But the way I see it half of them want to join groups together and half of them want to rip them apart There have always been "lumpers" and "splitters". others want to just take the whole group and rename it for the sake of it. Yeah, but they're not allowed to do that. The ICN sees to that! The only way you can rename a whole group is if it's become evident that they don't belong to the the rest of the group - they're just too different, or they've descended from a different ancestor, or they're more closely related to something in a different group, for example.
There's rules about names - for example, each genus has a type species, which was the species originally used to define the genus. So if the genus is split, it's the group containing the type species that keeps the genus name (at least it is on the zoological side, and I think it's the same on the botanical side) even if that's the smallest group.
I still think it's better dealt with by discussion between the taxonomists studying the particular groups.
The trouble with viewing it all from the gardening side is a) the horticultural business is slow to change, much preferring to stick with known names, and it has a bias towards splitting - why sell one species when you could sell two? b) only a small proportion of each genus is actually in horticultural use, so any changes are more likely to seem arbitrary.
Aster seems to have been a hotch-potch of 600 species world wide. "According to molecular DNA studies, our native asters are more closely related to Solidago (goldenrods) and Erigeron (daisy fleabanes) species than the Eurasian asters." That seems fair enough. It's not all the asters that have moved, it's just the ones that didn't belong, even if they make up the bulk of what we are familiar with. There's 180 asters still in Aster.
I suppose it's what's always been happening - rue, wall rue, goats rue, meadow rue, so named because they all looked like each other, but now in completely different genera (and one is a fern) because we now know that similarity of leaf does not necessarily mean the plants are related. Now, with DNA, we're seeing that the physiological similarities which used to be all we could go on as recently as the 1980s aren't enough to be sure that plants are related.
Rosemary always was in the Salvia family - both Salvia and Rosmarinus were in Lamiaceae. It's the genus which has changed.
"To restate this more botanically: in a study published in 2017 (citation below) Bryan Drew and his coauthors compared DNA sequences in plants then classified in the genera Salvia, Rosmarinus, Dorystaechas, Meriandra, Perovskia and Zhumeria. The DNA showed all were equally related. The authors proposed putting all the plants into a single genus, Salvia. They made the case carefully in the paper: either the plants should all be Salvia or Salvia must be broken into several genera because it currently includes subgroups that cluster the way those five small named genera do. Breaking up Salvia would require new names for something over 700 plants. In contrast, by merging the five small genera into Salvia, a total of 15 species are being renamed. The other renamed species do not have much of a fan club, only rosemary's renaming has created a stir."
Salvia rosmarinus was first proposed for Rosemary in 1852, hence was the name chosen once Rosemary had to move.
And for further confusion, the Portuguese plant Rosmaninho is in fact a Lavender - the Portuguese lavender that looks rather like Lavendula stoechas (but I think is a different species). Rosemary is Alecrim.
I think the basic answer is there is an orderly underlying structure, but with exceptions possible if obeying the rules would create more mess than breaking them. I have a dim memory of reading a paper in the 60s proposing a breaking of the rules since to follow them would involve something of the magnitude moving "dog" out of Canis.
"An EMM on ee" But the tongue doesn't want to do it, hence the prevalence of "An ENemy" And not Annie Moan, which is how I think of it when I have to spell it 