The game AI of the ghosts is affected by a different glitch because of delta X and delta Y being treated as a single 16 bit offset, and Toru Iwatani kept that bug in because he preferred the effect! If you make a port of the game that does not have that bug, it plays differently from the arcade. What makes you think he did not approve of the normal ghost collision detection (ghost and pacman matching their 8*8 pixel position) being different from the blue-ghost collision (a collision box) he may have liked the occasional miraculous escape, and may have commented on that in Japanese magazines in the 1980s or later... (how would you check?) and how would you reliably reproduce it without frame-by-frame accuracy of movement (it is not truly random IIRC, but the movement of pacman and the ghosts depends on speed bitstrings of about 50% duty cycle, which I can't remember if they are reset on each game or level. Lets say you can run through a ghost one time in 16? Are you arguing against using patterns in Pacman? Whole books were written on Pacman patterns in the 1980s. Ms Pacman OTOH uses the "R" register of the Z80 for some AI choices so does not have reliable patterns like original Pacman which would make you happier.