What kind of obsession produces an effort, not just to complete Pokemon Platinum, which after all was sold to kids with the expectation that they would be able to beat it eventually. No, what about an effort to finish every possible game of Pokemon Platinum, using a script that works on every possible random seed, of over four billion, that the game can generate? And also operates mostly on “Nuzlocke Challenge” constraints, where any defeated Pokemon (here, after the first battle) have to be released? But that’s okay, because after that first fight, the player is never defeated?

That this is possible at all is because of Pokemon Platinum’s use of a PRNG, a pseudo-random number generator. While figuring out how, mathematically, to beat over four billion possible games is a formidable challenge, it’s still better than beating every possible conceivable random sequence of events, which can’t ever be done conclusively.

So, that’s what MartSnack did. They found out how to swim through the deep-yet-discrete sea of probability to obtain just the Pokemon, and Pokemon stats, they needed to complete the game, regardless of any random event the game could throw at them, with the same sequence of button presses. It’s a journey that requires frequent synchronization, to make sure no one possible play breaks free of the others, sending that branch of fate down a rogue path. How is this possible at all, I leave it to you to discover in their Youtube video, an interesting hour and five minutes found by MeFi user (and former owner) cortex, here:

Beating Every Possible Game of Pokemon Platinum At The Same Time (Youtube, 1h5m)

rodneylives

Pac-Man is a game that is vulnerable to patterns: if you do exactly the same thing each time on the same level, the same results will occur. There is one pseudo-random element in Pac-Man though: when vulnerable ghosts reach an intersection, the code picks an arbitrary address from a range of memory addresses, then uses that value to pick a direction to decide which route to take. Two implications of this: vulnerable ghosts are most likely to head left at intersections and least likely to go up, and if any byte in that range changes the behavior of the game slightly changes too, even if it’s not an executable byte. Patterns still work in Pac-Man, despite this pseudo-random function, because the seed is reset at the start of every level, so if you do exactly the same thing, vulnerable ghosts will still have the same information fed to their movement routines.

Ms. Pac-Man has other sources of randomness: the ghosts, in Scatter mode, use a different source of pseudo-randomness to decide where to go, one that isn’t so easy to manipulate; and which fruit appears and which of four predefined routes (three for one of the mazes) it’ll take through the board.

Ms. Pac-Man doesn’t have its ghosts scatter periodically through the level like they do in Pac-Man. They only scatter at the start of the board. It’s not much randomness, but it’s enough to upset rote pattern creation, since each ghost has the opportunity to make several decisions of which path to take during that period. The way the randomness is handled is interest itself. The ghosts pick one of the corners of the board, much like they would in original Pac-Man, but randomly, when making their choice of target to home in on.

So there! Now you can amaze your friends, if it were 40 years ago and your friends were then able to be impressed by your knowledge of Ms. Pac-Man! You’re retroactively welcome!

Random Elements of Ms. Pac-Man (Retro Game Mechanics Explained on youtube, 21 minutes)

rodneylives