incentives

Why knowledge creation needs a reward system, and how tru makes contributing to the cybergraph more profitable than free-riding.

the free-rider problem

Knowledge creation is costly. Discovering a genuine connection between two ideas -- finding that a particular particle relates to another in a way nobody has noticed -- takes time, expertise, and attention. The benefits of that discovery, once published as a cyberlink, flow to everyone who reads the graph. The discoverer bears the cost; the collective reaps the reward.

Without incentives, rational agents free-ride. They consume the graph's knowledge without contributing their own. The result is an epistemic tragedy of the commons: the graph stagnates, the good links stop appearing, and noise fills the vacuum.

the key insight: reward proportional to Δφ*

tru's answer is a single principle: reward is proportional to the focus shift your action creates.

$$\text{reward}(v) \propto \Delta\phi^*(v)$$

φ* is the focus distribution -- the collective attention of the entire cybergraph, computed by the tri-kernel. When you add a cyberlink that shifts focus toward a previously overlooked particle, you have created structure. That structure is value. The protocol recognizes it and mints $CYB proportional to the shift.

This means creating knowledge IS creating value. There is no separate reward pool, no committee allocation, no inflationary emission schedule disconnected from contribution. New tokens appear if and only if the graph gains new structure. The protocol's inflation is literally evidence of knowledge creation.

the discovery premium

The first neuron to surface a valuable particle captures the largest Δφ*. When nobody has linked a particle, the potential focus shift is enormous. The second neuron to link the same particle earns less -- the marginal gain is smaller. The hundredth neuron earns almost nothing.

This is the attention yield curve: early, accurate discovery is maximally rewarded. Late consensus-following earns little. The mechanism creates a race to discover genuine relevance rather than copy existing links.

self-minting

Rewards are not computed centrally. Each neuron proves its own contribution and claims its own reward.

The process: create cyberlinks, compute the local focus shift $\Delta\phi^*$, generate a stark proof that the computation is correct, bundle everything into a cyber/signal, and submit. Any verifier can check the proof against the block header in O(log n) time. If valid and Δφ* > 0, the neuron mints $CYB.

A neuron on a phone can participate: buy a header, query the neighborhood state, create links, prove Δφ*, mint tokens. No mining pool, no centralized aggregator, no permission.

confirming links vs foundational links

Different kinds of knowledge earn differently over time.

A foundational link -- the first connection between two important but previously unlinked concepts -- starts with low Δφ* that grows over time as the graph builds around it. The reward trajectory rises slowly and persists. This is infrastructure work: the neuron who lays the first bridge between two knowledge clusters earns a long-term yield.

A viral link -- a connection to a particle that immediately attracts attention -- earns high Δφ* early but decays fast as focus saturates. Quick returns, short horizon.

A confirming link -- the second or third signal reinforcing an existing connection -- earns lower individual Δφ* but strengthens the axon weight between clusters. Credit is shared through Shapley attribution, which divides the joint focus shift fairly among contributors.

A semantic bridge -- a cross-module connection -- earns moderate, persistent rewards because it improves the graph's global connectivity.

the game

The rules produce a game where early + accurate = maximum return:

• early, accurate links to important particles earn the most (the attention yield curve)
• confirming links strengthen axon weight -- repeated signals build consensus, not noise
• neurons build long-term reputation through accumulated karma
• focus as cost ensures every cyberlink is a costly signal -- you must stake real $CYB to play
• staking cost filters noise, reward function amplifies signal
• a neuron must risk real tokens to earn rewards, ensuring alignment between economic interest and knowledge production

The evolutionary loop that emerges: contribute accurately → Δφ* reward → accumulate $CYB → stake on more links → accumulate karma → links carry more adjacency weight → earlier Δφ* attribution → more $CYB per contribution. The flywheel rewards sustained accuracy, not one-time luck.

see reference/rewards for the formal reward functions, attribution formulas, and self-minting protocol. see docs/markets for how epistemic markets complement the incentive layer. see docs/honesty for why honest signaling is the dominant strategy.