Geod-Ellipsiod separation dynamics both mean globally and on a plate by plate basis accounting for creep is a thing. As is epoch based modelling and forward and backward projection.

So measurements from harbour posts can be reconciled to WGS84 despite a bit of Californian bounce.