[tptacek]

This seems pretty academic: even if RSA can be kept alive at some enormous key size, the reason we use RSA and not lattices or isogenies is that RSA is more practical.

That equation flips if quantum computing becomes a real threat, and the numbers in the paper don't appear to change that at all: the key sizes theorized here are, for instance, far bigger than the keys we use in RLWE schemes.

As others here have noted: the paper we're talking about is not entirely serious.