The paper works in a setting where the usual axiom of choice fails, specifically in a carefully constructed mathematical universe called Truss's model (or a related Feferman-type model). In ordinary mathematics, the axiom of choice guarantees that you can always make infinitely many selections simultaneously, but logicians have long studied what happens when you weaken or remove this guarantee. The authors focus on a particular version of the Baire Category Theorem, which in its classical form says that in a nice topological space, the intersection of countably many dense open sets is still dense. Here they ask: what if you only assume the family of dense open sets can be well-ordered, rather than assuming full choice? They show that two natural principles of this restricted kind are actually equivalent to each other, and they characterize exactly which index sets allow the intersection result to hold in Truss's model.
The heart of the paper is a detailed study of the "well-orderable subsets" of the Cantor set (the space of infinite binary sequences) inside this special model. Remarkably, many different notions of smallness that are distinct in ordinary mathematics all collapse to the same thing here: a subset is well-orderable if and only if it has strong measure zero, is universally null, is Marczewski null, contains no perfect subset, or satisfies the Rothberger covering property in every finite power. This is a striking confluence of ideas from measure theory, topology, and combinatorics that would never happen if the full axiom of choice were available. The authors also show that this collection of small sets is closed under well-ordered unions but sits in an incomparable relationship with the meager ideal, meaning neither collection contains the other.
Beyond these characterizations, the paper establishes several additional regularity results that hold throughout the entire model. Every set of reals is Marczewski measurable (a broad measurability notion), every function into a separable metric space has a restriction to a perfect set on which it is continuous, and every set with positive outer measure contains a perfect subset. These are strong structural properties that classical mathematics cannot guarantee for all sets without extra assumptions. The paper also notes that the Hurewicz covering property, a close relative of the Rothberger property, does not coincide with well-orderability in this model, showing the equivalence has precise limits. Taken together, the results paint a detailed picture of how measure, category, and combinatorial covering properties interact when choice is available only in a restricted, well-ordered form.