegation is honest,
- Total delegated voting power: ~318M
- Apathetic votes: 95M ARB (30% of DVP)
- Quorum (current system): ~209M Quorum (4.5% of voteable tokens)
- DVP-Quorum: 50% of delegated voting power
In the current system, an adversary will need to acquire just more than ~223M voting power to pass any proposal in the DAO. This is because the adversary only needs to overcome the set of active (and honest) voters. The current quorum of ~209M has little impact on the adversary.
In DVP-Quorum, if the adversary attempts the same strategy and acquires just more than ~223M voting power, then they will also increase the total delegated voting power to ~547M (324M+223M). As quorum is computed at 50% of the total delegated voting power, then quorum will be ~273.5M. Quorum becomes the new line of defense against the adversary and will halt the attack unless they continue to acquire more ARB.
Put another way, the financial cost to perform a hostile takeover can be greater in DVP-Quorum than the current system. This is why it is not strictly fair to say the new model harms security when in fact, under certain circumstances, it increases the cost of attack.
Finally, in relation to the redelegation campaign, this is a separate endeavor that is not mutually exclusive from changing how quorum is computed. There is no reason why both approaches cannot be pursued in parallel.
Thank you for looking into this, @curia. The corrected historical average values are even closer to our recommended baselines.
