About Reversi

Reversi is a game of flipping discs that goes back to England in the 1880s. Its modern form, with a fixed start, came from Japan in the 1970s as Othello, and this version plays by Othello's rules. Learn more about it on Wikipedia.

How to Play

Objective: have more discs of your color on the board than your opponent when the game ends.

Setup: the board is eight by eight squares. Each disc is black on one side and white on the other. The game starts with four in the middle, two of each color, set on the diagonals.

Gameplay:

• Black moves first, then the players take turns.
• On your turn, put a disc of your color on an empty square so that it outflanks some of your opponent's discs: a straight line of them, across, down or diagonally, with your new disc at one end and another of yours at the other.
• Every disc you outflank flips over to your color, in every direction at once. Only discs in a straight, unbroken line between two of yours flip.
• A move has to flip at least one disc. If you have no such move, you pass, and your opponent moves again. If you do have one, you must play.
• The game ends when neither player can move, usually because the board is full. Whoever has more discs wins. Any squares left empty count for the winner, as in tournaments.

Tips

• Corners are worth a lot: a disc in a corner can never be flipped, and discs along the edges next to it soon can't be either.
• Be careful with the squares next to an empty corner, especially the diagonal one: a disc there can hand your opponent the corner.
• Having lots of discs early on matters less than having lots of moves. Leave your opponent few good choices, and they'll have to make bad ones.

The AI

It works like DeepMind's AlphaZero. A neural network learned the game by playing itself, with no human games or tips. On its turn, a tree search uses the network to size up moves and plays the best one it found.

It all runs in your browser. Expert thinks for 2 seconds a move, so it's a bit stronger on a faster device, and Very strong looks a few moves ahead. Near the end of the game, from about 16 empty squares, both stop estimating and work out every possibility, so they play the endgame perfectly. The other levels don't look ahead at all. Strong plays whatever looks best right now, and each level below it picks more at random, down to Random, which plays any legal move. The Analysis page shows what it makes of any position, down to the exact final score once the endgame is solved.

The squares are named as in chess: columns a to h from the left, and rows 1 to 8 from the bottom: a1 is the bottom-left corner, and f5, a common first move, is in the sixth column and the fifth row.