The issue is he then uses the approximations to do with math without calling them approximations. 1.10^7 is reasonably close to 2, but 1.1^21 is 7.4 which is a fair distance from 8.

He goes so far as asks someone to do the approximation across several hundred years of compounding. And sure it get’s a big number but one no even close to accurate.