Dream mathematics is an alternative foundation of mathematics first studied (as a piece of metamathematics?) by Robert Solovay?. Several ‘ill-behaved’ counterexamples in analysis fail to exist in it.
Dream mathematics is mathematics founded on ZF (or an equivalent structural set theory such as SEAR) with dependent choice and the following axioms (any of which contradict full choice) required of every subset $A$ of the real line:
A dream universe is any model of dream mathematics. The most well known (and the first known) is the Solovay model.
Solovay proved that dream mathematics is consistent if the existence of an inaccessible cardinal is consistent with ZFC. More precisely, Solovay showed how to construct a model of dream mathematics (now called the Solovay model) from any model of $ZFC$ with an inaccessible cardinal.
Saharon Shelah later showed that one could start with any model of $ZFC$ and construct a model of $ZF + DC$ in which every set of reals has the Baire property; on the other hand, Ernst Specker had already shown that an inaccessible cardinal must be consistent if the perfect set property is. Various intermediate consistency results for Lebesgue measurability are also known, but a complete characterisation is still elusive.
Besides the axioms themselves, other nice properties hold in dream mathematics. Examples include:
Every (total) linear function from a Fréchet space to any topological vector space is continuous (so every linear mapping between Banach spaces is bounded).
On a localisable measure space, the dual of the Lebesgue space $L^\infty$ is $L^1$ (so these two spaces are reflexive).
Any two complete norms (or even F-norms) on a given vector space over the real numbers must be topologically equivalent?, which nicely explains why you have never seen two specific inequivalent complete norms on a given real vector space.
Wikipedia (English); Solovay model.
HAF; end of Chapter 27.