Who defines the preset sound at MIDI velocity 64? ...

... a person or an algorithm?

A piano preset sound at MIDI velocity 64 is in the "middle" of a scale of 128 possible values. Who decides, which sound of a recorded piano represents this velocity 64 in an instrument preset?

Is it a person, who says: This recorded sound represents the medium character of that instrument the best to my ears, let's define it as vel 64!

Or is it an algorithm, that divides the "loudness" range in 128 steps from the quietest to the loudest playable sound of a note?

If there is no systematic in this process, it probably makes sense to customize the velocity curve for each instrument preset. Maybe even after a revoicing update.