> Decriptivism allows the language to evolve as fast as people wish to evolve it.
More importantly, taken to an extreme, descriptivism describes language in the way a map describes the territory. Any time a person speaks and is understood, no matter how badly, end-stage descriptivism has to allow their diction as syntactically and semantically valid in the language in which they spoke. The most you can say is that some expressions are rarer than others (you see "was done well" much more often than you see "was done good").
But this is also a wrong way of talking about language, just as much the old prescriptivist way was wrong. People are not static language replication machines, learning how to speak purely from imitation of their elders and community, and observed from on high by language anthropologists seeking to observe how they behave. They are concept-builders, rule-learners. They have a sense for not just how to speak in particular cases, but also what it is to speak well. It is this public sense of correct speech that is the subject of evolution over time, and is therefore also the proper target for descriptivist accounts of language.
Writing is similar to speech, but in writing most people are even more keyed to correctness, and less keyed to achieving the bare minimum of communication. Rules are stickier. We ought to understand this Wikipedia editor not as a noxious outsider to the evolution of language, who like the anthropologist inserts prescriptivist rules where they are unwanted, but as someone who is part of the normal evolution of language itself and therefore part of the terrain to be described! There have always been people who have been sticklers for particular rules.