ARB Staking: Use Surplus Fees to Align Governance

Constitutional / Non-Constitutional



We propose to implement an ARB staking mechanism that uses 50% of future surplus sequencer fees to improve the economic security of the DAO and incentivize more active participation in governance. ARB holders who stake and delegate their governance power to an active governance participant (either themselves or another entity) will receive surplus sequencer fees proportional to their share of total ARB staked.


The Arbitrum DAO has accumulated over $50M of ETH in surplus fees from Arbitrum One and Nova. Meanwhile, only about 10% of the circulating supply of ARB (roughly $324 million of ARB) is actively used in governance (see “delegated tokens” on Tally’s delegation page). The number of unique voters participating in the DAO has been steadily declining since DAO launch. As a result, it is becoming economically attractive to attack the DAO treasury. The upside of attacking the DAO treasury is increasing (more ETH in the treasury), while the cost of attacking the DAO is not increasing proportionally to defend against attacks. A more developed version of this dynamic exists in the ENS and Compound DAOs, both of which are actively fighting off governance attacks (ENS documented here).

ARB Staking can improve the Arbitrum DAO’s security by incentivizing ARB holders to be actively engaged in governance. Instead of accumulating ETH into an onchain treasury that inevitably becomes a honeypot for governance attacks, ETH is used to reward ARB token holders who are contributing to the long-term health of Arbitrum.


This proposal contributes to Arbitrum Community Values by making the Arbitrum DAO more sustainable and secure. ARB staking addresses an emerging economic vulnerability in the Arbitrum DAO by perpetually incentivizing ARB token holders to engage in effective governance.

Specifications and Steps to Implement


We propose that 50% of surplus sequencer fees be allocated to ARB stakers. The existing DAO treasury of ETH from surplus sequencer fees would be unchanged, only future fees would be affected.

If we assume a 1:1 conversion of the current delegated ARB (~324M) to staked ARB and assume 12K ETH (roughly the historical annual rate) in annual surplus fees accumulated, then at an ARB price of $1 and an ETH price of $3800 this would equal to roughly a 7% annual reward rate.

  • Annual surplus fees in USD: 12,000 ETH * 50% * $3,800/ETH = $22,800,000
  • Staked ARB in USD: 324,000,000 ARB * $1/ARB = $324,000,000
  • Annual Reward Rate: $22,800,000 / $324,000,000 = 0.07 or 7%

We expect all of the variables involved in the reward rate to evolve significantly over time. In particular, we think it’s important to consider the following items because they may have a material impact on rewards:

  • Currently, there is no direct incentive to delegate ARB tokens. ARB staking introduces a direct incentive to stake and delegate. As a result, it’s reasonable to assume that the amount of staked and delegated ARB will significantly increase over time.
  • In conjunction with the recent EIP 4844 upgrade, Arbitrum recently dramatically underpriced its gas fee. As a result, the amount of surplus ETH accumulating in the DAO treasury over time has decreased significantly. We believe this situation is temporary and is likely to be addressed by increasing the gas fee in the near future, but it’s worth noting that at this time the rate of surplus fee accumulation is significantly lower than the historical norm.

Arbitrum staking specification

Our implementation of Arbitrum Staking builds on top of Unistaker. In addition to customizing Unistaker for Arbitrum’s governance architecture and fee collection mechanism, we add features designed to improve governance outcomes:

  • We require that tokens be delegated to an active delegate.
  • We direct 1% of surplus fees proportionally to delegates, with both fee streams going to the token holder if they are delegated to themselves and active.

We define an active delegate as a delegate with a Karma Score over 80. We use an oracle to represent Karma Score within the staking system. Receiving an accurate Karma score requires delegates to sign a message and post it on the forum as a way to prove that the forum user is truly the owner of the delegation address. Karma Score is calculated according to the following formula:

((100) * ((Forum Activity Score * 1) + (Off-chain Votes % * 3) + (On-chain Votes % * 5))) / (Sum of Weights times Max Score Setting * 1)

Our staking implementation directs 1% of surplus fees to delegates to sustain the ongoing growth and development of Arbitrum, a large portion of which is driven by the DAO. In its current form, ARB staking is not meant to replace the delegate incentives program, but rather to complement it with a smaller source of incentives that are programmatically tied to ongoing network operations. In the future, we believe it could make sense to integrate the delegate incentives program with ARB staking so that, instead of getting delegate incentive funds from the ARB treasury, they come directly from DAO revenue.

Tally will build ARB staking into our existing Arbitrum DAO platform, so that users can easily stake and delegate in one place.

Integration with future Arbitrum staking systems

We anticipate that additional Arbitrum staking systems will be developed over time, perhaps to incentivize decentralized block production in BoLD or to create an efficient MEV market in Time Boost. We view multiple staking systems as complementary. Each system would ask the staker to do different work for Arbitrum, take different risks, and pay out different rewards. Having multiple systems lets ARB holders pick between different risk/reward payoffs and specialize in different types of work to secure the system.

Liquid staked ARB

Tally is building a governance-compatible LST to enable staked positions to participate in governance while being liquid. By staking ARB on the Tally Protocol, users will receive a liquid staked ARB token that auto-compounds rewards, is restakeable, and is compatible with DeFi. Tally’s ARB LST automates the process of accessing rewards for token holders by automatically claiming staking rewards and redepositing them back into the staking contract. We plan to work closely with Arbitrum ecosystem DeFi protocols and restaking systems to enhance composability and utility of the Tally ARB LST. The ARB LST will be an exchange token, which means the staking rewards will accumulate by increasing the exchange rate of the token (rather than by increasing the account balance).

Estimated Timeline

Post proposal on forum for feedback: June

Post temp check proposal on Snapshot: July

Begin smart contract development (if temp check passes): July

Submit smart contracts for audit: August

Submit onchain proposal on Tally including full ARB Staking implementation and retro funding: September

Overall Cost

This proposal will not be submitted onchain until ARB Staking is completed and audited. At that time, we will request a retroactive payment of 100,000 ARB to help cover the costs of development.


This proposal should not be relied on as legal, tax, or investment advice. Any projections included here are based on our best estimates and presented for informational purposes only.


This is quite interesting, and exactly how I thought Uni should have approached the fee switch problem, which is: you get a yield if you delegate to an active party. I really love this proposal.

I have a few questions here, that could be idiotic but maybe are worth exploring:

  1. I see a potential fringe case: there is a delegate, with maximum karma score, that goes away from his delegates duties without coming back. He will likely have a score above 80 for a while, and so people would delegate to him/her based How long is this while last?
  2. Has there been any thought on protocols who do vote (often but not always), without providing rationale of vote? I can see several arbitrum protocols not reaching karma but being currently target of delegation from their users. Likely this protocols might not have as of now time or energy to actively engage with the dao, or at least to engage up to a level that takes them above threshold.
  3. Has there be any thought about a potential gamification of this mechanism, such as giving users that delegate a slightly higher yield/bonus if whom they delegate to has a very high karma score? maybe with a karma score measure not for the very long term, but just on a shorter span of time (ie: 30 days). Which means: you get yield if you delegate to someone with a karma higher than 80, but if him/her reached 110 in the last 30 days (and maybe we measure this is a slightly different way) you get a small bonus?

This is a very very interesting proposal thank you Frisson!


I believe that if this system is to be an integral part helping secure Arbitrum’s governance, it is important that it is easily verifiable, secure and decentralised.
Is it possible for me to easily recalculate Karma Scores locally in my own machine to ensure that rewards are being distributed according to the specification? This would require software.
What oracle will be used to relay these scores to the staking system? Will there be some economic backing to this oracle? Will it be slashable? What if it goes offline?

To all delegates regardless of criteria? ie all addresses that have tokens delegated to them?
I assume proportionally given the amount of tokens delegated?


The long-awaited proposal, thanks!
I think it should be possible to earn a commission regardless of the base commission.
Let me explain:
This creates a conflict of interest:

  1. Users are interested in keeping the commission to a minimum.
  2. On the other hand, ARB stakers are interested in increasing commissions

Also, I support JoJo’s questions (although point 3 is controversial, which could lead to all users going to one delegate, this needs to be discussed, but the impulse itself is true)


I really like the underlying goal of trying to fend off governance attacks, increase ARB utility, and encouraging active participation in governance. However, I see some problems…

  • If we tie the token to insignificant revenue values, we introduce the risk of people valuing ARB relative to its revenue generation and could face compression of multiples. We see this in the perps landscape, with governance tokens that share revenue trading in line with the yield they generate. Even if we ignore the drastic decrease in DAO revenue since ArbOS Atlas 20 / 4844, $25M (50% of last years revenue, which is still overstating it because Base fees made up a significant portion of revenue over that period) on an $8B FDV token doesn’t look great… I understand it’s only on delegated ARB so it would look more attractive, but these dynamics will change quickly if there is yield attached to delegating tokens. If we do see price come down as a result of this proposal, it could actually make it cheaper to conduct a governance attack while also hurting the value of the DAO treasury.

  • Is it actually a good idea in practice to tie delegators/delegates yield to sequencer revenue? This could misalign incentives - fees will likely creep higher and higher over time as voters/governors/delegates look to increase their yield by increasing fees, which could lead to a worse UX and developer experience on Arbitrum.

  • We are about to “spend” half our ETH holdings on BoLD validator bootstrapping, which will also require a 4% yield on 3600 ETH bonds into perpetuity. Additionally, we need a treasury management program to generate passive yield on DAO holdings to achieve long term sustainability, and this proposal does the opposite. It continues to drain the DAO of its resources rather than build it up over time (Mantle is an example of an L2 that has done a superb job on this front).

  • Are we sure the Karma score can’t be gamed in any way? We need to be 100% sure on this one, as it will become an integral part of Arbitrum governance if implemented and is very novel in nature.

  • In general, DAO expenses are growing while revenue declines. We need non-native assets to weather a bear market to ensure we aren’t forced into selling ARB at depressed prices.

I think this proposal is just too ahead of its time. If we had $200M of stables and ETH in the treasury, I’d be all for this. However, I believe that Arbitrum is still too early in its growth stage to consider revenue share with token holders / delegators.


The proposed solution does represent a meaningful step forward in terms of incentive alignment between token holders, governance participants and Arbitrum, while continuing to enhance the economic security and sustainability of the ecosystem.

The proposal also does a great job in its formula proposition of Karma Score, which (in the event of this proposal passing) likely becomes an attack vector increasingly prone to manipulation and gamification. However, the proposed solution effectively combines both on-chain and off-chain metrics and selects those that accurately reflect, represent and measure the desired activity and scope of involvement.

On all levels, it is clearly a win-win proposal and outcome for everyone involved.

The only aspect which remains worth unpacking and exploring in greater depth is:

  • How should the surplus fee be distributed in a legally compliant manner that does not increase legal and regulatory risks for Arbitrum?

This is arguably one of the greatest reasons why better value-accrual and incentive mechanisms are not more widely adopted across the industry. Hence, it would be worth elaborating on the potential design of the surplus fee distribution and how it takes these aspects into account.


^This should be terrifying for anyone who cares about Arbitrum succeeding long term. The treasury grows, but the cost to capture it is rapidly going down. Something is drastically wrong here and Frisson has the right solution. Right now every ARB holder has to decide: participate in governance or participate in the economy of the network. We desperately need a way to combine these things. I think there are other avenues to pursue as well, such as custom vaults on lending, AMM, or CDP protocols like Open Dollar, where users can still delegate their tokens when they are deposited and take loans against delegated ARB.

That said, myself and the team at Open Dollar are strongly in favor of this proposal and the proposed timeline.

I would suggest the rewards math and calculations go through ARDC and be voted on separately from the rest of the process.


This opinion is solely my own and does not represent the views of SEEDGov, with whom I collaborate as an Advisor for LTIPP.

hey @Frisson thank you very much for the proposal. Overall, I like it and agree with the general principle that using DAO revenue to incentivize participation in governance is an excellent idea and should be implemented.

The problem I see, and I agree with all the points raised by @swmartin , is one of timing. It seems a bit premature, as the DAO is not yet mature and does not have a clear revenue flow to distribute.

I think it hits the mark in pointing out:

As shown in the graph shared by Entropy, the revenue from the sequencer has dramatically declined since the approval of EIP-4844 (note that the spike at the end is due to the LZO airdrop):

At this moment, the DAO does not have a defined plan on how to spend the ARB and revenue to maintain a healthy treasury for the medium and long term. Therefore, I believe the DAO should stop approving large expenditures (in this case, 50% of the profit is being requested) until a treasury manager is appointed or a treasury management strategy is defined.

On the other hand,

Although it is true that the number of unique voters has declined, the amount of $ARB delegated for voting has remained stable. I don’t say this is positive, as the supply has increased significantly in recent months, so the expectation would be that $ARB participating in governance should increase by the same percentage.

However, I disagree that it is urgent or that it has become “economically attractive” to attack the DAO’s treasury. In the latest tally proposal, which was relatively controversial (due to heated last-minute discussions), 213M $ARB and approximately 15,900 addresses voted.

So, if we take that recent vote, the threshold someone would need to consider to attack the DAO is 213M $ARB (and I believe a realistic estimate would indicate that the value is higher, as in the event of an attack on the DAO, more $ARB that is usually not active in governance would likely vote).

What I mean by this: I agree with the proposed value and the idea of using sequencer revenue to encourage participation in governance. In the medium term, this should be something implemented. However, I don’t think it is urgent, and the DAO is not yet mature enough to handle this expenditure.

Finally, I know this is not the main point, but I disagree with this narrative. I do agree that using the revenue to incentivize participation in governance is positive and will increase the security of the DAO. However, I disagree with the notion that there are no incentives to delegate the ARB token. In fact, the only incentive that exists today for holding it is to participate in governance or delegate that participation. I believe we need to strengthen this aspect.


Thanks @Frisson for the proposal. We are very excited to have this discussion around potential solutions to tackle the lack of the delegated tokens even for one of the most active governance-equipped DAOs.

We are aware of some issues pointed out by other delegates, but overall, in favor of improving it to be a feasible proposal for the DAO to experiment on. The most important thing that we believe we should try is to introduce a mechanism aligned with the values, effective to work for the price appreciation and governance participation and hard to game on.

We are curious how we can avoid having centralized delegates get staked too much. Just relying on Karma Score over 80 wouldn’t prevent token holders from staking their tokens into “popular” ones. For example, on popular explorers like Keplr for Cosmos DPoS, their UX suggests to avoid delegating tokens to Top X validators. Can we consider introducing such an improvement for more decentralization?


Thanks for your thoughtful comments (as always) JoJo.

  1. Good question. It depends on the delegate and how high their score was prior to the inactivity. We could try to model out different scenarios using the Karma Score formula.
  • ((100) * ((Forum Activity Score * 1) + (Off-chain Votes % * 3) + (On-chain Votes % * 5))) / (Sum of Weights times Max Score Setting * 1)
  1. I personally haven’t thought a lot about this yet. My first instinct is to say that, with the Delegate Incentives Program, it should be possible to hire folks to help with this. I feel like Treasure is doing a great job with this, for example.

  2. I like this thought. I think there is a lot of opportunity to tweak the mechanism to optimize governance, especially after it has been on for awhile and we have some historical data to work with.


These are great questions. I’ll work on documentation that explain in more detail how this part of the system in particular will work and share it here on the forum. I think we can get to a reasonable level of decentralization and security. With that said, it will never be as decentralized and verifiable as purely onchain data. I personally don’t think using onchain voting only is sufficient to capture delegate participation, because it’s too easy to game/automate.

This is proportionally given the amount of tokens delegated, and is also subject to the activity requirement discussed above.


Appreciate the thoughtful response, Sam. A few follow up notes:

  • I can’t speak to how the market will view ARB Staking, but I will say that I personally disagree with the framing you outlined re: valuation. I think of ARB Staking as a valuable starting point for ARB utility. I anticipate additional use cases and utility for ARB in BoLD and Time Boost. Also, ARB Staking in and of itself should open up new use cases and utility that drive demand (e.g. restaking).
  • I personally disagree with your framing re: delegates and fees optimization. I think delegates are just as likely to see the long term benefits of market share for Arbitrum as they are to optimize for short-term yield maximization up to a point that drives excessive price elasticity for users.
  • I would say that DAO treasury diversification is a separate problem and should be addressed on its own merits. I don’t think accumulating a ton of ETH via fees in the treasury without a fee distribution mechanism is good for the DAO. In fact, it has shown to be quite dangerous.
  • I’m going to share a more detailed spec of the Karma implementation per @fred’s comment. I personally think it is robust, but look forward to further examination with you and others.

Thank you for your thoughtful analysis, @pedrob

I addressed a few relevant points in my response to Sam here: ARB Staking: Use Surplus Fees to Align Governance - #12 by Frisson

I disagree with this framing on multiple levels. First, I would say that $213M is small compared to the potential upside of attacking the DAO, which controls all of the TVL on Arbitrum One and Nova, in addition to the $50M (and growing) of ETH in the treasury. Second, it is difficult to defend against such an attack in the manner you describe because once an onchain proposal voting period starts, it is no longer possible to delegate tokens that vote on that proposal (you have to delegate before the voting period starts to participate in the proposal).

I’m not sure I understand this point. Where does the current incentive to delegate come from? Delegating your ARB tokens does not come with any rewards that I’m aware of. On the other hand, it does take time (especially if you are doing it productively and pay some attention to what your delegate is doing).

1 Like

Thanks, Tane. We could certainly implement something like this in the UI layer on Tally. That’s a great idea. An additional option would be to introduce smart contract mechanisms that amplify or reduce rewards based on voting power concentration, which I think could be powerful but are more complex to predict and get right.


The offer is nice but I don’t see sustainability because Dao revenues will not be able to subsidize this payment in the future. So, I think this proposal should be based on the expectation of future Dao income rather than the current accumulated Dao treasure. For this, the fee rate that will go to $ARB stakers in the network can be determined. Current L2, L1 fees should be investigated and an additional fee such as a Staker fee may be added to bring the Arbitrum network closer to the median.

Arbitrum doesn’t need to be the cheapest L2. The important thing, according to Game Theory, is to create a system where projects, teams, VCs and $ARB investors can consistently gain profit in the long run. However, according to game theory, I see no advantage for investors and project is being funded with investors’ money.

1 Like

I really like pairing the idea of pairing the staking mechanism with an increase in governance participation. Governance attacks are rarely considered but this is a great way to get some of that voting power back from dead governance participants and increase Arbitrum security.

Although I believe the fees that Arbitrum made so far do not represent an incredibly large sum of money and yields on redistribution may not be very attractive at first its good to start planning a method of distribution which does more than just rewarding pure staking. (Hopefully with fee increase we will see a much bigger fee intake)

I’m not sure using karma score for cutoff is the best way to cut off non-active delegates as it greatly favours older active participants and could mean an even greater concentration of power amongst the few already top delegates. The formula should be revised a bit to skew the score towards active participation in a shorter time span (prev 6 months?)

1 Like

Something that is really exciting to me about this proposal is that it allows us to start building the infrastructure that can let other people reuse the security of the ARB token for other applications.

When paired with Symbiotic for example, we can create a collateral type from ARB that can now be used in Vaults for restaking.

This allows people to build really cool things:

  • Arbitrum security backed Oracle service
  • Shared Sequencing models with slashing for bad behavior
  • Faster decentralized bridges secured by ARB token.

Personally, I’m not so worried about the cost to the ARB DAO, as noted, currently the income from the sequencer is quite low (but I expect we will probably want to reevaluate this because we’re far far cheaper than Base at the moment). Also, this would be for revenue going forward, I don’t think this proposal suggests using half of the ETH the DAO already has.

This system helps get ARB holders to lock their tokens up for a long term productive use. This is positive for ARB holders, and the ecosystem, as our supply is expanding too fast, IMHO, and this is a net positive method for withdrawing ARB from the market.

As a builder, I’m personally most excited about creating an entirely new infrastructure layer on which to build tooling on Arbitrum. I see it more an investment in expansion of the Arb tech stack and market reach than a cost.


I wanted to give everyone a heads up that I’m going to be away from my keyboard Sunday afternoon through Wednesday afternoon MST of this upcoming week. If I don’t respond to a forum comment or Telegram message promptly, I will get to it on Wednesday. Please also feel free to reach out to @dennison of Tally, who has full context on the proposal. Thanks Arbitrum fam, I appreciate all of you.


Sounds interesting…
This mechanism not only rewards stakeholders but also addresses the issue of declining voter participation, thereby strengthening the DAO’s defense against potential attacks and promoting long-term ecosystem health.


Overall, I think this is a solid strategy that enhances economic security and encourages more participation in Arbitrum DAO governance.