VERIFIED against subtensor source code (run_coinbase.rs, subnet_emissions.rs)
1. Block emission: 0.5 TAO/block (logarithmic halving, DefaultBlockEmission=2.0, residual=2)
2. Alpha emission: 1.0 alpha/block (VERIFIED via taostats API: alpha_rewards = 1e9 rao)
3. Emission share: price_i * (1-miner_burned_i) / sum(...) [PR #3058 pending: removes burn weighting]
4. Hill gate: 1/(1+(theta/s)^h), applied per-subnet then renormalized
5. Alpha split: owner 18%, then 50/50 miners/validators. Root takes rootProp of validator half
6. Chain buy cap: root_proportion * alpha_emission (= rootProp * 1.0, VERIFIED against taostats)
7. Owner cut: 18% global, chain-set (not adjustable by subnet owner)
8. Root dividends: always sold for TAO (staker yield). Null = stakers sell individually. Curated = root pool sells + reinvests.
9. Injected TAO: total TAO inflow minus chain buys. Enters pool as liquidity paired with minted alpha. Absorbs sell pressure.
APPROXIMATIONS (cannot compute without full chain state)
A. Spot vs EMA: calculator uses spot price for SN25, on-chain uses SubnetMovingPrice (EMA). Overestimates speed of share changes.
B. Field gate = 1.0: on-chain gates each subnet individually. Weighted avg ~0.99 (active subnets pass, tail subnets ~0 weight). Error: ~1%.
C. Theta: linear scaling from calibration point. Valid when only SN25 price changes. Breaks if all prices scale together.
D. Field EMA: sum of all other subnet prices. If field changes, theta recalibrates.
E. Root dividend pool: 983 TAO/day (chain constant, may change with governance)
SIMULATION ASSUMPTIONS (user-set, not from chain)
S1. Miner sell %: no on-chain data for sell-through rate. Simulation parameter.
S2. Validator sell %: same. Based on FlowSniper research suggesting most alpha gets sold.
S3. Owner sell %: owner's choice to hold or sell.
S4. Root basket weights: future-state simulation for curated mode.
S5. Alpha emission = 1.0: verified via taostats API (alpha_rewards = 1e9 rao). If a subnet exceeds 21M alpha outstanding, emission may change.