You're actually right. His citation has the correct statistic in the first paragraph (replace 2 with 3).
This struck me as a misleading use of statistics, because Pareto likes to say hi everytime we try to correlate a repeatable event and a global population. Most of those tickets will actually go to the same drivers, and Pareto's Law even gives us handy empirical proportions. 80% of the tickets will probably go to the same 20% subgroup.
This means that we have ~8 million tickets to distribute more or less randomly/statistically (over let's say ~8 million american citizens), and then ~2 additional millions get the rest of the tickets, for a total of 10 million ticketed people per year, or roughly 3-5% of the population.
Which is still huge and doesn't exactly change the author's point.