Revisit the time series granularity examples in https://www.mongodb.com/docs/manual/core/timeseries/timeseries-granularity/ for clarity and accuracy against the internal system.buckets collection behavior.
Clarify why setting the granularity too low is a bad idea, per Jeff's review: "Do you think it'd be appropriate to briefly explain why this is bad, kind of how we did for the description for hours where we mentioned the longer traversals / slower queries?"