I didn't know about FRACTRAN, that's pretty cool: https://en.wikipedia.org/wiki/FRACTRAN
> And it follows that FRACTRAN games are undecidable. This doesn’t speak directly to the Collatz conjecture, because (1/2)n | 3n + 1 is not a FRACTRAN game. But it does follow that arbitrary Collatz games are undecidable, since the set of all Collatz games includes the set of all FRACTRAN games.
[0] https://raganwald.com/2020/05/03/fractran.html#why-fractran-...