One thing missing from the whole bimodal vs. normal distribution debate is: what is the population that you use as a base?
If you take all humans as the base, the vast majority for any skill that isn't nearly universal is clustered at the zero. Of the 9 billion people on earth, nearly nobody can write programs to any meaningful degree, or play the piano to any meaningful degree, or build a telescope, or a wrist watch, or whatever.
So I'd assume that programming ability distribution looks a bit like a normal distribution centered at zero, and cut off at zero (no values for skills less than zero).
Which, ironically, makes basically everybody who can write a single line of code "above average". Which doesn't make them rock star programmers at all.
If you take all humans as the base, the vast majority for any skill that isn't nearly universal is clustered at the zero. Of the 9 billion people on earth, nearly nobody can write programs to any meaningful degree, or play the piano to any meaningful degree, or build a telescope, or a wrist watch, or whatever.
So I'd assume that programming ability distribution looks a bit like a normal distribution centered at zero, and cut off at zero (no values for skills less than zero).
Which, ironically, makes basically everybody who can write a single line of code "above average". Which doesn't make them rock star programmers at all.