One critique that I am noticing is that I think your formula does not take into account that the rewards claims needs to be recomputed with the curve. The formula is actually

R(r) * (reward pool / claims)

where claims is roughly the sum of R(r) over all pending posts (the actual implementation is that they are added to claims right before payout and a per-block exponential decay is added to phase those balances out).

This means that it is pretty dependent on the distribution of posts to begin with and can greatly influence your analysis, especially when it comes to sigmoid curves whose derivative tapers off.

But, in any case the analysis on what happens is roughly right in penalizing the low end as an immediate effect. However, what happens after will affect the payout distribution again, and it's hard to tell how it will shift. But we can certainly make guesses if there's less of "x" kind of posts rewarded and "y" downvotes apply to the top, etc etc.... But that's a very large rabbit hole....