Now build me a selection algorithm out of mergesort.
Just the other day I sped up a process by an order of magnitude because someone was sorting to do weighted median selection. Which is easy to build on top of std::nth_element. Top-k is another common need.
Quicksort is still worthy today also because of this.
Or you could just have people study QuickSelect and gain the same benefit without wasting on a slower sorting method when taking into account cache etc.
Just the other day I sped up a process by an order of magnitude because someone was sorting to do weighted median selection. Which is easy to build on top of std::nth_element. Top-k is another common need.
Quicksort is still worthy today also because of this.