Nah, division algebras are very frequent in mainstream research, it is just that they are not a good guiding rule and usually found a posteriori. It is similar that Connes's theory; if it "panned out", it would immediately be absorbed into the corpus. This is hinted when you see criticism about most of the results being trivial or well-known. I have liked specially the comment about 12 being half of 24... but then 14 is described by Motl as being 2/3 of 21, and not half of 28. Point being, there is a lot of representation theory to juggle with. “Octonions are to physics what the Sirens were to Ulysses,” says Pierre Ramond, in the Quanta Magazine article. But there are sirens within sirens, and what is interesting about Furey work is that they are being carefully avoided: Bott periodicity, Evans supersymmetry, Atiyah CP2, Tits–Freudenthal Square, Duff brane scan ... to name some.