37. The more number of diverse subnets a node connects to, the more benchmarks it can run.
38. The more benchmarks a node can run on data, the better it can predict its validly.
39. The better nodes can predict validity, the better data it can rely on.