In Season 7, we continued our experimental approach to Citizenship design by inviting a diverse group of Guest Voters to participate in governance alongside Citizens. This experiment built upon our previous learnings from Rounds 5 and 6, with a focus on understanding how different stakeholder groups approach resource allocation decisions.
Experiment Design
The Season 7 Guest Voter Experiment invited participants from across the Superchain ecosystem to join Citizens in selecting evaluation algorithms for Retro Funding. Eligibility was based on contribution data from OP Atlas, ranging from onchain activity to community participation.
To gather insights, we collected data through a survey designed to measure participants’ attitudes, incentives, preferences, and decision-making styles. This polling data became a valuable source for testing hypotheses tied to governance design—such as how funding preferences vary across roles, what characteristics predict certain governance behaviors and to what degree values are factored into strategic opinions.
Key Research Questions
The experiment sought to answer several important questions:
- Do participants naturally group into distinct stakeholder clusters with different perspectives and priorities?
- Do participants’ characteristics drive their strategic opinions and budgeting decisions?
- How do different stakeholder groups approach resource allocation?
- What factors predict engagement and participation in governance?
What We Found
Stakeholder Clustering
Using unsupervised machine learning on participants’ survey responses, we identified two distinct clusters rather than the four we had hypothesized:
- Cluster A (114 participants): Tended to align more closely with the Collective’s intent and set more conservative budgets
- Cluster B (38 participants): Showed stronger preference for supporting public goods over growing the Superchain, set significantly higher budgets (especially for out-of-scope categories), and expressed interest in spending more time on governance (11 hours/week on average vs. 6 hours for Cluster A)
Several factors predicted which cluster someone belonged to, with the most significant being:
- OP Stack Score (worse scores increased probability of being in Cluster B)
- Amount received in Retro Funding 6 (higher amounts correlated to being in Cluster B)
- GitHub linking status (participants with a GitHub account were more likely to be in Cluster A)
Interestingly, the strongest predictor of someone’s allocation preferences was their agreement with the statement that “public goods should be prioritized over growing the Superchain.” This suggests that fundamental values may be more important than stakeholder category in determining governance behavior.
Strategic Opinions
Contrary to our expectations, participants’ characteristics mostly did not explain their opinions on strategy and values. Most stakeholders agreed that driving measurable growth on the Superchain was the most important goal, but disagreed on how to achieve this or how other values should factor in.
A surprising finding was that even individuals working at chains and protocols seemed to express personal values on the survey rather than purely strategic organizational interests. This suggests that to capture genuine organizational perspectives, organizations themselves need direct representation.
Budgeting Decisions
When asked to allocate budgets for various missions and operational categories, several patterns emerged:
- Participants tended to agree with already-approved budgets for existing missions and operations
- There was evidence that participants increased budgets for categories that would directly benefit them
- Agreement with prioritizing public goods over Superchain growth was a statistically significant predictor of higher budgets for categories outside of Season 7 scope (p<0.05)
- Foundation and Labs employees showed more critical assessment of specific budget allocations compared to non-employees, particularly for governance services, educational content, and grants, suggesting that more context results in stronger opinions. This is something we also learned in Retro Funding 5.
Participation and Voting Patterns
As expected, turnout was lower from builders and higher from governance enthusiasts and community contributors. This participation bias is an important consideration for governance design, as it can lead to overrepresentation of certain perspectives.
While there were patterns in turnout among groups, participant characteristics did not predict which algorithm they selected. This suggests that the algorithms didn’t favour a particular group or align with a particular worldview, or perhaps the differences between the algorithms weren’t meaningful or easy enough to understand. The latter aligned with feedback gathered in interviews with voters.
Due to low turnout from builders, there were only three projects that had multiple voters from the same organization. Of those, two had all of their contributors vote the same way across both missions. This suggests that the teams coordinated to maximise their voting power.
Recommendations
Based on these findings, we’ve identified several recommendations for future governance design:
- Lower participation barriers: Gather feedback and improve the voter experience to decrease required effort and encourage representation of broader interests, especially from the groups with lower turnout in Season 7.
- Enable organizational representation: Create mechanisms for organizations to vote directly rather than inviting multiple team members from organizations.
- Balance stakeholder representation: Actively counteract participation bias by balancing the voting power of different stakeholder groups.
- Use expertise signals: The OP Stack score could serve as a useful signal for identifying voters who tend to approach problems in a strategic rather than values-driven way.
- Empower the Budget Board: Since voters rarely criticize proposed budgets, move budget proposals to optimistic approval, allowing the Budget Board to leverage their expertise while still remaining accountable to stakeholders.
- Prevent mission scope creep: Develop guardrails to prevent approval of Missions that don’t directly impact the Collective Intent.