It's not hard to reason about from a formal POV-- by "self-modifying" code, you're basically editing the continuation of the current function/statement. It's j

It's not hard to reason about from a formal POV-- by "self-modifying" code, you're basically editing the continuation of the current function/statement. It's just not very useful-- by definition, the extra performance that you might gain from directly self-modifying the program's binary code can't be any higher than the overhead of a good interpreter/JIT. That's just not interesting.