This is an interesting argument; thanks for sharing it. I didn't understand it right away and my intuition was that you must be incorrect. But after playing with some actual numbers I've come to understand. In an attempt to restate your claim in a way that's more intuitive to me:
If a group is reproducing above the replacement ratio, then they haven't hit peak child yet. So if only one subgroup remains that hasn't hit peak child, then that subgroup is also the only one that's growing (all others are shrinking). Assuming that trend continues (read: assuming subgroups have birth rates that reflect their culture, and that culture will be preserved in the next generation), this subgroup will grow without bound.
The pressure-release on a system like this (one which the article you linked points out in its conclusion) is that birthrate within a subgroup isn't fixed. The birthrate among the Amish appears to be declining slightly, perhaps influenced by economic forces in the US. In your example, while it's true that _tommorow_ there will be more Mormons and fewer Italians than today, that doesn't guarantee that eventually the world will be populated entirely by Mormons, because the birthrate in that subgroup isn't wholly 'inherited' (in the cultural sense) -- it may change in ways that we can't forecast today.
Not really, because subgroup membership is flexible: people can enter and leave a subgroup. But even ignoring that objection, calculating that the Amish will eventually outnumber everyone else is a crazy bad extrapolation, whereas predictions of world population are based on much more modest and justified extrapolations.
If you took some old population data you could perhaps extrapolate that Shakers were destined to increase exponentially, but what happened to them? I'd say that the destiny of a single religious group is typically a short-term and local phenomenon that doesn't much affect the long-term development of world population.
Proof: Assume the opposite, that there is a subgroup S that never hits a maximum amount of children, but the overall population has a maximum amount of children X. Since S does not hit a maximum amount of children, S at some point has more than X children. This contradicts the initial assumption. QED
By the same token, if a certain group keeps reproducing at a high pace, but the majority of the world keeps having fewer and fewer children, the overall number of children will decrease, so the curve will be past the peak.
No, it just has to make less children than it did eighteen years before.
Well yeah, obviously as long as the population keeps growing [1] (and for that it is sufficient that a subgroup does indeed), there can not be a peak child.
I was assuming that the article posited the existence of a population peak in the future, and was speculating about how many years would it be preceded by a child peak.
1. or rather, renew generations, since the population can grow without renewing generations.
Looking at overall birth rates might make it seem that way. But logically, in order to hit peak child, every subgroup will also have to hit peak child. Will every single country hit peak child? Will every religious group hit peak child? All it takes is one of those groups to keep expanding, and the world will not hit peak child. There are so many subgroups with drastically different birth rates and retention rates, I don't believe we will hit peak child. There will just be a transition period as the world starts to contain fewer Italians and more Mormons.
For an interesting take on a single subgroup's demographics, the Amish, check out this article: https://medium.com/migration-issues/how-long-until-were-all-...