Abstract
This report investigates the distribution of voting power within the Optimism Collective, aiming to provide an updated analysis of decentralization within the Collective. Building upon prior research, including the Foundation’s Mission Request on measuring power concentration, the study also defines “whales” (1) functionally through the seasons and the changes in their voting power (2). Through this lens, we aim to better understand whether Optimism’s governance truly reflects the values of openness and collective input it seeks to uphold.
It is part of a deep research document that offers an amplified breakdown of the data below.
Team: @delphine, @ElizaPancake, @Fehz, @Joxes , mrwildcat @Pumbi
Introduction
Have you ever wondered who holds power in the Collective? What does it mean to capture the protocol? Why does the space talk about whales? Driven by the need to better understand the inner workings of the Collective, this report explores its distribution of voting power. We intend to offer a more up-to-date perspective on an issue that remains a key indicator of a DAO’s decentralization.
According to the DAO Design Principles, “Power structures cannot be avoided. It is better to acknowledge that and publicly define them than to pretend they don’t exist. Power can be held accountable only when it is transparent”.
At SEEDGov, we are committed to offering the Collective, both new and experienced delegates, and anyone interested in Optimism governance, reports that help clarify how governance functions and identify critical areas where further decentralization and improvement may be beneficial. We hope this report contributes to fostering a healthy and constructive discussion in the forum to strengthen Optimism’s governance.
Theoretical Framework
Before conducting the analysis, let’s go back to basics and share some definitions that we will dive deep into later in the document:
Capture
Governance capture happens when a single actor or coordinated group accumulates enough voting power to unilaterally shape a protocol’s decisions, governance processes, and roadmap to serve their own interests.
Optimism’s total OP supply stands at 4.29 billion tokens. Of this, only about 116M OP—approximately 2.7 percent of the circulating supply (3)—has been delegated for on-chain governance. With such a modest delegation rate, one could imagine a relatively small portion of tokens being sufficient to influence governance decisions (2.bis), giving rise to various hypothetical capture scenarios:
Token House Capture
Actors wield direct control over the Token House—where most votes originate—and may exploit that position as follows:
- Single-Entity Dominance: A lone delegate holds voting power ≥ (total votes cast − their own vote), enabling unilateral ratification or rejection of any proposal.
- Multi-Entity Collusion: A coordinated subset of delegates whose combined power ≥ (total votes cast − their collective votes), effectively dictating outcomes without leaving formal traces of coordination.
- Abuse of Privileged Rights: Entities granted special on-chain permissions (e.g. by the Foundation or Security Council) deploy those rights in contravention of the Operating Manual or Law of Chains, thereby circumventing standard governance channels.
Citizen Capture
Within Optimism’s bicameral framework, the Citizens’ House holds veto authority over Token House decisions.
- Veto Capture: Concentration of veto-granting power in one or more delegates within the Citizens’ House, such that they can nullify any Token House approval irrespective of broader participation.
Other Capture typologies
Beyond chamber-specific vectors, additional mechanisms of influence include:
- Quorum Capture: Control ≥ 30 % of delegated OP (Q = 0.30·V), permitting a lone actor or coalition to unilaterally pass or block any vote.
- Approval Capture: Holding > 50 % of the votes cast, thus determining outcomes independently of turnout.
- Parametric Capture: Securing votes on foundational parameters (inflation schedules, delegation rules, quorum (4) percentages) to systematically recalibrate governance rules.
- Economic Capture: Pre-emptive acquisition of large OP holdings immediately before critical votes, skewing delegation incentives and centralizing power among well-resourced actors.
- Institutional Capture: Embedding allies within formal governance bodies (e.g. ACC, Security Council) so as to entrench influence via procedural or interpretive levers.
Though these scenarios remain hypothetical, it’s prudent to frame them—arming ourselves with all tools to steadily strengthen our collective. This report explores the potential Quorum Capture scenario.
Methodological Definitions
The analytical framework, set to promote clarity and theoretical fidelity, is outlined below.
-
The analytical scope and data extraction are explained here. All the code is open access in SEED Gov Dune Dashboard to see the underlying code.
-
Dataset: our dataset spans from Season 3, when on-chain voting on Agora first launched, through the current Season 7. The timeline of the seasons was extracted from Optimism’s official public governance calendar.
-
Delegate Stratification by Voting Power: To better illustrate voting power distribution, delegates are divided into six tiers. Considering the current Votable Supply, the segments were chosen to find a balance (percentage-wise) of VP across different Tiers. The ranges are fixed for all seasons. Adjustments based on increase of Votable Supply were not considered for this iteration.
- 0-1 OP
- 1–250 K OP
- 250 K–1 M OP
- 1 M–1.5 M OP
- 1.5 M–5 M OP
- ≥ 5 M OP
Note that this granular stratification may be adjusted in future reports.
- Seasonal Aggregation Protocol: Recognizing governance as an iterative process, each metric was characterized by:
- All on-chain votes recorded from Agora from Season 3 onward.
- Seasonal Average to enable comparative analysis of intra-seasonal
Deep Into the Ocean
Dimensioning a current Whale
In the marine world, a whale commands its environment through sheer presence. The concept of a whale in governance includes multiple concepts, such as the concentration of voting power, the potential for influencing outcomes, and the visibility of these actors within the ecosystem. Influence is not just about the quantity of tokens a delegate holds but what they can accomplish with that power.
This report adopts a functional and dynamic definition of whales. A whale is not merely a large holder, but an actor—or group of actors—capable of meaningfully shaping or unilaterally reaching quorum in decision-making processes. Influence is relational and context-bound, not absolute. In other words, a fixed threshold is not used to define “whales”; instead, the term denotes the dynamic setting. In practice, this group typically comprises those holding at least 1.5 M OP—most frequently the delegates exceeding 5 M OP. The current distribution of voting power will be examined to illustrate this.
The voting power distribution in number of delegates as of May 2025 is as follows:
In other words:
-
≥ 5 M OP: 6 addresses (0.0025 % of delegates) — including the Anti-Capture Commission (ACC) — each control more than 5 million OP.
-
1.5 M–5 M OP: 7 addresses (0.003 %) hold between 1.5 and 5 million OP.
-
1 M–1.5 M OP: 26 addresses (0.011 %) occupy the 1 to 1.5 million OP range.
-
250 K–1 M OP: 30 addresses (0.012 %) control between 0.25 and 1 million OP.
-
1–250 K OP: 45,972 addresses (18.911 %) control between 1 and 0.25 million OP.
-
0–1 OP: 197,050 addresses (81.060 %) each have at most 1 OP.
Only 69 addresses (0.029 %) hold at least 250 K OP tokens, and 39 addresses (0.02 %) hold at least 1M OP tokens. The ratio of delegates in the lowest tier to those in the highest tier exceeds 32,841:1, highlighting an extraordinarily skewed voting-power landscape. This minute subset of whale delegates, despite representing under 0.03 % of all participants—carries the potential to determine quorum formation and steer vote outcomes across the Optimism governance system.
The log-scale chart below allows us to visualize orders-of-magnitude disparities at a glance: just six addresses (including the ACC) hold over 5 M OP, 7 hold 1.5–5 M OP, 26 hold 1–1.5 M OP and 30 hold 250 K–1 M OP—while 242 022 delegates (≈ 99.97 %) control under 250 K OP. That 32 841 : 1 ratio between the smallest and largest tiers starkly illustrates how a tiny cadre of whales can dominate quorum formation and sway voting outcomes.
As of May 2025, the quorum was set at 30% of the votable supply (5)—around 35 million OP tokens. Achieving this required just 7 delegates from the top 100 to meet quorum independently, representing 0.0029% of the total 243k delegates (but of those just 46k have at least 1 OP of Voting Power). With the inclusion of the Anti-Capture Commission (ACC), this number drops to 5 addresses, further reducing the proportion to 0.0021%.
A better close up:
Note: Due to the extreme skew in the distribution of voting power among delegates, a logarithmic scale is used on the y-axis to improve readability.
Whales over time: Understanding the Quorum Tide and the Emergence of Whale Pods
Evolution of Circulating Supply and Quorum over Time
If whales in the Collective are defined not simply by how much OP token they were delegated, but by their capacity to influence outcomes, then the question becomes: influence in relation to what? In the Optimism governance framework, the answer is quorum.
To assess the pod of whales empirically it’s necessary to examine historical data: how many delegates—and how much voting power—have typically been required to meet quorum? Quorum itself is not fixed; it flows with the tides of each season. Starting in Season 2, the Collective established quorum as 30% of the votable supply. That figure has grown steadily with the ecosystem.
Across Seasons 3 to 7, we calculated the quorum for each opening vote of the season, extracted from Agora. The average quorum per season followed a clear upward trajectory:
The table below shows the First-Vote and Average Governance Metrics by Season:
Between Season 3 and Season 7, votable supply rose steadily from about 21.5 million OP at the start of Season 3 to roughly 115 million OP at the beginning of Season 7, and quorum thresholds increased in near-perfect lockstep, from approximately 6.46 million OP to 34.51 million OP. This tight, positive relationship is captured by a Pearson correlation of 0.9543, which means that over 90 percent of the variation in quorum levels can be explained simply by changes in votable supply, exactly as expected under a fixed 30% quorum rule.
Number of Whales that matched the quorum through the seasons
To understand how concentrated this power is—and how many actors are needed to reach the quorum threshold—we need to examine not only raw token distribution, but also patterns of participation and coordination among top delegates. That is where the concept of pods of whales becomes analytically useful.
Borrowing again from marine biology, where a pod refers to a coordinated grouping of whales swimming together, we define a pod of whales in governance as the smallest group of top voting-power holders whose combined votes are sufficient to meet quorum in a given voting cycle.
This increase reflects both higher votable supply and sustained participation, but it also raises the bar for what it means to have decisive influence. As quorum grows, the number of delegates required to meet it could, in principle, grow too—unless concentration increases proportionally.
What we observe, however, is the opposite: as quorum increases, the number of whales needed to meet it remains stable. Starting in Season 5, just five delegates, consistently, were enough to reach quorum when the Anti-Capture Commission was included. Even when excluding the ACC, quorum could generally be met by six to seven whales, all within the top 15.
This exposes a dual phenomenon: the collective’s design ensures a rising quorum in step with delegation, but actual voting power is increasingly centralized among a shrinking core of actors.
Below is an analysis of how the top ten delegates have alternated across seasons.
The data reveal that turnover is relatively low and is driven more by the entry of new participants than by reshuffling among incumbents. In other words, the most influential delegates tend to retain their positions over time.
Centralization over time: Echoes Through the Seasons
Season 3 – The Season of Distributed Coordination
In Season 3, governance was still finding its rhythm. The average quorum was under 11M OP, and no single actor held more than ~2.3M OP. It took at least four or five whales working in concert to reach quorum. The average number of delegates required was around 5.36 when including the top holder. Governance decisions depended on broad coordination among similarly sized actors—a fragile but pluralistic structure. In other words, at that stage the collective was still evolving: there weren’t delegates holding especially large amounts of voting power, and influence was more evenly spread. The more current distribution pattern—such as the 1 M OP delegations to top delegates introduced in Season 5—had not yet solidified.
Season 4 – Rising Quorum
Season 4 saw a sharp increase in votable supply and quorum. Despite this, power was not yet consolidated: it now took 7.67 top delegates on average to meet quorum. In some cycles, quorum required up to 12 distinct actors. The pod of whales had grown, but not yet tightened.
Season 5 – The Emergence of ACC
Season 5 marked a turning point. For the first time, a single actor— the ACC—could cover over 40% of quorum alone, holding upwards of 10M OP. When ACC voted, the total number of whales required to pass a proposal dropped to five. Without it, quorum still only required ~6.3 delegates. This season crystallized a shift: governance no longer required plural consensus—it could be steered by a dominant node and a handful of satellites.
Season 6 – Dependency Deepens
The same pattern held in Season 6. Despite a quorum nearing 30M OP, just five whales, when including the top actor, could meet the threshold. Excluding it, around seven whales sufficed. The system became structurally reliant on one actor’s presence*. The quorum kept growing, but its weight increasingly rested on a central whale’s back.
*Note: Please note that the ACC’s support for protocol upgrades acts as a helpful backstop—bringing votes closer to the required quorum and offering additional validation from experienced delegates, including a single actor with over 10 M OP. However, since these votes have always surpassed the quorum by a comfortable margin, this reliance is unlikely to be critical in practice.
Season 7 – Concentrated Stability
Season 7 extended this trend. Quorum exceeded 34M OP, but the number of delegates needed to reach it remained stable. With the ACC: five. Without it: ~6.4. The size of the pod stopped changing, even as the voyable supply around it swelled.
In earlier seasons 3 and 4, reaching quorum required coordination among multiple delegates with similar voting power—no single actor held decisive influence. Since Season 5, this changed: a single delegate gained enough voting power to cover 40–45% of the quorum, enabling decisions to hinge on a much smaller group.
Season to Season Summary
To examine the progression of delegate participation over time, we analyzed seasonal voting-power data to see how each tier contributed, treating each season as 100 % of the vote. The graphic below shows a marked decline in participation from the 1.5–5 M OP segment beginning in Season 5–6, while the over-5 M OP tier grew correspondingly.
This raises the next question: what if we examine the voting power average per proposal rather than each season’s relative share? As we can see below, the over 5 M OP tier captures a substantial increase in voting power participation, and the 1–1.5 M OP tier also gains notable influence, an effect obscured in the previous chart, where its percentage appeared steady.
Metrics of Concentration
Nakamoto (NI) (6)
Data were filtered by start date and assigned to Seasons 3 through 7 according to the first and last proposals in each period. For each season, the Nakamoto Index of the opening proposal and the average index across all proposals were computed. Percentage changes in seasonal averages were then calculated relative to the preceding season.
The table below shows, for each season, the Nakamoto coefficient of the first proposal voted in that season and the arithmetic mean of all coefficients within the season, with the relative change versus the preceding season.
To better illustrate this, the chart below plots seasons 3 through 7 along the x-axis and the Nakamoto Index on the y-axis. Each vertical bar—colored uniquely per season—indicates the opening NI at the first vote. A green line with circular markers traces the average NI across all on-chain votes for each season.
A close up:
-
Season 3: NI rose from 6.0 to 6.5 (+8.3%), suggesting a modest intensification of voting-power centralization in Optimism’s early governance stages.
-
Season 4: Both opening and average NI held at 7.0 (0.0%), reflecting stable delegate influence.
-
Season 5: Started at an opening NI of 4.0, then climbed to ≈5.67 (+41.8%)—the largest intra-season shift—likely driven by expanded participation and the launch of the Anti-Capture Commission.
-
Season 6: Peaked at an opening NI of 8.0 before settling at ≈7.46 (–6.7%), indicating an initial reconsolidation of power that partially retracted by season’s end.
-
Season 7: Grew from 6.0 to 6.6 (+10.0%), underlining a renewed push toward re-diversification of voting power.
Notes:
- Across Seasons 3–7, opening NI values spanned 4.0–8.0, while season averages fluctuated between ≈5.67 and ≈7.46.
- The Anti-Capture Commission’s introduction in Season 5 corresponds with the most pronounced intra-season shift, embedding a key meta-governance check into the protocol’s power dynamics.
Note: It’s worth mentioning that for this index we refer to the Top 100 delegates for illustrative purposes.
Gini (7)
The graphic below displays the Lorenz curve of the distribution of voting power among Optimism delegates.
See the graphics here: https://dune.com/seed_org/optimism-governance
A Gini index of 0.998 means that almost all the voting power is concentrated in the hands of a few. The Lorenz curve for this case would show that most voters have almost no power, while a small number control nearly everything. In simple words, this tells us the DAO is highly centralized in practice.
Disclaimer: However, we need to be aware that the Gini coefficient offers only a narrow perspective on the collective’s decentralization. In the DAO context, its implications are often overstated. Uneven token holdings do not inherently imply governance failure. As Vitalik Buterin said, “a high Gini coefficient could just as easily indicate a large passive user base, not necessarily a dangerous centralization of power” (Buterin, 2021). In other words, whales may hold significant voting power on paper, but actual influence depends on alignment, participation, and coordination. At most, you might find a few whales exerting influence over part of governance—often less than half—while the broader community continues to shape outcomes through mechanisms of deliberation, delegation, and social consensus.
Analysis of Whale Voting Participation
Quantitative Analysis
The following chart reveals the participation percentages for every vote across all seasons, stratified by delegates’ voting-power tiers.
The graphic shows that across Seasons 3–7, the 0–250 K OP tier contributes under 15 % of total voting power, despite comprising over 99 % of delegates.
Also, it shows that the three highest tiers (1–1.5 M, 1.5–5 M, and > 5 M OP) together account for more than 85 % of voting power, revealing a strong concentration among relatively few delegates.
Qualitative Analysis
While understanding how many whales are required to meet quorum offers insight into the system’s structural concentration, it tells us little about who those whales are and how often they vote. To fully understand the role whales play in the Collective, we must focus on their dynamic behavior—tracking patterns of participation across seasons and proposal types.
This section introduces a qualitative layer to the analysis: identifying when and why whales with over 5 million OP engage, and what types of proposals prompt them to surface.
A Typology of Governance Proposals
To analyze whale engagement with granularity, we developed an interpretive typology of on-chain governance proposals. Rather than relying solely on technical proposals, this typology frames the diversity of decision-making and groups proposals into eight categories:
- Protocol Upgrades
- Seasonal Intents
- Elections
- Budgets
- Mission Requests *
- Governance Body Amendments
- Special Proposals
- Ratifications
Note: It is important to note that Mission Requests are no longer operative during Season 7.
The detailed classification and voting examples are here.
Biggest Whale Preferences by Proposal Type
The classification above serves to visualize the types of proposals whales engage most. Among whales with >5M OP, Elections drew the highest total engagement, with 44 whale votes across 25 proposals—including high-profile decisions like Grants Council Reviewer Elections or Security Council selections.
Next came Upgrades (30 votes across 19 proposals), followed by Budgets, which, while fewer in number (8 proposals), showed the highest average whale turnout per vote (2.25 whales/proposal). This suggests that whales selectively emerge for critical operational decisions, particularly those involving financial stewardship. In other words, when leadership or council seats are at stake, whales pay close attention. Framing budget asks or charter changes in the context of “mandates” or “leadership vision” could boost engagement by tapping into that existing interest.
-
Elections account for 37.3% of all whale votes, indicating that major stakeholders focus proportionally on these decisions. This raises an open question: to what extent are elections within DAOs merely popularity contests?
-
Upgrades capture a solid 25.4% of whale votes, demonstrating strong whales alignment around protocol improvements.
-
Budget proposals, though only eight in total, account for 15.3% of whale votes and have the highest average whales per proposal (2.25), suggesting that high-VP holders prioritize funding allocations.
-
Mission Requests and Governance Body Amendments attract fewer whale votes (11.9% and 10.2%, respectively) and have lower average whales per proposal.
Alignments
To assess overall alignment among the Top 10 delegates in each season, we calculated a similarity index (8) defined as the fraction of votes on which two delegates agreed. In other words, the similarity index is a quantitative measure of how often two delegates vote the same way.
The charts below illustrate voting correlations by season. A value of 0—shown in light pink—represents minimal agreement (divergent votes), while a value of 1—shown in dark burgundy—represents complete agreement (identical votes). The predominance of dark burgundy across seasons indicates a strong coordination and alignment among top delegates.
Season 5 exhibits the greatest diversity in voting patterns, which corresponds to a period marked by multiple elections, budget decisions, and the formation of new councils with many nominations. This variation suggests that in elections, delegates display more diverse voting behavior—each voter weighs preferences for members, budgets, and charters uniquely—whereas for protocol upgrades and ratifications, votes are far more uniform. In other words, the community rallies behind advancing OP development in a largely unidirectional, consensus-driven manner.
When Do the Whales Surface (Vote) ?
We tracked the Top 15 delegates vote timing each season using a 0–1 scale (0 = start of voting window, 1 = end). We employed the vote_progress_normalized metric (0 = opening of the voting window; 1 = closure) to quantify the temporal distribution of ballots casted. The resulting seasonal is shown in parentheses:
- Season 3 (avg 0.50): Votes landed right around the midpoint.
- Season 4 (avg 0.73): Voting skewed heavily toward the deadline.
- Season 5 (avg 0.74): Late‐stage voting persisted, as whales waited on others.
- Season 6 (avg 0.66): A modest shift back toward earlier voting, but still post‐midpoint.
- Season 7 (avg 0.63): Continued moderation, with votes cast earlier than in Seasons 4–5, yet still late.
The chart below illustrates the temporal distribution of voting activity by the Top 15 delegates across the successive seasons.
Below is an overview of the overall voting pattern of the Top 15 delegates.
As shown in the graphics, whales often vote late. This aligns with claims that some top delegates intentionally hold back to avoid prematurely influencing outcomes, granting space for smaller delegates to weigh in first. In later seasons, there is a moderate rebalancing, though whales still tend to vote after the midpoint.
Conclusion
The Five-to-Seven-Stable Whale Pod
- Despite quorum quintupled between Seasons 3 and 7, it is still met by a small group of 5–7 key “whales,” which can be interpreted as growing centralization of voting power.
A brief overview: between Seasons 3 and 7, the number of delegates required to reach quorum flattened even as quorum itself quintupled (6,46 M → 34,51 M OP = 5,34×). In Season 3, five to six evenly balanced delegates sufficed; by Season 4, fragmentation pushed that up to seven. From Season 5 onward, however, a small “pod of whales” led by the Anti-Capture Commission restored stability—quorum passes with just five when the ACC participates, and still needs only around six to seven top holders without it. Statistical measures underscore this centralization: the Nakamoto Index fell from 6.0 to 4.0 in Season 5 (recovering only to a 6.6 average by Season 7), and a Gini coefficient of 0.998 reveals extreme voting-power asymmetry.
While high concentration doesn’t inherently imply malicious capture, it means that meaningful influence now resides with a tight cluster. At the same time, fragmentation creates friction: newcomers and smaller actors face entrenched hubs of power that shape outcomes long before a broad swath of delegates can weigh in. Without structural adjustments, this dynamic risks discouraging broader, sustained participation.
Some reflections on the relationship between circulating supply, votable supply, and quorum
- As of 2025, only about 2,7% of the circulating supply is actually votable—a surprisingly low proportion. Despite steady increases in both votable supply and the quorum required to pass a vote, this shift has not been accompanied by a growth in the number of whales.
Over Seasons 3–7, the Optimism Collective’s votable supply has risen steadily, signaling deeper institutional maturity and broader community engagement. From roughly 22M OP in Season 3 to about 115M OP by Season 7, supply more than doubled in Season 4 (+107 %), then grew by +22 % in Season 5, +17 % in Season 6, and +9 % so far in Season 7. As seen earlier, key governance milestones helped drive these gains: Season 4’s Intents and Missions frameworks empowered community-defined priorities, catalyzing unprecedented growth. Season 5 added core bodies—the Security Council, Anti-Capture Commission, Developer Advisory Board, and Code of Conduct Council—and launched (in Season 6) a Chain Delegation Program to onboard new OP chains.
These structural bodies boosted both visibility and participation, but still the pace of growth is slowing. The increase in votable supply may reflect macro supply trends rather than a fresh influx of participants. Targeted incentives—such as the Chain Delegation Program—are crucial to sustaining ongoing growth and preserving organic momentum as the system matures.
It’s important to note that, although growth rates have moderated, the Collective has not stagnated. Continued, if decelerating, increases signal a governance system entering a more mature and stable phase. One possible interpretation is that the next frontier—the collective’s next challenge—is qualitative: incorporating delegates through reputation, trust, aligned incentives, and legitimacy.
Are We Moving in the Right Direction?
- The data invites us to consider how to keep decentralization vibrant. Since quorum can be achieved by just five or six large delegates, it may be worthwhile to explore whether our delegation incentives and governance design could do more to foster broader, consistent participation.
Addressing power concentration does not mean reducing whale influence; it invites us to reimagine governance by rewarding active engagement rather than token volume. Continuing along the path of decentralization entails crafting structures that shine a light on consistently involved delegates—whether they command millions of OP or just a few thousand—and that shift incentives from passive holding toward sustained participation. It implies building in regular “delegation health checks” so token holders routinely realign their votes with evolving community priorities, and designing a quorum rule that responds to actual engagement—turnout, proposal diversity, cross-chain representation—rather than a fixed percentage of tokens. In this view, decentralization becomes not a static split of OP balances but a living reflection of who shows up, how often, and with what impact.
Appendix
Data Sources
- SEED Gov Dune: https://dune.com/seed_org/optimism-governance
Sources
- Vitalik Buterin, The Gini Coefficient is Not Enough, July 29, 2021, https://vitalik.eth.limo/general/2021/07/29/gini.html
Definitions
For definitions, please refer to the appendix here.