> I imagine that the top 100 of anything (that a lot of people do) that brings in income or fame will still be there in 5 years.

This sort of blanket assumption is exactly what the parent is arguing against. The mortality rate of top-n things is relatively easy to measure, and should be baselined first. Then we can compare recent performance vs historical performance, and actually say if something has changed. There's no need to start with the assumption "not much changes over 5 years" -- it can be measured instead.

If we use the S&P 500 as a proxy for the top 500 companies, I would guess that a lot more than 0% will be gone by the end of 5 years.

It's hard to find non-paywalled sources for business analysis, but from what I can find it'd be about 20%.