The law of large numbers

Hello Rabbi Michi,
If I understand correctly, the law of large numbers says (in a not-so-precise definition) that if you observe an infinite series of random observations that are independent of each other, all of which describe the same phenomenon, then the average of the series will gradually approach a constant value.
For example, in a large enough sample of people with a certain trait, the average will be similar to the general population from which the sample was taken.
Nadav Shnerb once wrote about this in connection with education (in the first person): "The educational action that these people imagine can only be done by people who are themselves noble people, and it seems that everyone understands that in a system of mass education, with hundreds of thousands of teachers, there is no possibility of achieving this goal. The law of large numbers ensures that teachers will be more or less a representative sample of society (perhaps minus the really poor classes, but not much more than that), and therefore the typical teacher will not be a more moral person or serve as a better personal example than the average parent."
The questions are as follows:
A) It is clear that the value of the sample is percentage and not absolute, so what is the threshold of sample size at which it would be possible to make Shnerv's claim?
B) Is it possible to influence the law through an education system, incentives, etc.? Let's take engineers for example. If, through education and economic and social incentives, etc., we cause all 10% (let's assume this is the threshold for question 1) of the population who have more than average engineering skills to become engineers, then does this mean that the average ability of an engineer will be equal to the average engineering ability of the population? Doesn't that go against logic?
If the answer is yes, then Shnarb's argument falls apart. If we succeed in making most teachers come from the more appreciative percentiles, then the appreciative average will be greater than the appreciative average of the population. And even if we say that this is illogical, at the very least we will reduce the sample size and only try to take the principals and educators from the more appreciative percentiles of the population.
 
thanks
N
Link to Shnarev's statement