I think they were designed at the time of less powerful machines so they had to be designed better. Nowadays there is not as much push to eke out every last bit of performance because there is loads of power at everyone's disposal and developers are pushed to focus on features first without being given time to refine performance because features mean adoption. So the bloat creeps up, and hardware makers keep designing more powerful machines which further enables the bloatiness. It is a vicious cycle.