Onchain governance for decentralized autonomous organizations (DAOs) was one of the earliest purported use cases for crypto. However, as the landscape and penetration of DAOs has shifted over the years, participation still appears limited to a core group of users. We find that there is little diversity in active voters relative to the number of token holders despite its globally accessible nature. 

All major governance tokens we examined had extremely high levels of concentration, though we found the more pressing issue to be the participation rate. Token concentration distributions, while skewed, were mostly in line with those of other crypto assets like USDC and wrapped ether (WETH), neither of which have voting rights. However, most onchain voting is currently performed by a few dozen delegate voters despite hundreds of thousands of token holders. More importantly, forum participation within earlier stages of the governance process also appears fairly limited to only a few hundred users (or less) per year. 

Beyond participation, DAO treasuries also remain largely dilutive in nature. Most treasuries are composed of the native governance token itself, even though expenditures and grant requests are mostly denominated in USD. This dilutive effect may diminish the face value of many large treasuries in our view. That said, new governance structures are aiming to improve DAO decision making abilities, and recent regulatory trends have given DAOs legal standing in some jurisdictions. We think these are constructive for the future of DAOs, though the challenge of actively engaging more of the token holder base remains.

A Refresher on Onchain Governance

Protocol governance follows a multistep process, much of which actually takes place offchain. Although the specific process varies across protocols and DAOs, the major phases are encapsulated below:

• Forum Discussion: New governance decisions typically begin with a post to a public forum, aiming to gather widespread community feedback in a transparent manner. In some ecosystems, like Uniswap and Aave, this process is known as a Request for Comments (RFC). 
• Temperature Check (Optional): After incorporating feedback from the forums, some protocols opt to use an offchain voting tool to gauge token holder sentiment before implementing any protocol changes onchain. This is commonly done via a service like Snapshot[1], which requires linking to a crypto wallet to sign offchain messages. Benefits of temperature checks include gas-free voting, no smart contract overhead, and more flexible voting options. Some DAOs may iterate via forum discussions and temperature checks multiple times, while others may bypass this step and proceed directly to onchain voting.
• Onchain Voting: Onchain voting is the core governance action that takes place onchain. The type of vote, vote options, and potential code executions are all assigned at the smart contract layer by a proposer (who is typically required to hold a certain amount of the relevant governance token). Token holders and delegates can then vote via a dedicated protocol governance portal or via governance aggregators like Tally or Boardroom[2] (which link to the same onchain voting contracts). Oftentimes, there is a time lock between when the vote is first scheduled onchain to when voting begins to ensure sufficient preparation time for stakeholders. 
• Execution: Once the vote is passed, there is usually an additional waiting period before the executed code changes go live to better protect against potentially erroneous or malicious actions.

Throughout this process, the role of Discord (and also Telegram) cannot be understated, particularly for DAOs with smaller communities. Numerous discussions and sometimes even temperature checks take place on Discord servers. That said, governance processes in principle aim to be as inclusive of various community stakeholders as possible across all stages, though this can sometimes lead to platform centralization – particularly for smaller groups without the means to build redundant tooling.

Variations in the governance process partially originate from differences in the implementation of the governance contracts themselves. Many protocols on the Ethereum Virtual Machine (EVM) leverage OpenZeppelin’s Governor contracts as their core governance framework. Prior to its August 2021 release, however, the GovernorAlpha and subsequent GovernorBravo frameworks created by Compound were widely used. (Uniswap leverages Governor Bravo, for example.) Other early protocols like MakerDAO, now Sky, also created customized governance contracts that they continue to use today.

More recently, alternative governance frameworks like Aragon, Moloch, and DAOstack have also gained traction among EVM protocols. On Solana, the Solana Program Library (SPL) provides a standard governance framework, though other frameworks like Tribeca are growing in usage too. The dispersion of governance structures makes it impossible to draw perfect parallels across protocols, and the programmatic scope of governance powers also varies. Some protocols limit governance change to parameter updates, while others allow full contract upgrades. Certain DAOs are focused on treasury spending, and many enable some combination of the above. That said, we think the differences in technological “plumbing” currently matters relatively little to investors compared with participation rates and other forms of perceived utility such as future revenue sharing.

(Un)Equitable Token Distributions

At the core of most onchain DAOs is a governance token that represents a weighted voting mechanism where one token is worth one vote. (Though there are variations like quadratic voting or reputation-based one person one vote systems.) Given that many DAOs operate in a direct majority decision making system, the concentration of token holders can be important. In particular, if an active cohort of tokenholders can coordinate a majority of actively voted stake, there is a possible “tyranny of the majority” threat where a small number of individuals can pass proposals that may only benefit themselves. 

Indeed, there have been multiple scenarios where communities were concerned that large investors and key figures controlled a disproportionately large amount of tokens. (Similar to challenges with the “big three” shareholders in traditional equities.) Beyond these concerns, full fledged governance attacks have actually occurred, similar to more traditional “hostile takeovers” where entities directly acquire more vote tokens. The programmatic nature of DAOs and the existence of flash loans has even led to more malicious takeovers via temporarily inflated (borrowed) share count due to a lack of governance guardrails. 

Both of these problems – outsized influences and hostile takeovers – can theoretically be remedied by more equitable token distributions that allow for a better balance of power. However, most governance tokens follow extremely high levels of concentration. Chart 1 shows the Lorenz Curves and Gini Coefficients for several major governance tokens on Ethereum with wrapped ether and USDC and as baseline references. The Gini Coefficients are high on an absolute level, and less than 0.1% of total holders own more than 50% of the total supply for all tokens measured).

Critically, however, governance token distributions don’t significantly deviate from other token distribution curves. In fact, the governance token distributions for many large protocols are more equitably distributed than USDC and WETH (both of which lack voting rights). We think these distribution skews result more from the nature of crypto assets generally, rather than from impending attempts at governance attacks. That is, the absolute concentrate levels of these holdings aren’t directly indicative of imminent governance pitfalls. 

Despite a small percentage of holders controlling the majority of the supply, the number of addresses holding each token is quite large. This makes majority collusion non-trivial (though still feasible) in our view. MKR, for example, has 57.6K addresses controlling $10 or more of the token on Ethereum, with the top 18 addresses needing to collude to control more than 50% of the supply. (Note that the “real” number of holders needing to collude is likely greater than 18 given that some of the top holding addresses are vaults with multiple owners.)

Furthermore, many DAOs also have countermeasures for unwanted behavior similar to traditional equity structures where poison pills and  staggered boards of directors can be used to defend against hostile takeovers or other unwanted attention. Onchain equivalents include programmatic timelocks, quorums, veto abilities, or emergency security councils. 

The Crux: Participation and Activity

While 51% collusions of the total token supply is not an imminent issue in our view, we think relatively low levels of participation – combined with a small quorum threshold – do pose greater challenges today. For example, a vote on Aave requires 320K tokens voting for the winning option in order to meet quorum. That is only 2% of the 14.9M circulating AAVE supply and is much easier to influence by a small number of well capitalized individuals. (Note that most quorums are based on the number of tokens voting, not on the number of voting addresses.)

The quorum threshold itself has also previously posed challenges. Many of the large DeFi protocols including Aave, Compound, and Uniswap have all faced situations where non-contentious votes failed as a result of low participation rates. Today, this has been largely resolved by the implementation of delegation functions where less involved token holders can delegate their voting power to a trusted representative that is more involved in ongoing operations. For protocols with a highly active governance process, this has been instrumental in enabling a core community to drive active decision making that is inaccessible to most token holders. For example, Compound has had more than 150 onchain votes YTD (more than 3 per week on average) and more than 90% of the votes have had less than 30 participating addresses. 

Beyond participation in onchain voting (the last step of the governance process), we also think participation in protocol forum participation is equally important to track. In addition to being earlier in the governance process, forum discussions lack a delegation mechanism. That is, each user on a forum is likely an active stakeholder in the protocol, which better highlights the size of the core decision making community. Based on forum data indexed by x23.ai, we find that forum activity tends to vary considerably between protocols. See Chart 2.

Interestingly, protocol forum participation may also reflect changing narratives and industry focus. Throughout the previous bull cycle in 2021-2022, MakerDAO (orange) had the most active forum discussions by a wide margin, with an average of four to five active topics each day. More recently, however, DAOs focused on layer-2s (L2s) such as Arbitrum (blue) and Optimism (red) have led forum activity. We think this is due to the relative nascency of their governance communities as well as their critical role in Ethereum’s scaling roadmap. 

Even as MakerDAO participation has dropped, however, community participation in Aave (purple) began to meaningfully increase at the start of 2023, flipping MakerDAO in daily active topics. The surge in Aave activity preceded Aave’s GHO stablecoin launch, which was also later followed by Aave flipping MakerDAO in total value locked (TVL). While not necessarily a direct result of more involved community activity, we do think that increased community activity is generally a positive signal.

Beyond forum topic count, we think the number of unique authors also provides insight into the size of each protocol’s core community. Most protocol forums only have 100-300 core participants each year despite having tens or hundreds of thousands of token holders onchain. Chart 3 shows the number of distinct authors for each protocol by year (including post initial topic posts as well as replies). Here, the surge of L2 governance activity since 2022 is also clear. Meanwhile, protocols like Uniswap have actually seen the overall number of forum participants drop.

We think the relatively concentrated number of both voting delegates and forum participants speaks to the natural accumulation of influence into a handful of actively involved parties. Of the above protocols studied, the number of forum participants in 2024 YTD represented between 0.1-0.6% of the total token holders and the number of final voting delegates was often an order of magnitude lower than that. 

The Treasury Quandary

Beyond major protocol upgrades, perhaps the most critical power of a DAO is their ability to direct treasury spending. Most treasuries, however, are composed primarily of the governance token itself. For example, Uniswap has the largest onchain treasury which is 99.9% its own UNI token. Meanwhile, Lido and Ethereum Name Services (ENS) treasuries are 49% and 88% their own tokens respectively. See Chart 4.

This creates a unique mechanism where a large portion of treasury “spending” is more akin to dilutive token issuance rather than cash reserve expenditures. Indeed, grant requests are often denominated in USD in order to justify cost breakdowns, and the entities receiving the grants often need to immediately sell (or hedge) the received tokens to fund the external costs associated with the proposal. In our view, this diminishes long term importance of native-token treasury assets as they are mostly dilutive in nature.

Thinking About DAOs

Despite the distributed and global nature of onchain governance, centralized points remain, both in terms of technology and participants. What makes DAO tooling unique, however, is the multiple platform interfaces that prevent monopolistic chokeholds from forming (e.g. Tally and Boardroom for onchain voting). Furthermore, the open nature of its forums gives a voice to all motivated stakeholders. 

Still, the number of these motivated stakeholders remains far lower than the number of total token holders. Less than 1% of token holders are active on forums (for large protocols), with even fewer participants in the final voting process. Much of the final voting responsibility has been aggregated into the hands of a few delegates. 

Other shortcomings persist, such as the dilutive nature of most DAO treasuries and the time lag for decision making. Newer challenges have also surfaced, such as the tension between equity holders for the “Lab” (developer corporation) and the token holders for the “Foundation” (decentralized governance) with sometimes murky distribution of power between the two.

That said, continuous improvements and experiments are being made with governance. Experiments with quadratic voting (Gitcoin) and market based voting (Futarchy) could lead to better decision making, and improved legislation could broaden the legal standing of DAOs. Indeed, states like Wyoming already have regulations giving DAOs similar rights as LLCs. At its core however, it remains to be seen if the permissionless nature of DAOs can ultimately lead to new breakthroughs in expanded participation and improved decision making in the long term.