Flow-Based Emissions

Flow-Based Emissions
Context	TAO reserve injection
Signal	Net TAO flow
Scope	Across subnets
Boundary	Before TAO and alpha rewards

Bittensor emissions include TAO and subnet-specific alpha tokens. Flow-based emissions are the subnet-level allocation model that compares net TAO movement before TAO reserve injection is assigned across subnets. The term belongs to the injection side of emissions, before within-subnet reward distribution.

References: Emission: Distribution Across Subnets, Emission: Distribution

Injection Context

Bittensor emissions can be read in two stages. Injection adds liquidity to subnet pools, while distribution allocates value to subnet roles. Flow-based emissions belong to the injection stage, where net TAO movement informs TAO reserve injection.

That makes the term narrower than emissions as a whole. It does not rank one miner, calculate one validator dividend, or describe an entire incentive mechanism. It sets a subnet-level reserve-injection context.

The term therefore belongs near TAO reserve injection. Net TAO flow helps determine which subnet contexts receive reserve injection, while later distribution logic decides how value is handled inside each subnet.

References: Emission: TAO Reserve Injection, Emission: injection and distribution

Net Flow Signal

The signal compares TAO entering a subnet through staking with TAO leaving through unstaking. The net direction of that movement is the demand signal used when subnets are compared.

The important idea is the direction of flow. Net inflow supports stronger injection-share context, while net outflow weakens it.

That makes flow-based emissions different from a price-only model. The comparison is based on net TAO movement into and out of subnet pools, not just a static reserve-ratio snapshot.

This also separates flow-based emissions from alpha price and reserve ratio. Those terms describe pool relationships; flow-based emissions describe how net TAO movement influences the injection share assigned across subnet pools.

References: Emission: Distribution Across Subnets, Understanding Subnets, Emission: TAO Reserve Injection

Smoothing Context

Flow signals can be noisy, so smoothing helps prevent brief bursts from dominating subnet comparison. Bittensor uses exponential moving averages in several contexts for smoothing changing signals.

Here, smoothing belongs to subnet comparison. It is not the final reward table inside a subnet and not a replacement for the Yuma Consensus path that interprets validator signals.

Smoothing also keeps the term tied to subnet comparison rather than a single block’s movement. The flow signal is interpreted as part of an allocation model rather than as an isolated staking or unstaking event.

References: Exponential Moving Averages, Emission: TAO Reserve Injection, Yuma Consensus

Negative Flow Context

When smoothed net flow is negative, the emissions documentation connects that condition with a zero-injection outcome for the affected subnet.

Zero injection is an injection-stage result. It is not a separate penalty rule for an individual miner, validator, or staker.

The condition is still scoped to subnet-level injection. A subnet can have a weak or zero injection context without that sentence explaining every later scoring, weight, or distribution outcome inside the subnet.

Zero injection is also separate from deregistration. Flow-based emission context can reduce or remove injection for a subnet, while subnet-removal logic remains a different tokenomics and subnet lifecycle topic.

References: Emission: Distribution Across Subnets, Subnet Deregistration

Distribution Handoff

After subnet-level injection is resolved, reward allocation still depends on within-subnet logic. Yuma Consensus explains how validator evaluations feed miner incentives and validator-side outcomes inside the distribution stage.

Flow-based emissions therefore compare subnets. They do not replace consensus, weights, or within-subnet reward allocation.

That means the same emission page can contain several layers. Flow-based emissions explain cross-subnet injection; distribution explains role-level allocation; Yuma Consensus explains the evaluation method used inside a subnet.

References: Yuma Consensus, Emission: Distribution, Emission

Development Stage Context

Bittensor separates mainnet, testnet, and localnet environments, so flow examples from one environment do not carry over automatically to another (Bittensor Networks, Introduction to Bittensor: Subnet development).

Subnet reserves, staking flow, and emission state can differ across environments. Localnet examples are isolated development examples. Testnet examples are shared non-production examples. Mainnet flow-based emission interpretation concerns production Dynamic TAO behavior on the active network.

Relationship to Yuma Consensus

Flow-Based Emissions and Yuma Consensus describe related parts of Bittensor’s incentive system. Yuma Consensus is the on-chain process that aggregates validator weight signals within a subnet into miner incentives and validator dividends, applying consensus clipping, bonding, and emission calculation (Yuma Consensus).

For readers, flow-based emissions names a specific part of that incentive picture, while Yuma Consensus names the consensus process that turns validator weights into the resulting incentives and dividends.

Reader Boundary

Flow-based emissions are tokenomics vocabulary for subnet comparison. They are not a staking tutorial, miner-ranking formula, return forecast, or complete explanation of all Bittensor emissions. The stable point is that net TAO movement helps determine subnet-level reserve injection before rewards are allocated inside the selected subnet (Emission: Distribution Across Subnets, Yuma Consensus).

Flow-based emissions compare subnets at the injection stage. Yuma Consensus and within-subnet distribution explain later role-level allocation; zero injection is an injection-stage result, not a separate penalty rule for an individual miner or validator (Emission: Distribution Across Subnets, Subnet Deregistration).