Gov-Acc Pipeline.

1 Evaluating DAO Sustainability and Longevity Through On-Chain Governance Metrics Silvio Meneguzzo, Claudio Schifanella, Valentina Gatteschi, Giuseppe Destefanis Abstract—Decentralised Autonomous Organisations (DAOs) In this paper, we examine how DAOs compare to tra- automate governance and resource allocation through smart ditional organisational models and analyse the blockchain- contracts, aiming to shift decision-making to distributed token based mechanisms that influence participation and governance. holders.However,manyDAOsfacesustainabilitychallengeslinked By evaluating real-world DAOs through Key Performance to limited user participation, concentrated voting power, and technical design constraints. This paper addresses these issues by Indicators (KPIs), we identify critical governance challenges identifying research gaps in DAO evaluation and introducing a related to user engagement and decision-making. framework of Key Performance Indicators (KPIs) that capture To address these issues, we define the following research governance efficiency, financial robustness, decentralisation, and questions: community engagement. We apply the framework to a custom- RQ1: Which challenges most significantly affect DAO sustain- built dataset of real-world DAOs constructed from on-chain data and analysed using non-parametric methods. The results reveal ability and longevity, particularly regarding user participation? recurring governance patterns, including low participation rates Rationale: DAOs face multiple obstacles, but inadequate andhighproposerconcentration,whichmayunderminelong-term engagement is consistently identified as a primary factor viability.TheproposedKPIsofferareplicable,data-drivenmethod influencing decentralisation and effectiveness. forassessingDAOgovernancestructuresandidentifyingpotential RQ2: Which Key Performance Indicators (KPIs) can be used areasforimprovement.Thesefindingssupportamultidimensional approachtoevaluatingdecentralisedsystemsandprovidepractical to evaluate DAO sustainability, including financial stability, tools for researchers and practitioners working to improve the governanceprocesses,decentralisation,andcommunityengage- resilience and effectiveness of DAO-based governance models. ment? Index Terms—Decentralized Autonomous Organizations, DAO, Rationale: Establishing measurable indicators allows for com- Blockchain, Voting Mechanisms, Decentralization, Governance, parability and helps identify organisational and technical Sustainability, Longevity, Key Performance Indicators, User weaknesses in DAOs. Participation. RQ3: How does applying these KPIs to real-world DAOs This work has been submitted to the IEEE Transactions on reveal governance issues and inform strategies for improving Computational Social Systems for possible publication. sustainability and longevity? Rationale: Beyond theoretical measures, the study aims to I. INTRODUCTION demonstrate how KPI-based analysis provides structured rec- DECENTRALISED Autonomous Organisations (DAOs) ommendations for improving DAO governance. introduce a governance model that replaces centralised We make three principal contributions. First (I), we identify decision-making with blockchain-based smart contracts and persistent open challenges in DAO governance, focusing voting mechanisms [1]. DAOs enable collective decision- specifically on the lack of consistent user engagement as a making by users without the need for central authorities or limiting factor for sustainability and effective decentralization. intermediaries [2]. This structure is based on decentralisation, Second (II), we develop a set of empirically grounded KPIs transparency, and automated decision-making, making DAOs spanning social, economic, and procedural dimensions of DAO applicable to various collaborative systems. Despite these governance.TheseKPIsofferareplicableandstructuredmeans characteristics, unresolved challenges remain [3], particularly ofassessingDAOsacrossmultipledimensions,includingvoting low user participation in governance [4]. When participants do activity, treasury management, automation, and distribution not vote or engage in the decision-making process, a DAO’s of power. Third (III), we apply these KPIs to a curated ability to function effectively and maintain its decentralised dataset based on on-chain data, comprising 50 active DAOs structure is weakened [5]. Addressing participation issues is and demonstrate how the ensuing analysis reveals critical necessary to ensure that DAOs operate as intended and support governance asymmetries, especially low voter turnout and long-term sustainability. concentration of proposer authority, and highlights paths for structural improvements. By establishing this integrated Manuscript created March, 2025; This work was developed by Silvio Meneguzzo (Corresponding author) is with the Department of Computer KPI framework, we aim to support both researchers and Science,UniversityofTurin,Italy(e-mail:silvio.meneguzzo@unito.it). practitioners in diagnosing governance shortfalls and designing ClaudioSchifanellaiswiththeDepartmentofInformatics,Universityof more resilient, community-driven DAO ecosystems. Turin,Italy(e-mail:claudio.schifanella@unito.it). ValentinaGatteschiiswiththeDepartmentofAutomationandComputer ByapplyingtheproposedKPIframeworktoreal-worldcases, Science,PolitecnicodiTorino,Italy(e-mail:valentina.gatteschi@polito.it). we show how inclusive governance structures, more equitable Giuseppe Destefanis is with the Department of Computer resource allocation, and user-friendly voting mechanisms can Science, Brunel University of London, United Kingdom (e-mail: giuseppe.destefanis@brunel.ac.uk). significantly boost engagement and decentralisation. In this 5202 rpA 51 ]YC.sc[ 1v14311.4052:viXra 2 sense, our findings suggest that structured, data-driven indica- smart contract code [6]. Governance-related difficulties also tors enable DAOs to develop more participatory and automated remain, particularly the concentration of voting power and governancemodels,ultimatelystrengtheningtheirsustainability. low participation rates [4] and related security issues in the Therestofthispaperisstructuredasfollows:SectionIIreviews governance process [17]. Early analyses stressed that DAOs related work on DAOs, governance mechanisms, and ongoing must balance on-chain rules with off-chain social coordination challenges. Section III describes the research methodology, [18],butthisapproachisincontradictionwithafulltransparent including KPIs development and the creation of a Multi-chain and decentralized desired behaviour. on-chain retrieval pipeline. Section IV presents the results based on the analysis of existing DAOs. Section V discusses C. DAO Governance Mechanisms theimplicationsofthesefindings,followedbythreatstovalidity Governance is a central aspect of DAOs, with voting in Section VI. Finally, Section VII provides conclusions and mechanisms forming a key part of their structure alongside directions for future research. blockchainandsmartcontracts.Fanetal.[19]examinedseveral We provide a replication package including all the nalysis votingmechanismsusedinDAOs,includingApprovedRelative scripts and results at this link1 to support reproducibility and Majority, Token-Based Quorum, Quadratic Voting, Liquid verification. Democracy, Weighted Voting, Rage Quitting, and Holographic Consensus. Ding et al. [20] also described Conviction Voting, II. BACKGROUND whichadjustsvoteweightbasedonpreferenceandtime.Dimitri A. DAOs: Definition and Evolution [21] highlights how different voting schemes, such as approval votingorrank-basedvoting,impactoutcomelegitimacyinDAO Decentralised Autonomous Organisations represent a gov- governance. Beck et al. [22] propose a blockchain governance ernance model that uses blockchain technology and smart model highlighting the tension between decentralization and contracts to automate decision-making processes [2]. Unlike the need for effective coordination. traditional organisations, which rely on centralised authorities, Despite the range of governance models, Feichtinger et al. DAOs distribute decision-making power among members [4] found that voting power remains concentrated in most through token-based voting systems [6]. DAOs. In their study of 21 governance systems, 17 were The concept of DAOs emerged with Buterin’s [1], [7] intro- controlled by fewer than 10 participants. Common measures duction of Ethereum, the first blockchain platform supporting of voting power distribution include the Gini Coefficient and smart contracts capable of encoding organisational rules. In NakamotoCoefficient[23],whichassessinequalityandcontrol 2016, “The DAO” was launched as the first large-scale concentration,respectively.Whilethesemechanismsrepresenta implementation,raisingover$150millionbeforeavulnerability diverse range of voting protocols, prior empirical studies often in its smart contract led to a major security breach [8]. This overlook the interplay between on-chain user participation, event demonstrated how flaws in smart contract code could treasury management, and the degree of proposal automation. result in governance failures [9]. Interest in DAOs was later As a result, comprehensive frameworks for capturing overall revived when MakerDAO introduced an on-chain governance DAOsustainabilityremainunderdeveloped.Thisgapunderpins systemin2018,followedbytheadoptionofsimilarapproaches the need for an integrated KPI-based approach, as we discuss across various DeFi (Decentralized Finance) protocols [10]. in Section III. B. DAO Characteristics and Challenges D. Evaluation Approaches for DAOs DAOs are built on three core principles: decentralisation Existing research on DAO evaluation often focuses on (distributed network-based management), automation (code- individual aspects rather than complete frameworks. Park et al. based governance), and organisation (transparent operating [24]introduceddimensionsforassessingdecentralisation,while rules via smart contracts) [11]. Additional attributes include Faqir-Rhazoui et al. [25] conducted a comparative study of transparency, immutability, resistance to manipulation, interop- DAO platforms on Ethereum. Although these studies provide erability, incentives for participation, and operational efficiency useful insights, they do not offer an integrated method for [2], [12]. assessing sustainability. Wang et al. [6] and Qin et al. [26] Despite these characteristics, DAOs encounter significant proposed architectural models to examine DAO structures, challenges. Security vulnerabilities in smart contracts can have covering technological, execution, coordination, organisational, severe consequences due to blockchain immutability, as seen and application layers. These models provide a conceptual in “The DAO” hack [6]. Privacy concerns arise from the full understanding of governance structures but do not translate transparency of all transactions [9]. Legal uncertainties persist into practical evaluation metrics. in many jurisdictions, raising questions about liability [12], despite the fact that some jurisdictions have begun formalizing E. Research Gap DAO-friendly statutes, including Vermont [13], Wyoming Despite the growing body of literature on DAOs, there [14], the Marshall Islands [15], and Utah [16]. Technical remains no widely accepted framework for comprehensively constraints include the gap between legal frameworks and assessing their sustainability and longevity. Many studies rely on off-chain or aggregator data that fail to capture on- 1The replication package is hosted on Figshare: https://figshare.com/s/ 2cf646c67f23ea917ac1 chain details and some of the existing works tend to focus 3 narrowly on specific issues such as security vulnerabilities [8], iousaspectsofDAOs,allowingreliablequantitativeassessment governance limitations [4], or technical implementations [6], of their structures and activities. while neglecting broader social and economic dimensions. For We interpreted our quantitative findings within the context example, Park et al. [24] propose decentralisation indicators of existing research on DAO governance models, established focused on organisational structure, and Faqir-Rhazoui et theoretical frameworks, and current organisational patterns al. [25] compare DAO platforms (Aragon, DAOstack, and [29], [30]. This combination ensured our KPIs captured DAOhaus) primarily in terms of feature sets. Although such both measurable indicators and broader aspects of DAO studies contribute valuable insights into decentralisation ratios, sustainability. treasury sizes, or code security, they rarely address procedural and community-oriented factors such as user participation, A. Key Performance Indicators voting fairness, and long-term engagement within a unified To assess the sustainability and longevity of DAOs, we evaluative model. defined four Key Performance Indicators covering core aspects Furthermore, the lack of a complete dataset incorporating suchascommunityparticipation,financialcapacity,governance detailed on-chain governance events further hampers progress. processes, and decentralisation. These KPIs were shaped by Many existing datasets offer incomplete views of token existing literature [4], [19], [24], [25], [31] and insights from distributionandvotingactivities.Arecentempiricalstudybased our curated dataset. Together, they offer a structured way to onoff-chaindatafromtheSnapshotplatform[27]analysed581 evaluate organisational resilience over time. DAO projects over more than three years, providing valuable 1) KPI 1: Network Participation: This KPI reflects the insights into DAO performance at scale. However, while such extent of member engagement within the DAO. Participation approaches capture broad activity metrics, they do not include in voting and proposal creation is essential to decentralised the fine-grained, verifiable on-chain details, such as precise governance and helps ensure representative decision-making, token transfers and smart contract event logs, needed for an adaptability and helps mitigate oligarchic tendencies [4], [31]. integrated assessment of governance performance. In response, We define the Participation Rate as: our work introduces a dataset, constructed directly from raw blockchain data, which enables a more accurate and integrated (cid:18) Active Members (cid:19) evaluation of DAO performance. Participation Rate= (1) Total Members Many DAOs suffer from persistently low voting turnout, An Active Member is any address that has cast a vote or centralised token ownership, or ambiguous legal standing [9], created a proposal. Total Members are token holders with [12], but a thorough, empirical framework that captures these on-chain voting rights. Drawing on literature reporting low varied challenges is still lacking. In particular, while indicators participation levels, we adopt the following classification: like the Gini Coefficient [23] or token-based voting measures [19] help quantify power concentration, they do not account • Low: <10% for social incentives, governance processes, or community • Medium: 10–40% resilience; factors that are equally critical to a DAO’s viability • High: >40% over time. These categories reflect empirical observations of voter turnout Toaddressthesegaps,thispaperintroducesaKPIframeworkto in prominent DAOs [4] and distinguish between minimal, evaluateDAOsacrossfourdimensions—participation,financial moderate, and widespread participation. Prior analyses indicate stability, voting efficiency, and decentralisation—linking social, that below 10% turnout, governance tends to be dominated by economic, and procedural factors to identify governance issues a handful of token holders, undermining true decentralization and support long-term sustainability. of partecipation. 2) KPI 2: Accumulated Funds: This KPI captures a DAO’s financial capacity. We consider two aspects: III. METHODOLOGY Treasury Size denotes the total assets held in smart contracts We used a data-based approach to develop and apply Key under DAO control. Larger treasuries allow for sustained Performance Indicators for evaluating the sustainability and contributor rewards, protocol development, and other activities. longevity of DAOs. Our methodology combined quantitative Circulating Token Percentage measures the proportion of data analysis with qualitative reasoning to ensure that our governance tokens in circulation: KPIs were both empirically grounded and conceptually sound. Existing studies often focus on isolated aspects of DAO (cid:18) (cid:19) Circulating Supply performance (e.g., decentralisation ratios, treasury sizes, or Circulating Token Percentage= (2) Total Supply codesecurity)andrelyonincompleteaggregator-baseddatasets, which limits analyses of voting power distribution and user Tokens held by the treasury or locked in vesting contracts are engagement [4], [6], [8], [11], [28]. excluded from circulation. High circulation suggests broader To address these limitations, we extracted raw data from multi- economic participation, while low circulation may indicate ple sources, including Smart Contracts’ ABI from blockchain concentration. We combine both aspects to categorise financial explorers (e.g., Etherscan) and direct queries to blockchain status: networks via providers like Infura and Alchemy. This process • Low: Treasury <$100M provideddetailednumericalandcategoricalinformationonvar- • Medium-Low: $100M–$1B, circulation ≤ 50% 4 TABLE I: KPIs and Their Level Divisions with Assigned • Medium-High: $100M–$1B, circulation > 50% Scores • High: Treasury >$1B Treasury size thresholds at $100 million and $1 billion KPI Level Description Score Low Participationrate<10% 1 reflect typical boundaries observed in leading DeFi DAOs NetworkParticipation Medium Participationrate11%–40% 2 High Participationrate>40% 3 like MakerDAO, Compound, and Uniswap, where financial Low Treasury<$100millionUSD 0.75 resources can significantly influence governance dynamics and AccumulatedFunds Medium-Low T 50 re % asury$100million–$1billionUSD,circulatingtokens≤ 1.5 Medium-High T 50 re % asury$100million–$1billionUSD,circulatingtokens> 2.25 system resilience. High Treasury>$1billionUSD 3 Low Approvalrate<30%and/orvotingduration<3days 1 3) KPI 3: Voting Mechanism Efficiency: This KPI considers VotingMechanismEfficiency Medium Approvalrate30%–70%,votingduration3–14days 2 High Approvalrate>70%,votingduration3–14days 3 governance effectiveness based on proposal approval rate and Low Largestholder≥66%ofresources 0.6 Medium-Low Largestholder33%–66%ofresources 1.2 voting duration: Decentralisation Medium L au a t r o g m es a t te h d o d ld e e c r isi 1 o 0 n % s –33%,mediumparticipation,notfully 1.8 (cid:18) Approved Proposals (cid:19) Medi H u i m gh -High L a L u a a t r r o g g m e e s s a t t te h h d o o l l d d d e e e c r r is < 1 io 0 n 1 % s 0 – % 33 o % f , re m so e u d r i c u e m s /highparticipation,fully 2 3 .4 Approval Rate= (3) Total Proposals weights to avoid introducing arbitrary prioritisation, though (cid:80)n Voting Duration future work may adjust this based on further analysis. Average Voting Duration= i=1 i (4) By combining these four KPIs, each based on our harmonised n on-chain dataset, we developed a multidimensional view of Short durations may indicate rushed decision-making, while DAO sustainability and enabled comparative analysis across longer voting windows can hinder timely execution. Based on different governance structures. empirical observations [19], we define: • Low: Approval <30% and/or duration <3 days B. Dataset Construction • Medium: Approval 30–70%, duration 3–14 days • High: Approval >70%, duration 3–14 days Our initial data collection included 5,999 DAOs from aggregator sources, most notably, the DAO Analyzer dataset These levels help distinguish between inefficient governance, from Kaggle. However, after applying strict criteria for on- functional deliberation, and overly streamlined approvals. chain voting and recent governance activity, most entries 4) KPI 4: Decentralisation: This KPI addresses the concen- failed to meet our thresholds for meaningful participation2.In trationofresourcesandthedegreeofautonomyinoperations.It particular, only a negligible number of DAOs (2 DAOs) from combines token distribution, member activity, and automation. the Kaggle dataset exhibited the required level of governance Inspired by work on DAO structures [24], [32], we assess activity; hence, we opted to exclude this dataset from our final decentralisation based on the largest token holder’s share, analysis. To address this, we developed the custom pipeline whether there is sufficient participation, and whether decisions described in Section III-C, creating a harmonized on-chain are executed automatically. We label a DAO’s decisions as dataset comprising 50 DAOs that demonstrated: “fully automated” when successful on-chain proposals directly triggercontractexecutionwithoutrequiringoff-chainsignatures • Robust on-chain governance mechanisms or multi-sig confirmations. • Transparent participation profiles • Low: Largest holder >66% • Key activity metrics (e.g., average voting duration, pro- posal frequency, treasury size) • Medium-Low: 33–66% • Medium: 10–33%, with at least medium participation, no We classified DAOs into four activity categories: automation 1) Highly Active: DAOs with at least 5 governance-related • Medium-High: 10–33%, with medium/high participation, transactions in the last 30 days, showing consistent full automation involvement from multiple members. • High: <10% 2) Moderately Active: DAOs with at least 1 transac- Our on-chain dataset allows precise measurement of token tion/proposal in the last 90 days, maintaining regular distributions and automation status. These categories differen- community activity. tiate between DAOs with strong individual control and those 3) Minimally Active: DAOs with transactions/proposals exhibiting distributed ownership and autonomous operations. older than 90 days and low recent activity. 5) Scoring System: Each KPI level is mapped to a numeric 4) Potential Test or Dormant: DAOs with fewer than 2 score from 0 to 3, with all four KPIs equally weighted. transactions since creation, possibly representing experi- The total score ranges from 0 to 12. This uniform scheme mental deployments or abandoned projects. avoids arbitrary prioritisation and reflects our assumption These thresholds follow conventions in blockchain governance that DAO sustainability results from balanced performance literature, which often uses monthly and quarterly assessments acrosscommunity,financial,procedural,andstructuraldomains. to track participation trends [25]. Table I summarises the scoring. Before calculating KPIs, we performed several validation This scoring method reflected our assumption that DAO checks: sustainability depends on a balanced combination of social • Removal of duplicates (participation), financial (funds), procedural (voting efficiency), and organisational (decentralisation) factors. We used equal 2https://github.com/smeneguz/data-analyzer-dao-ecosystem.git 5 Fig. 1: Multi-chain on-chain retrieval pipeline • Consistency checks across sources out repeated or malformed entries. The fourth panel of • Standardisation of timestamps Figure 1 illustrates how we transformed raw blockchain data • Verification of event logs by comparing proposal creation into structured event files (events decoded, eventName.json, and execution events eventName2.json), making them suitable for analysis. Any inconsistencies prompted targeted queries to blockchain Cross-linking Governance and Token Data (Step 5 in Fig- nodes or re-examination of aggregator data, helping ensure a ure1):Toassesshowvotingpowercorrelatedwithparticipation, reliable dataset. we supplemented event data with historical token transfers retrieved via RPC nodes, enabling us to track token holder distribution over time and identify concentrated voting power. C. Data Collection Pipeline As shown in the fifth panel of Figure 1, we processed token information into standardized JSON formats that captured Understanding DAO dynamics required thorough data ex- ownership patterns and voting activity. traction and analysis from multiple sources. Our multi-chain DAO Data Extraction and Analysis (Step 6 in Figure 1): on-chain retrieval pipeline, illustrated in Figure 1, consisted of We integrated governance events with token distribution data the following steps: to create a consistent view of each DAO’s activity. The sixth DAO and Token Address Enumeration (Step 1 in Figure 1): panelofFigure1showshowwecombinedthesedatasetsintoa We identified candidate DAOs on Ethereum and other EVM- unified format that enabled consistent analysis across different compatible networks (e.g., Polygon, Arbitrum, BNB Chain) by DAOs and governance models. scanning known governance contracts and verifying that they Harmonised Dataset Output and Visualization (Step 7 supported on-chain voting transactions. in Figure 1): Finally, we generated standardized files for As shown in the leftmost panels of Figure 1, we processed each DAO, including key fields such as total proposals, governancecontractsandtokencontractsseparately,organizing voter addresses, top holders, and execution outcomes. This their addresses and interfaces into config files for subsequent structuredoutput,representedintherightmostpanelofFigure1, processing. formed the basis for KPI calculation, statistical analysis, and Setup for Multi-chain Access (Step 2 in Figure 1): We con- visualization through customised dashboards. figured connections to multiple blockchain networks through node providers and explorers. As shown in the second panel of Figure 1, this involved setting up environment configurations (.ENV files) to access Ethereum, Polygon, Arbitrum, BNB IV. RESULTS Chain, and Optimism networks, ensuring broad coverage of DAO activity. Our sample consists of 50 DAOs spread across Ethereum, Smart Contract Event Retrieval (Step 3 in Figure 1): For Polygon, and Arbitrum. Together, they account for 6930 eachDAO’sgovernancecontract(s)andassociatedtoken(s),we proposals, 317317 unique voting addresses and 4524205 total queried block explorers and node providers to fetch all event members, providing a robust basis for evaluating governance logs from the deployment block until April 2025. In parallel, patterns. We applied the Shapiro–Wilk test for normality and we retrieved transactions from the associated governance token Levene’s test for variance homogeneity to assess the suitability smart contracts to capture token transfers and other economic of parametric methods. Based on the results of these tests, we events. As shown in the third panel of Figure 1, this process used one-way ANOVA for normally distributed groups with created separate datasets for smart contract events and token homogeneous variances, and the Kruskal–Wallis test for cases transactions, which were stored in dedicated databases for later where these assumptions were violated. To complement the analysis. statistical analysis, we generated box plots, violin plots, scatter EventDecodingandDataNormalisation(Step4inFigure1): plotswithregressionlinesandradarcharts.Thesevisualisations Wedecodedeachrawlogusingthecontract’sABItomapevent illustratetherelationshipsamongtheKPImetricsandfacilitate signatures into human-readable records. We then converted the empirical evaluation of the proposed framework using our timestamps to UTC, normalized token amounts, and filtered newly compiled dataset of 50 DAOs.

6 A. Statistical Approach 4) InterpretationofTestOutcomes: Toguidetheapplication of statistical tests, we established a decision process based on We structured the data analysis as a sequence of standard the outcomes of normality and variance homogeneity checks. tests to verify statistical assumptions and ensure that group Table II summarises the interpretation criteria for the Shapiro– comparisons were conducted appropriately. Wilk, Levene’s, ANOVA, and Kruskal–Wallis tests, along with 1) Shapiro–Wilk Test for Normality: For each KPI category the corresponding implications for group comparison methods. (e.g. Low, Medium, High), we first evaluated whether the data followed a normal distribution. The Shapiro–Wilk test [33] TABLE II: Interpretation of statistical test outcomes and was used, with the test statistic W computed as: corresponding analysis decisions. (cid:16) (cid:17)2 Condition Interpretation (cid:80)n i=1 a i x (i) Shapiro–Wilkp-value<0.05 Datadeviatefromnormality;ANOVAnotused. W = , Levene’sTestp-value<0.05 Variancesdiffer;ANOVAnotused. (cid:80)n (x −x¯)2 ANOVAp-value<0.05 Atleastonegroupmeandifferssignificantly. i=1 i Kruskal–Wallisp-value<0.05 Atleastonegroupdistributiondiffers;post-hoctestsapplied. where x denotes the i-th order statistic (i.e., the sample (i) This procedure ensures that each KPI category is analysed sorted in ascending order), x¯ is the sample mean, and a i using appropriate statistical methods, reducing the risk of are constants derived from the covariance matrix of a normal misinterpretation due to violations of parametric assumptions. distribution.Ap-valuebelowacommonlyusedthreshold(0.05) The following sections present the results for each KPI, led to rejection of the null hypothesis of normality. Given the including the applied thresholds and visualisations. Shapiro–Wilk test’s sensitivity to small sample sizes, we report its results only for categories with n≥3. 2) Levene’s Test for Variance Homogeneity: We used Lev- B. Network Participation ene’s test [34] to assess whether variances were homogeneous Definition. As introduced in Section III, Network Participation acrossgroups.Forkgroups,letX denotethej-thobservation ij measures the proportion of active members, those who cast in the i-th group, with i = 1,...,k. The test statistic is at least one on-chain vote or submitted a proposal, relative to computed as: total membership. We classified DAOs into three categories: Low (<10%), Medium (10–40%), and High (>40%). W = (N −k) (cid:80)k i=1 n i (cid:0) Z i· −Z ·· (cid:1)2 , Findings. Figure 2 presents the relationship between total Levene (k−1) (cid:80)k (cid:80)ni (Z −Z )2 membership (log-scaled x-axis) and participation rate (y-axis). i=1 j=1 ij i· The visualisation highlights an inverse pattern: smaller DAOs where Z =|X −X¯ | (or, alternatively, the median may tend to show higher participation, with thresholds at 10% ij ij i be used in place of the mean), Z is the mean of group i and 40% marking the category boundaries. Summary statistics i· after transformation, and Z is the overall mean of all Z . N indicate a low median participation rate (4.16%) and high ·· ij denotes the total number of observations, and n is the size variability, with a few outlier values capped at 100% i of group i. A significant p-value (typically <0.05) indicates Figure3showsnotchedboxplotsacrossthethreeparticipation heterogeneity of variances across groups. categories. The notches represent 95% confidence intervals for the medians. The Shapiro–Wilk test returned p-values 3) Parametric vs. Non-parametric Group Comparisons: below 0.05 for the Low and High categories, and Levene’s (1) One-Way ANOVA: If all groups satisfied the normality test indicated unequal variances. Based on these results, we assumption (Shapiro–Wilk) and exhibited homogeneous vari- applied the Kruskal–Wallis test, which identified significant ances (Levene’s), we applied one-way Analysis of Variance group differences (H = 30.45, p < 0.01). Post-hoc analysis (ANOVA) to test for differences in group means. The ANOVA using Dunn’s test with Bonferroni correction confirmed that F-statistic was evaluated against the null hypothesis that all the High group had a significantly higher median participation group means are equal [35]. rate(median=98.29%)thantheLowgroup(median=2.47%), (2) Kruskal–Wallis Test: When either normality or homo- confirming a substantial gap in user engagement across the geneity of variances was not satisfied, we used the Kruskal– DAO ecosystem. Wallis test [36]: 12 (cid:88) k C. Accumulated Funds H = n R2 − 3(N +1), N(N +1) i i Definition. Accumulated Funds reflect the DAO’s financial i=1 capacity, incorporating both treasury size and the proportion where N is the total sample size across k groups, n is the of circulating tokens. As defined in Section III, DAOs were i samplesizeofgroupi,andR istheaveragerankwithingroup classified into four categories: Low, Medium-Low, Medium- i i. The null hypothesis states that all groups follow the same High, and High, based on combined thresholds for these two distribution. A significant p-value indicates that at least one dimensions. group differs. Post-hoc pairwise comparisons were conducted Findings. Figure 4 visualises the distribution of DAOs by usingDunn’stestwithBonferronicorrectiontoidentifyspecific plotting treasury value (in log scale to handle wide differences group differences. from$1milliontobillions)againstcirculatingtokenpercentage. 7 Fig. 4: Scatter plot of treasury value (log scale) vs. circulating Fig. 2: Scatter plot showing the relationship between total token percentage. members (x) and participation rate (y) Fig. 5: Notched box plot of treasury sizes, grouped by Fig. 3: ”Notched“ box plot for Network Participation. circulating token status. Threshold lines at $100million, $1billion, and 50% token test indicated non-normality in the Low (p = 0.0373) group, circulation delineate category boundaries. whiletheMedium(p=0.4003)andHigh(p=0.0804)groups Figure 5 shows notched box plots of treasury sizes grouped by satisfied the normality assumption. Levene’s test confirmed circulating token status. DAOs in the Low category (treasury variance heterogeneity (p=0.0043). Accordingly, we applied < $100 million) exhibited high variance, possibly reflecting the Kruskal–Wallis test, which returned H =9.4687 and p= early-stageoperationsorinconsistentfundingcycles.Normality 0.0088. Although the High group showed the highest median tests (Shapiro–Wilk) returned p-values below 0.05 for all approval rate (88.24%), the differences were not statistically four categories, and Levene’s test indicated heterogeneity of significant at the 5% level. variances. These results support the use of non-parametric Visualisation Support. Figure 7 displays approval rate against methods for group comparisons. average voting duration, offering a joint view of the two Visualisation Support. We observed a weak and statistically dimensions that define this KPI. While no direct statistical non-significantcorrelationbetweentreasurysizeandcirculating inference is drawn, the pattern suggests that extremely short token percentage (Pearson’s r ≈−0.13, p≈0.37). or excessively long voting windows may undermine effective governance. These empirical observations are consistent with D. Voting Mechanism Efficiency prior findings [19] recommending moderate voting periods in DAO settings. Definition. Voting Mechanism Efficiency captures the balance between approval rates and average voting duration, grouping DAOs into Low, Medium, or High efficiency categories. This E. Decentralisation KPI reflects the trade-off between swift decision-making and Definition. Decentralisation encompasses economic distribu- thorough deliberation. tion, participatory engagement, and the degree of on-chain Findings. Figure 6 presents notched box plots of approval automation [24]. We classified DAOs into five categories: Low, rates across the three efficiency categories. The Shapiro–Wilk Medium-Low, Medium, Medium-High, and High, based on the 8 Fig. 6: Notched box plot of approval rates among the three Fig. 8: Notched box plot of proposer concentration across five efficiency categories decentralisation categories Fig. 7: Scatter plot of approval rate versus average voting Fig. 9: Scatter plot of largest holder percentage versus duration participation rate 33%, and 66% (economic decentralisation) and at 10% and largest holder’s token share and the presence of automated 40%(participation)providereferenceboundaries.Whileaslight governance processes. rank-based trend is observable, the visual evidence does not Findings.Figure8showsnotchedboxplotsofproposerconcen- indicate a strong predictive relationship between largest-holder tration across the five decentralisation categories. The Shapiro– share and participation rate. Wilk test indicated that the Medium group (p = 0.0018) deviated from normality, while other groups did not violate normality assumptions. Levene’s test showed no significant F. Composite Metrics and Overall Patterns variance heterogeneity (p = 0.2338). We therefore applied To summarise DAO performance across dimensions, we the Kruskal–Wallis test, which returned H = 15.10 and constructed composite scores based on the four KPIs: Net- p = 0.0045, indicating statistically significant differences in work Participation, Accumulated Funds, Voting Mechanism proposer concentration among the decentralisation categories. Efficiency, and Decentralisation. The Low decentralisation group (largest holder > 66%) Figure10presentsradarplotsforasubsetofDAOs,illustrating exhibited the highest mean proposer concentration (45.41), the trade-offs and balance across the four governance dimen- while the Medium group (10–33% ownership, no automation) sions. DAOs with more balanced profiles, showing consistent had the lowest mean (9.51). These results suggest that high scoresacrossallKPIs,tendtoachievehighercompositescores, tokenconcentrationdoesnotnecessarilylimitproposalactivity, suggesting more resilient governance structures. In contrast, whereas intermediate ownership levels may correspond to DAOs that perform strongly in one or two KPIs but poorly reduced proposer diversity in this dataset. in others often display structural imbalances that may affect Visualisation Support. Figure 9 maps the largest holder’s long-term sustainability. percentage against participation rate. The Pearson correlation For example, Uniswap and Lido DAO demonstrate high finan- coefficient was weak and non-significant (r = 0.09, p = cial capacity but score lower in decentralisation. By contrast, 0.5449),indicatingnolinearrelationship.However,Spearman’s DAOs such as Public Nouns, Lil Nouns, and Union achieve rankcorrelationshowedamoderatemonotonictrend(ρ=0.27, high composite scores through more distributed governance p = 0.0644), suggesting a mild association between lower structures and consistent performance across participation and concentration and higher engagement. Threshold lines at 10%, voting dimensions. DAOs with lower composite scores, such 9 moderate to large treasuries were combined with automated on-chain governance mechanisms. These patterns suggest that decentralisationisshapednotonlybytokendistributionbutalso by social and procedural dynamics such as proposer diversity, governance automation, and participation incentives. Notably, variations in the largest-holder percentage did not consistently suppress network participation. This finding com- plicatesassumptionsaboutcentralisationeffects,suggestingthat concentrated capital, when coupled with effective delegation or automation, may coexist with active governance. C. Voting Mechanism Efficiency While DAOs classified as High in voting efficiency generally exhibited shorter but adequate voting durations and higher approval rates, not all observed differences across groups were statistically significant (Section IV-D). This outcome points to theinfluenceofcontextualfactors,suchasproposalcomplexity Fig. 10: radar composite plot - 10 daos or urgency, on governance behaviour. Shorter decision cycles do not inherently imply more effective outcomes. Scatter plots of approval rate versus average voting duration as HAI, Open Dollar, and Unlock, often reflect a combination suggested that overly short windows may hinder deliberation, of centralised ownership and limited user engagement. while excessively long ones can reduce engagement. A more These aggregated comparisons reinforce the core premise dynamic approach, calibrating voting duration based on prior of the study: assessing DAO sustainability requires a multi- proposal complexity or participation history, could improve dimensional perspective. Strong performance in one area does legitimacy without sacrificing efficiency. These observations not necessarily compensate for weaknesses in others. A com- suggest that future refinements to the boundaries for KPI 3 bined evaluation of participation, financial health, procedural may strengthen its interpretive value. efficiency, and decentralisation provides a more complete understanding of organisational robustness. D. Implications for DAO Governance V. DISCUSSION The KPI framework enables a structured diagnosis of DAO A. Interpretation of KPI Findings governance strengths and weaknesses. DAOs with limited The four KPIs, Network Participation, Accumulated Funds, participation but substantial financial reserves may benefit Voting Mechanism Efficiency, and Decentralisation, yielded fromchangestovotingaccessibilityorcommunityengagement statistically significant group differences in most cases, based strategies. Conversely, DAOs with strong participation but primarily on non-parametric testing. DAOs classified as High weaker financial capacity may need to diversify treasury in Network Participation and Accumulated Funds exhibited structures or enhance economic sustainability mechanisms. more consistent engagement and greater financial capacity, Governance reforms targeting concentration risk, such as token respectively.ThosewithMedium-HightoHighdecentralisation lockups,quadraticvoting,orpartialdelegation,mayhelpreduce levels showed broader proposer distributions and lower voting disproportionate influence without discouraging large token concentration. In contrast, correlations such as that between holders from participating. Effective design in this area can largest-holder percentage and participation were weaker or facilitate broad-based input while preserving capital efficiency marginal, indicating that concentrated token ownership does [37]. not necessarily reduce member activity. Morebroadly,aligninggovernancepracticeswithDAO-specific DAOs scoring highly across multiple KPIs demonstrated operational profiles (e.g. adjusting voting durations based on structuralfeaturesassociatedwithmoresustainablegovernance. proposal type) may help ensure decision quality and continuity These findings support the broader view that sustained commu- overtime.Ourfindingsconfirmthatpersistentlowparticipation nity involvement, procedural efficiency, and equitable resource remains a core vulnerability. This aligns with prior studies distribution jointly contribute to the resilience of decentralised indicating that under 10% voter turnout can lead to oligarchic organisations. outcomes [4]. In particular, large token holders often propose andpassinitiativeswithminimalcommunityinput.Weshowed B. Degree of Decentralisation thatDAOswithmoreequitabletokendistributionandmoderate The results show that many DAOs exhibit partial decentralisa- voting windows (3–14 days) achieve higher median approval tion, with voting activity or proposal initiation often concen- rates and more balanced proposer activity. Hence, introducing trated among a small number of addresses. However, several tiered quorums or partial delegation (as in Liquid Democracy) casesofhighdecentralisationwereobserved,particularlywhere could further diversify proposer authority. 10 E. Comparisons with Existing Literature Measurement Accuracy. Node queries and event de- coding can be affected by parsing errors, outdated The observed patterns of partial decentralisation are consistent ABIs, or contract anomalies. Despite validation (e.g. with prior research on governance concentration in DAOs ProposalCreated/Executed checks), some modules de- [4], [6]. While many DAOs aim to implement community-led viate from expected schemas. We addressed this via schema- models, token distribution and proposer activity often remain basednormalizationandcurateddata,reaching99.8%coverage. uneven.Thisstudybuildsonpreviousworkbyquantifyinghow treasury size, participation rates, and voting structure relate to C. External Validity sustainability indicators. The moderate association between treasury size and participa- Generalisability of Findings. Our analysis focuses on tion mirrors earlier findings by [25], where financial capacity Ethereum and EVM-compatible networks, which follow alone did not guarantee user engagement. These findings ERC-20/721 governance models. Governance in Tendermint-, reinforce the view that sustainable governance depends on Cosmos-,orSubstrate-basedDAOsmaydiffer,sofindingsmay multiple interacting factors, not isolated metrics. not generalise beyond EVM contexts. TemporalContext.ThedatasetcapturesDAOgovernanceasof April2025.Givenevolvingrules,tokenomics,andparticipation, F. Limitations and Opportunities for Future Research KPI scores may change. A longitudinal approach would better Fewlimitationsshouldbenoted.First,thedatasetincludesonly reflect governance stability and change over time. DAOs meeting certain on-chain activity thresholds, excluding organisationswithlimitedoroff-chaingovernance.Second,the D. Reliability analysis provides a snapshot as of April 2025; DAO structures Reproducibility of the Pipeline. The analysis relies on third- and participation trends may evolve over time. party services (e.g. Infura, Alchemy, Etherscan), which may Future work could incorporate longitudinal analysis to observe face downtime or updates. Script or API changes can affect governance changes over time, introduce weighting schemes future runs. While version control and pinned contracts were basedongovernanceoutcomes,orexpandcoveragetooff-chain used, replication may still be impacted by ABI or ecosystem processesthroughcommunityforumsandgovernanceplatforms. changes. Analysing the gap between the realities of DAO governance Thresholding and Scoring Variations. An equal-weighted and its ideals [38] represents another avenue for future work. scoringschemewasusedtoreducebias,butalternativeweights Extending the pipeline to non-EVM-compatible chains, or or finer scoring could shift results. Open-source code and clear integrating data across bridging protocols, would improve definitions support replication and comparison across models. generalisability across DAO ecosystems. Finally, refining score thresholds and incorporating adaptive metrics could make the While the study ensures robustness through direct data collec- framework more responsive to DAO-specific use cases, such tionanddefinedKPIs,somelimitationsremain:thresholdtrade- as fast-moving DeFi projects or socially-driven communities. offs,exclusionofoff-chainactivity,andlimitedgeneralisability to EVM-based DAOs. Still, the data supports strong links between governance sustainability and participation, decentral- VI. THREATSTOVALIDITY isation, financial robustness, and procedural efficiency. Future A. Construct Validity work integrating off-chain and multi-chain data can expand DefinitionofKPIs.ThefourKPIsprovideastructuredviewof coverage and address these gaps. DAOgovernancebutareproxiesforbroaderqualities.Off-chain VII. CONCLUSION deliberation, community sentiment, and informal leadership are not captured in on-chain data. As such, participation and This study introduced a data-driven framework for evaluating voting metrics may underestimate engagement in DAOs that DAO governance, combining four empirically grounded KPIs: rely on off-chain activity. Network Participation, Accumulated Funds, Voting Mechanism Boundaries of KPI Thresholds. Thresholds (e.g. 10% and Efficiency,andDecentralisation,intoaunifiedanalyticalmodel. 40% for participation; $100 million and $1 billion for treasury Analysis of on-chain data across a diverse set of 50 DAOs size) are based on empirical patterns and conceptual rationale. revealed that higher sustainability scores tend to be associated However,smalldifferences(e.g.9%vs.10%)canshiftcategory with broader participation, decentralised control, and balanced placement. Token-based decentralisation thresholds may also financial and procedural structures. These findings reinforce notapplyuniformlyacrossDAOtypes,introducingsubjectivity the view that sustainability is shaped by the interaction of into classification. social, economic, and procedural factors, rather than any single attribute. The framework offers a replicable basis for comparative analysis and may support both academic research B. Internal Validity and practitioner decision-making in decentralised governance. Data Completeness. The dataset includes DAOs with active Evaluating DAOs as complex socio-technical systems, rather on-chain governance, excluding those led off-chain. This thanthroughisolatedmetrics,providesamoreaccurateaccount reduces reliance on aggregator dashboards but may bias the oftheirgovernancedynamics.Futureworkcanrefinethismodel sample toward more formal or transparent governance models, through longitudinal evaluation, off-chain data integration, and influencing KPI trends. cross-chain extensions. 11 REFERENCES [21] N.Dimitri,“VotinginDAOs,”ACMSIGecomExchanges,vol.21,no.2, pp.39–48,2023. [1] V.Buterin,“DAOs,DACs,DAsandmore:Anincompleteterminology [22] R.Beck,C.Mu¨ller-Bloch,andJ.L.King,“Governanceintheblockchain guide,”EthereumBlog,2014. economy:Aframeworkandresearchagenda,”JournaloftheAssociation [2] S.HassanandP.DeFilippi,“Decentralizedautonomousorganization,” forInformationSystems,vol.19,no.10,2018,https://aisel.aisnet.org/ InternetPolicyReview,vol.10,no.2,pp.1–10,2021,https://doi.org/10. jais/vol19/iss10/1,DOI:10.17705/1jais.00518. 14763/2021.2.1556. [23] R.Fritsch,M.Mu¨ller,andR.Wattenhofer,“Analyzingvotingpowerin [3] J.Tan,T.Merk,S.Hubbard,E.R.Oak,H.Rong,J.Pirovich,E.Rennie, decentralizedgovernance:Whocontrolsdaos?”Blockchain:Research R.Hoefer,M.Zargham,J.Potts,C.Berg,R.Youngblom,P.DeFilippi, and Applications, vol. 5, no. 3, p. 100208, 2024. [Online]. Available: S.Frey,J.Strnad,M.Mannan,K.Nabben,S.N.Elrifai,J.Hartnell,B.M. https://www.sciencedirect.com/science/article/pii/S2096720924000216 Hill,T.South,R.L.Thomas,J.Dotan,A.Spring,A.Maddox,W.Lim, [24] H.Park,I.Ureta,andB.Kim,“Developingdimensionsandindicators K.Owocki,A.Juels,andD.Boneh,“Openproblemsindaos,”arXiv tomeasuredecentralizationindecentralizedautonomousorganizations,” cs.CY,vol.2310.19201,June172024,https://arxiv.org/abs/2310.19201. Administrative Sciences, vol. 13, no. 11, 2023. [Online]. Available: [4] R. Feichtinger, R. Fritsch, Y. Vonlanthen, and R. Wattenhofer, “The https://www.mdpi.com/2076-3387/13/11/241 hidden shortcomings of (d)aos – an empirical study of on-chain [25] Y. Faqir-Rhazoui, J. Arroyo, and S. Hassan, “A comparative governance,” in International Conference on Financial Cryptography analysis of the platforms for decentralized autonomous organizations andDataSecurity. Springer-Verlag,2023,pp.165–185,https://doi.org/ in the ethereum blockchain,” Journal of Internet Services and 10.1007/978-3-031-48806-1 11. Applications, vol. 12, pp. 1–20, 2021. [Online]. Available: https: [5] J. Han, J. Lee, and T. Li, “A review of dao governance: //doi.org/10.1186/s13174-021-00139-6 Recent literature and emerging trends,” Journal of Corporate [26] R. Qin, W. Ding, J. Li, S. Guan, G. Wang, Y. Ren, and Z. Qu, Finance, vol. 91, p. 102734, 2025. [Online]. Available: https: “Web3-based decentralized autonomous organizations and operations: //www.sciencedirect.com/science/article/pii/S0929119925000021 Architectures,models,andmechanisms,”IEEETransactionsonSystems, [6] S.Wang,W.Ding,J.Li,Y.Yuan,L.Ouyang,andF.-Y.Wang,“Decen- Man,andCybernetics:Systems,pp.2073–2082,2022. tralizedautonomousorganizations:Concept,model,andapplications,” [27] Q. Wang, G. Yu, Y. Sai, C. Sun, L. D. Nguyen, and S. Chen, IEEETransactionsonComputationalSocialSystems,vol.6,no.5,pp. “Understandingdaos:Anempiricalstudyongovernancedynamics,”IEEE 870–878,2019. TransactionsonComputationalSocialSystems,pp.1–19,2025. [7] V. Buterin, “Ethereum white paper: A next-generation smart contract [28] C.-C.Tsai,C.-C.Lin,andS.-W.Liao,“Unveilingvulnerabilitiesindao: anddecentralizedapplicationplatform,”Ethereum,2014. Acomprehensivesecurityanalysisandprotectiveframework,”in2023 [8] Q. DuPont, “Experiments in algorithmic governance: A history and IEEEInternationalConferenceonBlockchain(Blockchain),2023,pp. ethnographyof“thedao”,afaileddecentralizedautonomousorganization,” 151–158. inBitcoinandBeyond. Routledge,2017,pp.157–177. [29] O. Rikken, M. Janssen, and Z. K. and, “Governance impacts of [9] L.Liu,S.Zhou,H.Huang,andZ.Zheng,“Fromtechnologytosociety: blockchain-baseddecentralizedautonomousorganizations:anempirical An overview of blockchain-based dao,” IEEE Open Journal of the analysis,”PolicyDesignandPractice,vol.6,no.4,pp.465–487,2023. ComputerSociety,vol.2,pp.204–215,2021. [Online].Available:https://doi.org/10.1080/25741292.2023.2270220 [10] H. Altaleb and R. Zolta´n, “Decentralized autonomous organizations [30] A.Pen˜a-Calvin,J.Saldivar,J.Arroyo,andS.Hassan,“Acategorizationof review,importance,andapplications,”in2022IEEE26thInternational decentralizedautonomousorganizations:Thecaseofthearagonplatform,” ConferenceonIntelligentEngineeringSystems(INES),2022,pp.000121– IEEETransactionsonComputationalSocialSystems,vol.11,no.6,pp. 000126. 8143–8155,2024. [11] Q.Wang,G.Yu,Y.Sai,C.Sun,L.D.Nguyen,X.Xu,andS.Chen, [31] G. Liu, “The illusion of democracy— why voting in decentralized “Afirstlookintoblockchaindaos,”inICBC,2023,pp.1–3.[Online]. autonomous organizations is doomed to fail,” NYU Law and Available:https://doi.org/10.1109/ICBC56567.2023.10174961 Economics Research Paper, pp. 24–13, 2024. [Online]. Available: [12] A. Wright, “The Rise of Decentralized Autonomous Organizations: http://dx.doi.org/10.2139/ssrn.4441178 Opportunities and Challenges,” Stanford Journal of Blockchain Law [32] V. Lo Monaco, P. Momtaz, and S. Vismara, “Distributed governance &Policy,jun302021,https://stanford-jblp.pubpub.org/pub/rise-of-daos. and value creation in decentralized autonomous organizations: [13] T. V. S. Online, “Subchapter 012: Blockchain-based limited liability Evidence from a regression discontinuity design,” Economics companies,”2018,https://legislature.vermont.gov/statutes/section/11/025/ Letters, vol. 248, p. 112233, 2025. [Online]. Available: https: 04173?ref=internetnative.org. //www.sciencedirect.com/science/article/pii/S0165176525000709 [14] I. N. Organization, “Wyoming dao jurisdictions,” 2023, https:// [33] S.S.ShapiroandM.B.Wilk,“Ananalysisofvariancetestfornormality internetnative.org/wyoming-dao/. (completesamples),”Biometrika,vol.52,no.3/4,pp.591–611,1965. [15] N. I. Organization, “Marshall island dao jurisdictions,” 2023, https: [Online].Available:http://www.jstor.org/stable/2333709 //internetnative.org/marshall-islands-dao/. [34] H. Levene, “Robust tests for equality of variances,” in Contributions [16] R. B. Lamb, “Utah passes innovative dao legislation,” 2023, https:// to Probability and Statistics: Essays in Honor of Harold Hotelling. natlawreview.com/article/utah-passes-innovative-dao-legislation. Stanford University Press, Palo Alto, 1960, pp. 278–292. [Online]. [17] J.Ma,M.Jiang,J.Jiang,X.Luo,Y.Hu,Y.Zhou,Q.Wang,andF.Zhang, Available:https://api.semanticscholar.org/CorpusID:117424234 “Understanding security issues in the dao governance process,” IEEE [35] L. Sthle and S. Wold, “Analysis of variance (anova),” Chemometrics TransactionsonSoftwareEngineering,pp.1–16,2025. and Intelligent Laboratory Systems, vol. 6, no. 4, pp. 259–272, 1989. [18] Y. Liu, Z. Gao, Y. Xia, and Z. Zhang, “From technology to society: [Online]. Available: https://www.sciencedirect.com/science/article/pii/ An overview of blockchain-based DAO,” in 2019 IEEE 1st Interna- 0169743989800954 tionalConferenceonCivilAviationSafetyandInformationTechnology [36] W. H. Kruskal and W. A. Wallis, “Use of ranks in one- (ICCASIT). IEEE,2019,pp.518–522. criterion variance analysis,” Journal of the American Statistical [19] Y.Fan,L.Zhang,R.Wang,andM.A.Imran,“Insightintovotingin Association,vol.47,no.260,pp.583–621,1952.[Online].Available: daos: Conceptual analysis and a proposal for evaluation framework,” http://www.jstor.org/stable/2280779 Netwrk. Mag. of Global Internetwkg., vol. 38, no. 3, p. 92–99, Jun. [37] D.Rozas,A.Tenorio-Forne´s,S.D´ıaz-Molina,andS.Hassan,“When 2023.[Online].Available:https://doi.org/10.1109/MNET.137.2200561 ostrom meets blockchain: Exploring the potentials of blockchain for [20] Q. Ding, W. Xu, Z. Wang, and D. K. C. Lee, “Voting schemes commonsgovernance,”SAGEOpen,11(1).,2021. in dao governance,” p. 2350004, 2023. [Online]. Available: https: [38] A.Alawadi,N.Kakabadse,A.Kakabadse,andS.Zuckerbraun,“Decen- //doi.org/10.1142/S2811004823500045 tralizedautonomousorganizations(daos):Stewardshiptalksbutagency walks,”JournalofBusinessResearch,vol.178,p.114672,052024.