I don't either & have already said as much earlier back in
post 76. The 5e & current playtest packet version of smite however applies a deceptive mechanic to provide a large number of charges in a small value. That small value allows it to justify a big punch while hiding behind "it maxes out at 4 charges & only a handful of slots compared to $otherClass' spell slots or $otherClass charges of their abilityB" when in reality those numbers are much bigger because $otherClass needs to consume the slot or charge
before the dice get rolled.
That would be much better & significantly mitigate a big part of the problems caused by retcon smites. I suspect it could lead to other problems by introducing a precedent like regaining a spell slot if the save fails as a mechanic that might wind up in the hands of more classes & magic items or interact with possible multiclass combos.
If paladins need specifically more charges to compensate for miss chance then they should have more charges, given a bonus to the smite save/attack they declared prior to rolling or be given an always on ability like radiant strikes. Something like one of those three would be a much better mechanic than a retcon because those work within the basic framework of the system rather than using the GM as glue for a weird inversion of the basic action declaration ->action resolution for one small chunk abilities mostly limited to a single class..