I really don't know why M. complicates the matter to this extent. Hume's problem is really a problem regarding the objective but possibly unknown regularity of nature. Popper's misunderstanding lies at that, while he accepts that the present explanation of a regularity might be breached, he never thought it possible that the objective regularity of nature as an ontological property can be subject to change over time. Epistemological a regularity is always assumed, and science needs to find this regularity; human science might fail, but regularity itself is never believed to be able to fail suddenly. Now, as for Kant, Hume's problem is epistemological solved by means of elevating this regularity of objective and external nature to a regularity that stems from the very possibility of constructing the phenomenal world: without this regularity the phenomenal world cannot be constructed.

Hence M.'s thesis is simply that, what if one of the fundamental assumption of general relativity, namely the general covariance of spacetime, is false, and that moreover or better still, what if in principle there's no way to account for the essential irregularity of world. And he thinks it is possible to challenge Kant's thesis that consciousness itself, which structures the phenomenal world, guarantees and posits the existence of objective regularity of the world, since, plainly, extro-science fictions factually exist and are completely comprehensible to its readers. These are broadly regular, approximately regular, worlds, that do not need to even obey a statistic or a probabilistic law, where scientific method can only penetrate to some extend, but never fully chart. Moreover, a non-regular world is not a chaotic one, but sometimes or in some aspect regular, but some other times and in some other respect irregular one. Thus, these are fundamentally non-regular that are still capable of being constructed.

This is ALL!