Bittensor · interactive mechanism simulation

Miners submit AI work → validators score it and submit weight vectors → Yuma Consensus clips outliers and aggregates stake-weighted → TAO emissions split to miners, with the rest restaked into the subnet's dTAO alpha pool, which recycles through registration churn back into consensus. Edit the parameters and watch the incentive loop change.
Miners produce AI work Validators score work, submit weights Yuma Consensus 41% of emission → miners ⚠ weight-copying detected Alpha pool (dTAO) 0.0 TAO staked Subnet demand threshold: 120 TAO Reward queue pending payouts Registration churn 0 deregistered
Weight-setting rounds
0
TAO emitted (session)
0.0 TAO
Miners deregistered
0
Signal quality
100%
Alpha pool staked
0.0 TAO
Parameters — edit me
8
45%
41%
120 TAO
Controls

Illustrative simulation. Defaults mirror the researched Bittensor mechanism (stake-weighted Yuma Consensus clipping validator weight vectors, TAO emission split between miners and a subnet's dTAO alpha pool, registration churn recycling slots), but round timing and amounts are randomized for visualization — not live onchain data. The "weight-copying attack" button illustrates the documented failure mode where validators copy the consensus vector instead of scoring independently: emissions keep flowing even as signal quality collapses. Drag the sliders to explore how the incentive loop responds. Part of The Onchain Experiment Atlas.