Free variation is incredibly rare in general. And even thenStephen Carlson wrote:What can we say about the diachronic development of the distribution of forms? When one form is advancing and the other is retreating, would this imply, at least on the margin, that there isfree variation (if not synonymy)? Or should one be looking at register or regional variations? Could there be a change in markedness (e.g., the aorist passive used to be marked for change of state and the aorist middle, but then the aorist middle became marked for volition and the aorist passive unmarked)?
Allan & Rijksbaron's view for the sense of the -θη forms is that they mark a higher degree of subject affectedness, which is why you don't see them in direct reflexive constructions...and I generally agree, but reject subject affectedness as a semantic feature that can explain all usage. It's a very useful feature that explains very much and is generally sufficient for introducing concepts, but that's all. In any case, when grammatical forms involving degree undergo grammatical change, where one form replaces another, the pattern of language change is fairly clear. The -θη forms are undergoing analogical leveling by means of semantic bleaching.
The change is roughly the same that has taken place in English with the words "welm" and "overwelm." At this point in history, "welm" has essentially dropped out of the language except in a few very, very specific contexts. But historically, the situation is thus:
Originally, there was only the word "welm," but at some point in history, the form "overwelm" was introduced to express a higher degree of "welmedness." Rather slowly overtime, the sense of higher degree disappeared, such that today, for the most part, "overwelm" is the form that remains. But in the middle period, both forms existed together. One might say that that they're in free variation and to some extent that's true. A speaker might have rather arbitrarily choosen to use one and then the other in different contexts without any different at all. But during that period of "free variation" the moment someone put both forms beside enough in a sentence: I'm not just welmed by every that needs to get done, I'm completely overwhelmed, then free variation and synonymy is no longer a useful explanation. When a speaker *chooses* to make a contrast, The forms themselves become contrastive even though they normally aren't.
We can be skeptical all we want about the two aorist and future forms and their relationship between them. But skepticism doesn't explain tokens like Genesis 1:3.