I think in all 4 cases, there is enough variability within each company that speaking about "hackers from big company $X" in general is not particularly meaningful -- that is, the mean at Google may be slightly higher, but the variance is still dominant.
I think we all realize this. I think we all realize that "Nine-year-olds are taller than eight-and-a-half-year-olds" doesn't deny the fact that there's more variance among than between. But if you have two random samples from each group, are you going to bet that variance will swamp mean, or that mean will show up despite variance?
