learning incentives

learning is mining: every link you cast mints your stake in the graph. teaching is staking: when the focus of others flows through your links, the reward compounds. collective intelligence converges on what genuinely matters — without voting, without moderators, without any central authority — and pays the people who taught it.

every reward in cyber traces to one quantity: the impulse — the proven change in collective focus $\Delta\phi^*$ a neuron delivers by adding cyberlinks. a neuron computes the shift locally on its neighborhood, proves it with a zheng proof against the BBG root, and self-mints $CYB proportional to its fair share of the shift. minting is bounded by the actual global $\Delta\phi^*$ (conservation), so inflation is evidence of knowledge creation — there is no emission without demonstrated contribution. creating valuable structure is creating value: no designed loss function, the physics of convergence defines what deserves reward.

the full specification — reward functions, Shapley attribution, self-minting, token operations, and the settled design — lives in tru/specs/rewards. see cyber/tokenomics for system-level economics and collective learning for group dynamics.