One reason would be if you are using an interface with a native bit rate of 48k (think Creatve/Audigy or others). No 'on the fly' converting needed then.

It is more likely one would rather change to 24 bit as opposed to 16 bit though, which still needs a conversion.

There is more difference between 16/24 than there is between 44/48 (if the using the same 16/24 bit rate on each).
In other words, usually more to gain going from 16/44 to 24/44 than there is going from 16/44 to 16/48.
I don't see where anyone mentioned going to 48, did you assume that? Or see it in a different thread?

As for the original question; why 16/44? That's where RTs started. It was the 'common best' at the time. There are still systems in use that can only support that.

If they changed now they'd have to go back and redo thousands of recordings in order to gain nothing but additional file storage requirements. This may include redoing all the magical markers that make RTs work.

Make your sound your own!
.. I do not work here, but the benefits are still awesome