This paper describes the methodology behind ChainSights' Governance Vitality Score (GVS) and the DAO Governance Index (DGI). GVS is a composite score (0–10) measuring the governance health of a single DAO across four dimensions: participation quality, delegate engagement, power distribution, and community breadth. DGI aggregates GVS scores into an ecosystem-wide benchmark, analogous to the S&P 500 for DAO governance health. We cover all scoring components, their formulas, data sources, confidence thresholds, data quality measures, the Vote Quality Score (VQS) for individual delegates, and known limitations. As of March 2026, 47 DAOs are tracked and 4,279 individual voters analyzed.

Decentralized Autonomous Organizations (DAOs) present a novel governance challenge: decision-making power is formally distributed across token holders, but the on-chain record rarely reflects genuine decentralization. Voting is dominated by a small number of wallets, many proposals pass with single-digit participation, and "community consensus" frequently means three whales agreed.

Existing governance analytics focus on raw participation numbers — total voters, total proposals, total voting power — without distinguishing who is voting, how thoughtfully, or how concentrated power has become.

ChainSights takes a different approach: measure governance health through signals that correlate with genuine decentralization and meaningful participation, not vanity metrics.

GVS (Governance Vitality Score) is that measure for a single DAO.
DGI (DAO Governance Index) is the ecosystem benchmark derived from GVS scores.

All scoring is based on publicly available on-chain and off-chain governance data:

All scores are computed from raw data with no reliance on third-party governance ratings or curator judgment.

Update frequency: All GVS scores are recalculated daily via an automated pipeline. DGI benchmarks are recalculated immediately after individual DAO scores complete.

Governance data contains noise. Treasury wallets, exchange cold wallets, and bridge contracts technically "vote" on-chain but do not represent human community participation. Including them inflates voter counts and distorts participation rates.

ChainSights applies a classification filter before computing any participation-based metric:

• Treasury wallets — DAO treasuries holding protocol funds

• Exchange wallets — CEX hot/cold wallets (Coinbase, Binance, Kraken, etc.)

• Bridge contracts — Cross-chain bridge addresses (Polygon, Optimism, Arbitrum, Wormhole, etc.)

• Protocol contracts — Routers, factories, token contracts, and other smart contracts

• Individual wallets — All personal EOAs participate fully

• Fund/VC wallets — Venture capital and investment funds are legitimate governance participants

• Multisig wallets — Flagged for transparency, but retained in calculations

Wallet labels are sourced from the eth-labels dataset (115,000+ labeled Ethereum addresses) supplemented by a manually curated override file for major wallets not in the dataset. Labels are cached in a wallet_labels database table and refreshed periodically.

The filter is applied in: HPR voter counts, DEI delegate identification, PDI voter-power distributions, and GPI small-holder participation.

GVS is a composite score on a 0–10 scale representing the governance health of a single DAO. It is composed of four component scores, each measuring a distinct dimension of governance quality:

GVS formula:

GVS = (HPR × 0.35) + (DEI × 0.25) + (PDI × 0.20) + (GPI × 0.20)

Each component is normalized to the 0–10 scale before composition.

Weight: 35%

HPR measures what proportion of a DAO's voters are genuine human participants, as opposed to bots, treasury wallets, or coordinated sybil clusters.

Formula:

HPR = (Likely Human Voters / Total Unique Voters) × 10

Where Likely Human Voters = total unique voters minus those identified as bots, protocol contracts, exchange wallets, or treasury addresses (see Section 3).

Important distinction: HPR measures voter quality (are voters human?), not voter quantity (what % of followers voted). A DAO can have 10/10 HPR but only 0.5% absolute participation rate. Both metrics matter — participation rate data is provided separately in the Deep Dive Report.

Why 35% weight: Sybil resistance is the foundation of meaningful governance. A vote where half the "voters" are controlled addresses is not a legitimate vote, regardless of the outcome.

Weight: 25%

DEI measures how actively top delegates participate in governance proposals using a Weighted Participation Rate.

Delegate identification: Top 20% of voters by voting power (minimum 3 delegates, maximum 50)

Lookback window:

• Standard: last 90 days (up to 30 proposals)

• Adaptive: if fewer than 5 proposals in 90 days, the window extends to 180 days and recency weighting is disabled

Recency weights:

Formula:

DEI_raw = Σ (delegate_i_participationRate × recencyWeight) / Σ recencyWeight
DEI = DEI_raw × 10

Why it matters: Delegation is only effective when delegates are active stewards. A delegate holding large voting power who votes on 20% of proposals is less valuable than one who votes on 80%. The Weighted Participation Rate penalizes both inactivity and stale participation.

Weight: 20%

PDI measures voting power concentration and governance decentralization.

Sub-signals:

1. Gini Coefficient (40% of PDI) — Measures inequality in voting power distribution. Gini = 0 means perfectly equal, Gini = 1 means one address holds all power.

2. Nakamoto Coefficient (30% of PDI) — Minimum number of addresses needed to control >50% of total voting power. Higher = more decentralized.

3. Concentration Score (30% of PDI) — Percentage of total voting power held by the top 5 voters.

PDI formula:

PDI = (giniScore × 0.4) + (nakamotoScore × 0.3) + (concentrationScore × 0.3)

All sub-signals are normalized to the 0–10 scale (higher = more decentralized) before composition.

Why it matters: High power concentration means a small number of whales can override community consensus. True decentralization requires broadly distributed voting power.

Weight: 20%

GPI measures the participation rate of smaller token holders — the "bottom 80%" of the token distribution by holdings.

Formula:

GPI = (Active Small Holders / Total Small Holders) × 10

Where Small Holders = all token holders outside the top 20% by holdings.

Why it matters: Governance that only involves large token holders is not community governance. High GPI indicates the DAO's rank-and-file members are engaged, not just the wealthy.

Each GVS score is accompanied by a confidence level indicating data completeness and reliability:

Low confidence does not mean a DAO is performing poorly — it means there is less data to analyze. New DAOs naturally start with lower confidence until they build governance history.

Historical scoring: DAOs with no recent Snapshot activity retain their most recent GVS score, labeled as "historical." Historical scores are excluded from DGI benchmark calculations.

While GVS measures a DAO's overall governance health, the Vote Quality Score (VQS) measures individual delegate behavior. It answers: "How thoughtful is this voter's participation?"

Each delegate receives a VQS from 0 to 10, composed of up to four signals.

Current coverage: 24 of 47 tracked DAOs (~51%), covering 4,279 analyzed voters.

Measures how long a voter waits before casting their vote after a proposal opens. Voters who consistently vote within minutes may not be reading the full proposal. Higher deliberation time suggests more thoughtful engagement.

Normalization: Median deliberation time across all of the voter's votes, scaled to 0–10 based on ecosystem distribution.

Measures diversity of a voter's choices across proposals. Voters who always select the same option (e.g., always "For") demonstrate less independent judgment than voters who evaluate each proposal on its merits.

Normalization: Entropy of the voter's choice distribution across their votes, scaled to 0–10.

Note on ranked-choice DAOs: In DAOs using ranked-choice or weighted voting (Aavegotchi, Balancer, Frax, StakeDAO, CVX), single-choice votes do not exist. Independence cannot be meaningfully calculated and is shown as N/A. The VQS composite is recalculated using only available signals (see Section 6.6).

Measures whether a voter engages across diverse proposal categories (treasury, governance, technical) rather than only voting on one type. Broader categorical engagement indicates deeper governance involvement.

Normalization: Number of distinct proposal categories voted on, scaled to 0–10.

Measures how independently a voter decides compared to the largest token holders (whales). Voters whose choices closely mirror whale voting patterns may be following rather than forming independent opinions.

Normalization: Divergence from top-5-voter consensus, scaled to 0–10.

Note: Same N/A handling as Independence for ranked-choice DAO voters.

When all four signals are available, each contributes 25% to the VQS:

VQS = (Deliberation + Independence + Focus + Originality) / 4

When some signals are unavailable (N/A), the weight is redistributed equally among remaining signals:

VQS = Σ(available signals) / N_available

Example: For a ranked-choice DAO voter where Independence and Originality are both N/A:

VQS = (Deliberation + Focus) / 2

This ensures no signal inflates the score via a placeholder value.

Score distribution: Red (<2): 3.5% · Orange (2–4.9): 73.5% · Yellow (5–7.9): 18.5% · Green (≥8): 4.4%

Across 24 DAOs, three distinct voter quality patterns emerge:

Pluralistic (Median VQS ≥ 6)
Delegates vote independently across diverse proposal types. High deliberation, varied choices, strong originality.
Examples: CVX (7.8) · Aavegotchi (7.1) · Frax (6.7)

Mixed (Median VQS 4–6)
A broad middle ground — some engaged, deliberate voters alongside more routine participants.
Examples: Arbitrum (4.2) · Balancer (4.4)

Consensus (Median VQS < 4)
Delegates tend to vote in alignment with each other and with large token holders. Common in mature protocols with strong community convergence.
Examples: ENS (3.2) · Uniswap (3.6) · Lido (1.8)

These archetypes reflect governance culture, not quality rankings. A Consensus DAO is not worse than a Pluralistic one — the context and protocol type matter.

The DAO Governance Index (DGI) is an ecosystem-wide benchmark that aggregates GVS scores across all qualifying DAOs. Where GVS answers "how healthy is this DAO's governance?", DGI answers "how does this DAO compare to the ecosystem?"

Analogy: GVS = individual company score. DGI = the S&P 500.

The DGI uses an equal-weighted average of all qualifying DAOs' GVS scores:

DGI = (1/n) × Σ GVS_i    for all qualifying DAOs i = 1..n

Why equal-weighted?

• Philosophically consistent — no single DAO dominates the benchmark

• No dependency on treasury size or token market cap

• Avoids circular bias (large DAOs should not get more weight and score higher)

• Transparent and reproducible by anyone with the data

DGI is a family of 9 indices:

Composite: DGI Composite — Average GVS across all qualifying DAOs. The headline number.

Category indices: DGI DeFi · DGI Infrastructure · DGI Public Goods · DGI Social

Component indices: DGI-HPR · DGI-DEI · DGI-PDI · DGI-GPI

DAOs that fail these criteria are tracked but excluded from DGI calculations until they meet the threshold. A grace period applies before exclusion to avoid penalizing brief data gaps.

Snapshot-only coverage. VQS currently analyzes Snapshot data. DAOs using on-chain governance via Tally or Governor contracts are not covered for VQS.

Quantitative signals only. Proposal quality, forum discussion depth, constitutional design, and off-chain coordination are not captured.

High score ≠ perfect DAO; Low score ≠ bad DAO. A Consensus DAO may be making excellent decisions with high alignment.

Wallet classification is imperfect. eth-labels and manual overrides do not cover all protocol-controlled wallets.

Historical scores. DAOs with no recent Snapshot activity are scored based on their last available data.

This is a tool for improvement, not a verdict. Use GVS and DGI as a starting point for governance review, not as a substitute for deeper analysis.

No financial advice. GVS and DGI scores are governance health indicators only. Nothing in this paper or on chainsights.one constitutes financial, investment, or legal advice.

No endorsement. Inclusion in ChainSights rankings does not constitute endorsement of any DAO, protocol, or token by ChainSights or its affiliates.

GDPR compliance. ChainSights processes publicly available on-chain data in accordance with GDPR Art. 6(1)(f) (legitimate interests). DAOs may request removal: hello@chainsights.one.

Opt-out. Any DAO may request removal from all ChainSights rankings by contacting hello@chainsights.one. Removal is processed within 24 hours.

• ChainSights GVS Methodology: https://chainsights.one/rankings/methodology

• ChainSights DGI Methodology: https://chainsights.one/methodology/dgi

• Snapshot.org API: https://docs.snapshot.org/graphql-api

• eth-labels dataset: https://github.com/dawsbot/eth-labels

• Introducing the DGI (blog): https://paragraph.com/@chainsights/introducing-the-dgi

• Delegate Vote Quality (blog): https://paragraph.com/@chainsights/delegate-vote-quality-75-percent

©️ 2026 ChainSights. Published under Creative Commons Attribution 4.0 (CC BY 4.0).
For questions or corrections: hello@chainsights.one

This paper describes the methodology behind ChainSights' Governance Vitality Score (GVS) and the DAO Governance Index (DGI). GVS is a composite score (0–10) measuring the governance health of a single DAO across four dimensions: participation quality, delegate engagement, power distribution, and community breadth. DGI aggregates GVS scores into an ecosystem-wide benchmark, analogous to the S&P 500 for DAO governance health. We cover all scoring components, their formulas, data sources, confidence thresholds, data quality measures, the Vote Quality Score (VQS) for individual delegates, and known limitations. As of March 2026, 47 DAOs are tracked and 4,279 individual voters analyzed.

Decentralized Autonomous Organizations (DAOs) present a novel governance challenge: decision-making power is formally distributed across token holders, but the on-chain record rarely reflects genuine decentralization. Voting is dominated by a small number of wallets, many proposals pass with single-digit participation, and "community consensus" frequently means three whales agreed.

Existing governance analytics focus on raw participation numbers — total voters, total proposals, total voting power — without distinguishing who is voting, how thoughtfully, or how concentrated power has become.

ChainSights takes a different approach: measure governance health through signals that correlate with genuine decentralization and meaningful participation, not vanity metrics.

GVS (Governance Vitality Score) is that measure for a single DAO.
DGI (DAO Governance Index) is the ecosystem benchmark derived from GVS scores.

All scoring is based on publicly available on-chain and off-chain governance data:

All scores are computed from raw data with no reliance on third-party governance ratings or curator judgment.

Update frequency: All GVS scores are recalculated daily via an automated pipeline. DGI benchmarks are recalculated immediately after individual DAO scores complete.

Governance data contains noise. Treasury wallets, exchange cold wallets, and bridge contracts technically "vote" on-chain but do not represent human community participation. Including them inflates voter counts and distorts participation rates.

ChainSights applies a classification filter before computing any participation-based metric:

• Treasury wallets — DAO treasuries holding protocol funds

• Exchange wallets — CEX hot/cold wallets (Coinbase, Binance, Kraken, etc.)

• Bridge contracts — Cross-chain bridge addresses (Polygon, Optimism, Arbitrum, Wormhole, etc.)

• Protocol contracts — Routers, factories, token contracts, and other smart contracts

• Individual wallets — All personal EOAs participate fully

• Fund/VC wallets — Venture capital and investment funds are legitimate governance participants

• Multisig wallets — Flagged for transparency, but retained in calculations

Wallet labels are sourced from the eth-labels dataset (115,000+ labeled Ethereum addresses) supplemented by a manually curated override file for major wallets not in the dataset. Labels are cached in a wallet_labels database table and refreshed periodically.

The filter is applied in: HPR voter counts, DEI delegate identification, PDI voter-power distributions, and GPI small-holder participation.

GVS is a composite score on a 0–10 scale representing the governance health of a single DAO. It is composed of four component scores, each measuring a distinct dimension of governance quality:

GVS formula:

GVS = (HPR × 0.35) + (DEI × 0.25) + (PDI × 0.20) + (GPI × 0.20)

Each component is normalized to the 0–10 scale before composition.

Weight: 35%

HPR measures what proportion of a DAO's voters are genuine human participants, as opposed to bots, treasury wallets, or coordinated sybil clusters.

Formula:

HPR = (Likely Human Voters / Total Unique Voters) × 10

Where Likely Human Voters = total unique voters minus those identified as bots, protocol contracts, exchange wallets, or treasury addresses (see Section 3).

Important distinction: HPR measures voter quality (are voters human?), not voter quantity (what % of followers voted). A DAO can have 10/10 HPR but only 0.5% absolute participation rate. Both metrics matter — participation rate data is provided separately in the Deep Dive Report.

Why 35% weight: Sybil resistance is the foundation of meaningful governance. A vote where half the "voters" are controlled addresses is not a legitimate vote, regardless of the outcome.

Weight: 25%

DEI measures how actively top delegates participate in governance proposals using a Weighted Participation Rate.

Delegate identification: Top 20% of voters by voting power (minimum 3 delegates, maximum 50)

Lookback window:

• Standard: last 90 days (up to 30 proposals)

• Adaptive: if fewer than 5 proposals in 90 days, the window extends to 180 days and recency weighting is disabled

Recency weights:

Formula:

DEI_raw = Σ (delegate_i_participationRate × recencyWeight) / Σ recencyWeight
DEI = DEI_raw × 10

Why it matters: Delegation is only effective when delegates are active stewards. A delegate holding large voting power who votes on 20% of proposals is less valuable than one who votes on 80%. The Weighted Participation Rate penalizes both inactivity and stale participation.

Weight: 20%

PDI measures voting power concentration and governance decentralization.

Sub-signals:

1. Gini Coefficient (40% of PDI) — Measures inequality in voting power distribution. Gini = 0 means perfectly equal, Gini = 1 means one address holds all power.

2. Nakamoto Coefficient (30% of PDI) — Minimum number of addresses needed to control >50% of total voting power. Higher = more decentralized.

3. Concentration Score (30% of PDI) — Percentage of total voting power held by the top 5 voters.

PDI formula:

PDI = (giniScore × 0.4) + (nakamotoScore × 0.3) + (concentrationScore × 0.3)

All sub-signals are normalized to the 0–10 scale (higher = more decentralized) before composition.

Why it matters: High power concentration means a small number of whales can override community consensus. True decentralization requires broadly distributed voting power.

Weight: 20%

GPI measures the participation rate of smaller token holders — the "bottom 80%" of the token distribution by holdings.

Formula:

GPI = (Active Small Holders / Total Small Holders) × 10

Where Small Holders = all token holders outside the top 20% by holdings.

Why it matters: Governance that only involves large token holders is not community governance. High GPI indicates the DAO's rank-and-file members are engaged, not just the wealthy.

Each GVS score is accompanied by a confidence level indicating data completeness and reliability:

Low confidence does not mean a DAO is performing poorly — it means there is less data to analyze. New DAOs naturally start with lower confidence until they build governance history.

Historical scoring: DAOs with no recent Snapshot activity retain their most recent GVS score, labeled as "historical." Historical scores are excluded from DGI benchmark calculations.

While GVS measures a DAO's overall governance health, the Vote Quality Score (VQS) measures individual delegate behavior. It answers: "How thoughtful is this voter's participation?"

Each delegate receives a VQS from 0 to 10, composed of up to four signals.

Current coverage: 24 of 47 tracked DAOs (~51%), covering 4,279 analyzed voters.

Measures how long a voter waits before casting their vote after a proposal opens. Voters who consistently vote within minutes may not be reading the full proposal. Higher deliberation time suggests more thoughtful engagement.

Normalization: Median deliberation time across all of the voter's votes, scaled to 0–10 based on ecosystem distribution.

Measures diversity of a voter's choices across proposals. Voters who always select the same option (e.g., always "For") demonstrate less independent judgment than voters who evaluate each proposal on its merits.

Normalization: Entropy of the voter's choice distribution across their votes, scaled to 0–10.

Note on ranked-choice DAOs: In DAOs using ranked-choice or weighted voting (Aavegotchi, Balancer, Frax, StakeDAO, CVX), single-choice votes do not exist. Independence cannot be meaningfully calculated and is shown as N/A. The VQS composite is recalculated using only available signals (see Section 6.6).

Measures whether a voter engages across diverse proposal categories (treasury, governance, technical) rather than only voting on one type. Broader categorical engagement indicates deeper governance involvement.

Normalization: Number of distinct proposal categories voted on, scaled to 0–10.

Measures how independently a voter decides compared to the largest token holders (whales). Voters whose choices closely mirror whale voting patterns may be following rather than forming independent opinions.

Normalization: Divergence from top-5-voter consensus, scaled to 0–10.

Note: Same N/A handling as Independence for ranked-choice DAO voters.

When all four signals are available, each contributes 25% to the VQS:

VQS = (Deliberation + Independence + Focus + Originality) / 4

When some signals are unavailable (N/A), the weight is redistributed equally among remaining signals:

VQS = Σ(available signals) / N_available

Example: For a ranked-choice DAO voter where Independence and Originality are both N/A:

VQS = (Deliberation + Focus) / 2

This ensures no signal inflates the score via a placeholder value.

Score distribution: Red (<2): 3.5% · Orange (2–4.9): 73.5% · Yellow (5–7.9): 18.5% · Green (≥8): 4.4%

Across 24 DAOs, three distinct voter quality patterns emerge:

Pluralistic (Median VQS ≥ 6)
Delegates vote independently across diverse proposal types. High deliberation, varied choices, strong originality.
Examples: CVX (7.8) · Aavegotchi (7.1) · Frax (6.7)

Mixed (Median VQS 4–6)
A broad middle ground — some engaged, deliberate voters alongside more routine participants.
Examples: Arbitrum (4.2) · Balancer (4.4)

Consensus (Median VQS < 4)
Delegates tend to vote in alignment with each other and with large token holders. Common in mature protocols with strong community convergence.
Examples: ENS (3.2) · Uniswap (3.6) · Lido (1.8)

These archetypes reflect governance culture, not quality rankings. A Consensus DAO is not worse than a Pluralistic one — the context and protocol type matter.

The DAO Governance Index (DGI) is an ecosystem-wide benchmark that aggregates GVS scores across all qualifying DAOs. Where GVS answers "how healthy is this DAO's governance?", DGI answers "how does this DAO compare to the ecosystem?"

Analogy: GVS = individual company score. DGI = the S&P 500.

The DGI uses an equal-weighted average of all qualifying DAOs' GVS scores:

DGI = (1/n) × Σ GVS_i    for all qualifying DAOs i = 1..n

Why equal-weighted?

• Philosophically consistent — no single DAO dominates the benchmark

• No dependency on treasury size or token market cap

• Avoids circular bias (large DAOs should not get more weight and score higher)

• Transparent and reproducible by anyone with the data

DGI is a family of 9 indices:

Composite: DGI Composite — Average GVS across all qualifying DAOs. The headline number.

Category indices: DGI DeFi · DGI Infrastructure · DGI Public Goods · DGI Social

Component indices: DGI-HPR · DGI-DEI · DGI-PDI · DGI-GPI

DAOs that fail these criteria are tracked but excluded from DGI calculations until they meet the threshold. A grace period applies before exclusion to avoid penalizing brief data gaps.

Snapshot-only coverage. VQS currently analyzes Snapshot data. DAOs using on-chain governance via Tally or Governor contracts are not covered for VQS.

Quantitative signals only. Proposal quality, forum discussion depth, constitutional design, and off-chain coordination are not captured.

High score ≠ perfect DAO; Low score ≠ bad DAO. A Consensus DAO may be making excellent decisions with high alignment.

Wallet classification is imperfect. eth-labels and manual overrides do not cover all protocol-controlled wallets.

Historical scores. DAOs with no recent Snapshot activity are scored based on their last available data.

This is a tool for improvement, not a verdict. Use GVS and DGI as a starting point for governance review, not as a substitute for deeper analysis.

No financial advice. GVS and DGI scores are governance health indicators only. Nothing in this paper or on chainsights.one constitutes financial, investment, or legal advice.

No endorsement. Inclusion in ChainSights rankings does not constitute endorsement of any DAO, protocol, or token by ChainSights or its affiliates.

GDPR compliance. ChainSights processes publicly available on-chain data in accordance with GDPR Art. 6(1)(f) (legitimate interests). DAOs may request removal: hello@chainsights.one.

Opt-out. Any DAO may request removal from all ChainSights rankings by contacting hello@chainsights.one. Removal is processed within 24 hours.

• ChainSights GVS Methodology: https://chainsights.one/rankings/methodology

• ChainSights DGI Methodology: https://chainsights.one/methodology/dgi

• Snapshot.org API: https://docs.snapshot.org/graphql-api

• eth-labels dataset: https://github.com/dawsbot/eth-labels

• Introducing the DGI (blog): https://paragraph.com/@chainsights/introducing-the-dgi

• Delegate Vote Quality (blog): https://paragraph.com/@chainsights/delegate-vote-quality-75-percent

©️ 2026 ChainSights. Published under Creative Commons Attribution 4.0 (CC BY 4.0).
For questions or corrections: hello@chainsights.one