Rocket Pool

Rocket Pool is a decentralized liquid-staking protocol on Ethereum, governed by RPL holders and a set of trusted oracle nodes. It issues rETH, a reward-bearing token that appreciates against ETH as staking rewards accrue. Where the largest staking protocols run a curated, DAO-approved set of professional node operators, Rocket Pool's defining choice is a permissionless operator set: anyone can run a node by posting an ETH bond and RPL collateral, with no allowlist. That decision shapes both the product and the governance, which Rocket Pool splits across two distinct DAOs. Sources throughout link to Rocket Pool's own documentation and its RPIP improvement-proposal register.

rETH and the permissionless node-operator model

Solo staking on Ethereum requires 32 ETH and continuous uptime. Rocket Pool lets anyone deposit from 0.01 ETH upward for rETH – a token whose exchange rate rises as the underlying validators earn rewards – while a separate class of participants, node operators, actually run the validators. An operator spins up a megapool validator, posting a personal bond of 4 ETH that the contracts pair with 28 ETH of liquid-staker deposits to make up a 32 ETH validator. Saturn I took that shape to mainnet on 18 February 2026, and with it the change that matters most to this page: RPL is no longer part of the bond.

The documentation now reads that an operator “may optionally stake RPL to their node to qualify for additional rewards and to gain voting power within the protocol DAO” (read 17 September 2026). Running a validator and holding governance power are separate purchases now, where one used to require the other. Because operators are permissionless, Rocket Pool's node set is far more numerous and geographically diffuse than a curated one – the Lido centralization critique is the exact problem Rocket Pool designs against. The tradeoff is that permissionless operators can't be trusted with protocol admin, which is precisely why governance is split in two. See the protocol docs for the full megapool and rETH mechanics.

Two DAOs: the pDAO and the oDAO

Rocket Pool separates who sets policy from who reports trusted facts:

• Protocol DAO (pDAO) – the RPL-holder governance body that shapes protocol direction: parameters, treasury/inflation, upgrades. Voting power is a function of effective RPL stake held by node operators (see the next section).
• Oracle DAO (oDAO) – a small set of well-regarded staking-ecosystem members running special nodes that supply data the smart contracts cannot compute for themselves (validator balances, the rETH exchange rate, reward Merkle trees) and execute mandated administrative steps. Its remit, membership, and slashing rules are fixed by RPIP-24, the oDAO Charter.

This is a deliberate answer to a hard problem: a permissionless validator set needs some trusted party to attest to off-chain reality, but that trusted party should not also control policy. The oDAO is intentionally narrow – an oracle and executor, not a legislature – while the pDAO holds the political power. Rocket Pool has openly worked to reduce the residual centralization in how the two interact.

On-chain pDAO governance: square-root weight and layered votes

The Houston upgrade moved the pDAO fully on-chain (RPIP-33), replacing pure Snapshot signalling with binding contract execution. Two features stand out:

• Square-root voting weight. A node operator's pDAO power is scored on the square root of their effective RPL stake – so the larger the stake, the less voting power per RPL. The explicit goal, per Rocket Pool's design notes, is to temper large operators and let blocs of smaller operators actually move a vote. This is a rare on-chain anti-whale weight, distinct from the linear token-weighted voting most protocols use.
• Snapshot-and-override voting. Eligibility, delegation state, and power are frozen at a snapshot when a proposal is created. There are two voting periods: first delegates (and direct voters) cast votes, then members who had delegated their power get a window to override their delegate's choice. A veto quorum lets the community instantly defeat a proposal – and the proposer forfeits their bond – to deter spam and proposals that skipped off-chain signalling.

Proposal creation and challenges are bonded throughout, so putting a bad proposal on-chain has a real cost. The full rules live in RPIP-4 and the governance guides.

The security council and upgrade guardrails

A fully on-chain, permissionless governance layer needs a circuit breaker. Rocket Pool elects a security council whose members can pause the protocol quickly if something goes wrong and can make certain parameter changes without waiting out the normal delay. The council is itself controlled by the pDAO, which sets the required quorum and can replace members. RPIP-60 formalizes these protocol upgrade guardrails: upgrades authorized by a pDAO vote pass through a mandatory delay during which the security council can veto, and are then executed on-chain by the oDAO.

Saturn I has now run that pipeline end to end, so it is a record rather than a plan: halving the operator bond to 4 ETH, introducing megapools, express and standard deposit queues, adjustable commission and an RPL fee switch, it was authorized by pDAO vote, executed by the oDAO after the mandated delay, and has been live on Ethereum mainnet since 18 February 2026, the date the upgrade’s own site gives beside its statement that Saturn is live.

How Caper approaches this

Rocket Pool's answer to plutocracy is a concave function of capital: square-root the RPL stake so whales get less power per token, add a trusted oracle set for facts, and elect a council to catch emergencies. It is clever, and it needs no participation history – power is still bought with RPL, just at a diminishing rate. Caper makes a different bet. A caper's vote weight grows with the member's earned soulbound votes, not with held tokens alone, and the same weight sets a member's treasury share on exit. The vote token is earned only by casting a ballot and is non-transferable (its depositor role is locked to the DAO itself), so no member can ever sell another the participation they have already earned – and capital cannot mint its own without voting.

This is not a bagless system – t is a genuine multiplier, so stake still counts – but the decisive factor a well-funded buyer cannot acquire is the earned soulbound record, not a curve applied to their capital. And there is no separate oracle DAO or elected council to trust: typed PAYOUT/INVEST/DIVEST proposals execute directly against a protocol-controlled treasury vault. Honestly, that means a caper forgoes exactly the professional operator layer and emergency council Rocket Pool provides; the wager is that earned, untransferable weight plus a minimal on-chain surface beats a trusted set you have to keep honest.

References

• Rocket Pool – protocol site and rETH overview.
• Rocket Pool documentation – megapools, node operation, and Houston governance guides. A node operator’s responsibilities is the source for the 4 ETH bond, the 28 ETH of borrowed deposits, and optional RPL.
• RPIP-33 – implementation of an on-chain pDAO; RPIP-4 – community resolutions and voting.
• RPIP-24 – the Oracle DAO Charter; RPIP-60 – protocol upgrade guardrails.
• Saturn I – live on mainnet since 18 February 2026 (megapools, 4 ETH bond, queues, RPL fee switch).
• Rocket Pool governance forum – live pDAO discussion and proposals.