Rain Chain / Validators
Seats are earned by stake, not sold.
Anyone can bond. The top validators by total stake form the active set, voting power is stake-weighted and capped, and misbehaviour is slashed.
The validator set
7 at genesis, up to 26 by stake
The active set is the top 26 by total stake (self-bond plus delegations). Each validator's counted voting power is capped at 10%; stake above the cap still earns rewards at a 50% rate. At most two validators enter or leave per 24-hour epoch, which prevents sudden set capture and leaves time for the randomness-beacon key generation. The chain refuses to produce blocks with fewer than 5 active validators.
Requirements
What it takes
| Parameter | Value |
|---|---|
| Minimum self-bond | 25M RAIN, locked for 1 year from activation |
| Commission | Min 5%, max 20%. Increases ≤ 1 pp per 7 days, announced 7 days ahead; decreases are instant |
| Unbonding | 21 days — longer than any slashing evidence window and the longest bridge delay |
| Uptime | Rewards scale with signed blocks; uptime factor is 0 below 90% in an epoch |
| Keys | Consensus keys on remote signers with double-sign protection; bridge keys in HSM; cold keys on hardware wallets |
| Topology | Validator behind sentry nodes; validator IP not public |
| Duties | Propose and vote on blocks, sign bridge withdrawals, contribute randomness-beacon shares |
Hardware
Estimated node requirements
Estimates from the specification; final requirements will be published with the testnet.
| Role | Purpose | Hardware (estimate) |
|---|---|---|
| Validator | Votes, proposes, holds consensus and bridge keys | 16–32 cores ≥ 3.5 GHz, 128 GB RAM, 2× 2 TB NVMe, 10 Gbit |
| Sentry | Shields the validator, gossips | 8 cores, 32 GB RAM, 1 TB NVMe |
| Full node | Non-voting replica for apps | 16 cores, 64 GB RAM, 2 TB NVMe |
| RPC / API | REST, WebSocket, EVM JSON-RPC | Full node + load balancer |
| Archive | All history | 16 cores, 128 GB RAM, 10+ TB NVMe |
Nodes export Prometheus metrics, structured logs and OpenTelemetry traces. Snapshots every 10,000 blocks target a full-node sync in ≤ 30 minutes.
Rewards
Rewards and commission
- Each block, the reward pool — staking-reserve release, protocol inflation and RAIN priority fees — is distributed to active validators in proportion to counted stake × uptime factor.
- Each validator takes its commission on delegators' rewards; delegators receive the rest pro rata.
- The 30% staker share of trading fees, paid in stablecoins, is distributed the same way once per epoch.
See the staking-yield model for how APR varies with the staking ratio.
Delegation
Anyone can stake
1,000 RAIN
Delegation is open to any address.
Instant
Once per 7 days per delegation.
21 days
Delegations share their validator's slashing risk while bonded and unbonding.
Slashing
Misbehaviour costs stake
| Offence | Penalty |
|---|---|
| Double-sign (equivocation) | 5% of validator and delegated stake; tombstoned (permanent removal) |
| Correlated double-sign | min(100%, 3 × fraction of power that double-signed in the same 21-day window) |
| Downtime | Jailed after missing 95% of blocks in a 10,000-block window (~33 min); 0.1% slash; unjail after 1 h |
| Invalid bridge signature | 100% of self-bond and 5% of delegations; tombstoned |
| Beacon withholding | Counted as downtime |
Slashed RAIN is burned, not redistributed, to remove any incentive to frame validators.
Security
Why buying seats doesn't buy the chain
A common critique: "26 seats at $1M each = a $26M takeover." It assumes seats carry equal votes. They don't.
Votes are stake-weighted
Controlling the chain needs more than 2/3 of counted bonded stake; halting it needs more than 1/3. The number of seats an attacker holds is irrelevant.
An attacker must add stake on top of honest stake
With honest bonded stake B, control requires buying A > 2B and halting A > B/2 — before price impact. At just 10% of supply staked, control means acquiring around 20% of all RAIN.
The 10% cap forces many validators
Each needs its own 25M RAIN self-bond, and the entry limit of two per day gives the community days of visible warning.
Slashing makes attacks self-destructive
Correlated double-signing by ≥ 1/3 of power loses 100%, and the 21-day unbonding keeps the attacker slashable afterwards.
The social layer
A safety violation is publicly provable; validators coordinate recovery that burns the attacker's stake under a pre-published procedure.
Get involved
Genesis operators
The genesis set targets at least 4 of 7 independent operators, with the active set opening toward 26 in Phase 3. Validator onboarding opens with the public testnet; a public application process will be announced here.
Coming soon Testnet validator programme