Indeed, I feel like I'm reaching towards a generalized understanding of context.

It seems to me that unless the logical languages are restricted from reasoing about themselves they become either incomplete or worse, inconsistent.

I have a sensation that on the computer science side of things, unless "reasoning about terms themselves" is somehow 'handled' (restricted) one can end up with undecidable systems, or worse...

But I still have questions about complexity so I cannot really undesrtand how completeness and soundness realte to decideability.