Live from the chain
Subnet pools
Reserves, spot alpha price and pool depth read directly off the Finney chain — the mechanism, shown live.
The instrument
The Observatory
An observation post. We watch what the network does — what changed, when it changed, and what else moved at the same moment — and publish what we can show. Every figure carries its source and its age. Where the cause of something is not knowable from the data, we say that instead of inventing a reason for it.
01
The TAO journey
Every price chart tells you what happened and then quietly implies why. This one tries to separate the two. Each marker is a day big enough to matter, and beside it sits Bitcoin's move on the same day — because the first question about any spike is whether anything happened at all, or whether the whole market moved and TAO simply moved further. Markers in red are days the market does not explain. That is a statement about arithmetic, not a claim that we know the cause.
02
Where the network stands
Reserves and spot price are read straight from Finney: the price is the pool ratio SubnetTAO ÷ SubnetAlphaIn, and depth is the TAO side of that pool. This table is rendered from the same snapshot file the written pieces cite — so the number you see here is the number a report can be held to.
Raw view — the pool table
The same snapshot as a table. Kept because a chart is a summary and a table is not: if you want to check one subnet rather than watch the network, this is the honest form for it.
| Subnet | Price (τ) | TAO in pool | Alpha in pool | Depth (τ) | Locked α | Emission |
|---|
Reading the latest snapshot…
Raw data: data/latest.json · the append-only log is in data/snapshots/. If the table above does not render, the numbers are one click away rather than hidden behind a failure.
03
Where the stake sits
Staking TAO is not one decision. It can sit in root — the network's default position — or in a subnet, where it shares that subnet's fortunes. This panel tracks how much TAO sits in root, and which way it has been moving. Both series are storage reads from the chain, sampled at historical blocks from an archive node, so any point can be re-derived rather than believed.
The same period as a direction, month by month. Copper bars are TAO moving into root; grey bars are TAO leaving it. A level line alone cannot tell you this — a flat-looking line can hide a quarter of inflow and outflow at the same time.
What this measures, and what it does not
- Root pool is SubnetTAO[0] — the TAO reserve of the root subnet's own pool. After dTAO that is the direct reading of "staked to root".
- The quantity did not exist before dTAO. SubnetTAO[0] is absent at block 4,920,350 and present at 4,920,351, so the copper line has no earlier history to draw. The grey total-stake line is there to carry that period.
- The pool was seeded, not grown. At activation it opens at 5,864,261 τ — identical to the total stake read from the block immediately before. We check that in the data and state it, rather than letting it imply root accumulated that size organically.
- Deliberately not shown: root alpha outstanding. On root, SubnetAlphaOut[0] reads exactly equal to SubnetTAO[0], which is not how a normal subnet behaves and which we cannot yet explain. It stays off the page until it is understood.
- Sampling: weekly before dTAO and daily after, at fixed block intervals, using each block's own timestamp — so the x-axis is the chain's time, not our fetch time. Monthly bars are the difference between the first and last sample in each month.
Sourced feed
Network activity
Registration, lock-mode and emission changes, shown with the exact timestamp of the underlying source.
Historical
How things changed
Time-series built from archive-node reads, so long-range claims can be reproduced rather than believed. Currently: how much TAO sits in root, and which way it has been moving.
Downloads
Snapshots
The raw data behind each published report, exported with its query and a content hash. The log is already building, one row per six hours.
A note on freshness
Some third-party feeds republish on their own schedule and can lag by hours. Rather than pretend otherwise, this page states, on every figure, when the underlying source last changed — not merely when we last fetched it. Delayed data is still useful data, as long as nobody is told it is live. The snapshot above is taken every six hours; its age is computed in your browser each time you look, so a stale snapshot cannot quietly present itself as a fresh one.