Specification for Rewards Calculation

This document serves as a formal specification for the way that the rewards intervals and the values within are calculated as part of the Redstone rewards system.

Version

This describes v8 of the rewards calculation ruleset.

This version implements RPIP-30 following the suggestions here

Changes since v7

The following updates have been made from v7 of the spec.

Major Updates

  • Following the Protocol DAO vote to approve RPIP-30, Collateral Rewards rewards are now weighted on a curve, favoring 8 Ether Minipools and creating diminishing returns as additional RPL above the minimum is staked.
    • This change is phased in linearly over 5 intervals, starting with interval 18, and will be fully effective in interval 23.
  • There is no longer a maximum node collateral.
    • RocketDAOProtocolSettingsNode.getMaximumPerMinipoolStake() is no longer called and a constant of 150% is used instead when calculating the phase-in for RPIP-30.

Minor Changes

  • The Rewards File now includes a totalNodeWeight field which represents the sum of the rewards weighting.

Clarifications

  • Disclaimers / Notes section has been updated to specify multiplication precedes division to preserve accuracy.
  • Attestation eligibility surrounding the Deneb hard fork was clearly defined.

Disclaimers / Notes

  • All arithmetic described here is intended to be integer arithmetic. There are no floating point values, and floating point division is not used to allow for maximum portability and elimination of floating point errors.
    • Multiplicative operations, except where specifically grouped by parentheses, should precede division operations when otherwise reorderable through the associative property, to preserve accuracy.
  • Unless explicitly specified, the following rules about data formats apply:
    • Timestamps are represented as Unix timestamps, and are provided as a total number of seconds since the Unix Epoch.
    • Token quantities are represented in wei.
    • The code samples here are all presented in pseudocode format and will not compile to a known language. Extrapolate to your system of choice accordingly.

Scheduling and Target Blocks

As with the original protocol, rewards in Redstone are minted and recorded in discrete intervals. Each interval is determined by an on-chain setting that specifies the amount of time that must pass from the previous interval until the next interval is ready to be generated.

Time of Eligibility

This value marks the start time, called startTime (as a Unix timestamp, the number of seconds since the Unix Epoch) of the currently active interval:

RocketRewardsPool.getClaimIntervalTimeStart()

If this value is 0, use the start time from the RPL token (also a Unix timestamp) as startTime:

RocketTokenRPL.getInflationIntervalStartTime()

This value marks the amount of time in seconds, called intervalTime, that must pass since startTime before a new interval is ready to be recorded:

RocketRewardsPool.getClaimIntervalTime()

The current time is determined by the timestamp field in the header of the latest block of the Execution client - also a Unix timestamp. When the current time is greater than intervalTime plus startTime, a new interval is eligible.

More specifically, the number of eligible intervals at any given time is calculated by the latest block's timestamp, minus startTime, divided by intervalTime (using integer division):

latestBlockTime := latestBlockHeader.Time
timeSinceStart := latestBlockTime - startTime
intervalsPassed := timeSinceStart / intervalTime

This quantity is multiplied by intervalTime and added to startTime to produce the most recent eligibility timestamp, known as endTime:

endTime := startTime + (intervalTime * intervalsPassed)

endTime is used to determine which Beacon slot and Execution block to snapshot the states for when calculating rewards (see below).

Missed / Multi-Period Intervals

If the number of intervalsPassed is greater than 1, it indicates that the Oracle DAO failed to adequately report a rewards checkpoint during one interval window and has entered into the window for a subsequent interval.

When this happens, RPL is still minted on schedule (e.g., at the start of the new interval window). The missed rewards checkpoint is simply rolled into the new one. RPL and ETH rewards pending distribution have accumulated during this time, so the rewards checkpoint will use these pending figures as the amounts to distribute as part of the interval and add them to the Merkle Tree accordingly.

In other words, "missing" a rewards interval does not cause a loss of rewards. It merely delays their distribution.

The number of intervals passed is recorded in the event emitted during a rewards tree submission to account for this occurrence.

Target Beacon Slot

Each rewards period is ultimately defined by a target Beacon slot and corresponding Execution block which serve as chain state reference points for rewards calculation.

Once the system detects that a new rewards interval is eligible, it will calculate the Beacon Epoch that endTime falls in using the Beacon chain's genesis configuration. The slot used is the first slot after endTime; that is, the first slot where slotTime > endTime.

genesisTime := eth2Config.GenesisTime // Unix time of Beacon chain's genesis
secondsPerSlot := eth2Config.SecondsPerSlot
slotsPerEpoch := eth2Config.SlotsPerEpoch

totalTimespan := endTime - genesisTime
targetBcSlot := math.Ceil(totalTimespan / secondsPerSlot) // First slot *after* endTime
targetSlotEpoch := targetBcSlot / slotsPerEpoch

Once the target Epoch is determined, the target slot becomes that last slot in that epoch. If that slot was missed (it has an empty Block because the proposer failed), then the target slot will become the previous slot. If that is also missed, use the slot before that, and so on until the slot was not missed.

This slot will become the targetBcSlot which will be used for state snapshotting.

The intent is to use the latest state of the Beacon chain prior to the Epoch boundary, as the Epoch boundary signals the beginning of the next interval.

Target Execution Block

Once the targetBcSlot has been found, the corresponding block on the Execution layer can be determined.

Pre-merge, this will be the last block before targetSlotEpoch's timestamp. Note that this could potentially be an EL block that was added to the Execution chain after targetBcSlot was added to the Beacon chain.

Post-merge, this will be the Execution block corresponding to targetBcSlot.

This block will be the targetElBlock.

Timing of State Collection

Once targetBcSlot and targetElBlock have been determined, the user will need to wait until the Epoch after targetSlotEpoch has been finalized. This is because the rewards calculation will involve analyzing the attestation performance of validators up to the targetBcSlot.

Due to EIP-7045, attestations are valid when they are assigned:

  • In Epochs less than DENEB_FORK_EPOCH as defined in the consensus spec (e.g., 269568 on Mainnet) and included up to 32 slots after the assigned slot.
  • In Epochs at least DENEB_FORK_EPOCH and included before the end of the Epoch following the Epoch containing the assigned slot.

The Epoch following the attestation slot must also be finalized so the attestation performance can be tracked.

For example, if a rewards interval occurred on Epoch 63, the user would have to wait until Epoch 64 was finalized before generating the rewards tree.

Note that this relies on the ability to query the state of both targetBcSlot and targetElBlock well past their submission timestamps; users may need access to archive nodes if their clients cannot look far enough back.

RPL Rewards

This section describes the calculation for the RPL rewards distributed to each node operator and the Protocol DAO treasury.

RPL Amounts per Group

The amount of RPL to be distributed during a rewards checkpoint can be found with the following contract function:

RocketRewardsPool.getPendingRPLRewards()

This accounts for all of the RPL minted since the last rewards interval submission. If multiple intervals have gone by and are being rolled up into this period, all of the RPL minted across each of them will be inherently aggregated into this value.

RPL rewards are divided into three groups, the fraction of which can be retrieved by the following contract methods:

  • Collateral rewards for normal Node Operators

    collateralPercent := RocketRewardsPool.getClaimingContractPerc("rocketClaimNode")

  • Oracle DAO rewards

    oDaoPercent :=RocketRewardsPool.getClaimingContractPerc("rocketClaimTrustedNode")

  • Protocol DAO (treasury) rewards

    pDaoPercent := RocketRewardsPool.getClaimingContractPerc("rocketClaimDAO")

Each of these values is a percentage, given in wei, where 100% (1.0) corresponds to 1 x 10^18 wei. For example, if the collateral fraction collateralPercent had a value of 700000000000000000 (7 x 10^17), this corresponds to a percentage value of 70% (0.7).

Thus, the total expected amount per group is as follows:

_100Percent := 1e18
pendingRewards := RocketRewardsPool.getPendingRPLRewards()

collateralRewards := pendingRewards * collateralPercent / _100Percent
oDaoRewards := pendingRewards * oDaoPercent / _100Percent
pDaoRewards := pendingRewards * pDaoPercent / _100Percent

If the total amount of pending rewards is 0 for any reason, this interval cannot be used for rewards submission. Simply wait for the protocol to roll over to the next interval and try again.

Note that these will not be the final values attributed to each of these groups due to division truncation; they are simply starting points when calculating the actual values per group. The final amounts are discussed later in this section.

Collateral Rewards

Start by acquiring the complete list of node addresses using the following contract functions:

nodeCount := RocketNodeManager.getNodeCount()
nodeAddresses := address[nodeCount]
for i = 0; i < nodeCount; i++ {
    nodeAddresses[i] = RocketNodeManager.getNodeAt(i)
}

For each node, retrieve the effective RPL stake. This should be calculated as follows.

For each minipool belonging to the node, get its current state:

state := minipool.getStatus()

Ignore minipools that are not in the staking state.

Define eligibleBorrowedEth as the total amount of ETH borrowed by the node operator from the staking pool for eligible minipools. Define eligibleBondedEth as the total amount of ETH the node operator has bonded for its eligible minipools. Start with both set to 0.

For each staking minipool, check if it was not exited from the Beacon Chain at the end of the interval:

  1. Get the status of the validator from the Beacon Chain for targetBcSlot (e.g., /eth/v1/beacon/states/<targetBcSlot>/validators?id=0x<pubkey>). If the validator did not exist at targetBcSlot, ignore it and continue.
  2. Get the exit_epoch for the validator.
  3. If the validator's exit_epoch is after targetBcSlot's epoch (exit_epoch > targetSlotEpoch), it is eligible.
    1. Add the amount of ETH borrowed by the node operator for this minipool to eligibleBorrowedEth: 
      borrowedEth := minipool.getUserDepositBalance()
eligibleBorrowedEth += borrowedEth
    2. Add the amount of ETH bonded by the node operator for this minipool to eligibleBondedEth: 
      bondedEth := minipool.getNodeDepositBalance()
eligibleBondedEth += bondedEth

Next, calculate the minimum and maximum RPL collateral levels based on the ETH/RPL ratio reported by the protocol:

ratio := RocketNetworkPrices.getRPLPrice()
minCollateralFraction := RocketDAOProtocolSettingsNode.getMinimumPerMinipoolStake() // e.g., 10% in wei
maxCollateralFraction := 1.5e18 // i.e., 150% in wei. Hard-coded by RPIP-30's phase-in rules.
minCollateral := eligibleBorrowedEth * minCollateralFraction / ratio
maxCollateral := eligibleBondedEth * maxCollateralFraction / ratio

Note that minCollateral is based on the amount borrowed, and maxCollateral is based on the amount bonded.

Now, calculate the node's effective RPL stake (nodeEffectiveStake) and weight (nodeWeight) based on the above:

nodeStake := RocketNodeStaking.getNodeRPLStake(nodeAddress)
if nodeStake < minCollateral {
    nodeEffectiveStake := 0
    nodeWeight := 0
} else {
    if nodeStake > maxCollateral {
        nodeEffectiveStake := maxCollateral
    } else {
        nodeEffectiveStake := nodeStake
    }
    nodeWeight := getNodeWeight(eligibleBorrowedEth, nodeStake, ratio)
}

getNodeWeight() is defined in the getNodeWeight section.

Next, scale the nodeEffectiveStake and nodeWeight by how long the node has been registered. This prorates RPL rewards for new nodes that haven't been active for a full rewards interval, so they only receive a corresponding fraction of the rewards based on how long they've been registered.

For example, if the rewards period were 6300 Epochs (28 days) and a node registered 10 days ago, their nodeEffectiveStake and nodeWeight would be reduced to 35.7% (10 / 28) of its true value.

The node's registration time can be retrieved with the following contract method:

registrationTime := RocketNodeManager.getNodeRegistrationTime(nodeAddress)

This should be subtracted from the timestamp of targetElBlock to determine the node's age. It should then be compared to intervalTime to determine the prorated effective stake:

nodeAge := targetElBlock.Timestamp - registrationTime
if (nodeAge < intervalTime) {
    nodeEffectiveStake = nodeEffectiveStake * nodeAge / intervalTime
    nodeWeight = nodeWeight * nodeAge / intervalTime
}

Finally:

  • Sum each nodeEffectiveStake to retrieve the totalEffectiveRplStake across the entire network.
  • Sum each nodeWeight to retrieve the totalNodeWeight across the entire network.

If the totalEffectiveRplStake or totalNodeWeight is 0 (i.e., none of the nodes are eligible for RPL rewards), add collateralRewards to pDaoRewards. Otherwise, if any node is eligible for rewards, perform the following steps instead.

Now, calculate the the cycle factor, C, of RPIP-30's phase-in. Define C to be on the closed range [1, 6] and calculate it:

C := min(6, interval - 18 + 1)

You can now calculate the collateral RPL per node from nodeWeight, totalNodeWeight, nodeEffectiveStake, totalEffectiveRplStake, C, and collateralRewards.

nodeCollateralAmount := (collateralRewards * C * nodeWeight / (totalNodeWeight * 6)) + (collateralRewards * (6 - C) * nodeEffectiveStake / (totalEffectiveRplStake * 6))

Sum the nodeCollateralAmount for each node to arrive at the totalCalculatedCollateralRewards. As a sanity check, compare this to the original collateralRewards value using either the total number of nodes or the total number of minipools, whichever is higher, as a delta value:

epsilon := max(numberOfNodes, numberOfMinipools)
if collateralRewards - totalCalculatedCollateralRewards > epsilon {
    // Raise an error because the calculation has excessive error
}

getNodeWeight

Calculate stakedRplValueInEth:

stakedRplValueInEth = nodeStake * ratio / 1 Eth.

Calculate percentOfBorrowedEth:

percentOfBorrowedEth = stakedRplValueInEth * 100 Eth / eligibleBorrowedEth

If percentOfBorrowedEth <= 15 Eth, return 100 * stakedRplValueInEth.

Otherwise, return ((13.6137 Eth + 2 * ln(percentOfBorrowedEth - 13 Eth)) * eligibleBorrowedEth) / 1 Eth.

ln is specified in the next section.

ln

ln(x) is defined as log2(x) * 1 Eth / 1442695040888963407

log2 is defined in the next section.

log2

log2(x) is approximated using iterative approximation for 60 inner loops.

Define result = 0

Calculate exponent of the highest power of two that, when multiplied by 1 Eth, is less than or equal to the input x. That is, if x is 40 Eth, the highest power of two that is less than 40 Eth is 32 Eth, and the most significant bit of 0b0100000 (32 in binary) is at index 5 (counting the least signifigant bit as index 0), so exponent is 5.

Multiply the exponent by 1 Eth and add it to the result:

result = result + exponent * 1 Eth

Next, calculate the iterative approximation's y term:

y = x >> exponent

If y is 1 Eth, return result.

Otherwise, define delta = 1 Eth.

Loop 60 times. In each loop:

  1. Divide delta by 2, delta = delta / 2
  2. Square y, y = y * y / 1 Eth
  3. If y >= 2 Eth:
    • Add delta to result, i.e. result = result + delta
    • Divide y by 2, i.e. y = y / 2

After 60 loops, return result.

Oracle DAO Rewards

Start by acquiring the list of Oracle DAO node addresses using the following contract methods:


oDaoCount := RocketDAONodeTrusted.getMemberCount()
oDaoAddresses := address[oDaoCount]
for i = 0; i < oDaoCount; i++ {
    oDaoAddresses[i] = RocketDAONodeTrusted.getMemberAt(i)
}

Next, scale the amount of RPL earned by each Oracle DAO node by how long the node has been part of the Oracle DAO. This prorates RPL rewards for new nodes that haven't been a member for a full rewards interval, so they only receive a corresponding fraction of the rewards based on how long they've been a part of the DAO.

For example, if the rewards period were 6300 Epochs (28 days) and a node joined 10 days ago, their share would be reduced to 35.7% (10 / 28) of its true value.

The node's join time can be retrieved with the following contract method:

joinTime := RocketDAONodeTrusted.getMemberJoinedTime(nodeAddress)

This should be subtracted from the timestamp of targetElBlock to determine the time since joining. It should then be compared to intervalTime to determine the prorated rewards.

One way to accomplish this is to use the number of seconds the node participated in the interval as an analog to the "effective stake" in the Collateral RPL calculation above:

odaoTime := targetElBlock.Timestamp - joinTime
participatedSeconds := intervalTime
if (odaoTime < intervalTime) {
    participatedSeconds = odaoTime
}

When finished, add each of these to retrieve the totalParticipatedSeconds for all oDAO nodes.

With this in hand, you can now calculate the Oracle DAO RPL per node by taking the original oDaoRewards value, multiplying by the participatedSeconds, and dividing by the totalParticipatedSeconds:

oDaoAmount := oDaoRewards * participatedSeconds / totalParticipatedSeconds

Sum the oDaoAmount for each node to arrive at the totalCalculatedODaoRewards. As a sanity check, compare this to the original oDaoRewards value using either the total number of nodes or the total number of minipools, whichever is higher, as a delta value:

epsilon := max(numberOfNodes, numberOfMinipools)
if oDaoRewards - totalCalculatedODaoRewards > epsilon {
    // Raise an error because the calculation has excessive error
}

Protocol DAO Treasury Rewards

Unlike the other two groups, the amount of RPL awarded to the Protocol DAO Treasury is not calculated on its own. Rather, the Protocol DAO Treasury is simply used as a buffer for all of the remaining RPL that's unaccounted for:

actualPDaoRewards := pendingRewards - totalCalculatedCollateralRewards - totalCalculatedODaoRewards

You may want to compare this amount to the original pDaoRewards calculation earlier to log how much of a delta is being handled, but the sanity checks in the previous two steps will prevent this from being too far from the expected value.

Smoothing Pool Rewards

The Smoothing Pool's current balance is distributed to all of the nodes that have been opted into the Smoothing Pool for some (or all) of this rewards interval and are eligible, with one exception.

Interval 0

The first rewards interval using the Redstone rewards system will not produce Smoothing Pool rewards. This is because the Smoothing Pool's rewards calculation depends on the time that the previous interval occurred as a way to determine each minipool's eligibility and prorating status, and the event containing that data is only emitted upon a successful rewards snapshot.

For the first interval using Redstone's system (interval 0), ignore the Smoothing Pool calculation. Its balance will be rolled over into interval 1.

Balance and Start Blocks

Start by getting the current balance of the Smoothing Pool contract:

smoothingPoolBalance := RocketSmoothingPool.Balance()

If the balance is 0 (e.g., because nobody has opted into the Smoothing Pool), simply end here.

Next, get the rewards event emitted for the previous interval:

previousIntervalEvent := RocketRewardsPool.RewardSnapshot(currentIndex - 1)

From this event, you can get the bnStartBlock and the elStartBlock for this interval.

The bnStartBlock is the first non-missed slot in the Epoch after the Epoch that previousIntervalEvent.ConsensusBlock belonged to.

Pre-merge, the elStartBlock is simply previousIntervalEvent.ExecutionBlock + 1. Post-merge, the elStartBlock is the EL block that corresponds to bnStartBlock.

Node Eligibility

For each registered node (the gathering of which was shown previously in the RPL calculation), observe the status and penalty count on each of its minipools:

Next, look at the minipools for the node with the following contract methods:

nodeMinipoolCount := RocketMinipoolManager.getNodeMinipoolCount(nodeAddress)
nodeMinipools := address[nodeMinipoolCount]
for i := 0; i < nodeMinipoolCount; i++ {
    minipool := RocketMinipoolManager.getNodeMinipoolAt(nodeAddress, i)
    state := minipool.getStatus()
    penaltyCount := RocketNetworkPenalties.getPenaltyCount(minipoolAddress)
}

If the state is staking and the penaltyCount is 3 or more, this node is a cheater and is not eligible for Smoothing Pool rewards. Remove it from the list of eligible nodes and ignore it.

If the node has at least one staking minipool, then it is eligible for calculation. Otherwise, remove it from the list of eligible nodes and ignore it.

(Optional) Minipool Eligibility

In addition to filtering out ineligible nodes, minipools can also be filtered to speed up calculations. This is done by removing minipools that exited before the interval starts, or are scheduled to activate after the interval ends.

For each staking minipool in each eligible node, check the activation_epoch and exit_epoch for that minipool's validator:

  1. Get the status of the validator from the Beacon Chain for targetBcSlot (e.g., /eth/v1/beacon/states/<targetBcSlot>/validators?id=0x<pubkey>).
    1. If the validator does not exist at that slot (its status is empty), or if its status is pending_initialized or pending_queued, it is not eligible for any rewards. Ignore it in the following calculations.
  2. If the validator's activation_epoch is after targetBcSlot, it is not eligible. Remove it.
  3. If the validator's exit_epoch is before bnStartBlock, it is not eligible. Remove it.

Node Opt-In / Out Timing

For each eligible node, determine the opt-in time and opt-out time. These will be used during attestation performance to determine if a given attestation should count towards the Smoothing Pool rewards or not.

Start by retreiving the opt-in status and the last time of status change for the node:

isOptedIn := RocketNodeManager.getSmoothingPoolRegistrationState(nodeAddress)
statusChangeTime := rocketNodeManager.getSmoothingPoolRegistrationChanged(nodeAddress) // The contracts provide the Unix timestamp, in seconds

Use these details to determine the opt-in and opt-out time; for example:

farPastTime := 0 // Some arbitrary timestamp that occurred before the start of the interval; the Unix epoch is fine for this
farFutureTime := 1e18 // Some arbitrary timestamp that will occur far after the end of the interval

if isOptedIn {
    optInTime = statusChangeTime
    optOutTime = farFutureTime
} else {
    optInTime = farPastTime
    optOutTime = statusChangeTime
}

Calculating Attestation Performance and Minipool Scores

Start by defining the following variables:

  • totalMinipoolScore, which cumulatively tracks the aggregated minpool scores for each attestation, starting at 0
  • successfulAttestations which tracks the number of successful attestations that were eligible for Smoothing Pool rewards, starting at 0
  • minipoolScores, a map of minipools to their individual cumulative minipool scores

For each eligible minipool in each eligible node, make a note of its status and status change time:

status := minipool.getStatus()
statusTime := minipool.getStatusTime()

For duties to be eligible for rewards inclusion, the minipool must be in the staking status at the time of the attestation duty assignment. You may use the state of the chain at the time of the duty assignment or any state after the duty assignment to assess this. This is used because status is one of the final states of a minipool (the other being dissolved, which is mutually exclusive with staking) and statusTime indicates the time at which the minipool entered staking status. Thus, if a minipool's status is staking, it will always be staking and you can determine when it entered that state by using statusTime.

Note that the finalized flag is not a true state and does not overwrite staking; it is a separate boolean value.

Next, process the attestation performance of the minipool to gauge its minipoolScore. Attestation performance is calculated on an Epoch-by-Epoch basis, from the first Epoch to the last Epoch of the interval, as follows for each Epoch:

  1. Get the attestation committees for the Epoch (e.g., /eth/v1/beacon/states/head/committees?epoch=<epochIndex>)
  2. Traverse the list of slots and committees, noting the slotIndex, committeeIndex, and position of an attestation assignment for the minipool (where position is the 0-based index of the entry in the response's list of validator indices). Ignore validators that do not correspond to eligible Rocket Pool minipools.
  3. Get the block at slotIndex (e.g., /eth/v2/beacon/blocks/<slotIndex>).
  4. Get the time of the block: 
    blockTime := genesisTime + secondsPerSlot * slotIndex
  5. For the minipool corresponding to position:
    1. If blockTime occurred before the parent node's optInTime or after the parent node's optOutTime, this attestation is not eligible for Smoothing Pool rewards. Ignore it.
    2. If the minipool is not in staking status by the time of this attestation, it has not performed any eligible attestations yet so this duty should be ignored.
    3. If blockTime occurred before the minipool's statusTime, it was not in staking status during the attestation duty so this duty should be ignored.
      1. Note that this check will only be relevant for solo staker migrations, as conventionally-created minipools will enter staking long before they begin attesting whereas solo staker migrations will be attesting prior to entering staking status.
  6. Look at the attestations in the subsequent blocks with matching slotIndex, committeeIndex, and position. Start at the block directly after slotIndex, and look up to 1 Epoch away (BeaconConfig.SlotsPerEpoch) from slotIndex.
    1. If one was recorded in any of these blocks, this attestation was successful. Calculate the minipoolScore for this attestation as described below.
    2. If the attestation was not found, it was missed. Add it to a running list of missedAttestations.
    3. The boundary is inclusive, so if an attestation for slot n is found in slot n + BeaconConfig.SlotsPerEpoch then it was successful. If it was found in slot n + BeaconConfig.SlotsPerEpoch + 1, it is too late and should be considered missed.

When a successful attestation is found, calculate the minipoolScore awarded to the minipool for that attestation:.

  1. Add the attestation to a running list of goodAttestations for the minipool.
  2. Get the amount of ETH bonded by the node operator and the commission (node fee) for this minipool on this block (the block corresponding to the attestation duty assignment) by using the block's timestamp and the timestamp of the minipool's last bond reduction: 
    currentBond := minipool.getNodeDepositBalance()
currentFee := minipool.getNodeFee()
previousBond := RocketMinipoolBondReducer.getLastBondReductionPrevValue(minipool.Address)
previousFee := RocketMinipoolBondReducer.getLastBondReductionPrevNodeFee(minipool.Address)
lastReduceTime := RocketMinipoolBondReducer.getLastBondReductionTime(minipool.Address)

fee := currentFee
bond := currentBond
if lastReduceTime > 0 && lastReduceTime > blockTime {
     // If this block occurred before the bond was reduced, use the old values
    bond = previousBond
    fee = previousFee
}
  3. Calculate the minipoolScore using the minipool's bond amount and node fee: 
    minipoolScore := (1e18 - fee) * bond / 32e18 + fee // The "ideal" fractional amount of ETH awarded to the NO for this attestation, out of 1
  4. Add minipoolScore to the minipool's running total, and the cumulative total for all minipools: 
    minipoolScores[minipool.Address] += minipoolScore
totalMinipoolScore += minipoolScore
successfulAttestations++

(Optional) Removing Idle Minipools

Once each minipool's score has been determined, you may optionally remove minipools with zero score (i.e., no successful or missed attestations). These minipools are considered idle and will just waste calculations in the final tally.

While skipping this step won't affect the final calculation, if you are logging the records of your calculation (such as with the "minipool performance" file included in the canonical Rocket Pool rewards intervals) then idle minipools may be omitted from the final report.

Calculating Node Rewards

Start by calculating the "ideal" amount of ETH that would go to node operators (normalizing successfulAttestations into an ETH amount in the process), based on their cumulative fractional scores:

totalNodeOpShare := smoothingPoolBalance * totalMinipoolScore / (successfulAttestations * 1e18)
totalEthForMinipools := 0

Here we also define a variable totalEthForMinipools that will contain the cumulative total ("actual") amount of rewards for all node operators, which is initialized to 0.

Next, for each minipool, calculate the minipool's share of this ideal ETH total and add it to the cumulative total:

minipoolEth := totalNodeOpShare * minipoolScores[minipool.Address] / totalMinipoolScore
nodeEth[minipool.OwningNode] += minipoolEth
totalEthForMinipools += minipoolEth

where nodeEth is the true amount of ETH awarded to each node.

Now, calculate the final "actual" pool staker balance (which will act as a buffer and capture any lost minipool ETH due to integer division):

poolStakerEth := smoothingPoolBalance - totalEthForMinipools

Constructing the Tree

With all of the above values, you can now create the Merkle tree for this interval using the tree specification.