There's another, easier-but-geekier explanation. Based on what I've learned from rose hybridizers and other readings, Miniatures typically are heterozygous for an incompletelly dominant dwarfing mutation. Basically, what makes a Miniature is having one copy of this dwarfing mutation. If the rose has two copies (in the case of diploids), three copies (in the case of triploids), or four copies (in the case of tetraploids) -- in other words, the rose is homozygous for the dwarfing mutation -- the rose is a Micro-Miniature, like 'Si'. As far as all the Micro-Miniatures I've looked up, every one had Miniatures for both parents. If a Micro-Miniature was crossed with any other rose, all the offspring would be Miniature.
ETA -- I think the "Patio" or "Mini-Flora" roses may be in-between because they have only one copy of the dwarfing mutation and are triploid or tetraploid -- in other words, they have only a 1/3 or 1/4 "dose" of the mutation, the rest being "normal", and so their dwarfing is less than in Miniatures, which would be 1/2 or 2/4. So diploid Miniatures would have one copy of the dwarfing mutation, tetraploid Miniatures would have two, or maybe three.
So what does this have to do with two Miniatures having a Floribunda seedling? Well, first of all, a little bit of history for the Miniatures. The earliest ones arose as mutations in China roses -- I think 'Rouletii' was one of the first, and others descended from it. Some were bred with small Polyanthas, of which the really tiny ones likely also have the dwarfing mutation. So the early Miniatures were basically tiny Chinas and Polyanthas. But as the 20th Century moved along, hybridizers like the late Ralph Moore bred these miniatures with Hybrid Teas and Floribundas, with the idea of creating Miniatures with "classic Hybrid Tea form." Since Miniatures were heterozygous for dwarfism, this meant that, on average, half of the offspring with Floribundas and Hybrid Teas also received a copy of the dwarfing mutation, and thus were also Miniatures. Ralph Moore, in particular, did a lot of these crosses, and his Miniatures have quite a diverse genepool as a result of him crossing Miniatures with all sorts of types. If the non-Miniature offspring seemed worthwhile, they were introduced as well.
So, now let's look at what would happen if two modern Miniatures were crossed. Both parents being heterozygous for the dwarfing mutation, some of their seedlings would be Miniatures (heterozygous for the dwarfing mutation), some would be Micro-Miniatures (homozygous for the dwarfing mutation), and some wouldn't be Miniatures at all (no copy of the dwarfing mutation). It works sort of like how two brown-eyed people can have a blue-eyed child, if both of those parents were heterozygous for blue eyes.