TAOSCOPE / Reference

Data sources

Where each number comes from, how current it is, and where it can be wrong. If you are making a decision that costs money, read this page.

Our sources

SourceWhat it provides
Bittensor chain RPCDirect queries against the Subtensor network, including subnet metagraphs — neuron UIDs, hotkeys, coldkeys, incentive, trust, consensus, stake, axon endpoints, and last-update blocks.
TaoScope indexerIndexed subnet, validator, pool, and transaction data, including historical series that would be expensive to reconstruct from raw blocks.
On-chain subnet identitySubnet names, descriptions, logos, and the repository, Discord, and website links shown on each subnet page.
On-chain account identityColdkey display names and logos attached to neurons, used on miner leaderboards when a miner has published one.
OpenTensor delegate registryCommunity display names for well-known hotkeys, used when on-chain identity is empty.

Miner leaderboards and reward distributions come from the chain metagraph rather than from an indexer, because incentive is the value miner earnings depend on and we want it as close to the source as possible. Miner names are filled from on-chain coldkey identity first, then the public delegate registry. Most miners have not published a name, so the truncated hotkey is shown instead.

Freshness and caching

TaoScope caches upstream responses. Caching keeps the site responsive and keeps us inside third-party rate limits, at the cost of some latency behind the chain head.

  • Cached responses are served for a short interval that varies by endpoint. Market data refreshes more frequently than structural subnet data.
  • When an upstream provider rate-limits us, we serve the most recent cached response rather than failing or showing zeros. That data may be older than usual.
  • Transaction lists stream from the TaoScope indexer as finalized blocks are written to TaoScope's Neon PostgreSQL explorer database. Browsers never open a Subtensor RPC socket.
  • Background producer jobs query chain RPC for metagraph/current-state data and publish shared Redis/Neon snapshots; normal page requests read those prepared results.

Do not treat displayed values as the chain head

For anything time-sensitive — most importantly the registration cost immediately before you register — confirm against the chain with btcli.

What we derive

Most values pass through unchanged. These are the exceptions, and each states its inputs.

Derived valueFormulaAssumption
Estimated daily emission ratePer-block emission × 7,20012-second blocks, which is the network's target block time.
Miner pool per dayDaily emission × 0.41The protocol emission split of 18% owner, 41% miners, 41% validators.
Per-miner earningsMiner pool per day × that miner's normalised incentiveIncentive is normalised across the subnet, so it is already a share.
TAO from RAORAO ÷ 10⁹Fixed chain constant.
Neuron scoresRaw u16 ÷ 65,535Chain encodes these scores as unsigned 16-bit integers.
Slot utilisationOccupied UIDs ÷ max_allowed_uidsBoth values read from the chain.

TaoScope publishes no editorial scores. There are no difficulty ratings, no hardware tiers, no task categories, and no opportunity indices. Anything of that kind would be our opinion presented in the visual language of measured data.

Known limitations

Missing values

Not every subnet reports every field. Where a value is absent we display “—”. Filters that sort on a missing value exclude that subnet rather than treating it as zero, which would otherwise place unknown subnets at the top of any ascending sort.

Snapshot timing

Different endpoints are captured at slightly different moments. Two values on the same page can therefore come from blocks a short distance apart, which is usually invisible but occasionally explains a small inconsistency.

Metagraph availability

Miner leaderboards depend on a chain RPC call that can fail or time out under load. When that happens the section reports that data is unavailable instead of rendering an empty or partial list.

Miner display names

A public miner name exists only when the coldkey has on-chain identity or the hotkey appears in the OpenTensor delegate registry. Most miners have neither, so the leaderboard shows the truncated hotkey rather than inventing a label.

Historical registration cost

The registration cost chart is built from snapshots recorded in your own browser as you use the site. It is not a complete on-chain history, and it starts empty on a new device.

Historical returns

APR and APY figures are derived from recent observed returns. They describe the past and are not a projection.

Verifying independently

For decisions with money attached, check the chain yourself. The CLI is the most direct route:

  • btcli subnets show --netuid <N> for live subnet parameters, including the current registration cost.
  • btcli subnets list for a network-wide overview.
  • btcli wallet balance for your own balances.

Found a discrepancy?

If a TaoScope figure disagrees with the chain, we want to know. Report it via the contact page with the subnet or address and roughly when you observed it.

Last updated August 27, 2026