ideally we would also collect a "running standard deviation" besides collecting the average of a given endpoint

[mircealungu]

stack overflow discussion on the topic: https://stackoverflow.com/questions/1174984/how-to-efficiently-calculate-a-running-standard-deviation

that would allow us to detect "outliers" scientifically :)

[kloostert]

Do we still want this running standard deviation to be implemented?