Planetarium No. 06 · Solana mainnet-beta

Every validator,
a star in one sky.

Validome maps Solana's validator set as a live night sky — who holds the stake, how few it takes to halt the chain, who has gone dark, and what software they run.

Stars in the dome—
Nakamoto coefficient—
Dark (delinquent)—
Epoch—

Bright cyan stars and the rose constellation = the superminority (smallest group holding >33.3% of stake). Star size ∝ √stake. Live data, not a mock-up.

I · The observatory

Live validator set

Pulled straight from getVoteAccounts, getBlockProduction and getEpochInfo on public Solana RPC. Cached in your browser for 10 minutes so we don't hammer a shared node.

Loading the sky…

Active stake

—

SOL delegated to voting validators

Validators voting

—

vote accounts in "current"

Nakamoto coefficient

—

validators needed to pass 33.3% of stake

Delinquent

—

vote accounts not voting recently

Median commission

—

stake-weighted mean: —

Cluster skip rate

—

leader slots without a block, this epoch

Try:

Cumulative stake curve

Validators ranked by stake (x) against the share of total stake they hold together (y). The rose line marks 33.3% — the halting threshold.

Stake bands

How many validators sit in each stake band, and how much stake each band holds.

Commission spectrum

Validator count by commission rate. Bar fill shows stake share at that rate.

The superminority constellation

The smallest set of validators that together hold more than one third of stake. If all of them went offline together, the network could not finalize blocks.

    Dark stars (delinquent)

    Vote accounts the RPC node reports as delinquent — they haven't voted on a recent slot. Stake on a delinquent validator earns no rewards while it is dark.

      Client census

      Which validator software is running the stake? This uses getClusterNodes, a ~1.5 MB response, so it only loads when you ask.

      Star catalogue

      Solana validators, sortable by column
      Status
      Loading…
      —

      II · Mechanism

      How the dome is drawn

      1. 1. Ask the cluster

        Your browser sends one JSON-RPC request to a public Solana node for every vote account, its delegated stake, commission and recent vote credits. One request, ~300 KB, cached for 10 minutes.

      2. 2. Rank by stake

        Validators are sorted largest to smallest. We walk down the list adding stake until the total passes one third — that count is the Nakamoto coefficient, and those validators form the superminority.

      3. 3. Measure behaviour

        getBlockProduction tells us, per identity, how many leader slots each validator was given this epoch and how many blocks it actually produced — the gap is its skip rate. Vote lag compares its last vote to the current slot.

      4. 4. Optional census

        Press "Load client census" and we fetch the gossip table (getClusterNodes), match each validator's identity key, and tally stake by client software and version.

      III · Uses

      What you can learn here

      01

      Check a validator before you stake

      Paste its vote key to see commission, stake, recent vote credits, skip rate and whether it is delinquent right now.

      02

      Watch decentralisation

      The Nakamoto coefficient and superminority list show how concentrated stake is — and whether your delegation adds to the concentration.

      03

      Spot validators going dark

      The delinquent list surfaces vote accounts that have stopped voting. Stake there earns nothing until they come back.

      04

      Compare commissions fairly

      See where a validator's commission sits in the whole-network spectrum, both by count and by stake weight.

      05

      Track client diversity

      A network where one client runs most stake is exposed to that client's bugs. The census shows the mix.

      06

      Teach consensus visually

      The dome and stake curve are a quick way to explain to newcomers why "one third of stake" matters on Solana.

      IV · The star atlas

      A field guide to Solana validators

      Vote account vs. identity

      Every validator has two important keys. The identity key (the "node pubkey") signs blocks and gossip messages and pays the voting fees. The vote account is a separate on-chain account that records the validator's votes, its commission and the credits it earns. Stake accounts delegate to the vote account, not the identity — which is why staking tools always ask for the vote key. Validome lets you search by either.

      Stake and leader slots

      Solana is proof-of-stake: the more SOL delegated to a validator, the more its votes weigh in consensus and the more leader slots it is assigned in the leader schedule, which is computed per epoch (432,000 slots, roughly two to three days). A leader slot is a ~400 ms window in which that validator is expected to produce a block.

      Skip rate

      If a leader fails to produce a block in its slot — because it was offline, slow, or its block didn't propagate — the slot is skipped. Skip rate = 1 − blocks produced ÷ leader slots. Some skipping is normal across the whole cluster; a validator with a skip rate far above the cluster average is losing its own block rewards and fees.

      Commission

      Inflation rewards are paid per epoch to stake accounts in proportion to the vote credits their validator earned. The validator keeps its commission percentage and passes the rest to delegators. 0% commission is common for validators funded in other ways (for example MEV tips or block fees); 100% means delegators get none of the inflation reward. Commission can be changed by the validator, so it is worth re-checking.

      Vote credits

      Validators earn credits for voting on blocks that become rooted, with more credits for faster votes (timely vote credits). The epochCredits field lists, for the last few epochs, the cumulative credits at the end and start of each epoch; the difference is what the validator earned that epoch. Similar credits across validators mean similar performance — a big drop is a red flag.

      Delinquency

      An RPC node marks a vote account delinquent when its last vote is more than a set number of slots (127 by default) behind the tip. Delinquent stake still counts as delegated, but earns no rewards while dark, and it doesn't help consensus.

      Superminority & the Nakamoto coefficient

      Solana's Tower BFT consensus needs more than two thirds of stake to confirm blocks. So any group controlling more than one third can stop finalization by going offline together. The smallest such group is the superminority; its size is the Nakamoto coefficient. A higher number means it takes more independent parties to halt the chain. Note: several vote accounts may be run by the same operator, so the real number of independent parties can be lower than the count shown.

      Client diversity

      Validators run different software: Anza's Agave, Jito's Agave fork with MEV block-building, Firedancer/Frankendancer from Jump, and other forks. If one client holds more than a third of stake, a single bug in that client could stall the network; more than two thirds, and a bug could finalize something wrong. A spread across clients is healthier.

      Epoch
      432,000 slots; stake changes and rewards apply at epoch boundaries.
      Slot
      A short, sub-second window with one scheduled leader (leaders get 4 in a row).
      Activated stake
      Delegated SOL that is fully warmed up and counts this epoch.
      Root
      The latest slot a validator considers finalized.
      Lamport
      1 SOL = 1,000,000,000 lamports. RPC reports stake in lamports.

      V · Questions

      Frequently asked

      Limits & honesty

      • No names. Validator names require a method the public RPC blocks, so only keys are shown.
      • One node's view. All figures come from solana-rpc.publicnode.com. Another RPC node may disagree slightly about delinquency at any instant.
      • Vote accounts ≠ operators. One operator can run many vote accounts; the Nakamoto coefficient here counts accounts, which flatters decentralisation.
      • Census coverage. Some validators don't appear in the gossip table at the moment of the query or report no client ID; they are shown as "unknown".
      • Skip rate is epoch-to-date. Early in an epoch it is based on few leader slots.
      • Rate limits. The node is shared and free. If requests fail, wait a few seconds and retry.