You are correct and wrong.

In a standard distribution, a bigger dataset will contain more more extremes and more extreme extremes. But the size doesn't change the distribution. For every extremely fit person, you would expect a propositional amount of non-fit people. So while you might encounter more extremes in specific locations like a bodybuilder event, in daily life, you would realistically encounter the same subset of people, than in a smaller population.

If we assume that the author is not seeking out locations with extremes. While the author is more likely to meet the actually fittest person in America than in e.g. singapore, as the actually fittest person is more likely to live there. it is vastly more likely that the author met "their" fittest person in the country that they spend most of their time in.

So variance due to size and more extreme extremes, doesn't explain the perception. His location would. Or if the distribution would be different in e.g. America than in e.g. singapore.