I can tell you the units you supplied are off. You need at least 2 more dimensions to get a volume of topsoil, otherwise it's a meaningless number. do they mean 3cm^3 of topsoil, because I guarantee that doesn't take 1000 years. Is it 3cm*earth_surface_area in a thousand years. Maybe... but my understanding is topsoil by erosion is the slowest way to get topsoil. I read a paper some years ago on biopurturbatuon, which changed my thoughts on topsoil mining. Basically worms, root structures, and other organisms are very good at creating topsoil. So you could very much produce huge amounts of the stuff in human time spans if the economic incentives were there.

It was an interesting paper, and I wonder if it didn't gain much traction, or was heavily refuted. If I was more cynical, I'd think Semedo was purposely ignoring it for political reasons.