1 out of 6 programmers can't solve fizzbuzz? That seems way too harsh. 2 standard deviations would mean 2.5%, or 1 in 40 -- that seems more likely, at least in my DX
This is a well-known sampling bias. The number of interview candidates who can't solve fizzbuzz is very high, because those are the people who aren't in jobs and are therefore interviewing. Interview candidates are a skewed sample of the entire population.
