Hi,
that is considered a good comprimise between fast convergence and CPU resource protection.
The faster the root sends BPDUs the longer the CPUs along the path will be busy processing them (causing issue if a switch in the path is too busy with other cpu intensive processes).
The longer they are sent the slower will be convergence in case of link or switch failure.
It is a convention and you are free to change it, even though this is NOT recommended.
Riccardo