Gov-Acc Pipeline.

Blockchain:ResearchandApplications5(2024)100208 ContentslistsavailableatScienceDirect Blockchain: Research and Applications journal homepage: www.journals.elsevier.com/blockchain-research-and-applications Research Article Analyzing voting power in decentralized governance: Who controls DAOs? Robin Fritsch∗, Marino Müller, Roger Wattenhofer ETHZurich,8092,Zurich,Switzerland A R T I C L E I N F O A B S T R A C T Keywords: We empirically study the state of three prominent DAO governance systems on the Ethereum blockchain: Decentralized autonomous organization Compound, Uniswap and Ethereum name service (ENS). In particular, we examine how the voting power is DAO governance distributed in these systems. Using a comprehensive dataset of all governance token holders, delegates, propos- Liquiddemocracy als, and votes, we analyze who holds the voting power and how this power is being used to influence governance Blockchain decisions. While we reveal that the majority of voting power is concentrated in the hands of a small number of addresses, we rarely observe these powerful entities overturning a vote by choosing a different outcome than that of the overall community and less influential voters. 1. Introduction somewhat controversially voted to create a lobbying group called the “DeFi Education Fund” and allocated $20 million of treasury funds to it Individuals with conflicting interests need to make joint decisions in July 2021.2 frequently in many parts of society: in political elections, at party The novel designs of DAO governance systems and the large amount conventions, or at shareholder meetings. How can such communities of power they hold make it compelling and relevant to study how these govern themselves and reach joint decisions in the most effective and systems work and how decisions are reached. Who holds the voting inclusive way? Research on this question goes back centuries. In 1884, power in these systems, and how do the actors influence the decisions? Charles Dodgson (better known as Lewis Carroll, the author of Alice’s In this paper, we study three of the most prominent and influential Adventures in Wonderland) suggested a transitive voting system in which governance systems on the Ethereum blockchain: Compound, Uniswap, candidates can transfer received votes to other candidates[1,2]. Today, and Ethereum name service (ENS). Compound, a decentralized lend- this concept is commonly referred to as liquid democracyand can be seen ing protocol [5], is one of the first protocols to introduce a gover- as a hybrid between direct and representative democracy. It has found nance system and has served as a role model for many other projects practical use in politics with, for instance, Germany’s Pirate Party us- [6]. For instance, Uniswap, the second governance we examine, has ing liquid democracy for internal voting [3,4]. Even more adaption and adapted Compound’s governance contracts. Uniswap is the clear mar- development in this area has, however, emanated from communities ket leader among decentralized exchanges and one of the most-used around popular blockchain-based projects. Many such communities and applications in decentralized finance (DeFi). Besides governing the pro- stakeholder groups have formed so-called decentralized autonomous or- tocol, Uniswap’s governance also controls a multibillion dollar treasury. ganizations (DAOs) in order to govern their projects and make decisions Finally, we study ENS, which is a vital part of Ethereum infrastructure about the project’s future in a decentralized manner. Such decisions and boasts one of the largest voting communities. In order to facilitate could concern changes to a protocol or the use of the treasury funds voting in a decentralized and pseudonymous setting, DAOs typically is- of a project. With DAOs becoming more and more influential, so do sue governance tokens and distribute them among stakeholders. These their governance systems: Some DAOs including Uniswap manage siz- tokens can initially be distributed in different ways: Uniswap, for in- able treasuries of multiple billion USD.1 Moreover, their governance stance, distributes its UNI tokens among the team, early investors and decision can be highly influential: Uniswap’s governance, for example, early users via an airdrop [7]. Later, the tokens are freely tradable, mak- * Corresponding author. E-mail addresses:rfritsch@ethz.ch(R.Fritsch), muemarin@ethz.ch(M.Müller), wattenhofer@ethz.ch(R.Wattenhofer). 1 For up-to-date numbers, see https://deepdao .io /organizations. 2 https://app .uniswap .org /# /vote /1 /1. https://doi.org/10.1016/j.bcra.2024.100208 Received 8 November 2023; Received in revised form 29 January 2024; Accepted 24 May 2024 Availableonline4June2024 2096-7209/©2024THEAUTHORS.PublishedbyElsevierB.V.onbehalfofZhejiangUniversityPress. ThisisanopenaccessarticleundertheCCBYlicense (http://creativecommons.org/licenses/by/4.0/). R.Fritsch,M.MüllerandR.Wattenhofer Blockchain:ResearchandApplications5(2024)100208 Table1 Key figures of the dataset. Holders Delegates Proposals Totalvotes Compound 188.198 1.766 84 3.431 Uniswap 306.580 5.784 10 1.660 ENS 57.511 11.597 –a – a Up until 1 March 2022, two proposals were voted on by ENS governance, however thesewerenotpartofourdataset. distribution and voter turnout. The analysis in our paper goes deeper in multiple ways; we additionally examine voting behavior and delegation network structure. Voting systems similar to those currently used by many DAOs have a long history of academic discussion. They are commonly referred to as delegative democracy[18]or liquid democracy[1,8]. The governance systems we study use slightly simpler versions of these concepts with a main difference being that they only allow for voting rights to be Fig.1.A visualization of the delegation network of token holders (blue nodes) delegated once from a token holder to a delegate. In liquid democracy, and delegates (red nodes) in Compound governance. it is often possible for a delegate to further (recursively) delegate the tokens they received to another delegate. The practical application of liquid democracy has been examined in a case study of the internal ing it possible to buy tokens and thereby voting power. In this sense, voting system of the German Pirate Party [3,4]. governance tokens have similarities with shares in companies that come with voting rights at shareholder meetings. The governance tokens can then be used to vote on governance pro- 3. Overview of the dataset posals: Each token counts as one vote. To vote, a simple form of liquid democracy [8]is used: To be able to vote, a token holder must delegate We collect comprehensive data on the token holders, delegates, their tokens to a delegate. Delegates can then vote on proposals with the votes and proposals for the Compound, Uniswap and ENS governance number of tokens delegated to them. A visualization of the network of systems from the Ethereum blockchain. The three DAOs were chosen delegations for Compound is shown in Fig. 1. Note that both holders for being among the most influential DAOs on Ethereum at the time and delegates are addresses on the blockchain. In particular, multiple of writing (besides MakerDAO, which is studied in Ref. [11]): While addresses can be controlled by the same person or organization. Compound’s governance model and contract have become a role model Most DAO governance processes, including the three we study, copied by many other DAOs, and Uniswap DAO governs the by far most involve multiple steps, only the last of which takes place on the prominent DeFi application with a multibillion-dollar treasury, ENS has blockchain. For example, the Uniswap governance process first requires one of the largest communities of governance token holders. a temperature check and a consensus check, both of which are off- Table1provides an overview of the state of these governance sys- chain and require a certain threshold of yes-votes. Only when these tems as of 1 March 2022. The analysis encompasses a more than two- two thresholds are reached is a proposal put to a vote on-chain. In this year period between Ethereum blocks 9.000.000 (25 November 2019) paper, we concentrate on the final on-chain votes where the actual de- and 14.300.000 (1 March 2022). To gather the data, we use the index- cisions are made. ing protocol, The Graph, which allows querying Ethereum blockchain In this paper, we analyze how voting power is distributed and used data. Specifically, we utilize three subgraphs that index governance- in DAO governance systems. In the process, we evaluate how decentral- related transactions of the three DAOs in question.3 Since more such ized these systems actually are (as this is a major goal of blockchain- subgraphs are being created and open-sourced, this allows for extend- based applications). Furthermore, we examine the structure of the del- ing this approach to more DAOs in the future [19]. egation networks and study the voting behavior of different types of The delegation networks of the governance systems are visualized in delegates. Finally, we discuss why it is not always possible to deter- Figs.1and 2using the Python library NetworkX[20]. An edge between mine the exact distribution of power since multiple addresses on the a holder (blue node) and a delegate (red node) indicates a delegation. blockchain can be controlled by the same entity. The sizes of the nodes are proportional to the number of tokens held or represented (with a minimum size for small holders and delegates). 2. Related work A clear difference between Compound and Uniswap on the one side and ENS on the other side is visible at first glance: For ENS, delegates A first attempt at creating a DAO was made with “The DAO” in the tend to have numerous holders of similar size delegating to them. For early days of the Ethereum blockchain in the year 2016 [9]. However, Compound and Uniswap, on the other hand, many delegates receive due to a hack, The DAO never actually operated as intended. One of the most of their voting power from a single large token holder delegat- first governance systems for a blockchain-based application was intro- ing to them. Note that in such cases, it is possible that the holder and duced by the stablecoin protocol MakerDAO in 2018 [10]. A case study delegate addresses are both controlled by the same entity (and that the of MakerDAO’s governance can be found in Ref. [11]. Since Compound delegate address was set up for voting purposes only). We investigate set up its token voting governance system in early 2020 [6], many other the differences between the governance systems in more detail in Sec- projects have followed suit. Some early analyses of the systems, mainly tion5. focusing on the ownership distribution of governance tokens of DeFi protocols, have been carried out in Refs. [12–14]. (For an overview of DeFi, see, e.g., Refs. [15,16].) In a recent, more detailed study, Bar- 3 https://thegraph .com /hosted -service /subgraph /arr00 /compound - bereau et al. considered the governance systems of nine DeFi protocols: governance -2, https://thegraph .com /hosted -service /subgraph /arr00 /uniswap - Uniswap, Aave, MakerDAO, Compound, SushiSwap, Synthetix, Yearn governance -v2, https://thegraph .com /hosted -service /subgraph /ianlapham / Finance, 0x, and UMA [17]. In their work, they focused mainly on token ens -governance. 2 R.Fritsch,M.MüllerandR.Wattenhofer Blockchain:ResearchandApplications5(2024)100208 Fig.2.Delegation networks: Blue nodes represent holders and red nodes represent delegates. The sizes of the nodes are proportional to the number of tokens held or represented. Each edge indicates a delegation. Table2 Gini and Nakamoto coefficients of the token distributions among holders, dele- gates, and delegates who voted at least once. All the numbers are rounded to 3 digits after the decimal point. Compound Uniswap ENS GiniHolders 0.998 0.996 0.977 GiniDelegates 0.987 0.995 0.908 GiniDelegatesVoted 0.964 0.971 – Fig.3.Distribution of voting rights among delegates: Delegates are sorted by NakamotoDelegates 8 11 18 size, and the width of each box indicates the share of votes delegated to them. Finally, considering only delegates who voted at least once gives a 4. Distribution of voting power slightly more equal distribution once again. The Nakamoto coefficient [23] is a measure of how decentralized In the following, we study who holds the power in the governance a system is. Applied to our setting, it counts the minimum number of systems. We examine this in three steps: 1) How are voting rights dis- delegates that together hold more than 50% of the voting power. The tributed? 2) How frequently are voting rights executed? 3) How much Nakamoto coefficients (see Table2) reveal extreme centralization in all do the executed votes influence the final governance decision? three governance systems: In the case of Compound, only 8 delegates can dictate any governance action using their majority. For Uniswap, 4.1. Distribution of voting rights this number is 11; for ENS, it is 18. Considering the Gini coefficient over time (Fig. 5(a)) reveals that As a first step, we examine how voting rights are distributed in the the strong inequality in the voting power distributions has persisted three governance systems. Fig. 3 shows the distribution of delegated for an extended time period and even seems to be rising. On the other tokens among delegates. The visualization clearly illustrates the concen- hand, increasing Nakamoto coefficients (Fig.5(b)) do indicate that the tration of voting rights among a relatively limited number of delegates. systems are becoming slightly more decentralized over time. To precisely quantify this concentration, we consider two inequality measures: the Gini coefficient and the Nakamoto coefficient. 4.2. Governance participation The Gini coefficient [21]is the most commonly used index to mea- sure inequality and ranges from 0 (perfect equality) to 1 (maximum A second crucial aspect in analyzing voting power is how often del- inequality). It measures the area between the Lorenz curve of a given egates participate in voting, i.e., how often delegates use their voting distribution and the curve of perfect equality [21]. Fig.4displays the power. Lorenz curves of the distribution of voting rights among delegates. For The upper plots in Fig. 6 show the fraction of the total supply of the remainder of this section, we consider the distributions of tokens at governance tokens which is delegated (the total supply is 10 million block 14.300.000, which occurred on 1 March 2022. for Compound and 1 billion for Uniswap). Since delegating is a pre- The precise Gini values in Table2indicate that ENS’ token distribu- requisite for voting, this is the maximum amount of tokens that could tion is less unequal than that of Compound and Uniswap, which have have participated in governance at that time. Furthermore, the figures Gini values around 0.99. Nonetheless, the values prove a high degree of show how many delegated tokens were actually used to vote for each inequality in the distribution of voting rights in all three cases. In par- proposal (each dot represents a proposal). ticular, the inequality is higher than that in the general distribution of Note that only about two-thirds of the total supply of COMP and wealth in the world: The Gini coefficient of the wealth distribution in UNI tokens circulated at the end of our observation period according 2020 is estimated to be 0.850 in the US, 0.814 in Europe, and 0.704 in to CoinMarketCap.4 The remaining tokens will come into circulation China [22]. Moreover, we also see that the distribution of tokens among dele- gates is slightly more equal than the distribution among token holders. 4 https://coinmarketcap .com/. 3 R.Fritsch,M.MüllerandR.Wattenhofer Blockchain:ResearchandApplications5(2024)100208 Fig.4.Lorenzcurvesofthevotingpowerdistributionsamongdelegates. Fig.5.GiniandNakamotocoefficientsfordelegatesovertime. Fig.6.Governance participation: The upper figures show the number of delegated tokens over time and how many of them were used to vote in each proposal. The lower figures plot the total number of delegates and how many of them voted. in the future. As only circulating tokens are available for delegation, The figures show that, on average, in both governance systems less the delegation rate of about 20% for Uniswap translates into 30% of than 10% of total tokens (or 15% of circulating tokens) participate in all circulating tokens. As a consequence, 15% of circulating governance votes on proposals. Hence, 7.5% of circulating tokens would have al- tokens can currently take control of Uniswap’s governance. Note that ready sufficed to win an average vote in the past. On the other hand, circulating supply has increased over time, which could in part explain it should be noted that more than half the delegated tokens regularly the increase in delegated tokens. vote. 4 R.Fritsch,M.MüllerandR.Wattenhofer Blockchain:ResearchandApplications5(2024)100208 Fig.7.Votingparticipationofdelegates. For comparison, participation in voting at shareholder meetings of come of a proposal vote, i.e. how often the result would have been traditional companies (which arguably share some similarities at least different without the delegate’s votes. It is defined as w c co o i r m t d h p i n U a g n n i i t e s o s w i R a s e p a f . b a o n [2 u d 4 t C ] 7 , o 0 t m % h p e . o p u a n r d ti c g i o p v a e ti r o n n a n r c a e te ) i a s m s o ig n n g i fi sh ca a n re tl h y o l h d i e g r h s e r o : f A U c S - 𝑝𝑒 𝑑 𝑥 = { 0 1 i e f ls ( e 𝑉 𝑉 (𝑦) (𝑦 + ) 𝑉 −𝛿 (𝑛 𝑑 ) 𝑣 − 𝑑 𝑣𝑑 >1 2 ) ≠ ( 𝑉(𝑦 𝑉 )+ (𝑦 𝑉 ) (𝑛) >1 2 ) . The lower plots in Fig.6show the total number of delegates and how many of them participated in each vote. We find that only a very small Again, the overall exercised power is then defined as the mean over all number of delegates actually participate in voting (note the logarithmic 𝑝𝑒 𝑑 𝑥 . 𝑦-axis). This again shows how these governance systems are centralized. For Compound, the number has been falling and the average is currently Fig.8(a) plots the potential and exercised voting power of delegates around 20, while it has lately been less than 100 for Uniswap. in Compound’s governance system. The chart reveals that a substantial Finally, we consider how the participation rate of delegates depends portion of the potential voting power is shared among just four of the on the number of delegated tokens they have received. Fig.7plots how largest delegates. Notably, one of them could have altered the outcome often delegates vote against the number of tokens delegated to them. of five out of the nine proposals they voted on. This observation is once Each dot represents a delegate, and the dashed line represents the re- again evidence of how much power lies in the hand of a select few gression line. We observe that delegates with more tokens delegated to individuals. However, we also find that these individuals do not exercise them tend to vote more often. A notable exception is the second largest their power often; their actual voting activity falls significantly short of Uniswap delegate, who has been delegated almost 12.8 million UNI to- their potential influence. kens (from a single address) worth more than $13 million on 1 March The same is true for Uniswap (Fig. 8(b)): Large delegates only ex- 2022, but has never voted in a governance decision. ercise their voting power rarely. This is examined in more detail in Section5.2when we compare how differently large and small delegates 4.3. Potential and exercised voting power vote and find that large delegates often vote in line with small dele- gates. As the final step, we examine how the votes that are cast actually Note that most potential and exercised voting power of small del- influence the outcome of the governance decisions. It is one thing for egates that can be seen in the figures stems from two proposals with delegates to have the possibility to change votes. A completely differ- very close outcomes (which means that they could have been influ- ent matter is whether they actually choose to do so. To measure this, enced by many delegates): Proposal 16 for Compound5 and Proposal we use the following two metrics inspired by Kling et al. [4], which 1.2 for Uniswap.6 are designed to quantify the potential and exercise the voting power of 5. Delegate classification and voting behavior delegates. In this section, we investigate the voting behavior exhibited by var- Definition 4.1 (Potential Voting Power). We measure the potential vot- ious categories of delegates. ing power of a delegate by counting how often the delegate could have Section 3 hints at a difference in the structure of the ENS’ delega- changed the outcome of a proposal vote by changing their own vote. tion network. In the following, we set out to quantify this difference. To Formally, the potential power 𝑝 𝑝𝑜𝑡∈{0, 1} of a delegate who partici- achieve this, we categorize the delegates into two distinct groups: Dele- pated in a proposal vote is defined as gates who receive at least 50% of their voting power from a single token 𝑝𝑝𝑜𝑡 = { 1 if𝑣 𝑑 >1 2 ( 𝑉(𝑦)+𝑉(𝑛) ) −𝑉(𝑦)+𝛿 𝑑 𝑣 𝑑 >0 , h cl o a l s d s e ifi r e a d re a l s a c b o e m le m d u a n s i t s y in d g e le le h g o a l t d e e s. r delegates. The remaining delegates are 𝑑 0 else where 𝑉(𝑦) and 𝑉(𝑛) are the number of Yes-and No-votes on the pro- 5.1. Single holder delegates vs. community delegates v p o o t s e a s l , , a re n s d p e 𝛿 c 𝑑 ti ∈ ve { ly 0 . , 1 F } ur i t n h d e i r c m at o e r s e , w 𝑣 h 𝑑 eth d e e r n o d t e e l s e g d a e t l e e g 𝑑 at v e o 𝑑 te ’s d n Y u e m s b (1 e ) r o o r f com F m ig u . n 9 ity s h d o e w le s g a th te e s . s h T a h r e e d o i f ff d er e e le n g c a e t e b d e tw to e k e e n n s g o th v a e t r n i a s n c c o e n s tr y o st ll e e m d s b i y s No (0) on the proposal. apparent: For ENS, about 60% of voting rights are in the hands of com- The overall potential power of a delegate is then defined as the mean of 𝑝𝑝𝑜𝑡 over all proposals. munity delegates, while it is less than 10% for Compound and Uniswap. 𝑑 Definition 4.2 (Exercised Voting Power). The exercised voting power 5 https://compound .finance /governance /proposals /16. measures how often the vote of a delegate actually changed the out- 6 https://app .uniswap .org /# /vote /1 /2. 5 R.Fritsch,M.MüllerandR.Wattenhofer Blockchain:ResearchandApplications5(2024)100208 Fig.8.Potentialandexercisedvotingpowerofdelegates. narrow one: On proposal 1.2, large delegates are 50.5% in favor, while small delegates oppose it (34.5% are in favor). 6. Address clustering During all the analyses, it is important to keep in mind that multiple addresses—holder as well as delegate addresses—can be controlled by the same entity. This fact can make it challenging to determine the actual distribution of power in a governance system and can lead to the illusion of greater decentralization in the system than what truly exists. Moreover, the situations of a single entity controlling multiple ad- dresses are actuallyfavored by the design of the delegation rules of the governance systems: It is only possible for a token holder to delegate all of their tokens to a single address. A token holder wishing to delegate to multiple delegates must split their tokens across multiple addresses Fig.9.Share of delegated tokens controlled by community delegates (i.e. dele- and delegate separately from these addresses. gates who receive less than 50% of their votes from a single address). A situation in which an entity requires multiple addresses is a large token holder with a delegation program. An example of an entity run- While the fraction of voting rights held by community delegates has al- ning such a program is venture capital firm a16z[25]. ways been low for Compound, community delegates historically had To identify such scenarios, we consider transfers of governance to- more influence in the early stages of Uniswap governance. kens between addresses. When tokens are sent from one address to The above observations illustrate that delegations are not utilized another, it strongly suggests that both addresses are under the con- to a great extent in Compound and Uniswap governance. Most voting trol of the same entity. Examining these transfers creates a network power lies in the hands of delegates who receive most of their voting of addresses with links between them. From this network, we extract power from a single token holder. It is possible that in many such cases, the connected components. Throughout the process, it is necessary to the two addresses are controlled by the same entity. In this regard, exclude certain addresses, like those associated with centralized ex- Compound and Uniswap governance show similarities to shareholder changes. Using this method, we find a number of groups of addresses of meetings, where large investors represent their interests. The ENS, on Compound and Uniswap token holders that are potentially controlled the other hand, is more similar to a decentralized community. by a single entity. As a next step, we compare the voting behavior of community del- Fig. 10 shows two examples of such groups. It is conceivable that egates and single holder delegates. We find that both groups vote sim- these addresses are part of a16z’s delegation program[25]. Of course, ilarly: They only voted differently on three Compound proposals and it is important to keep in mind that this method is not bullet-proof, one Uniswap proposal. More precisely, only in these four cases did the and we refrain from making definitive claims about the structure of majority of community delegates vote for a different outcome than the control of the addresses shown in the figure. With these examples, our majority of single holder delegates. goal is to illustrate that blockchain-based governance systems are not always entirely transparent, and that the power within them can be 5.2. Large vs. small delegates more centralized than it initially perceived. In the following, we compare the voting behavior of large and small 7. Conclusion delegates. Large delegatesare defined as the largest delegates that col- lectively hold at least 90% of the voting rights at the time of a proposal, Our analysis reveals a significant concentration of voting power in while all other delegates are categorized as small delegates. all three governance systems, with a majority controlled by a limited It is clear that large delegates, defined in this way, hold all the number of addresses. Additionally, the often low degree of participation power in the governance system and can decide any vote in their favor. in voting leads to even more influence of those with significant voting However, the critical question is how frequently they use this power to power who actively participate. We state these observations without override the minority of small delegates. judgment. Nonetheless, they do illustrate the challenges of building a Our findings show thatthis only rarely happens: In Compound gov- truly decentralized governance systems. ernance, large delegates only vote differently from small ones on six out In other aspects, we find noteworthy differences between Compound of the 84 proposals. For Uniswap, the picture is similar: Here, large and and Uniswap on the one side and ENS on the other side. Besides the vot- small delegates only disagree on a single proposal, and even that is a ing power being slightly more decentralized for the latter, community 6

R.Fritsch,M.MüllerandR.Wattenhofer Blockchain:ResearchandApplications5(2024)100208 Fig.10.Addressespossiblycontrolledbyasingleentitybasedontokentransfers. delegates hold much more voting power in ENS governance. In con- [7]U. Labs, Introducing uni, https://uniswap .org /blog /uni, 2019. (Accessed30 May trast, for Compound and Uniswap, almost all the voting power lies in 2024). the hands of delegates that receive the majority of their voting power [8]A.Paul in,Anoverv iewoften yea rsofli quiddemocr acyresear ch,in :Th e21st from a single address. In this sense, their governance systems are sim- Annu alIn ternational Conference onD igital GovernmentR esearch,Dg .O’2 0,A ssoci - ationforComputingMachinery,NY,2020,pp.116–121,https://doi.org/10.1145/ ilar to shareholder meetings of traditional companies, where a small 3396956 .3396963. number of large investors represent their interests. ENS governance, on [9]Gemini, What was the dao?, https://www .gemini .com /cryptopedia /the -dao -hack - the other hand, more resembles a decentralized community. makerdao, 2022. (Accessed30 May 2024). Notably, we find that those large and powerful delegates rarely ex- [10]MakerDAO , Found ati on proposal v2, https://med ium .com/@MakerDAO/ foundation-proposal-v2-f10d8ee5fe8c,2018.(Accessed30May2024). ercise their voting power to overturn a vote. In most cases, they decide [11]X.Sun,C.Stasinakis,G.Sermpinis,Decentralizationillusionindefi:evidencefrom in the same way as the larger community, i.e., smaller delegates or com- makerdao, arXiv, 2022, preprint, arXiv :2203 .16612. munity delegates. [12]K. Stroponiati, I. Abugov, Y. Varelas, et al., Decentralized governance As the importance of DAOs continues to rise, continuing research in defi: examples and pitfalls, https://static1.squarespace.com/static/ along the lines of this paper will become increasingly compelling. The 596 6eb2ff7e0ab 3d29b 6b55 d/t/5f 989987fc086 a1d8 482ae70/1603837124500/ defi_governance_paper.pdf,2020.(Accessed30May2024). future will present opportunities for exploring a broader array of DAO [13]J.R. Jensen, V. von Wachter, O. Ross, How decentralized is the governance of governance systems and analyzing larger numbers of proposal votes. blockchain-based finance: empirical evidence from four governance token distri- butions,arXiv,2021,preprint,arXiv:2102.10096. CRediT authorship contribution statement [14]M.N adler,F. Sc här,Dece ntralized finan ce,centralize do wnership? Anit erativ emap- pingprocesstomeasureprotocoltokendistribution,J.BlockchainRes.1(1)(2022) 29–36,https://doi.org/10.4310/JBR.2022.v1.n1.a3. Robin Fritsch:Conceptualization, Investigation, Methodology, Su- [15]S. Werner, D. Perez, L. Gudgeon, et al., Sok: decentralized finance (DeFi), in: pervision, Visualization, Writing – original draft, Writing – review & Proceedings of the 4th ACM Conference on Advances in Financial Technologies, editing. Marino Müller: Data curation, Investigation, Visualization. AFT ’22, Association for Computing Machinery, NY, 2023, pp. 30–46, https:// Roger Wattenhofer:Supervision, Writing – original draft. do i.org/10.11 45/ 355853 5.3 559780. [16]S.Aramonte,W.Huang,A.Schrimpf,DeFirisksandthedecentralisationillusion, BISQuarterlyReview,https://www.bis.org/publ/qtrpdf/r_qt2112b.htm,2021.(Ac- Declaration of competing interest cessed30 May 2024). [17]T.Barbereau,R.Smethurst,O.Papageorgiou,etal.,Decentralisedfinance’stimo- The authors declare that they have no known competing financial crati cg overnan ce:thed istributiona nd exer ciseo f tokenise dvoti ngrights,Technol. Soc.73(2023)102251,https://doi.org/10.1016/j.techsoc.2023.102251. interests or personal relationships that could have appeared to influence [18]B.A.Ford,Delegativedemocracy,https://infoscience.epfl.ch/record/265695,2002. the work reported in this paper. (Accessed30 May 2024). [19]R.Feichtinger,R.Fritsch,Y.Vonlanthen,etal.,Thehiddenshortcomingsof(d)aos– References anempiricalstudyofon-chaingovernance,in:FinancialCryptographyandData Security.FC2023InternationalWorkshops,SpringerNatureSwitzerland,Cham, 2024,pp.165–185,https://doi.org/10.1007/978-3-031-48806-1_11. [1]J.Behrens,Theoriginsofliquiddemocracy,Liq.Democr.J.Electron.Particip.Col- [20]A.A.Hagberg,D.A.Schult,P.J.Swart,Exploringnetworkstructure,dynamics,and lect.Moderat.VotingSyst.(5)(2017),https://liquid-democracy-journal.org/issue/ functionusingNetworkX,in:Proceedingsofthe7thPythoninScienceConference, 5/The_Liquid_Democracy_Journal-Issue005-02-The_Origins_of_Liquid_Democracy. 2008,pp.11–15,https://doi.org/10.25080/TCWV9851. html. [21]GiniIndex,in:TheConciseEncyclopediaofStatistics,SpringerNewYork,NewYork, [2]L.Carroll,ThePrinciplesofParliamentaryRepresentation,HarrisonandSons,Lon- NY,2008,pp.231–233,https://doi.org/10.1007/978-0-387-32833-1_169. don,1884. [22]A. Shorrocks, J. Davies, R. Lluberas, Credit suisse research institute global [3]B. Swierczek, 5 years of liquid democracy in Germany, Liq. Democr. wealth databook 2021, https://www.credit-suisse.com/media/assets/corporate/ J. Electron. Particip. Collect. Moderat. Voting Syst. (1) (2011), https:// docs/about-us/research/publications/global-wealth-databook-2021.pdf,2021.(Ac- liquid-democracy-journal.org/issue/1/The_Liquid_Democracy_Journal-Issue001- cessed30May2024). 02-Five_years_of_Liquid_Democracy_in_Germany.html. [23]B.S. Srinivasan, L. Lee, Quantifying decentralization, https://news.earn.com/ [4]C.C.Kling,J.Kunegis,H.Hartmann,etal.,Votingbehaviourandpowerinon- quantifying-decentralization-e39db233c28e,2017.(Accessed30May2024). linedemocracy:astudyofliquidfeedbackinGermany’spirateparty,in:M.Cha, [24]K.E.Zachariadis,D.Cvijanovic,M.Groen-Xu,Free-ridersandunderdogs:participa- C. Mascolo, C. Sandvig (Eds.), Proceedings of the Ninth International Confer- tionincorporatevoting,EuropeanCorporateGovernanceInstitute–FinanceWork- ence on Web and Social Media, AAAI Press, Calif., 2015, pp. 208–217, http:// ingPaper(649),SSRN,2020,preprint. www.aaai.org/ocs/index.php/ICWSM/ICWSM15/paper/view/10566. [25]J.Amico,Opensourcingourtokendelegateprogram,https://a16z.com/2021/08/ [5]R.Leshner,G.Hayes,Compound:themoneymarketprotocol,https://compound. 26/open-sourcing-our-token-delegate-program/,2021.(Accessed30May2024). finance/documents/Compound.Whitepaper.pdf,2019.(Accessed30May2024). [6]R. Leshner, Compound governance, https://medium.com/compound-finance/ compound-governance-5531f524cf68,2020.(Accessed30May2024). 7