this is not quite a definition, but a criterion. There are more criteria, but these are not arbitrary, or not totally arbitrary. I´d say that the arbitraryness has more to do with the distinction between close related groups being somewhat blurred, then for practical reasons you got to draw a line eventually.
Higher taxa is a bit more arbitrary, I guess... however... still not totally, and the arbitraryness imposed by classic linlean system tend to collapse as taxonomists make efforts to make accurate classifications, respecting the diversity rather than pre-established taxonomic groups.
Linne started with something like Kingdom, phylum, class, family, genus, and species.
Nowadays not only there are more of these than Linne first proposed, but there are more categories, such as infraorder, suborder, infraclass, subclass, superfamily, subfamily, infratribe, tribe, subtribe, subgenus, subspecies.
And sometimes more... ...some people think that should be more gradation between phyla too, or even abandon using this rank (and some people are in favor of abandoning rankings in general).
Then the whole "a mosquito changes but is still a mosquito" argument conveniently ignores that is possible to walk from mosquito to "inframosquitos", then to "subflies", flies, "infraflies", etc, all the way to insects or higher taxa, and then you can descend again to another species.
This may be not so easy to visualize with insects, but I think that is easier with animals related to dogs, when everybody has in mind (and hopefully accepts) that dog breeds are modified descendants of wolves, and the nature of this modification does not differ from the nature of changes between wolves and foxes, and etc.
