CIP-Draft: Simplifying the operations of the CoW DAO bonding pool

Governance
1 
CIP: Draft
title: Simplifying the operations of the CoW DAO bonding pool
author: @harisang, @notsoformal
status: Draft
created:  2026-05-16

SIMPLE SUMMARY

We propose to simplify the rules around service fees and setting up reduced bonding pools within the CoW DAO bonding pool. Concretely, we propose that only the following conditions are imposed on solvers within the CoW DAO bonding pool:

  • A service fee of 10% in the solver and quote rewards (whenever positive) is imposed from day 1 on all solvers that are part of the CoW DAO bonding pool; this 10% will be withheld and not be distributed.

  • The size of a reduced bonding pool is set to be either (i) [$200k in stables or ETH (yield bearing)], or (ii) [$100k in stables or ETH] and [0.5M COW]. Moreover, the concept of a minimal initial deposit that needs to double in size within 1 year of its creation is removed.

  • We propose that every solver within the CoW DAO bonding pool has a dedicated Safe, whose sole owner is the Mainnet payouts safe, currently operated by the core team, and whose purpose will be to receive 20% of the rewards each solver claims every accounting week (HODL Rule). If the balances on such a Safe reach the size of a reduced bonding pool, the corresponding solver will be allowed to operate their own “driver” component. We will refer to these safes as the “Solver Bond Safes” per solver.

  • The payouts process will automatically transfer 20% of the rewards (HODL Rule) to the Solver Bond Safe of every solver that stays in the CoW DAO bonding pool, during each payouts transaction, as long as the solver remains in the CoW DAO bonding pool.

  • All further requirements around withholding some percent of rewards, submitting via specific rpc endpoints (in the case of colocated solvers), and providing signatures attesting to the fact that a colocated solver won an auction, are lifted.

MOTIVATION

Each solver operating in the protocol needs to be attached to some bonding pool. The requirement for fully controlling a whitelisted solver address (in any chain) was specified in CIP-7, and roughly speaking, is to create a CoW DAO owned Safe with 0.5M in stables and 1.5M COW tokens. To lower the barrier of entry, CoW DAO created the CoW DAO bonding pool (CIP-15), and assigned its management to the core team.

Currently, when joining the CoW DAO bonding pool, the following are applied:

  • the private keys of the whitelisted solver submission addresses are controlled by the core team, that continues to operate the so-called “driver” component;

  • solvers are required to withhold 25% of the COW rewards they receive - soft locking;

  • after a solver has spent 6 months in the CoW DAO bonding pool, a service fee of 15% applies to the COW rewards, meaning 15% of these rewards is withheld and not distributed by the DAO;

  • solvers within the CoW DAO bonding pool have the option, as specified in CIP-44 and CIP-48, to take control of the submission address, if they create a reduced bonding pool, initially of size $50k/0.5M COW, that needs to grow over a period of 1 year to a pool of size $100k/1M COW. In such a case:

    • Service fees are lifted for the first 3 months after setting up a reduced bonding pool, but they are reinstated after this 3-month period.

    • Solvers are required to contribute 50% of the COW rewards towards growing that pool up to the desired size of $100k/1M COW.

    • Solvers that take over the control of their submission address are still required to submit their solutions via MEV Blocker.

We believe that all the above create significant overhead to both the core team and each solver team that is operating a reduced bonding pool, without necessarily providing any meaningful, in practice, benefits. Specifically, it is not trivial to track all these requirements, and coordinate with the multiple teams that are creating reduced bonding pools, so that all these requirements are consistently met. Moreover, failure to (temporarily) meet any of these requirements does not currently have any well-defined set of actions the core team needs to take.

For these reasons, we propose to significantly simplify the operations of the CoW DAO bonding pool and use a simple set of rules that will make the operations much more transparent and easier to verify, as well as naturally incentivize solvers to gradually build up a (reduced) bonding pool.

Besides easier verifiability, we hope such a simplification will make it easier for all solvers within the CoW DAO bonding pool to consider creating a reduced bonding pool. Such a direction continues to be important, as it reduces the pressure in the core team, eliminates single points of failure, and ultimately makes the protocol more flexible, decentralized, efficient and robust.

SPECIFICATION

We propose the following changes in the above rules:

  • A service fee of 10% is applied from day 1 to all solvers within the CoW DAO bonding pool; this 10% will be withheld and not be distributed.

  • The size of a reduced bonding pool within the CoW DAO bonding pool is set to set to be either (i) [$200k in stables or ETH (yield bearing)], or (ii) [$100k in stables or ETH] and [0.5M COW].

  • Each solver within the CoW DAO bonding pool has a dedicated Solver Bond Safe on Mainnet, whose sole owner is the mainnet payouts safe (eth:0xA03be496e67Ec29bC62F01a428683D7F9c204930), currently operated by the core team, and whose purpose will be to receive 20% of the rewards each solver claims every accounting week. If the balances on such a Safe reach the size of a reduced bonding pool, the corresponding solver will be allowed to operate their own “driver” component. We clarify that 20% of the rewards will continue to be transferred to that Safe in each payouts transaction, as long as the solver remains in the CoW DAO bonding pool

  • When the Solver Bond Safe reaches the size of a full Bond, the Core Team has a mandate to move said solver and whitelist it under its own bond, removing the service fee of 10% and migrating the ownership of the Solver Bond Safe to the CoW DAO safe.

  • All further requirements around withholding some percent of rewards, or submitting via specific rpc endpoints in the case of colocated solvers with a reduced bonding pool, are lifted.

  • Solvers with an existing reduced bonding pool are given a 3-month period in order to adapt their pools so that they are compatible with the new rules. Failure to do so will result in the solver switching to a non-colocated setup, with the core team taking over the ownership of the submission addresses of the solver.

  • CIP-44 referenced a potential setup where solvers with a reduced bonding pool would need to provide a signature given out by the autopilot attesting to the fact their solver indeed won the competition. The infrastructure for this has not yet been developed, so, given this has been long overdue, we propose that this condition is completely dropped.

2

The 25% soft lock rule made sense when rewards were subsidized. With current level of rewards, I’m worried that the 20% hodl policy will only hurt small solvers, making it difficult for them to survive. This could lead to reduced solver competition and consolidation among a few well-funded entities.

Solvers that don’t have the funds to put up the full bond will have a total of 10% + 20% = 30% taken out of their earnings.

The full bond requires about $750,000. This is a hefty amount. Based on earnings from the past few months, even for top-tier solvers, it will take about 16 years of 20% deductions to amass enough funds for reach the full bond; a solver will likely to go out of business long before this happens.

Finally, it seems like the 20% withholding can be circumvented by repeatedly leaving and rejoining the pool… Here is a scenario that is very likely to happen:

Solver goes broke, but has $100k in the pool → leaves pool → not broke anymore → rejoins pool

Sounds to me like this rule will cause a great deal of pain and do little to help attain the desired outcome.

Flash & Sector Finance team

3

Thanks for the response!

With current level of rewards, I’m worried that the 20% hodl policy will only hurt small solvers, making it difficult for them to survive. This could lead to reduced solver competition and consolidation among a few well-funded entities.

This is a valid concern. As you point out later on, this 20% is still owned by the solver, but it is locked and used towards building a large enough bond so that (i) the solver can first colocate, i.e., own their submissions, and (ii) eventually remove themselves from the CoW DAO bonding pool.

As you are all aware, there is significant risk by having lots of solvers in the CoW DAO pool, so there needs to be some incentives for solvers to work towards detaching themselves from the pool, as otherwise we have, in a way, a single point of failure. A large-scale malicious act by a single solver in the pool could result in slashing of the pool and tons of solvers stopping operations immediately, until the pool is replenished (which will probably require a DAO vote).

We should note that with the current state of things, many teams are not really incentivised to move out of the pool, and arguably we haven’t seen much activity for creation of other pools that could attract solvers. I don’t mean to say that we should make the terms so harsh so that solvers are forced out of the pool, but on the other hand there needs to be a plan for solvers within the pool to start building an inventory towards the creation of reduced pools, and eventually full ones.

Finally, it seems like the 20% withholding can be circumvented by repeatedly leaving and rejoining the pool… Here is a scenario that is very likely to happen:

Solver goes broke, but has $100k in the pool → leaves pool → not broke anymore → rejoins pool

I wouldn’t consider this a very likely scenario as solvers cannot automatically join the CoW DAO pool; there is screening by the core team involved, so such teams will be unlikely to be accepted for a second time.

Sounds to me like this rule will cause a great deal of pain and do little to help attain the desired outcome.

Would be great to hear other ideas around the topic, and I would personally not want to push for the CIP if there is no consensus/agreement among the solvers within the pool.

4

Why not allow COW token holders to delegate their tokens to solvers in exchange for a fee share, similar to how Livepeer works?

To participate in the network and perform solver work, operators would need delegated stake from token holders who find their fee structure attractive. This creates a more decentralized system, while also spreading slashing risk across many delegators instead of concentrating it entirely on the solver operator.

It would also give the COW token an additional utility layer. For example, I hold a meaningful amount of tokens, but I don’t have the technical expertise or infrastructure to run a solver myself. I’d still like to be able to contribute my tokens to the network and earn a small return in exchange for taking on part of the risk.

5

After reading Harisang’s reasoning more carefully, I now better understand the motivation behind this proposal and generally agree with its direction.

I still think the current solver market economics are challenging, especially for smaller or developing solvers. However, from a broader operational perspective, simplifying the bonding pool framework and reducing the operational overhead for the CoW DAO is a reasonable goal.

The current system has accumulated many special rules and operational requirements over time, which increases complexity for both the DAO and participating solvers. In the long run, a simpler and more standardized framework could improve the security and sustainability of the CoW DAO bonding pool while also reducing operational pressure and coordination costs.

I also think the proposal reflects an important long-term direction for the protocol: gradually moving solvers toward greater independence and responsibility through more structured bonding mechanisms.

That said, I still believe solver revenue concentration and the relatively low profitability of smaller solvers should be taken into consideration during implementation. A gradual rollout or some form of adaptive mechanism may help balance long-term security goals with maintaining healthy solver competition.

6

@karolak , that is a very reasonable idea. We had discussed it internally under the Value Distribution initiative and it was an idea (“community funded bonds”) that was put out of scope of this CIP, mostly due to:

  1. Being a higher complexity exercise that would remove the velocity of simplifying the operations;
  2. As per Aragon’s research and initial draft, this would require solvers to pay a cost of capital / incentive to token holders that delegate said tokens (thus impacting the solvers P&L, which is already a sensitive topic as discussed in this thread).
  3. Opens an attack vector under AML / KYC as malicious actors could use the bonds for money laundering under a fully decentralised solver acceptance mechanism.

So while I agree with you that it makes a lot of sense that idea, would consider it should be out of scope of the current CIP and a more medium term initiative research topic.

Since there has not been much new considerations in the last weeks, I would suggest this CIP is moved to voting stage.

7

Not sure if I missed anything, but has anything materially changed since the earlier concerns were raised, including whether the concern around the long timeline implied by the proposed numbers has been revisited?

I also did not see much consensus being reached during or after the solver meeting. Given that, I am not sure moving this CIP to a voting stage is appropriate yet.

Regarding the 20% HODL amount, I understand that the funds are still notionally owned by the solver and are intended to build up a future bond. However, the concern is not only about ownership, but about liquidity and operational usability.

Capital that is locked in a bond safe cannot be used to pay infrastructure costs, gas costs, development costs, operational risk, or team expenses. So from a cash flow and runway perspective, the 20% withholding is still a real reduction in usable income, even if the funds technically still belong to the solver. Combined with the 10% service fee, this means 30% of positive rewards would be unavailable for current operations, which is a material impact on solver economics and runway, especially for smaller or developing solvers.

This is especially relevant because current market conditions and several rounds of operational adjustments appear to have already made solver profitability more challenging. Treating locked bond capital as equivalent to liquid operating revenue does not reflect how solver operations actually work.

Since these points do not seem to have been fully addressed, and there does not appear to be clear alignment from the solver meeting, it would seem premature to move this CIP to voting at this stage.

8

Agree with simplifying the operations, and we understand the systemic risk of having too many solvers rely on the DAO pool long term.

We recently joined the protocol as a new solver team (Nexroute) on BNB Chain. As we work to stabilize and optimize our setup, we are experiencing firsthand the heavy infrastructure costs required to remain competitive. To keep contributing to the protocol’s execution quality, covering these out-of-pocket costs is our immediate priority. The proposed flat 30% cut from day 1 conflicts with the goal of attracting new solvers and pushing them toward self-bonding for two main reasons:

  1. Slowing down capital accumulation: The first 6 months are the most expensive for a new solver due to node setup, human resources, and optimization costs. New solvers also operate at lower matching efficiency early on, meaning thin or negative margins.
    • The 20% HODL from day 1 locks up critical operational cash right when expenses are highest. A solver cannot accumulate a bond if they run out of liquidity and go bankrupt in month 3.
    • The 10% service fee permanently extracts capital (going to the DAO instead of the Solver Bond Safe) that could otherwise be used to fund the bond. Combined, reducing available rewards to 70% from day 1 actually delays the time it takes to reach a reduced or full bond, keeping solvers dependent on the CoW DAO bonding pool for longer.
  2. No economic incentive for reduced bonds: Under the current draft, a solver that locks up 200k USD in stables/ETH of their own capital (or reaches it via HODL) pays the exact same 10% fee and 20% HODL as a solver with no initial bond. While they get driver autonomy, there is no financial incentive to tie up capital and take on slashing risk.

We can solve both issues without adding governance overhead by using a standardized ramp. Since reward distributions are already script-driven, applying a time-based rate based on a solver’s onboarding date is trivial to automate and adds close to zero manual tracking overhead for the core team:

For solvers in the CoW DAO bonding pool (without a reduced bond):

  • Months 1-6: 0% fee / 0% HODL (100% liquid) → gives new teams runway to bootstrap, optimize, and prepare for the accumulation phase.

  • Months 7-10: 10% fee / 10% HODL (80% liquid)

  • Month 11+: 10% fee / 20% HODL (70% liquid)

For solvers with a reduced bonding pool (200k USD in stables/ETH, or 100k USD + 0.5M COW):

  • 5% fee / 10% HODL (85% liquid). This recognizes the capital risk the solver is taking, and provides a clear financial incentive to self-fund or reach this tier quickly.

  • If a solver joins the pool with a reduced bonding pool from day 1, they get the same 6-month grace period (0% fee / 0% HODL) to cover setup, followed by the 5% fee / 10% HODL rate.

For solvers whitelisted under their own full bond:

  • 0% fee / 0% HODL (100% liquid), as proposed.

This keeps the rules simple and automatic, but ensures the onboarding funnel remains viable and rewards solvers who de-risk the DAO. Voting now without addressing these points seems premature.