game is evil timewaster. Some thoughts were here: http://stackoverflow.com/questions/22342854/what-is-the-optimal-algorithm-for-the-game-2048. The key implementation detail is that new tiles are in a random location with 90% probability of 2 and 10% of 4.
heuristics that work: keep largest tile in corner if possible, or next to 2nd largest in corner. Keep chains of growable numbers next to each other (ie 64 32 16 8) also preferably in last row. Avoid any risk of losing high row on the side by moving it. You can also treat the area on the far side of your high tile corner as your primary work area to build higher tiles from. The advanced heuristic I've not seen anyone on that thread use, include avoiding positions that might lead to a poor "forced" move. Major forced move is moving away from your high side row, but also compromising your secondary work area is a medium forced move, and minor forced moves are moves that don't clear a tile, or pile 4s and 2s together. ----- Original Message ----- From: Richard Donovan <[email protected]> To: J Programming <[email protected]> Cc: Sent: Friday, March 28, 2014 5:03:31 PM Subject: [Jprogramming] 2048 Has anyone tried the game "2048" on http://gabrielecirulli.github.io/2048/ with the possibility of writing a J routine to solve it? For those who haven't tried it it's a very compelling number / logic game that I think is possibly solvable in J but I don't have the necessary experience to do it. Thanks ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
