Treasury-funded incentive programs across major DAOs have consistently boosted delegate activity but failed to decentralize voting power. Three deep dives into Arbitrum, Uniswap, and ENS reveal a similar pattern: programs that reward existing delegates entrench incumbents, compress participation into “check-the-box” compliance, and leave macro-level voter turnout structurally fragile.
A strong counter-model is ENS’s Delegation Incentives Program, which directs 90% of rewards to delegators rather than delegates, treating governance incentives as a security investment against capture.
Cardano’s current landscape, where close to 60% of governance participants default to Abstain and roughly 11 DReps control slightly more than 50% of active voting power (see the Cardano Governance Health Dune dashboard for live figures), makes designing anti-concentration experiments highly urgent.
Below is an overview of mechanism designs, measured outcomes, and failure modes from three protocols, followed by five concrete treasury-funded experiment proposals for Cardano that prioritize decentralization alongside participation.
Part 1: Cross-Protocol Lessons from Three Incentive Models
We see three recurring findings across these protocols:
- Delegate-only compensation creates a “cartelized dividend”: A self-perpetuating loop where paid delegates vote to continue their own compensation (Uniswap forum user Userisky’s term).
- Participation quantity and quality diverge sharply: Once payments begin, quality drops. Arbitrum’s RAD program found rationale posting “reduced to a pointless matter of bureaucracy,” while Uniswap’s DRI achieved 100% voting rates among paid delegates but saw overall DAO turnout decline to its lowest-ever quorum margins.
- Economic stakes drive participation, not incentive programs: Uniswap’s fee-switch vote drew 125M UNI (3× normal turnout), and Cardano’s 2025 budget vote achieved near-universal participation among delegated stake.
The design lesson for Cardano is clear: reward the act of delegation itself (moving ADA from Abstain or non-participation to active DReps), not just the act of voting, and build in explicit anti-concentration mechanics.
Arbitrum: The Most Expensive Delegate Program, the Sharpest Controversies
Arbitrum has run the most complex and costly governance incentive stack of any major DAO, spending an estimated 437M ARB in cumulative commitments across protocol incentives (STIP, LTIPP, DRIP), delegate compensation (DIP, RAD), Foundation grants, and strategic partnerships since 2023.
The Delegate Incentive Program (DIP) launched in March 2024 under SEEDGov management with a 1.58M ARB budget. Version 1.0 enrolled 40 delegates, qualified 38, and used only 68.91% of its budget, a sign that activity was sufficient but not deep. DIP 1.5 (November 2024) dramatically scaled up to a $4.2M annual budget with tiered compensation: Tier 1 delegates earning up to $7,000/month required ≥85% participation, while Tier 3 delegates at ≥65% earned $3,000–$3,250/month. Enrollment peaked at 75 delegates in December 2024 with 67 qualifying.
A major failure occurred with DIP v1.6, when SEEDGov raised the new-delegate voting power minimum 10× to 500,000 ARB, cut delegate pay by roughly 40%, and requested a 25% raise for itself as program manager. The community pushed back heavily:
- Paid delegates dropped from 49 to 21 within two months (a 57% decline).
- Multiple delegates accused SEEDGov of “retroactive rule changes.”
- A formal proposal to sunset the DIP entirely was submitted, citing “subjective scoring, VP multiplier penalizing smaller delegates, and scope overreach.”
Both successor proposals (DIP 2.0 and “Triple Dip”) failed Snapshot votes in October 2025, leaving Arbitrum without any delegate incentive program for two months.
The Rewarding Active Delegates (RAD) program launched January 2026 as a simpler replacement, managed by Arbitrum’s OpCo rather than SEEDGov. RAD uses per-proposal budgets ($5,000–$15,000 per vote depending on type) with a 200,000 ARB minimum voting power requirement. In its first month (November 2025), RAD distributed $26,000 to 22 delegates across four votes. By February 2026, only 25 of 37 enrolled delegates received rewards, and total spending dropped to $5,000 for a month with only one proposal.
The cost-per-participation metrics highlight the issue. RAD averages ~$54,000/month for ~2.5 proposals versus DIP’s ~$67,000/month for ~5.6 proposals. In November 2025, each RAD delegate-vote cost approximately $295. But RAD enrollment (33–37 delegates) is significantly lower than DIP’s peak (78 delegates), partly because the 200K ARB threshold, worth ~$18,800 at current prices versus ~$41,100 when proposed, excludes small delegates. With ARB trading at $0.094 (a 95% decline from its $2.39 all-time high), the real-dollar value of ARB-denominated governance work has collapsed.
Protocol-level incentive programs have been larger but similarly mixed. STIP allocated 71.4M ARB to 56 projects, driving a 25% TVL increase within two weeks but also attracting governance fatigue (delegates reviewed 95+ applications) and documented Sybil activity identified by OpenBlock Labs. LTIPP distributed ~45M ARB via a council-mediated model that reduced delegate burden but still concentrated benefits among larger protocols. The current DRIP program commits 80M ARB over four seasons and has increased Arbitrum’s lending market share from 1% to ~9%.
Primary failure point: The “corporate bureaucrat” delegate problem, named by Tally.xyz, manifests when systems “incentivize delegate participation and communication, neglecting the crucial aspect of executing towards the protocol’s strategic objectives.” RAD Update #3 (March 2026) confirmed this is still an issue: delegates posted minimal rationales like “I voted for because I liked the proposal” to satisfy compliance requirements.
Uniswap: High Delegate Fidelity Masking Systemic Quorum Fragility
Uniswap’s Delegate Reward Initiative (DRI) is the cleanest experiment in delegate compensation, running four consecutive cycles from June 2024 through February 2026 with a total spend of approximately $1.9M in UNI.
The DRI pays up to $6,000/month per delegate split evenly: $3,000 for maintaining ≥80% voting participation, and $3,000 for timely vote rationales and community call attendance. Each cycle selects the top 15 delegates by a points-based scoring system that has grown more sophisticated over time. Cycle 1 (June–August 2024) scored delegates on voting only; by Cycle 4 (September 2025–February 2026), scores incorporated onchain and offchain voting (7 points), proposal authorship (3 points), community participation (1 point), UNI token ownership of ≥1,000 UNI (1 point, new), and perfect participation (1 point, new).
Individual delegate metrics are high: all 12 Cycle 1 delegates maintained 100% voting participation on both onchain and offchain votes. In Cycle 2, all 16 delegates stayed above 85%, with 10 achieving perfection. Community call attendance averaged 81.25% among paid delegates, with ~38 participants per call.
However, the macro metrics show a different trend. Uniswap’s quorum margin, the percentage above its 40M UNI threshold, has declined monotonically: 79.38% (2021) → 51.94% (2022) → 20.47% (2023) → 14.14% (2024) → 11.80% (2025). In 2025, 46% of proposals passed with less than 10% quorum margin. Without the DRI delegates’ participation, “quorums would not have been met” according to the Cycle 3 proposal text. But DRI addresses only the ~15 most engaged delegates; the structural threat is institutional actors like a16z, which retracted 20M+ UNI of delegations in June 2025, overnight removing voting power from Michigan Blockchain, she256, FranklinDAO, and six other university-affiliated delegates.
Voting power is heavily concentrated. A PANews four-year study found a Gini coefficient of 0.938, with the top 1% controlling 47.5% of voting power and the top 10% holding 91.4%. Only 42 wallets hold ≥1M UNI (proposal threshold); of those, only 14 have ever voted; only 5 have published proposals. The DAO effectively operates with ~13 “meaningful votes” per proposal.
The UNIfication fee-switch vote in December 2025 acts as a natural experiment for what drives turnout. The vote, activating protocol fees and burning 100M UNI from treasury, attracted 125M+ UNI in favor (99.9% support), the highest turnout in Uniswap history and nearly 3× the average 2025 proposal turnout of ~45M UNI. UNI price surged 35% within two hours of the proposal announcement. Takeaway: economic stakes drive broader mobilization than payment incentives, suggesting that programs should align with token holder value rather than just paying for compliance tasks.
Forum criticisms offer good lessons for Cardano. Penn Blockchain (FranklinDAO) argued DRI is “too focused on activity volume rather than value-added participation” and “measures whether delegates show up, not whether they deliver value.” Forum user Userisky warned: “A program like this is easy to start but difficult to stop, as the delegates are incentivized to keep voting for it.” DAOplomats noted that the Cycle 4 temp-check proposing significant changes to delegate incentives received only 3 replies before going to vote, suggesting the DRI community has become insular.
ENS: The Only Model That Rewards Delegators, and Why It Matters Most
ENS DAO’s Delegation Incentives Program, proposed by service provider blockful in January 2026, is a highly innovative governance incentive design. Unlike every other major protocol, ENS directs 90% of rewards to delegators (token holders who delegate) and only 10% to delegates (who vote). The underlying concept: governance security comes from tokens being actively delegated, not just from the delegate’s voting labor.
The mechanism design works as follows: A monthly reward pool R scales dynamically with month-over-month growth in active voting power, ranging from 5,000 ENS (0–10% VP growth) to 30,000 ENS (>100% VP growth). The total pilot budget is capped at 90,000 ENS (~$500,000 at ~$5.60/ENS) over three monthly rounds.
Delegator rewards (90% of R) are distributed proportionally to each delegator’s 180-day time-weighted ENS balance (the per-second average balance over the preceding 180 days, calculated from on-chain transfer events). This acts as the main anti-gaming mechanism: recently purchased or transferred tokens count for far less than tokens held over the full window. A Sybil attacker splitting their balance across wallets gains nothing since new wallets have near-zero historical balances. Each delegator’s payout is capped at 5% of R, with excess redistributed pro-rata. Payouts below 1 ENS enter a lottery mechanism (10 ENS prize, weighted by calculated payout, randomness from RANDAO of the last block of each round).
Delegate rewards (10% of R) go to “Active Delegates”, those who voted on at least 7 of the last 10 on-chain proposals, proportional to their average voting power. Per-delegate payouts are capped at 1% of R. As of January 16, 2026, blockful identified 55 qualifying delegates, of which only 26 held >10 ENS in voting power.
The anti-gaming stack includes: the 180-day time-weighted balance factor, per-delegate and per-delegator caps, deduplication of franchiser contracts (multi-wallet structures), on-chain-only vote counting, viewpoint-neutral rewards, and dynamic pool sizing. ENS founder nick.eth, while supportive overall, raised a sharp critique: caps are “pointless and counterproductive” because they’re trivial to circumvent and create incentives to split tokens across accounts, “permanently corrupting on-chain data about token distribution.”
The capture cost analysis provides a strong economic case for delegation incentives. Blockful calculated that the program could raise ENS DAO’s capture cost from $11M to $94M (at ~$10/ENS on January 16, 2026). Capture cost equals the market price of acquiring enough tokens to control >50% of active voting power. With low delegation rates, fewer tokens vote actively, making capture cheaper. By incentivizing delegation to active delegates, the denominator grows. Blockful’s Anticapture platform tracks this metric continuously.
ENS’s broader governance context is notable. The Nakamoto coefficient stands at 18, meaning 18 delegates can control a majority, compared to 8 for Compound and 11 for Uniswap. Roughly 60% of ENS voting rights are held by “community delegates” receiving tokens from multiple holders, versus 10% for Compound and Uniswap. This structural advantage stems partly from ENS’s airdrop distribution to domain holders, creating a broader initial base.
The ENS governance stack operates at roughly $5–$6M/year: three working groups with stewards earning $3,000–$4,500/month each, a secretary at $5,500/month, and a Service Provider Program (SPP2) funding 8 providers at $4.5M/year total. The EP 6.31 temp-check Snapshot vote closed February 9, 2026; circumstantial evidence suggests it passed (the forum lists it as “Active Proposal” and the timeline called for a February start), though no distribution reports had been published as of April 2026.
There are still outstanding issues: estmcmxci, despite voting in favor, warned the optimal strategy becomes “delegate, collect ENS, ignore governance,” attracting mercenary capital rather than durable participation. clowes.eth called the program “too complex” and argued the core problem is creating environments where people want to contribute, not financial inducements to participate.
Part 2: Cardano’s Governance Landscape in 2026
CIP-149
CIP-149 (“Optional DRep Compensation”), authored by Philip DiSarro of Anastasia Labs and created February 19, 2025, proposes an entirely off-chain, opt-in compensation standard requiring zero ledger changes. When a wallet builds a delegation transaction, it prompts users to select a basis-point percentage (stored as metadata label 3692). During subsequent reward withdrawals, the wallet adds an extra output sending the specified percentage to the DRep’s payment address (from CIP-119 metadata). The design explicitly serves as an MVP to gather data before committing to any ledger-level compensation system.
Current DRep Landscape
Cardano governance participation remains dominated by Abstain. Per the Cardano Governance Health Dune dashboard, close to 60% of governance participants default to the Abstain option, and only a modest share of circulating ADA reaches autonomous DReps. Roughly 11 DReps control slightly more than 50% of active DRep voting power, and the top of the distribution contains a mix of founding entities, wallet-affiliated DReps that benefit from default-delegation UX, and a small number of high-profile community DReps. Because exact figures on the Dune dashboard refresh continuously, any precise quantitative claim in this report should be cross-checked against the live dashboard at the time of proposal submission.
Anecdotal context: an existing initiative distributing voting power more broadly
As one example of a program that helps distribute voting power to a broader set of DReps, the Cardano Foundation has publicly delegated 360 million ADA to a range of community DReps across three cohorts focused on builders, adoption and operations. Programs of this kind illustrate that top-down delegation by large holders can seed a more diverse DRep set. The experiments proposed below are designed to run alongside such initiatives rather than be compared against them; their goal is to complement any top-down delegation with bottom-up, organic pathways into active governance.
The Governance Health Working Group published its KPI framework v1.0 on December 3, 2025, defining seven core metrics including DRep Gini coefficient and DRep correlation, but the report explicitly states it “does not yet provide live data or finalized code for every metric.” The Dune dashboard currently serves as the most comprehensive live view of governance health. The GHWG flags “herd behavior” as a monitored risk.
Cardano’s treasury, as of Epoch 625 (19 April 2026), holds ~1.6 Billion ADA. Between Epoch 531 (2024-12-30) and Epoch 605 (2026-01-04) Treasury increased from 1,637,199,824 ada to 1,672,567,635 ada, a change of 35,367,812 ada (+2.16%). Total additions: 292,133,962 ada and total withdrawals: 277,031,507 ada
Part 3: Five Treasury-Funded Experiments for Cardano
Design Philosophy: From Delegate Rewards to Delegation Security
These experiments share a core idea from ENS’s Anticapture framework: governance incentive spending is a security investment, not a reward. The primary objective is reducing capture cost by increasing the share of ADA actively delegated to DReps, not by enriching existing power-holders. Every experiment includes explicit anti-concentration mechanics, Gini/Nakamoto coefficient tracking, and kill switches triggered by power concentration.
All experiments assume the current tooling stack (GovTool, CF Voting Tool, Cexplorer, Adastat, CGOV, Cardanoscan, CIP-95 wallets, the Cardano Governance Health Dune dashboard) and complement, rather than duplicate, CIP-149’s voluntary delegator-to-DRep payments.
Experiment 1: Cardano Delegation Incentives Pilot (CDIP) (ENS Adaptation)
Hypothesis: Rewarding ADA holders for delegating to active DReps (rather than remaining on Abstain or non-voting) will increase the share of circulating ADA under active governance by ≥20% within 6 months, while reducing the DRep voting power Gini coefficient (centralization of voting power).
Mechanism design (directly adapted from ENS’s Delegation Incentives Program):
A monthly reward pool R scales with epoch-over-epoch growth in active delegated voting power, disbursed in ADA from a treasury withdrawal. The pool is split 90/10 between delegators and DReps.
Delegator rewards (90% of R): Distributed to all ADA holders delegated to a “Qualifying DRep” at the snapshot epoch, proportional to each delegator’s 90-epoch time-weighted ADA delegation balance (approximately 90 days, adapted from ENS’s 180 days to reflect Cardano’s 5-day epochs and pilot duration). The time-weighted balance is calculated as the per-epoch average ADA delegated to any Qualifying DRep over the prior 90 epochs, penalizing recent movements. Each delegator’s payout is capped at 3% of R to prevent whale concentration; payouts below 5 ADA enter a lottery pool (50 ADA prize per pool, weighted by calculated payout, randomness sourced from the epoch nonce).
DRep rewards (10% of R): Distributed to Qualifying DReps proportional to their time-weighted average voting power, capped at 1% of R per DRep. Qualification requires voting on ≥70% of governance actions in the prior 90 epochs AND having ≥10 unique delegators with ≥100 ADA each (prevents self-delegation-only DReps).
Dynamic pool sizing (per month):
| Active VP Growth (MoM) | Monthly Pool (ADA) | Delegator Cap | DRep Cap |
|---|---|---|---|
| 0–5% | 100,000 | 3,000 | 1,000 |
| 5–15% | 200,000 | 6,000 | 2,000 |
| 15–30% | 350,000 | 10,500 | 3,500 |
| 30–50% | 500,000 | 15,000 | 5,000 |
Treasury request: 2,000,000 ADA (covers maximum 6-month scenario of 350K/month with 200K operational buffer).
Eligibility criteria: Any ADA holder delegated to a Qualifying DRep may earn rewards. The 90-epoch time-weighting naturally favors long-term holders over speculators. The 3% delegator cap prevents any single wallet from capturing outsized rewards. Qualifying DRep thresholds (70% participation, ≥10 delegators) are deliberately low to include small active DReps.
Anti-concentration mechanics: To avoid reinforcing existing concentration, any DRep already sitting in the top 20 by voting power at the start of a monthly cycle is eligible for at most 25% of their otherwise-calculated DRep reward, with the balance redistributed pro-rata to the remainder of the qualifying pool. The delegator-side 90% pool has no such restriction: delegators are rewarded for actively delegating, regardless of which qualifying DRep they choose.
Success metrics:
- Active delegated VP increases ≥20% from baseline, measured at the epoch just prior to pilot launch.
- DRep voting power Gini coefficient decreases ≥0.02.
- Abstain delegation share decreases by ≥5 percentage points.
- Nakamoto coefficient (DReps needed for 50% VP) increases by ≥3.
- ≥500 new delegators move from Abstain to active DReps.
Failure criteria / automatic cutoff: Program terminates if:
(a) DRep Gini increases above baseline for two consecutive months,
(b) any single entity captures >20% of total distributed rewards via wallet splitting,
(c) active VP growth is <2% after three months despite maximum pool disbursement, or
(d) >50% of delegator rewards are sold within 7 days of distribution (monitored via on-chain analysis).
Timeline: 6-month pilot (Months 1–2 setup and tooling integration with existing governance tools, Months 3–8 active distribution).
Tooling: Calculation engine built on Koios/Blockfrost API data; distribution via CF voting portal claims or direct transfers from a pilot multisig; monitoring via the Cardano Governance Health Dune dashboard and the GHWG KPI framework.
Complement to CIP-149: CIP-149 enables voluntary delegator-to-DRep payments from staking rewards. CDIP provides treasury-to-delegator rewards for the act of delegation itself. These are complementary: CIP-149 compensates DReps for governance labor; CDIP compensates delegators for providing governance security through active delegation.
Experiment 2: Long-Tail DRep Bootstrap Matching Fund
Hypothesis: An inverse-weighted matching fund that provides larger per-delegator matches for DReps with lower voting power will increase the Nakamoto coefficient by bootstrapping DReps currently ranked 21–200, without increasing concentration among the top 20.
Mechanism design:
The Bootstrap Fund matches organic ADA delegations to small-and-mid-tier DReps using treasury ADA, with match ratios inversely correlated to existing DRep voting power. The fund does not delegate ADA itself: it augments rewards for delegators who choose to delegate to eligible DReps.
DRep eligibility tiers:
| DRep Tier | Voting Power | Match Ratio | Monthly Cap per DRep |
|---|---|---|---|
| Micro | <5M ADA | 3:1 | 75,000 ADA |
| Small | 5M–20M ADA | 2:1 | 50,000 ADA |
| Mid | 20M–50M ADA | 1:1 | 30,000 ADA |
| Large (>50M ADA) | Ineligible | N/A | N/A |
DReps must meet minimum activity requirements: voted on ≥50% of governance actions in the prior 60 epochs, registered for ≥30 epochs, and published at least one rationale (verified via on-chain metadata).
How matching works: When a delegator delegates ≥500 ADA to an eligible DRep and maintains that delegation for at least 10 epochs, the Bootstrap Fund distributes a matching reward to the delegator at the end of each 30-epoch cycle. Match reward = (net new ADA delegated to that DRep by this delegator in the cycle) × (match ratio for the DRep’s current tier), up to the DRep’s monthly cap. Rewards accrue to delegators, not DReps, keeping incentives aligned with the decentralization thesis.
Treasury request: 1,500,000 ADA for a 6-month pilot.
Anti-concentration mechanics: The inverse weighting is the core anti-concentration mechanism: it becomes unprofitable to delegate to already-large DReps. The >50M ADA exclusion directly prevents top-20 reinforcement. Per-DRep caps prevent any single DRep from absorbing disproportionate matching.
Success metrics:
- Number of DReps with >1M ADA voting power increases by ≥30.
- Nakamoto coefficient increases by ≥5.
- Median DRep voting power increases while mean stays stable (compression of distribution).
- ≥2,000 delegators participate in matching.
Failure criteria: Terminate if
(a) >30% of matched delegations are withdrawn within 5 epochs of reward distribution (gaming),
(b) the program results in net delegation moving from mid-tier to micro-tier DReps via strategic re-registration (DReps splitting to game tier thresholds), or
(c) total matching distributed exceeds 80% of budget in the first 2 months (unsustainable trajectory).
Timeline: 6-month pilot. Month 1 setup and outreach to DRep community; Months 2–7 active matching cycles.
Tooling: Delegation tracking via Blockfrost/Koios APIs; matching calculations run by a designated administrator (potentially Intersect Civics Committee); distribution via multisig.
Experiment 3: Abstain-to-Active Delegator Onboarding Campaign
Hypothesis: Targeted incentives for first-time delegators who move ADA from the Abstain default to active DReps will meaningfully reduce Abstain delegation and onboard ≥5,000 new governance participants within 6 months.
Mechanism design:
Two interlocking components target distinct barriers to participation.
Component A: First Delegation Bonus. Any wallet that
(1) has never delegated to an autonomous DRep (currently delegated to Abstain, No Confidence, or not delegated at all),
(2) delegates ≥100 ADA to any active DRep, and
(3) maintains that delegation for ≥15 epochs, receives a one-time bonus of 25 ADA. This is deliberately small: enough to cover transaction fees and signal that governance participation is valued, but too small to attract pure mercenary capital. Capped at 5,000 bonuses total (125,000 ADA).
Component B: DRep Discovery Matching. For the first 60 epochs of the pilot, delegators who delegate to any DRep ranked outside the current top 30 receive a 2× delegation bonus (50 ADA instead of 25 ADA for first-time delegators). This nudges new participants toward underrepresented DReps.
Treasury request: 500,000 ADA (125K first-delegation bonuses + 125K discovery matching + 250K operations, marketing, and contingency).
Anti-concentration mechanics: Component B explicitly incentivizes delegation to DReps outside the top 30. One-time bonuses prevent ongoing rent-seeking.
Success metrics:
- Abstain delegation share (per the Dune dashboard) decreases by ≥5 percentage points.
- ≥5,000 wallets make their first autonomous DRep delegation.
- ≥60% of new delegators maintain delegation after 30 epochs (stickiness).
- Net new delegators distribute across ≥50 distinct DReps (breadth).
Failure criteria: Terminate if
(a) >40% of bonus recipients undelegate within 5 epochs of receiving the bonus,
(b) >50% of new delegations concentrate in ≤5 DReps, or
(c) Sybil analysis (using time-weighted wallet age, transaction patterns) identifies >20% of claimed bonuses as fraudulent.
Anti-Sybil measures: Wallets must have existed for ≥90 epochs before the pilot start date. Wallet must hold ≥100 ADA continuously. One bonus per staking key. Sybil detection via on-chain clustering analysis (adapted from Gitcoin’s COCM methodology).
Timeline: 6-month pilot. Month 1 tooling development and marketing materials; Months 2–7 active enrollment.
Tooling: Integration with GovTool and wallets for delegation flow; Explorers for tracking and verification; community outreach via Intersect, Cardano Forum, and social media.
Complement to existing programs: CIP-149 addresses DRep compensation from existing delegators. This experiment uniquely targets the majority of governance participants currently sitting on Abstain, the largest untapped pool.
Experiment 4: Quadratic DRep Participation Fund
Hypothesis: A fund that rewards DReps based on the square root of their Sybil-resistant unique delegator count (rather than total ADA delegated) will incentivize DReps to attract many small delegators rather than few large ones, compressing the Gini coefficient of DRep voting power.
Mechanism design (inspired by Gitcoin’s quadratic funding mechanism):
Each month, a participation pool of ADA is distributed to Qualifying DReps using a quadratic formula. Rather than rewarding DReps proportionally to their voting power (which reinforces concentration), the formula weights rewards by the square root of unique active delegators, after Sybil filtering.
Formula: DRep_reward(j) = √(delegator_count(j)) / Σ√(delegator_count(k)) × Pool
Where delegator_count(j) = number of Sybil-filtered unique wallets delegating ≥50 ADA to DRep j, with delegation maintained for ≥10 consecutive epochs.
Qualification requirements: DRep must have voted on ≥60% of governance actions in the prior 60 epochs, published at least one rationale, and have ≥5 unique qualifying delegators.
Per-DRep reward cap: 5% of monthly pool. No DRep may receive more than 50,000 ADA across the entire pilot.
Monthly pool: 250,000 ADA (fixed, not dynamic: simplicity reduces administrative overhead).
Treasury request: 1,500,000 ADA (250K × 6 months).
Anti-concentration mechanics: The quadratic formula is inherently anti-concentration. A DRep with 100 delegators receives √100 = 10 “quadratic points,” while a DRep with 10,000 delegators receives √10,000 = 100: only 10× more despite having 100× more delegators. This rewards breadth over depth. The per-DRep cap further limits large-DRep capture.
Sybil resistance: the central design challenge
Quadratic mechanisms are only as strong as the Sybil resistance of their “unique delegator” count. Since Cardano has no built-in proof-of-personhood and UTXO splitting is cheap, this experiment must treat Sybil resistance as its primary engineering problem, not an afterthought. The Sybil stack layers five independent defences:
- Wallet-age gate: Only wallets that existed for ≥90 epochs before the pilot start count toward
delegator_count. This forces any attacker to either prepare Sybils 15+ months in advance or buy aged wallets on a secondary market. - Continuous minimum balance: Qualifying wallets must have held ≥100 ADA continuously (not just at snapshot) throughout the 90-epoch lookback. Creating 100 fake wallets each with 100 ADA held for 15 months represents a meaningful opportunity cost.
- Connection-oriented cluster matching (COCM, borrowed from Gitcoin GG18+): Before applying the quadratic formula, group delegators into clusters based on shared funding sources, near-simultaneous delegation events, and on-chain transaction graph proximity. Delegators who cluster tightly with many others have their voting weight in the quadratic sum reduced. This does not require identifying Sybils individually: it reduces the marginal value of each additional Sybil that pattern-matches with its siblings.
- Funding-source entropy score: For each delegator wallet, compute the entropy of its historical funding sources. Wallets funded entirely from a single source within a short window are down-weighted relative to wallets with diverse transaction histories.
- Rate-limit on DRep growth: A DRep whose Sybil-filtered delegator count grows by more than 50% within a single 30-epoch cycle is quarantined for that cycle: they still receive rewards based on their pre-spike count, and the spike is investigated before the next cycle. This prevents “flash Sybil” attacks that appear just before a reward snapshot.
An attacker attempting to game this system faces a layered cost: capital lockup (100 ADA × N wallets × 15 months), wallet-age procurement cost, clustering penalties that compound as more Sybils are added, and the rate-limit on growth. Combined, these make the expected return on Sybil spend decisively negative for realistic pool sizes. Importantly, this Sybil stack is also the reusable infrastructure Experiments 1, 2, and 3 rely on, so the engineering investment benefits the entire experimental portfolio.
Success metrics:
- Median Sybil-filtered unique delegators per DRep increases by ≥50%.
- DRep Gini coefficient decreases ≥0.03.
- ≥100 DReps qualify for rewards (vs. ~55 qualifying under ENS’s 7/10 threshold).
- Small DReps (those ranked 50–200 by VP) see delegator count growth outpacing top-20 DRep delegator growth.
Failure criteria: Terminate if
(a) the Sybil pipeline detects >15% fraudulent delegator wallets that evade filtering,
(b) >50% of rewards flow to top-10 DReps by VP, or
(c) DRep registration spikes >50% without corresponding Sybil-filtered delegator growth (indicating empty DRep shell creation).
Timeline: 6-month pilot. Month 1 tooling, formula validation, and Sybil pipeline stress-testing; Months 2–7 active distribution.
Tooling: Delegator counting via Blockfrost/Koios APIs; Sybil pipeline developed as an open-source module reusable across experiments; quadratic calculations by designated administrator; results published transparently on Cardano Forum and Intersect.
Complement to existing programs: This experiment rewards DReps based on their ability to attract genuine, Sybil-resistant community delegation: a fundamentally different signal from voting-power size, and one that rewards grassroots legitimacy.
Experiment 5: DRep Rationale Quality and Peer Review Stipend
Hypothesis: Compensating DReps for high-quality governance rationales (evaluated by peer review, not just submission) will increase rationale publication rates to ≥80% while maintaining meaningful content quality, avoiding the “corporate bureaucrat” failure observed in Arbitrum.
Mechanism design: This experiment explicitly addresses the quality-versus-quantity gap identified across all three comparator DAOs. Rather than rewarding DReps for merely voting or submitting rationales (Arbitrum’s failed approach), it creates a small peer-review ecosystem.
Rationale submission incentive: On Cardano, governance actions currently have a lifetime of 6 epochs, and DReps are expected to vote during that window and attach a rationale to their vote via CIP-100 metadata. Under this experiment, any DRep who votes on a governance action during its active lifetime and publishes a rationale (on-chain via CIP-100 metadata, or on the Cardano Forum with a verifiable DRep signature that is linked from the on-chain vote’s metadata anchor) receives a base payment of 200 ADA per governance action voted on with rationale. Rationales submitted after the action has left its voting window are ineligible. Maximum 20 qualifying rationales per DRep across the pilot.
Peer review layer: A rotating panel of 7 reviewers (selected from the DRep community via lottery, weighted by time-weighted delegation) evaluates each rationale on three dimensions: (1) substantive reasoning (does it engage with the proposal’s merits?), (2) transparency (does it disclose conflicts of interest?), (3) accessibility (can a non-expert understand the position?). Each dimension is scored 1–3. Rationales scoring ≥7/9 receive a quality bonus of 300 ADA. Reviewers receive 100 ADA per review, capped at 30 reviews per reviewer per pilot.
Anti-gaming: Rationales are anonymized for peer review (DRep identity removed). Reviewers who consistently score all rationales identically (variance <0.5 across >10 reviews) are removed and replaced. Rationales generated by AI without human editorial engagement are flagged via stylometric analysis and community reporting.
Treasury request: 1,000,000 ADA (base rationale payments ~200K, quality bonuses ~200K, reviewer compensation ~200K, administration ~200K, contingency ~200K).
Anti-concentration mechanics: The flat 200 ADA base payment (not proportional to voting power) means small DReps receive the same per-rationale compensation as large ones. The lottery-based reviewer selection ensures review power doesn’t concentrate. The 20-rationale cap per DRep prevents large DReps from dominating rewards.
Success metrics:
- DRep rationale publication rate increases from current baseline to ≥80%.
- Average peer review quality score ≥6/9.
- ≥50 unique DReps publish rationales.
- Delegator surveys indicate improved ability to evaluate DRep positions.
Failure criteria: Terminate if
(a) >60% of rationales score <5/9 for three consecutive months (check-the-box behavior despite peer review),
(b) fewer than 20 DReps participate after three months, or
(c) reviewer pool cannot maintain ≥5 active reviewers.
Timeline: 6 months. Month 1: recruit initial reviewer panel, build rationale submission pipeline (using CIP-100 metadata standard and Cardano Forum integration); Months 2–7: active operation.
Tooling: CF voting tool for voting and rationale submission; app.cgov.io for rationale reading; other tooling integration for rationale links on DRep profiles; Intersect Civics Committee oversight.
Complement to existing programs: CIP-149 compensates DReps financially; this experiment compensates them specifically for governance labor quality, creating a continuous quality signal that can inform any future selection processes the community runs for delegation, stewardship, or committee roles.
Anti-Centralization Comparison Matrix
| Experiment | Rewards delegators (not just DReps) | Favors long-tail DReps | Onboards new participants | Explicit anti-concentration metric | Caps top-DRep capture | Caps per-entity rewards | Complements CIP-149 |
|---|---|---|---|---|---|---|---|
| 1. CDIP (ENS adaptation) |  90% to delegators |
Low qualification thresholds |
Partial (incentivizes Abstain-to-active movement) | Gini + Nakamoto tracking; kill switch on Gini increase |
Top-20 DReps earn at most 25% of their DRep-side reward |
3% delegator cap, 1% DRep cap |
Treasury-to-delegator vs. CIP-149 delegator-to-DRep |
| 2. Long-Tail Bootstrap | Match rewards go to delegators |
Inverse weighting; >50M ADA DReps excluded |
Partial (attracts delegators to small DReps) | Nakamoto coefficient target +5 |
Large DReps ineligible |
Per-DRep monthly cap |
Grassroots, orthogonal to CIP-149 |
| 3. Abstain-to-Active | First-delegation bonus to delegators |
2× bonus for DReps outside top 30 |
Primary focus |
Distribution breadth across ≥50 DReps |
Partial (discovery matching biases away from top) | One-time bonus per staking key |
Targets the untapped Abstain pool |
| 4. Quadratic DRep Fund | Indirect (via Sybil-resistant count) | √delegators formula favors broad support |
Partial (incentivizes delegator outreach) | Gini target ≥0.03 decrease; Sybil-filtered metrics |
Partial (quadratic diminishes large-DRep advantage) | 5% pool cap per DRep |
Rewards community legitimacy, not VP |
| 5. Rationale Quality |  DRep-focused |
Flat payment regardless of VP |
DRep-facing |
Partial (flat payment prevents VP-proportional capture) | All active DReps eligible on equal per-rationale terms |
20-rationale cap per DRep |
Rewards quality; CIP-149 rewards financial labor |
Combined Budget and Sequencing
| Experiment | Budget (ADA) | Priority |
|---|---|---|
| 1. CDIP (ENS adaptation) | 2,000,000 | Highest: core decentralization mechanism |
| 3. Abstain-to-Active onboarding | 500,000 | High: feeds delegators into CDIP |
| 2. Long-Tail Bootstrap | 1,500,000 | High: structural Nakamoto improvement |
| 4. Quadratic DRep Fund | 1,500,000 | Medium: novel mechanism, higher complexity |
| 5. Rationale Quality | 1,000,000 | Medium: quality layer, can start later |
| Total | 6,500,000 ADA |
At 6.5M ADA total, this represents 1.9% of the 2026 Net Change Limit (350M ADA) and roughly 0.4% of the treasury balance: a modest pilot investment relative to the governance security value at stake. For context, 277 million ada were withdrawn from the Cardano Treasury in 2025-
Conclusion: Governance Security, Not Governance Rewards
Studying these three protocols shows that delegate compensation programs are necessary but insufficient, and when designed poorly, actively harmful to decentralization. Arbitrum’s DIP v1.6 controversy and Uniswap’s declining quorum margins despite $1.9M in DRI spending both demonstrate that simply paying people to participate does not solve the structural problem of concentrated voting power.
ENS’s Delegation Incentives Program takes a different approach: treating incentive spending as capture resistance infrastructure rather than delegate compensation. By rewarding delegators (the source of governance security) over delegates (the executors of governance labor), and by gating reward pools to actual growth in active voting power, ENS aligns individual incentives with collective security outcomes. Cardano’s Abstain dominance, close to 60% of participants per the Cardano Governance Health Dune dashboard, represents both the largest risk and the largest opportunity. Every ADA moved from Abstain to an active DRep directly reduces the cost of capturing DRep voting power.
The five experiments proposed here form an integrated stack: CDIP and Abstain-to-Active onboarding attack the demand side (why should I delegate?), the Long-Tail Bootstrap and Quadratic DRep Fund attack the supply side (which DReps deserve support?), and the Rationale Quality stipend addresses the legitimacy side (is governance participation meaningful?). All five include explicit kill switches triggered by concentration increases, ensuring that if any experiment inadvertently centralizes power, it self-terminates before causing lasting damage.
The critical metric here is the DRep Nakamoto coefficient: the smallest number of DReps whose combined voting power would suffice to carry or block an outcome in the DRep body. Today that number sits at roughly 11 for a simple DRep majority. These experiments aim to raise it meaningfully within 12 months. The DRep Nakamoto coefficient is just one axis of Cardano’s overall capture resistance. Under CIP-1694, no governance action is decided by a single body. Each action requires at least two of the three bodies (DReps, SPOs, and the Constitutional Committee) to vote in favor, and some (such as hard forks) require ratification by all three. Strengthening DRep decentralization therefore reinforces, rather than replaces, the structural checks that the CC and SPO bodies already provide. To ensure this robust decentralization, the community should prioritize funding these experimental pilots in the upcoming budget cycle.