We redesigned proof-of-compute rewards so the chain naturally becomes whatever the market needs it to be

Most blockchains hardcode reward splits at genesis. 55% miners, 20% storage, etc. Those percentages are a forever-bet on what the chain will be used for. BTCPC does it differently.

The model

Every pool (inference, storage, sensors, verification, services) gets a calibration target. Your actual activity divided by your calibration target gives a utilization score between 0 and 1. Your share of the reward pool is proportional to your utilization score relative to everyone else's.

No fixed percentages. The split emerges from what the network actually does.

Four layers that compose

• Layer D — tiny infrastructure base for clock nodes and testnet operators. Always fires, keeps the chain alive when idle.
• Layer B — main dynamic pools. If inference dominates, inference earns more. If storage dominates, storage earns more. Market decides.
• Layer A — long-term scalar. Dual EMA (7-day fast + 90-day slow) scales the reward ceiling from 70-100% of block reward. Sustained activity raises it; sustained quiet lowers it; always pulls back to 90-day average.
• Layer C — fee-driven boost. Previous epoch's verified fee volume pushes pools toward the ceiling. Three-layer protection against circular payment attacks: net flow accounting, 20% job slippage, and a two-epoch capital lock.

Nothing burns

Undistributed rewards always go to the recycle fund, never to a burn address. Recycle fund sustains era-5 rewards (~124 years from now) and tops up the testnet fund perpetually.

This is live on testnet. Calibration targets auto-adjust EIP-1559-style toward 50% utilization. Governance can nudge them if needed, but the chain self-corrects.

Not affiliated with anything. Just building.