[taeric]

Funny framing of it, but that isn't necessarily how the median works. Consider, what are the median number of fingers on a person's hand? How many people have less than the median number of fingers? :D

Is a fair point that it could be more common than I'd expect. I remember being a teenage driver and we were quite bad. Would love to see data on this.

[falcolas]

The normal way I hear it stated is "half of X are worse than average". But an average would not work with that statement in a pedantic environment. HN is a pedantic environment.

"On a distribution curve of driver skill, half of the drivers will be of lower skill than the driver at the median of the data set. Unless there's an even number of entries, in which case there is no median driver, just an inferred skill value based on the values to either side of the middle of the data set."

[taeric]

Before I finish typing this, I want to make it clear that I'm just having fun with the numbers/math/language here.

Even that statement doesn't really work outside of fully unique values across a specific distribution. Take samples where you have many duplicates, and you can easily have the vast majority of values flat out be the average/median when you have some distributions. Is why I picked average number of fingers. The VAST majority of counts there are the same value. The outliers being dwarfed to insignificance both high and low.

For more continuous distributions, https://stats.stackexchange.com/questions/326304/in-a-given-... is a pretty good quick google of what I mean.

Now, what is the distribution of driving ability? I have no idea. I see no reason to think it is a normal distribution, though.