It may be that any ultimate universal rules will have to look at what's possible, rather than what is, although there's still always the chance that a culture's perspective on the world could preclude a rule from surfacing in the actual language. Maybe there's a wide pool of rules possible in every language, but we have a limit on how many rules can actually be used. (That sounds like a way to cheat and call a nonuniversal rule universal, but I guess the way around that would be to see if you could teach the concept to someone whose own language does not have it-- and have them use it fluently--, as "what's possible" gets into cognition. It's almost like the pool of possible consonants, there are several, and each language uses certain sounds. Forget for the moment about a limit on how many can be used. But in theory anyone can learn how to pronounce any consonant. So the universal rule might not be "all languages use / s /," but rather "all languages use consonant sounds from this specific pool." That is horribly general, but that's what universal rules are, massive generalizations.)

If only we knew more about language history and proto-languages, which was the one (or were the few) from which other languages evolved, we could see how and when various proposed rules popped up. Wishful thinking, I know. Well, with efforts to save theories like glottochronology, maybe we'll eventually learn something....

Another theory for the hell of it-- maybe the Piranha language once had recursion, but eventually got rid of it because the concept wasn't needed, kind of like a language shedding noun declension endings because the prepositions alone were enough.