For example, over time, people notice that a great many theorems of several existing theories could be proven for all those theories at once, if one happened to introduce a suitably chosen common abstraction of all of them. So that new abstraction would have to catch enough details to still allow substantial proofs to be formulated.
There are books about these developments, I just haven't got any references handy. So this would be under the rubric: history of mathematics (but not beginning with Babylonian mathematics, perhaps).
The reason I ask is this: the introduction of a higher level abstraction (like for example topological spaces) does not at all necessarily make the lower level abstractions it subsumes (like metric spaces) completely superfluous. Also, we have groups, rings, and fields, but we still want to be able to talk about natural or real numbers, and their particular properties, do we not?
Higher levels of abstraction are developed all the time. As far as I know, the most abstract level we've reached so far is the discipline of Category Theory. Before that it was Abstract Algebra and Topology. Mathematical Logic gets pretty rarefied.
I admit to having little patience with those who insist on keeping an eye on practicalities. Go as far out into the wilds of pure thought as you like, wonderboy.
We have many people who like the practical and many more who like the abstract/theoretical in math and I applaud them both.