# From Apathy to Accountability: The Evolution of Delegates in Compound Finance

**Published by:** [blockful blog](https://paragraph.com/@blockful/)
**Published on:** 2026-07-20
**URL:** https://paragraph.com/@blockful/evolution-delegate-landscape

## Content

1. How It Was Before: The Original Delegation Model Compound's governance has been live since 2020. The system requires a minimum quorum of 400.000 COMP (4% of the total supply) for a proposal to pass, and at least 25.000 delegated COMP to create one. During the early years, delegation was entirely organic: holders delegated their votes to addresses they trusted, like researchers, developers, VC funds, with no financial incentive or formal accountability mechanism. The chronic problem: falling participation and apathy As the protocol grew and the volume of proposals increased, the number of voters declined consistently. In 2021, the average was 28 voters per proposal. In 2022, there was an atypical spike to 129 average voters, explained by a cluster of high-impact events that mobilized the community: the launch of Compound III (Proposal #116, August 2022) generated the highest participation ever recorded in the protocol (1.2M COMP voted); an oracle upgrade bug (Proposal #117) froze the cETH market for 7 days, requiring an emergency governance response; and the risk management proposals around the Ethereum Merge (September 2022) and the deprecation of the FEI market (TribeDAO collapse) brought in participants who would normally have stayed on the sidelines. Between 2023 and 2024, unique voter counts fell sharply, with averages dropping to 34 and then 25 voters/proposal - 2024 recorded the historical minimum with a median of just 22 voters. With the Delegate Race, 2025 saw a significant recovery to an average of 60 voters. In 2026, partial data from January to June points to stabilization around 28 voters, similar to 2021 levels, but with a fundamental difference: quorum is now guaranteed by design, not by chance. The decline from 2022 to 2024 shows that without an incentive to vote, and with the growing technical specialization of proposals (market parameters, integrations), participation concentrated into fewer addresses. Proposal volume exploded while participation fell One of the paradoxes of Compound's governance is that proposal volume grew enormously while per-proposal participation shrank. The protocol started with 36 proposals in 2021 and peaked at 175 in 2024, nearly 5x more in three years. This growth directly reflects the expansion of Compound III across multiple chains and markets: each new pool, parameter adjustment, or integration generated its own on-chain proposal. In 2025, volume fell to 135, and the first six months of 2026 already total 67, an annualized pace close to 2025's. Execution rates remained high throughout: in 2021, 78% of proposals were executed; in 2026, that figure rises to 94%, reflecting the quorum reliability provided by the Delegate Race - we will talk about this later. For delegates, the volume growth had a direct cost: in 2024, keeping up with and voting on roughly 15 proposals per month was the minimum required to meet participation thresholds. The catalyst: the July 2024 governance attack The turning point in Compound's governance was the attack orchestrated by the Goldenboys group, led by Humpy. In July 2024, after two failed attempts (Proposals 247 and 279), Proposal 289 passed, allocating 499.000 COMP to the goldCOMP vault, equivalent to approximately $25 million at the time. Humpy had accumulated over 81% of the required quorum (228,000+ COMP) across five wallets, exploiting exactly the kind of apathy that allowed quorum to be reached with very few votes. The episode exposed how a small number of active voters lowers the cost of a governance attack. 2. The Franchiser Contract: Origin and Adoption For the DAO to delegate COMP from its own treasury to external participants without giving up custody of the tokens, a mechanism was needed that didn't exist in the original system. A contract capable of transferring voting power without transferring the tokens themselves, and one that could revoke that delegation at any time if a delegate failed to meet their obligations. That mechanism is the Franchiser contract - and it was not created by Compound. The story begins two years earlier, in June 2022, when Noah Zinsmeister (Uniswap Labs) created the repository and developed the contract. Trail of Bits completed the security audit in August of that year. In September 2022, the Uniswap Foundation was the first to adopt it in production: the Uniswap Foundation creation proposal was executed on 09/08/2022 and included exactly 2.5 million UNI delegated via Franchiser, revocable by the DAO at any time. Compound only entered the picture in August 2024, when it approved Delegate Race Cycle 1 (Proposal 315) and allocated 300.000 COMP via Franchiser. In November 2025, Cycle 2 (Proposal 504) added another 310.000 COMP. The Franchiser enables three things the previous system did not: Delegate voting power without transferring tokens - COMP remains in the DAO's Comptroller and Timelock Revoke the delegation at any time via an on-chain proposal Create auditable sub-delegations with traceable addresses on the blockchain Compound therefore adopted infrastructure already tested by Uniswap for two years, reducing audit and implementation risk. 3. The Turning Point: What the Delegate Race Is and How It Works The problem that motivated the program The Compound Governance Working Group (CGWG) was established in mid-2024 discussions began in May, before the July attack. Proposed by StableLab, Arana Digital, and PGov in response to challenges already visible in governance, the group was created with a specific mission: to design a program that solved three problems simultaneously. First, the DAO needed enough active delegates to reliably guarantee quorum, without depending on large holders who could simply stop voting. Second, some accountability mechanism was needed, a way to require participation and revoke voting power from those who didn't meet their commitments. Third, all of this had to be done without permanently and irrevocably transferring COMP from the treasury. The Franchiser solved the third point. The other two were addressed by the design of the Delegate Race. What the Delegate Race is The Delegate Race is a competitive selection process run by the CGWG in which the DAO formally delegates COMP from its own treasury to organizations and individuals that demonstrate active governance participation. The "race" analogy is literal: candidates register publicly, present their history and commitments, and are evaluated against objective participation criteria - number of votes cast, forum contributions, proposals authored. What differentiates the Delegate Race from a simple token distribution is the revocation mechanism: all COMP delegated via Franchiser can be reclaimed by the DAO at any time through an on-chain proposal. Cycle 1 (Proposal 315, August 2024): establishing the base With 18 candidates enrolled, the CGWG evaluated participation history, quality of contributions, and technical capacity. Eight delegates were selected through the scoring process, plus 1 (Alpha Growth) ratified separately by a prior proposal, totaling 9 Franchiser contracts and 300.000 COMP delegated from the treasury, voting power sufficient to guarantee quorum even without the historical whales. The minimum requirement for Cycle 1 was simple: participate in votes. There was no voting power cap per entity and no financial compensation — the program focused on establishing a reliable participation base before adding complexity. The effects were immediate. 100% of proposals reached quorum from September 2024 onwards, and average participation jumped from 508.000 to 713.000 COMP comparing the first and second halves of 2024 - the largest semester-over-semester increase in Compound's governance history. In February 2025, the CGWG attempted to go further with the Delegate Compensation Program Pilot (Proposal 409), a 6-month program to compensate active delegates. The proposal was defeated. Humpy, who still concentrated enough voting power to block any initiative single-handedly, voted against it. All named delegates supported or abstained; the votes against came from the same addresses flagged during the July 2024 governance attack. The episode showed that Cycle 1 had solved quorum, but had not sufficiently diluted the concentration of power in one actor's hands. Cycle 2 (Proposal 504, November 2025): accountability by design Cycle 2 incorporated the lessons of the prior period. In addition to allocating 310,000 additional COMP (bringing the active total to 610,000 COMP) and adding 4 new delegates (Wintermute Governance, Tané, Curia Lab, and WOOF!), it introduced two structural rules that Cycle 1 lacked. The first was a cap of 60,000 COMP per entity (combining all sources - treasury plus any organic delegations received), preventing excessive concentration and leveling the playing field. The second was a mandatory semi-annual review with a minimum 80% participation threshold: every six months, the CGWG audits the on-chain voting participation of all treasury-delegated addresses. Those below the threshold have their delegation revoked by on-chain proposal, and the COMP is reallocated to candidates who demonstrated engagement during the period. 4. Participation: What the Data Shows The historical series tells a story about the profile of who governs, not just about how much was voted. The 999K average COMP in 2021 looks healthy, but conceals a fragile structure: participation was high because very few addresses concentrated enormous voting power and were still actively voting. The median of roughly 1M votes slightly above the average reveals that the distribution was tight: almost every proposal had similar participation, which seems positive but simply means the same few large holders showed up consistently. When those large holders gradually stepped back through 2022 and 2023, participation fell with them, with no mechanism to compensate for the departure, from 999K down to 702K in 2022 and 580K in 2023, each data point representing dozens of proposals with progressively less active voting power. In 2024, looking only at the annual average of 596K misses the most important detail. The first half registered just 508K, close to the minimum quorum of 400K, while the second half rose to 713K following Delegate Race Cycle 1. The difference between the two halves (205K more COMP per proposal on average) corresponds directly to the entry of delegates with a obligation to vote. The median of 575K, close to the average, indicates that the recovery was consistent throughout the semester, not concentrated in a few atypical proposals. In 2025, the average reached 695K, and partial 2026 data shows 835K, the highest average since 2021. But comparing to 2021 reveals the profile shift. The typical 2021 delegate was a large capital holder who voted when they judged it relevant, with no formal obligation. The typical 2026 delegate is a governance-specialized organization, with treasury allocation subject to semi-annual review - and one that loses voting power if they don't participate. 5. The Current Landscape: Who Governs in 2026 Top 15 delegates by voting power (July 2026) Current delegates fall into three groups: delegates receiving treasury allocations, the Foundation, and participants with their own tokens. Treasury-delegated delegates receive voting power through Delegate Race Cycles 1 and 2. Today, they are the primary contributors to helping Compound reliably reach quorum, particularly on more operational proposals, like parameter changes and market listings or delistings. Within this group, PGov stands out as the delegate with the most voting power. Other important contributors include Arana Digital, Tané, and WOOF. The Compound Foundation is a key stakeholder in the DAO's governance. With nearly 500K COMP in voting power, it is one of the strongest entities in governance. Its participation tends to be selective, but its support proved decisive in the approval of Proposals 580, 581, and 582, which established Compound v4 funding, the delegate compensation program, and improved treasury management. In that sense, the Compound Foundation is the most influential governance participant, not only through voting power but through its ability to coordinate with the treasury-delegated delegate group. Finally, Humpy is the most powerful single entity identified in Compound's governance. With 1.540.000 COMP consolidated across two identified addresses, he holds more than double the Foundation's voting power and more than 13x the cap applied to treasury delegates. The address 0x3B6431fb5C71105cB3EaB2Cf058B135d4cCFc9C5 is not openly acknowledged as his, but has documented links to his wallets, including humpy.eth. After the attack attempt in Proposal 289, there were concerns that Humpy might block Proposals 580, 581, and 582 given his opposition to them. However, through coordination between the Compound Foundation and all DAO stakeholders, they were approved with a historic vote total: 1.88M COMP - participation three times the average. The governance dynamic that emerges is one where delegates handle operational decisions, the Foundation coordinates on major strategic votes, and Humpy acts as an opposing force on more sensitive matters, such as the Compound v4 proposal. The Delegate Rebalancing - and blockful's Entry The accountability mechanism introduced in Delegate Race Cycle 2 was tested for the first time in May 2026. The review window covered Proposals 505 to 586, six months of votes since Cycle 2 passed in November 2025. The CGWG audited the on-chain participation of all 13 treasury-delegated addresses and found three below the 80% threshold: Michigan Blockchain had voted on 58 of 77 proposals (75.3%); ReservoirDAO/AlphaGrowth participated in just 35 (45.5%); and Sharp voted on none (0.0%). All three had their delegations revoked, freeing up 81.178 COMP for redistribution. The reallocation followed the waterfall method: the 81.178 COMP were distributed in descending order of participation among eligible candidates who had demonstrated consistent engagement during the same period. FranklinDAO, with perfect participation (100%), received the 8,262 COMP needed to reach its 60K cap. DAOplomats, at 85.7%, received 40,000 COMP. blockful, at 83.1%, received the remaining 32,916 COMP. The blockful case illustrates what the system was designed to enable: the organization had been participating in governance as an organic delegate with only ~440 COMP of its own - insufficient voting power to meaningfully influence any vote, but enough to demonstrate commitment. The rebalancing recognized that track record and elevated blockful to ~33.357 COMP in total voting power, making it one of the protocol's principal active delegates. It was the first time in Compound's history that voting power was revoked for low participation and redistributed on merit. This rebalancing will still be formally submitted as an on-chain proposal before being executed. The Compensation Program: the version that passed (May 2026) The second attempt to compensate delegates drew on the lessons of the February 2025 defeat. The new format was leaner and more demanding at the same time: instead of 16 delegates across two tiers with compensation proportional to participation, the 2026 program selected the 8 delegates with the highest participation across Proposals 491–582 and adopted a binary criterion, either a delegate reaches ≥ 95% participation in a given month and receives the full amount (up to $4,000), or receives nothing. Payment shifted to USDC (not COMP), eligibility reviews became quarterly (not semi-annual), and Snapshot votes began counting toward the criterion. The initial cohort comprised 8 delegates with participation rates between 87% and 100% over the evaluated period. Four had perfect participation (100%): Tané, Wintermute Governance, Curia Lab, and PGov. Arana Digital and allthecolors came in at 97.7%, FranklinDAO at 95.4%, and Avantgarde at 87.4% — the only one below 95% in the qualifying period, but still eligible under the historical selection criterion. In the first month of operation (May 2026), all 8 delegates achieved 100% participation across the 6 proposals in the period, and 7 of 8 met the communication requirements. The total distributed was $30,000 in USDC, the first time Compound formally compensated its delegates for governance participation. 6. The Future of Compound Governance Compound is moving toward a more formalized governance model, where delegates are paid for their participation and commit to voting on proposals, reviewing and approving (or rejecting) all operational changes on an ongoing basis. This is not a new model and functions well in some DAOs, such as Sky, though it has also shown weaknesses, as seen in Arbitrum. This model will require the DAO to spend several hundred thousand dollars annually, but it will also ensure that quorum is reliably reached in Compound. It is important that the CGWG, together with the Compound Foundation, continues to review and coordinate with the delegates receiving Franchiser voting power, to ensure the success of both the compensation and delegation programs. Blockful is pleased to be part of Compound's delegate group and to contribute to the protocol's long-term success.

## Publication Information

- [blockful blog](https://paragraph.com/@blockful/): Publication homepage
- [All Posts](https://paragraph.com/@blockful/): More posts from this publication
- [RSS Feed](https://api.paragraph.com/blogs/rss/@blockful): Subscribe to updates
- [Twitter](https://twitter.com/blockful_io): Follow on Twitter