Code for the paper Entombed: An archaeological examination of an Atari 2600 game by John Aycock and Tara Copplestone.
Messed around with the code to add coloured (ANSI) output, highlighting tiles where the values from the generation table aren't symmetrical to the inverse values.
I have also stuck random to always be 1 in order to highlight the impact of those tiles on the level's navigability better (worst-case scenario)
I don't know if any conclusions can be drawn from this. Maybe the table originally was completely symmetrical, and then those two special cases were picked to significantly increase a path's chance of continuation?