This memo is a response to Frisson’s proposal to introduce an ARB staking program as a means of defeating voter apathy and enabling the token’s value on the application layer. We believe the proposal is very well reasoned, complete with a specification for how this would be implemented and a concrete estimate of how much this is going to cost. But there are details that need to be figured out that pertain to the sustainability of such an initiative.
Core Motivation: Arbitrum is a Top Protocol, But ARB is Not a Top Token
The Arbitrum L2 is home to some of the most widely used DeFi apps. Only Ethereum and Solana realistically fall ahead in terms of volume, traction, and innovation. However, this sentiment is not reflected in the ARB token. ARB has one of the lowest MCap/TVL and FDV/TVL ratios across the top chains today.
A lot of this comes down to simple market dynamics. A significant amount of ARB has been unlocked over the past few months, resulting in a YTD performance of approximately -50%. In today’s landscape, investors expect to see some form of value accrual – whether it is direct (revenue share) or indirect (buyback and burn).
L1 PoS chains have a fairly simple path to value accrual – nodes stake tokens and validate blocks on the chain, and in return receive block rewards and transaction fees from user actions. L2s are extensions of L1s, and thus cannot implement the exact same value accrual mechanisms. For L2s, the most clear-cut revenue generation path is via sequencer fees i.e. the delta between the cost of running the sequencer + cost of posting data to L1, and transaction fees collected from L2 users. This proposal seeks to use this sequencer revenue to create financial value for ARB – admirably, with the express goal of improving voter turnout in the Arbitrum DAO.
A lack of financial value, and the perception of being “just another governance token” plagues all L2s today. This would likely be the first instance of an L2 attempting to bake revenue sharing into the native L2 token.
While on the whole a thoughtful proposal, we believe there is nuance to this argument – and thus, there are considerations for delegates to take into account here.
Understanding the Sequencer Figures
Pre-Dencun, L2s like Arbitrum enjoyed high on-chain profits from sequencer operations. Note that we are only able to consider on-chain costs (posting data to L1) and not off-chain costs (compute to operate the sequencer). But things have fallen off considerably in a post-Dencun Ethereum.
In the 30 days before the Dencun upgrade, the Arbitrum sequencer’s on-chain net profit was 835 ETH. In the 30 days after, that figure fell to 588 ETH – still a significant figure, but a 30% drop. However, we observe a continued drop post-Dencun.
LayerZero’s airdrop on June 20 is an anomaly in Arbitrum’s sequencer collections. Without June 20, 2024’s data point, Arbitrum’s sequencer net profit for June 2024 stands at 224.6 ETH.
In the 100 days preceding the introduction of Blobs, the median daily net profit for the sequencer stood at 25.25 ETH per day. In the 100 days after, the median daily net profit was 4.91 ETH per day.
The point of all of this is to draw Delegate’s attention to the fact that past sequencer collection numbers are not all that relevant in a post-blob world. Yes, the sequencer still has significant on-chain net profits MoM. And perhaps it does make sense to implement a staking program funded by sequencer profits. However, it should be noted that the rate of incremental growth to this pool of capital has slowed down and is likely to continue at a similar pace.
So if a version of this proposal is implemented, the community must evaluate a sustainable rate of incentivization taking the decrease in sequencer revenue into account.
Looking beyond purely sequencing revenue for Arbitrum One, Orbit chain fees could end up being a contributor towards a staking program. Even though fee collections on that front are not significant today, there does seem to be a strong BD push to attract more builders to spin up their own Orbit L3s. Between existing Orbit chains like Sanko Gamecorp and Xai, to initiatives like the Gaming Catalyst Program whose KPIs include overseeing the creation of new Orbit chains, there is scope for Arbitrum-native revenue generation stemming from here.
We recognize that initiatives around Orbit chains are still in their infancy, but this is a conversation worth having as sequencer revenue plateaus at a lower run-rate than what was set in 2023.
Impact Assessment
Delegate Incentive Program
Trying to evaluate the impact of a program that hasn’t been piloted or implemented yet is a difficult task that yields speculative results. The proposed staking program’s main goal is improving voter turnout in Arbitrum and getting a larger share of circulating supply to delegate their ARB.
The Delegate Incentive Program by SEEDGov provides us some basis for where to start, having been implemented with the same end-goal of improving governance participation. SEEDGov published their analysis of the Delegate Incentive Program. You can read the entire analysis here, but to summarize:
- Delegate participation rate improved by 12% on Tally and 13% on Snapshot compared to three months prior. Tally votes tend to be more important as they are binding and execute protocol upgrades.
- Total participating voter power decreased by 10M ARB compared to the previous year.
- The DAO must be cautious of increasing quorum requirements as ARB circulating increases. More ARB will be required to pass quorum as unlocks steepen, necessitating measures to incentivize more delegation.
The incentive program shows that there is an appetite to participate with more fervor when there is money on the table. The maximum incentive applicable is roughly 5,000 ARB per month per delegate, which at current prices ($0.70 per ARB) is about $3500. However, one should note that participating entities are the largest delegates in the Arbitrum DAO and not smaller fry. The program required delegates to have 50k+ of ARB voting power and a 60%+ total participation rate for eligibility. Over 97% of all ARB voting power is with delegates/voters in this cohort.
Of course, it’s not entirely apples to apples. A staking program with intrinsic delegation would be an incentive for non-participating ARB holders to allow their tokens to be used in governance. The Delegate Incentive program stands to encourage existing delegates to be more active; this proposal tries to get ARB on the sidelines to enter the arena.
However, a staking program where a commission model is implemented could hit 2 birds with one stone. For example, if 5-10% of total yield went to delegates and the remaining 90-95% went to delegators, this could act as a more organic form of the delegate incentive program – and one that is open to delegates of all sizes.
All in all, the staking proposal and Delegate Incentive program are synergistic and attempt to reach a similar end-goal by targeting different holder cohorts.
Compound Governance Attack
Just this week, Compound Finance fell prey to a governance attack orchestrated by Humpy and the Golden Boys. The proposal sought 499k COMP – or 5% of the DAO treasury – that the Compound DAO would allegedly still own, but would be deployed to the Golden Boys’ goldCOMP product and could potentially be used to vote in the Golden Boys’ interest. This would have given Humpy and the Golden Boys control over a large majority of delegated tokens.
There was a lot of outrage over this, but the proposal was ultimately canceled after reaching an agreement. Ironically, the agreement was for COMP to become a value accretive asset. Compound’s Growth Program has agreed to build out a staking program that rewards holders with 30% of market reserves (DAO’s portion of interest generated for protocol reserves) in exchange for Humpy canceling their proposal.
The incident raised concerns over governance apathy and a lack of stakeholder alignment in DAOs – something that is very relevant to the ARB staking discussion. It’s now clear that governance apathy can have a serious detrimental effect for DAOs. Compounds’ proposal 289, which is Humpy’s initial proposal, saw a total of 42 addresses vote; 23 for and 19 against.
Approximately 1.315m COMP was used to vote on the proposal, representing about 13% of total supply or 16% of circulating supply (per CoinGecko). A total of 2.64m COMP has been delegated per Tally as of the time of writing. This number likely reflects a significant increase in the aftermath of prop 289.
13% of total supply voting is not a bad turnout per crypto governance standards in DeFi. However, there were many larger delegates who didn’t vote – which is the real issue here.
One reason some of the larger delegates did not vote is because Humpy spread out their total tokens across a number of addresses. The tokens were used to vote and achieve quorum at the last minute, which minimized the DAO’s ability to launch a counter. However, if these delegates were still active and voted no, Humpy would not have been able to get the proposal passed.
Efforts to incentivize delegates to be more active definitely look like the need of the hour. And growing the delegation pie is also important, which is why pairing delegate incentives with “delegation incentives” (for ARB holders that are inactive in governance) seems sound.
The ENS Solution
In April 2024, researchers that are part of the ENS DAO identified governance attack vectors similar to the one carried out against Compound. Their mitigation response included setting up a new delegate that could only vote “no” on governance proposals. This delegate address is backed by ENS tokens owned by ENS Labs and affiliated people.
It’s a simple solution to the problem – use tokens in the DAO treasury only to ensure malicious proposals are stopped in their tracks. However, this does not address larger issues around governance tokens. Such as the apathy around the entire process and finding a long-term solution to grow a stimulated community base that is active in protocol governance.
The veto delegate could also take decisions that are subjective, in the case of contentious proposals that some may consider malicious and some may consider genuine. How do you objectively ascertain malicious intent or DAO detriment before casting a vote? The ENS veto delegate was set up to be controlled by a multi-sig of at least two people, which makes it prone to bias. While currently the veto delegate has 3.99m ENS of voting power, the plan is to build a more robust system where the DAO’s entire 10m ENS can be put to use in governance.
If there is significant concern over something similar happening with Arbitrum, the DAO should consider the viability of implementing something similar to provide near-term relief from the idea of a governance-based treasury raid.
Constitutional Provisions
One thing to note is that the existence of the Security Council, and the rights conferred to the council by the Constitution, can also prevent meaningful governance exploits.
If the Security Council has the ability to use the $2b of ARB in the treasury to thwart objectively malicious proposals with no real community approval, then situations like what happened at Compound can be averted.
Relying on the Security Council is not a long-term solution either. In an ideal situation, there is robust governance participation and these situations are organically quashed. We want the system to be naturally resilient to the highest degree possible, without the Security Council having to intervene. So the idea of delegate and delegation incentives are still relevant.
Porting Insights for Arbitrum
In reality, the cost of a governance attack is directly related to participation in the system. If quorum is 4% and only 6% of tokens are active in governance, then the cost of an attack is – at most – slightly above 6% of token supply. If tokens are liquid and can be easily purchased and sold, or even borrowed, then it exacerbates the situation by making the attack more accessible.
Another factor to consider is the sheer size of the Treasury. If Compound with approx. $480m in the treasury was subjected to such an incident, Arbitrum with $2b+ definitely is. Arbitrum requires 103m ARB (treasury) or 172m ARB (core) to hit quorum for proposals on Tally. 312m ARB have been delegated in total, resulting in a delegation/quorum ratio of 3 and 1.8 for treasury and core proposal respectively.
As mentioned above, ARB’s lengthy unlocks will increase circulating supply and thus quorum requirements as well.
Over the last 30 governance proposals on Tally, Arbitrum has had a mean of 164m votes and a median of 160m. Since the majority were treasury proposals, the average deviation of the median over required quorum is 55%. This is not terrible, but it is cause for some concern given the expectation that quorum will continuously increase.
There are two separate realities here. The first is that governance apathy does not come to be in isolation. It happens when there is apathy for the product too. Compound’s TVL never recovered after the blow off in 2022, while competitors like Aave have forged ahead. Arbitrum is very much in the crypto zeitgeist, and it’s difficult to see how even large ARB holders who are not active in governance could sit on the sidelines when a $2b+ treasury is potentially under attack.
The second reality is that you don’t want to play with smaller margins, especially when we know the voting quorum is going to increase. Growing the pool of delegated ARB is definitely a matter of concern. And while there are band-aids that can help mitigate these risk factors, more long-term oriented solutions are required.
We believe the substance and reasoning behind the staking proposal is valid and legitimate. We also think the call for staked ARB to exist in LST form is valid, as keeping token liquidity hostage in the name of governance only stands to birth protocols like Convex that manage duration. However, there are still open questions around implementation. For example:
- Implementing a commission model for delegates could act as delegate incentive + delegation incentive. What is the right split to ensure adequate incentivization?
- What is a sustainable yield that can be promised given the tapering off in sequencer revenues? Would ARB emissions be on the table?
- What other revenue sources can be tapped into for future incentivization? Orbit chain fees?
- Does the Arbitrum DAO have the ability and appetite to set aside ARB to delegate across the top delegators to vote on behalf of the Foundation/DAO’s mandate?
While staking could be lucrative for DAO governance, ARB holders, and the general perception around ARB and L2 tokens, there needs to be a clear-cut economic roadmap to ensure this isn’t done in a haphazard manner.