Ah... so it's the challenge, and not whether anyone will ever use it. In the same vein, there's an old DOS assembler trick from the late 80's through the mid 9

Ah... so it's the challenge, and not whether anyone will ever use it. In the same vein, there's an old DOS assembler trick from the late 80's through the mid 90's - remap INT 20h (which was a 1-byte opcode to exit the currently running program), to INT 21h (a 2-byte instruction for DOS, which includes a clean exit for the program as a built-in function). This could break how programs exit cleanly, but probably never did. In a substantial program, it saved a lot of bytes in DOS calls. It was mostly used competitively, but it was a pretty stable code size hack.