I generally am not interested in conlangs that come with diachronic versions–i.e. hypothetical versions where plausible or otherwise changes mutate the words from one pronunciation to another. A crude version of what plausible is, is a rule like “The sounds change to corresponding sounds a row lower on the IPA chart” This simplified rule was mentioned in a McWhorter lecture I watched. Your mileage may vary.
Why not? First, I’m not all that interested in tolkien-style conlangs, nor it’s development agenda, i.e. supports a fictional story or fake world, where the story might never be written, isn’t written for fan usage, often has “poison pills” that make fan usage unusually problematic, etc. Second, I prefer conlangs that a fan could actually learn. Not necessarily as an Esperanto style auxlang, but as a general purpose usable system of communication. A language that comes in several versions is going to be harder to use than one that doesn’t. If you put archaic, late, modern and future Fakese, then some fans will eventually try to write in all of them and the corpus is going to get more and more difficult for people to learn.
Why? But these tools exist and I think there is a use for them. My favorite conlang, toki pona, appears to be instable. I mean it looks like if a real speaking community were to use it, the words would almost instantly collapse into shorter words and certain minimal pairs would mutate to increase the distance between them.
So if we took these sound change applications, applied a set of plausible rules and the words didn’t change much, then we could say that the language is stable. If a small likely change cause the vocabulary to mutate beyond recognition, then there are problems. Similarly, if a small likely change results in excessive homophones, causes grammatical markers to become unsuable, then that is useful information for design of a single language.
Another possible methodology is to apply a variety of different plausible change and keep applying them over and over until the language reaches an equilibrium and doesn’t change anymore. Obviously some sound change rules will eventually result in all letter eroding to nothing. So this will only work if there is some offsetting force in the sound change rules to prevent degenerative states, such as all of the sounds disappearing from all words.
Here is one that I hope to try out someday.