:D ja, things like this happen but are more sound logic heuristics than predictions I guess.

Stake is subject to centralization because its input follows pareto. But proof-of-stake is a distribution-agnostic oracle while dPOS is a consensus mechnism. Proof-of-Stake can be implemented in a consensus mechanism which is resistant against centralization. See the Casper FFG Paper or Avalanche?