Suppose a team was given a goal of reducing variance in latency on an API. Getting to the root causes of variance and fixing them is hard. Injecting extra lat

Suppose a team was given a goal of reducing variance in latency on an API. Getting to the root causes of variance and fixing them is hard. Injecting extra latency on the fast queries is easy. Even if the query returns in 80ms, we just wait until 800ms have passed before returning the result. These are the same incentives given to administrators tasked with improving equity.