It might be interesting to see what mathematicians-in-the-wild think about higher category theory, which is viewed in a somewhat similar way by people who don't work in certain fields.[0]

http://mathoverflow.net/questions/169187/what-non-categorica...

I suspect the reasons are broadly analogous. To paraphrase a quote of Atiyah's, the usefulness of theory is that once a field is complex enough, a little bit of it can serve to provide "organizing principles" (quoting Qiaochu Yuan from the MO post above) that help both learners (who can exploit similarities in different fields so that unfamilar things feel less alien) and experts (for whom many things are made easier, or even possible, by working at higher levels of abstraction[1]).

Knowledge just keeps piling up, and abstraction is one way of increasing the "density" of information one takes in so that one doesn't have to spend one's whole life trying to learn enough to work -- or, in the researcher's case, getting to the frontiers of current knowledge.

[0]: Roughly, my understanding of it is that it's what happens when you don't require properties like associativity, but say that (ab)c should be "deformable" into a(bc) in some way. Complications arise, interesting stuff happens. The usual.

[1]: Akin to the birds of Freeman Dyson's "birds and frogs".

http://www.math.columbia.edu/~woit/wordpress/?p=1506