[Discussion] [OBRA] Split Delegation - Gnosis Guild

[Safe OBRA Proposal] [Gnosis Guild] Split Delegation

Aligned strategy:

Strategy 5: Increase governance participation

Funding request:

60.000 USDC

Relation to budget:



  1. Impact on governance participation (direct attributable increase in participation through tool):

  2. amount of tokens used for voting

  3. amount of individual voters

  4. distribution of voting power

Initiative description:

Currently, there are several large SAFE token holders who are likely unable or unwilling to participate in Safe governance, since doing so would mean delegating their substantial vote weight to a single entity and trusting it to act in the entity’s and Safe’s best interest. This is due to a fundamental limitation in the current delegation strategy available in Snapshot and used by SafeDAO: accounts are limited to delegating their full balance in a given context to only one account.

Gnosis Guild is currently building an updated delegate registry and Snapshot strategy, implementing three key features:

  1. Split Delegations: Accounts can delegate their vote weight to any number of other accounts, along with defining the ratio of their vote weight that each of their delegated accounts should receive. For example, Alice could delegate 80% of their vote weight to Bob and 20% of their vote weight to Charlie.
  2. Transitive Delegation: Accounts can delegate not only their own vote weight, but also any vote weight delegated to them. If Alice delegates to Bob, and Bob delegates to Charlie, Charlie’s total vote weight is the sum of the vote weights of Alice, Bob, and Charlie.
  3. Delegation Term Limits: Accounts and Snapshot spaces can optionally set limits to length of time for which any given delegation is valid. Spaces can also choose which, if any, term limit to respect when calculating vote weights.

This initiative supports the development of the registry, Snapshot strategy, Snapshot delegation updates, and backend tooling. It also includes the implementation, maintenance, and support of the updated delegation framework for the SafeDAO.

The outcome of this work should enable much more fine-grained delegation options for members of the SafeDAO, ultimately leading to improved vote-weight utilization (more SAFE tokens, from a greater diversity of holders, being represented in each proposal).

Current Status:

The second version of the delegate registry contract has been successfully implemented and audited. The backend services for calculating vote weights have also been implemented, along with the first version of the corresponding Snapshot strategy. However, a significant amount of testing is still required for both the vote weight service and the Snapshot strategy, as well as SafeDAO’s specific proposed setup. We anticipate additional dev work to address any issue that may arise during testing and to make any necessary modifications for Safe’s specific setup.


This initiative involved updates to SafeDAO’s Snapshot strategy, a critical governance component. As such, it should be approached with caution and be well tested before being enabled on SafeDAO’s Snapshot space.

Timeline and milestones:

Week Focus Outcomes USDC $SAFE
1-2 Implementation Complete implementation of the Snapshot strategy, updated Snapshot delegation dashboard, subgraphs, and backend services. 22.500 N/A
3-4 Testing Implement a test Snapshot space replicating SafeDAO’s setup with the new delegation strategy, and handle any issues identified while testing. 22.500 N/A
4-8 Support Resources available to provide support for issues that arise post-rollout. 15.000 N/A

Initiative Lead

Gnosis Guild


Gnosis Guild would assign a full pod (one designer/PM and two engineers) to this initiative for four weeks, and on-demand resources to respond to any issues that arise in the four weeks after the DAO starts using the new strategy.

Additional support/resources:



Nice proposal for the community.

Would this plan work for increasing the number of new contributors?

Or increase the participation rate of current token holders?

1 Like

The intent here is to increase vote weight utilization across the full spectrum of users. However, split delegation is probably most impactful on larger holders that are unwilling to delegate their entire vote weight to one individual.


In favor of this initiative.

The need for better delegation in SafeDAO and other DAOs is clear, and the Gnosis Guild team has proven capable of shipping solutions at the intersection of Safe and Snapshot.


In strong support of this proposal put forward by the Gnosis Guild.
The proposal addresses a crucial limitation in our current governance model. By enabling SAFE token holders to distribute their vote weight among multiple delegates, we are opening the door to more nuanced and representative decision-making. Such a system will surely give way to more diverse contributions at SafeDAO and eventually, a more robust delegate framework.

Upon implementation and after further testing and usage in SafeDAO, I’d love to see this adopted across other DAOs. I wonder, is this delegation strategy easily replicated in other Snapshot spaces?


Yes, this is built in a generalized way that will make it usable by other orgs leveraging Snapshot and Safe.


First of all, great initiative as partial delegation has been discussed multiple times in the past to unlock and diversify voting power. The governance framework had already referenced that such a possibility might be implemented in the future

I have two questions:

  • Can delegator Alice only choose ratios, e.g. 20%, or also absolute numbers, e.g. 25.000 Safe token to be delegated to Bob.

  • Can each delegate again partially delegate?

1 Like

A follow-up question: Does this implementation still allow the vote of token holders to take precedence over the vote of their delegates? (see section in the governance framework below)

How does this look like with transitive delegation (e.g. Alice has 10.000 Safe token voting power and delegates to Bob, Bob delegates to Charlie. Charlie votes for outcome A, Bob votes vor B, Alice votes for A. Which outcome got the voting power?)

1 Like

Delegates can only delegate ratios of their total vote weight, not absolute values. This is important since one’s balances might change over time in a way that could conflict with absolute delegations.

Yes, when one delegates (split or no), one is delegating ALL of one’s vote weight (one’s own and any delegated to one).

1 Like

The current implementation is similar to comp/OZ Governor delegation in this regard; if one has delegated their vote, they are unable to cast a vote themselves.

This was a design choice informed by implementation complexity. I would personally prefer it if we could maintain the current state where one has the option to override their delegate’s vote, if this is a requirement we may need to adjust the proposal to account for additional implementation time.

1 Like

Could you provide an estimate of how the implementation of a mechanism that allows delegators to cast votes themselves and superseding the votes of their delegates, would impact costs and the timeline?

Is there anything else that could be different from the current delegation via Snapshot, e.g. limitations to undelegation/redelegation?


No other differences that I’m aware of.

Will get back to you with an estimate on the additional time to implement changes.


Thanks @auryn for putting out this great initiative!

I’m excited about the shift towards partial delegation in SafeDAO’s governance. This significant enhancement is key to diversifying voting power and developing a more robust delegate framework.

If we do actually adopt this, I’m particularly interested in its wider implications, especially for governance analytics platforms like the one I’ve developed for the Safe community: SafeDAO Governance Dashboard

I’d like to bring up a relevant example from Optimism Collective, where recent upgrades to their Governor contract highlighted key challenges:

  1. Integration Challenges: Post-upgrade, various analytics platforms experienced functional issues, indicating the complexities of integrating new delegation features across different systems.

  2. Data Accuracy: Ensuring accurate governance data representation becomes more challenging with dynamic delegations.

  3. Smooth Transition: Optimism’s situation underscores the need for best practices in community-wide adaptation, especially regarding developer support and advance notice for significant upgrades.

These points from OP folks are highly relevant as we discuss similar upgrades in SafeDAO. What lessons can we learn, and how can we apply them to ensure a smoother transition for our own governance system? Here’s the thread on the OP forum for further context.

To sum up, these are some question I have:

  • How will new delegation features impact platforms built on top of our current delegation system?
  • What will be the definitive source for voting power data, and how can we ensure its accuracy with dynamic delegations?
  • What are the best practices for a community-wide adaptation to these changes?

Keen to hear your thoughts on these aspects. :smile:



I’m excited to see this proposal for partial delegation as it builds a specific and tangible tool that improves voting participation of SafeDAO and makes improvements to the distribution of voting power.


It’d be great if possible for Gnosis Guild to work with the Curia team (A Safe grants project) to track the metrics outlined in this proposal in their dashboard: safedao.curiahub.xyz

Owner override

The option to override a delegate’s vote is important long-term. Instead of increasing the requirements of the current proposal this could potentially by added to a future sprint/season.


The 60,000 USDC funding request is well outlined in the “Timeline and milestones” section across each week and area of focus.

  • Is this full-time work (40 hours/week) for the assigned pod (1 designer/PM and 2 engineers)?
  • How does the funding compare to the market rate for this work to be done if hired by another company?
    • For example, if this is full-time work for the “Implementation” phase that’d be $93.75 per hour per person = $22,500 / (40 hrs per person x 2 weeks x 3 people)
    • Broader question for OBRA is the goal to fund projects at the market rate? If so, one option could be to combine SAFE tokens as part of the funding in the future.
1 Like

I imagine there could be some ripple effects. Do you have some specific examples apps / users that should be taken into consideration before changes are implemented?

We’ve built an additional service to calculate vote weights. Anyone can run an instance of the service to validate the results, however someone needs to run the instance of it which will be queried by the Snapshot space. This could be Gnosis Guild, the SEF, or some other service provider. I don’t have a strong preference there.


Looks like that dashboard already serves up some relevant data for this proposal! :tada:

Yes, this would be the full-time focus of the pod for that period.

1 Like

Beside Curia, there’s also Dune Dashboard, and governance aggregators such as Boardroom that would need to adapt to the changes