This is a pretty poor criticism of the normal distribution, for one it doesn't contain a single mention of variance or standard deviation which is as important as the mean. Modelling something with a normal distribution does not say that any member of the sample actually recorded the average value, it simply states that the values cluster around the mean. If the average American household has 2.5 children, it's fairly obvious that there is not a household in the country that has the mean value. I haven't bothered to look at the underlying distribution of children but the principle is the same.
