In general, most intelligent people are aware of the dangers of sampling bias and try to avoid it. But what do you do when your sample size is really small? We extrapolate from what we have seen work, but only a handful of open source projects have had the longevity and success that Linux has had.

We see the same thing with decisions on what kind of movies to make. A handful of high budget movies are made every year, and when its time to decide which movies to fund studios extrapolate from past successes and failures. This includes wrong ideas like "no movies centred around female superheroes because Catwoman, Elektra etc" or "international audiences only connect with white leads, not with black ones". It goes the other way too, Black Panther's success suddenly means that international audiences love black actors.

Me personally, I don't think either of the conclusions wrt black actors is correct - there's insufficient data to support either hypothesis. But people make do with the data they have, which is the same thing with Linux. "Linux is wildly successful - what do they do differently? Hmm, must be Linus' management style"