From ethresearch by Wenxuan Deng, Tanisha Katara, and David Hamoui

AcknowledgementsAdditional thanks to Peter Liem 3 for his assistance with data fetching and to Mateusz Rzeszowski 1 for his insightful comments from a governance perspective.

Abstract

This research examines the impact of financial incentives on voter behavior in two significant blockchain ecosystems, Curve Finance and Polkadot, through a user behavior study of all their voters. As the first comprehensive examination of voter behavior in Web3, the study suggests that when combined with governance, financial incentives lead to longer staking preferences among different voter personas and increased voter turnout. The research will establish the critical decision-making and governance strategies adopted by Curve Finance and Polkadot, which allow the community to agree on significant network changes collectively. However, even in the momentary celebration of collective decision-making, the evolving governance structures built must solve either some form of centralization or extremely low voter turnouts.

Introduction

Blockchain technology has ushered in a new era of innovative governance mechanisms for decision-making. It shifts power from small, centralized groups to a broader community, democratizing and monetizing voting power. This transformation significantly boosts participation, enhances transparency, and secures the decision-making process within the community.

A pivotal element of this new crypto-native governance model is the financial incentivization tied to voting. Drawing on economist Robin Hanson’s concept of Futarchy, where betting markets are used to aggregate market information efficiently, financial rewards for voting play a crucial role in aligning community interests and governance outcomes.

As the fields of tokenomics and governance rapidly evolve, two protocols have emerged as particularly influential. Curve DAO pioneered the veToken model, introducing gauge voting and bribery mechanisms to decentralized finance (DeFi). Meanwhile, Polkadot’s governance system has been notable for its innovative approach to quorums, supported by a robust research framework. These protocols stand out as significant pioneers in enhancing governance conviction and shaping the future of decentralized systems.

Governance conviction, often referred to as the lockup time multiplier, is a mechanism used in decentralized governance models to enhance the influence or voting power of token holders based on the duration for which they are willing to lock up their tokens. This system operates under the principle that the longer a participant commits their tokens, the more conviction they demonstrate towards the decisions being made within the network. As a result, their voting power is multiplied by a factor corresponding to the lockup period. This incentivizes longer-term commitment and stability within the governance process, aligning participants’ interests with the long-term health and success of the platform.

Besides the implementation of governance conviction, both governance systems utilize permissionless on-chain execution. This means that decision-making is directly translated into canonical code, eliminating the need for an external authority or intermediary to implement changes voted on by token holders.

However, there is a notable difference between the two governance systems in terms of token lock-up mechanisms. Curve Finance employs a governance mechanism that encourages participants to actively engage in decision-making processes by allowing them to gain financial rewards for locking their tokens.

Let’s first briefly go over how these two governance systems work.

Curve Finance is a protocol that allows for the seamless exchange of ERC-20 tokens with minimal hassle and low costs. This is achieved through the use of Liquidity Pools, which require a sufficient number of tokens to ensure successful swaps and incentivize liquidity providers. Curve offers rewards to those who contribute, creating a win-win situation where users can easily exchange tokens while liquidity providers receive rewards.

To vote, CRV token holders must possess veCRV. veCRV represents CRV tokens that are locked for a certain period (Table 1). Users can lock their CRV for at least one week or up to a maximum of four years.

Polkadot is a protocol that aims to connect different blockchains, also known as parachains, to enable seamless communication, interoperability, and scalability within its network. It facilitates the creation of interconnected blockchains to allow them to work more efficiently and experience shared security.

For instance, if Alex has 100 tokens and locks them for 14 days, his voting power per the above formula is 100 (Dot Tokens Held) * 2 (Multiplier for 14 days, as mentioned in the above table) = 200. Therefore, Alex will have the voting power of 200 tokens.

However, we are still in the early stages of this game-theoretical decentralized governance system, and many aspects of cryptoeconomic design remain unclear. Although innovation and iteration in governance cryptonomics are rapid and continue to accelerate, there are few rigorous and empirical studies from the game-theoretic perspective on voter personas. Therefore, in this research paper, we not only provide a comprehensive glance at the decentralized voting system but also construct a new playground methodology to delve deeply into voter personas.

We will first discuss governance voter analysis (Section 2) and governance proposal analysis (Section 3). Then, we will try to understand the different voter personas in each system (Section 4). Finally, we will analyze voter behavior and seek to understand the governing principles of these complex environments. To do so, we will break down the types of market conditions and variations in voter behavior towards them. The objective is to provide valuable insights into the unique governance features of Curve Finance and Polkadot. To achieve this aim, we assess various factors such as voter turnout, proposal participation, voting power, and lock-up windows.

Voter Turnout Analysis

We did a comprehensive analysis on voter turnout for governance voter engagement. The calculation of voter turnout metrics is based on the number of veCRV tokens and DOT tokens used for voting over time, relative to the total veCRV and DOT tokens locked over the same period.

Curve Finance: On average, 65% of the circulating CRV is locked as veCRV. Out of the 65% locked, an average of 38% tokens have been used for voting. This highlights that although a significant proportion of CRV is locked, a relatively low percentage is used for voting. Further investigation is required to determine the exact factors contributing to the low percentage of utilized tokens.

Polkadot: In contrast, only 54.5% of the circulating DOT is locked, and out of the 54.5% locked, only 0.11% tokens have been used for voting. This highlights a significant disparity in voter engagement between the two blockchain ecosystems, with DOT showing a much lower level of voter engagement than CRV. The low percentage of utilized tokens for voting in DOT could be attributed to several factors, such as a lack of financial incentives.

The following graphs provide a visual representation of the amount of veCRV and DOT used for voting. Upon analyzing the data, it was observed that the average veCRV used for voting on Gauge proposals is significantly higher than the average veCRV used for Non-Gauge proposals. Conversely, the second graph shows that there is no notable difference between the amount of DOT locked for voting on treasury proposals and non-treasury proposals. These findings suggest that there are different financial incentives contributing to these variations. While Curve Finance incentivizes rewards through gauges, the incentives in Polkadot treasury proposals are limited to particular utilities.

Screenshot 2024-04-03 at 6.25.37 PM1498×906 133 KB

The voter turnout metrics shed light on the extent to which token holders actively engage in voting in both the ecosystems. This sets the stage for a deeper exploration into the nature of governance proposals and their implications on decision-making dynamics within the Curve Finance and Polkadot communities. By examining the types and frequencies of proposals submitted in governance, we can gain further insights into the priorities and interests driving the respective ecosystems’ governance mechanisms.

Governance Proposals Breakdown

Governance proposals play a crucial role in decision-making within any community. Our analysis of two ecosystems, Curve Finance and Polkadot, reveals that financial-centric proposals make up a significant portion of all proposals. Gauge proposals constitute approximately 70% of all proposals in Curve Finance, while Treasury proposals make up approximately 80% of all proposals in Polkadot.

Screenshot 2024-04-01 at 9.33.00 PM1412×704 106 KB

Screenshot 2024-04-01 at 9.33.08 PM1344×684 120 KB

However, in Curve Finance, we found that most proposals are initiated by individuals associated with the Curve.fi team (Table 3), particularly two wallets linked to the founder, Michael Egorov. The top wallet, identified as the Curve.fi deployer on the Arkham data platform, is notably active in both gauge and non-gauge proposal submissions. We have provided a table of top Curve Finance proposers, ranked in the order of participation.

On the other hand, we did not observe such patterns in Polkadot. The most referenda submitted by a single author amounted to only 6% of the total proposals.

Understanding the governance proposals and the voters who participate is important. Let’s dive deeper into the different types of voters.

Voter Personas and Respective Patterns

The following research analyzes the behavior of voter personas and their respective patterns concerning token holdings. The voter personas are categorized based on the size of their token holdings, and the hierarchy is defined as follows: the top 1% is labeled as Whales, the next 5% as Sharks, the next 10% as Dolphins, the next 20% as Fishes, and the remaining as Shrimps.

Curve Finance: In the context of Curve Finance governance, voter personas emerge vividly when we dissect them by the size of their veCRV token holdings. Over 58% of token holders choose to lock their tokens for four years. However, an intriguing trend surfaces among the different cohorts of holders.

As shown in the graph below, the x-axis displays the initial lock-up windows, which range from 7 days to 4 years, and the y-axis displays the percentage of user personas who have locked their tokens. Our more giant holders, the Whales, Sharks, and Dolphins, show mild hesitation to commit for more extended lock periods. Conversely, they slightly prefer shorter commitments, particularly those under six months. Although the margin is slim, it is a telling divergence.

It hints that more oversized holders may not need to lock up their tokens for extended periods to wield significant voting power. For them, the flexibility not to lock in for an extended time could be a strategic move to mitigate risk and maintain liquidity options.

Screenshot 2024-04-01 at 11.04.41 PM1170×1478 106 KB

Polkadot: In the Polkadot ecosystem, analysis yielded a similar yet valuable correlation to Curve Finance. 4% of DOT holders initially choose to lock their tokens for a maximum of 224 days (roughly seven months). Holders with more significant positions prefer shorter lock-up periods, and this pattern is glaringly apparent. Particularly striking in the Polkadot ecosystem is that about 93% of whale and 98% of shark holders tend to lock up their tokens for 14 days or less. Contrastingly, shrimp holders display markedly different behavior, with approximately 30% opting for an 8-week lock-up and about 5% committing to a 32-week lock-up.

image600×800 31.2 KB

Among different holder categories, from Whales to Shrimps, there is an incremental increase in the preference for longer lock-ups as we move down the scale of holdings. The distinctions and preferences between prominent and smallholders are consistent within each protocol.

However, there is a significant difference between the two protocols. As shown in the pie charts below, in Curve Finance, 67.2% of voters across all groups opt for a four-year lock-up, while in Polkadot, only 4% of the users choose the most extended lock-up window of 224 days. Even among the most minor stakeholders, the Shrimps, less than 5% chose the most prolonged lock-up period of 32 weeks.

This divergence could be attributed to the fundamental differences in the underlying rewards and incentives of Curve Finance and Polkadot. Curve’s gauge weight voting system incentivizes users to boost their voting power by locking their tokens for extended periods. This indicates that sustained rewards are crucial in incentivizing token holders to stay longer.

Screenshot 2024-04-01 at 11.04.50 PM1454×1360 90.3 KB

Screenshot 2024-04-01 at 11.05.01 PM1456×1398 85.6 KB

With this understanding of voter personas and their lock-up behaviors, it is essential to dive into how they accumulate voting power in different market conditions.

Voting Power Accumulation Patterns in Different Market Conditions

The study of voting power accumulation patterns and the associated dynamics between voters and their locked-up tokens are of paramount importance in understanding how voters use their tokens in upward and downward market conditions. In this regard, the two primary ways in which voters can augment their voting power (VP) are by purchasing additional tokens and locking them up, or by extending their lock-up period, resulting in an increased multiplier. VP calculation is derived from the multiplication of token balance and a multiplier based on the lock-up period.

Voting Power (VP) = token balance * multiplier based on lockup time

However, analyzing the behavior of voters in response to changes in token prices and market conditions presents a significant challenge. It is challenging to determine whether voter behavior is influenced by changes in token locked amounts and lock-up duration or whether they are affected by the daily routines of the average Externally Owned Account (EOA) wallet.

To overcome this challenge, a robust and reliable quantitative methodology has been developed to simplify this complex analysis. This approach is based on a decomposition of the changes in voting power into its constituent factors. By quantifying the changes in voting power (Δvp) over time, an analysis of the two constituent factors, changes in balance (Δb) and changes in conviction (Δc), can be made.

The change in voting power between two consecutive time points, t and t−1, can be derived through arithmetic calculations as follows:

Δvp=vp(t)−vp(t−1)

This equation can be expanded as:

Δvp=b(t)⋅c(t)−b(t−1)⋅c(t−1)=b(t)⋅[c(t)−c(t−1)]+c(t−1)⋅[b(t)−b(t−1)]=b(t)⋅Δc+c(t−1)⋅Δb

Here, b(t) and c(t) represent the balance and conviction at time t, respectively. The terms Δb and Δc denote the changes in balance and conviction from time t−1 to t.

For this study, each timestamp where a new transaction occurs on the blockchain is considered a discrete-time point. This approach captures the dynamic nature of voting power changes with high granularity. To represent the changes in voting power across different voters and time points, a matrix formulation is used. The matrix of changes in voting power (ΔVP) is defined as follows:

ΔVP∈NT×W

Here, T+1 represents the total number of time points, while W indicates the number of voters. The changes in voting power can be represented using the following formula:

ΔVP=B⊙ΔC+C(t−1)⊙ΔB(1)

The formula can be represented in matrix form as:

image1462×642 72.8 KB

The matrix ΔVP represents the change in voting power for each voter at each timestamp, while B represents the balance of each voter at time t, ΔC represents the change in conviction between t and t−1 for each voter, and C(t−1) represents the conviction of each voter at time t−1. The matrices ΔB and ΔC represent the changes in balance and conviction, respectively, for each voter between t and t−1. The symbol ⊙ represents the element-wise multiplication of matrices.

If Δvpt,i is non-zero, the voting power of voter i's wallet has been altered at time point t. A non-zero change in VP due to a change in conviction occurs because b⋅Δc is non-zero. Conversely, a change in balance results in a non-zero value because c(t−1)⋅Δb is non-zero. Typically, these two terms hold non-zero values simultaneously if the user changes the lock-up window and balance in the same transaction.

Here, the notation ||1 denotes the L1 norm, which essentially sums up the absolute values of all elements in the matrix.

In summary, this methodology for governance conviction provides a more in-depth and comprehensive approach to analyzing voting power accumulation patterns in different market conditions. Through the matrix formulation, the changes in voting power across different voters and time points can be represented and analyzed with high granularity, providing valuable insights into voter behavior dynamics.

Accounting for Market Conditions in Curve and Polkadot

In this study, we aim to assess the impact of market conditions on how voters accumulate voting power. To achieve this, we analyze upward and downward trends in the market by employing a combination of short-term and long-term moving averages, namely the 7-day moving average (MA7) and the 30-day moving average (MA30). We define an upward trend when the MA7 exceeds the MA30 and a downward trend when the MA7 falls below the MA30.

It is crucial to acknowledge that token behavior is circumstantial, and varying market conditions may elicit different responses from holders with varying stakes, thereby exhibiting diverse behavior patterns. Therefore, we adopt a nuanced approach, which considers these factors to provide a precise and insightful understanding of how balance and conviction impact the ebb and flow of voting power within the governance systems.

The charts presented above illustrate how the MA7-MA30 differential correlates with the token price. Our analysis leverages these definitions to explore how market trends affect voter behavior, specifically the influence of the token price and lock-up duration on the dynamic voting power within the governance frameworks of Curve Finance and Polkadot.

Findings: Voting Power Accumulation in Curve Finance and Polkadot

Curve Finance: In the case of Curve Finance, we encountered a challenge when studying shrimp voters due to their high number exceeding 12,000 and the high computational complexity of our method. As a result, we adopted a sampling strategy, randomly selecting 2000 shrimp voters in each experiment, and repeated this process 500 times. We calculated the log ratio of conviction impact to balance impact in each experiment and grouped the results by upward and downward market trends. The grouped histograms below showed distinct patterns.

Screenshot 2024-04-01 at 11.18.38 PM1474×698 72.1 KBIn the grouped histograms, we noticed distinct patterns:

During downward trends, the log ratio values were mainly concentrated between 0 and 0.5, displaying a distribution similar to a normal distribution. This suggests that shrimp behavior is more uniform in downward markets, and most log ratios exceeding 0 indicate a tendency among shrimp to increase their lock-up duration to alter their voting power.

During upward trends, the scenario was notably more complex, as three peaks around -0.3, 0.5, and 1 indicate that shrimp behavior is inconsistent during upward markets. However, most shrimps preferred changing their lock-up window, a tendency that was even more pronounced than during the downward trends.

Polkadot: When analyzing Polkadot’s market trends, we observed a deviation from the typical pattern observed in Curve Finance. Instead of a normal distribution, there was a noticeable long tail in the data. Upon closer inspection, we discovered a fascinating insight: a particular group of shrimp voters in Polkadot had a strong inclination towards altering their lock-up window rather than increasing their balance during bullish market conditions. This behavior was particularly prominent and suggestive of a unique pattern among this subset of voters.

Screenshot 2024-04-01 at 11.20.53 PM1332×620 40.8 KB

Conclusion

The research examined voter personas, governance proposals, and voting power accumulation patterns to determine key trends and patterns in voter behavior.

Firstly, the research highlights the importance of financial incentives in driving voter turnout, with Curve Finance exhibiting higher levels of voter turnout compared to Polkadot. Financial-centric proposals such as Gauge proposals and Treasury Proposals make up the majority of proposals in Curve Finance and Polkadot respectively. However, in Curve Finance, it was found that the majority of proposals are initiated by individuals linked to the Curve.fi team.

Additionally, the analysis reveals the presence of distinct voter personas with varying preferences and behaviors. Whales, Sharks, and Dolphins exhibited preferences for shorter lock-up windows. Furthermore, in Curve Finance, the majority of voters locked their tokens in the highest lock-up window. This trend stood in stark contrast with voters in Polkadot. Understanding these personas is crucial for designing effective governance mechanisms that cater to the diverse needs and motivations of token holders.

Lastly, the study highlights the impact of market conditions on voter behavior. During upward trends, token holders in both Curve Finance and Polkadot exhibit a propensity to adjust their lock-up durations to maximize their voting power. Conversely, during downward trends, there is a more uniform tendency among voters to increase their lock-up durations.

This research contributes to our understanding of decentralized governance in blockchain ecosystems and provides valuable insights for the design and optimization of governance mechanisms. Further research in this area will be crucial to ensuring the scalability and effectiveness of decision-making as the blockchain landscape evolves.

Appendix

Table 1: veCRV amount by lock-up period

Table 2: DOT conviction multiplier by democracy lock

Table 3: Top Proposal Address Labels on Curve.fi