I think the key is that the correlation to ability is merely perceived. Has this ever been shown? Do people who code side projects in their free time actually tend to provide more value at work than people who don't? That would be interesting to find out.