Yuma Consensus

Validators evaluate miners and submit their scores (weights ) to the blockchain. Yuma Consensus processes those weights to determine how ALPHA emissions get distributed. The key idea is that a single validator's opinion does not directly determine rewards, instead, the algorithm finds what the majority agrees on.

The blockchain calculates a stake -weighted median for each miner's score. Validators with more stake carry more weight in this calculation. The result is a consensus score for each miner, representing what most of the stake considers a fair evaluation.

Weights that exceed the consensus score get clipped (brought down to the consensus score). If a validator rates a miner significantly higher than the majority agrees on, that excess gets removed. Weights below consensus stay as they are. This clipping prevents any single validator from inflating a miner's rewards beyond what the others consider fair.

After clipping, the blockchain calculates a trust score for each validator. Validator trust measures how much of a validator's original weights survived clipping. If a validator's weights were already close to consensus, most of their weights remain intact and their trust is high. If their weights were frequently clipped, their trust is lower.

Validators with higher trust retain more influence over emission distribution. This creates a direct financial incentive to score honestly (validators whose weights consistently match the majority earn more dividends over time).

Validators create a "trust score" called V-Trust by comparing their ratings. When validators rate miners similarly, they earn higher trust. This method stops people from giving unfair or random scores. Validators whose ratings match the network's overall view have more influence.

Additionally if used features like liquid alpha and commit-reveal incentivize validators to be honest and to not copy weights from others.

Bonds track the relationship between validators and miners over time. Each time Yuma Consensus runs, the bonds between validators and the miners they score get updated using an exponential moving average. This means bonds build up gradually, rewarding validators who consistently score miners rather than those who change their scoring frequently.

Validator dividends are calculated from these bonds. A validator earns more dividends from miners they have strong bonds with, meaning miners they have been consistently scoring over time.

Once consensus, clipping, and bonds are calculated, the algorithm produces two emission scores for each participant: incentive for miners and dividends for validators. These scores are normalized and multiplied by the total ALPHA emission for that interval to determine the actual amount each participant receives.

If no valid weights exist (all validators are inactive or no weights have been set), the emission defaults to being distributed proportionally by stake.