A game is solved when the result of perfect play by both sides is known. That is a statement about the game, not about any program: once checkers was solved, no future engine can ever beat another perfect engine at it. Every game in the list below has been settled by a computation, most of them by years of one.
Being solved does not make a game boring for people — it makes it boring for machines. You will still lose to a perfect opponent, but you would have lost anyway; the difference is that now the loss is a theorem.
| Game | Solution | Who and when | Perfect play ends in |
|---|---|---|---|
| Tic-tac-toe 3×3 on this site | strong | known long ago; the search is trivial | draw |
| Nim on this site | strong | Charles Bouton, 1901 | the side to move from a non-zero XOR wins |
| Connect Four 7×6 on this site | strong | Victor Allis and James Allen, independently, 1988 | first player wins |
| Nine Men's Morris on this site | strong | Ralph Gasser, 1993 | draw |
| Gomoku 15×15 on this site | weak | Victor Allis, 1993 | first player wins (free-style rules) |
| Kalah 6×4 (mancala) on this site | weak | Geoffrey Irving, Jeroen Donkers and Jos Uiterwijk, 2000 | first player wins |
| Awari (mancala) | strong | John Romein and Henri Bal, 2002 | draw |
| Kuhn poker on this site | strong | Harold Kuhn, 1950 — solved on paper, no computer involved | the second player wins 1/18 of an ante per hand; optimal play is randomised |
| Pig (dice) on this site | strong | Todd Neller and Clifton Presser, 2004 — value iteration | the win-maximising strategy is known; the dice still settle each game |
| English draughts 8×8 on this site | weak | the Chinook team, Jonathan Schaeffer, 2007 — eighteen years of computation | draw |
| Othello 8×8 on this site | weak | Hiroki Takizawa, 2023 | draw |
| Chomp 8×5 on this site | ultra-weak | David Gale, 1974 — a strategy-stealing proof; small boards computed | first player wins, though the theory names no moves |
| Sim (the Ramsey game) on this site | strong | proposed by Gustavus Simmons, 1969; solved by exhaustive search | second player wins — and a draw is impossible by Ramsey's theorem |
| Hex on this site | ultra-weak | John Nash, 1949 — a strategy-stealing proof; boards up to 9×9 computed | first player wins, though on large boards nobody knows how |
| Dots and Boxes on this site | partly | Joseph Barker and Richard Korf, 2012 — boards up to 4×5 | depends on the board; the general case is open |
| Go | partly | 5×5 — Erik van der Werf, 2002; 19×19 unsolved and likely to stay so | on 5×5 the first player wins; on 19×19 nobody knows |
| Chess | partly | only endgames up to seven pieces are fully computed (2012) | unknown; a draw is considered most likely |
| Heads-up limit hold'em | weak | Cepheus, Michael Bowling's group, 2015 | essentially solved: the gap is unmeasurable in a lifetime |
| Backgammon on this site | open | unsolved: the dice get in the way | no solution in the usual sense exists |
| Russian draughts on this site | open | unsolved | unknown — forced captures and flying kings widen the tree |
| Ataxx 7×7 on this site | open | unsolved | unknown |
| Minesweeper on this site | open | not a two-player game; Richard Kaye, 2000 — deciding a safe move is NP-complete | a fast way to always play safely most likely does not exist |
The bars are powers of ten: how many positions the game has. Each step to the right is ten times more than the step before, so the distance between checkers and chess is not a gap you close with a faster machine.
Nine of ours are settled. Nine Men's Morris is a draw with perfect play, Gomoku on a free-style board is a first-player win, and Othello was shown to be a draw in 2023. Connect Four and Kalah 6×4 both go to whoever moves first. Kuhn poker was solved on paper back in 1950, and Pig in 2004 — solved in strategy, that is: the cards and the dice still decide each single game. English draughts is a draw too — but the game here is Russian draughts, which is not solved. In Chomp the first player wins and in Sim the second one does — both of those seats are handed to you. Hex stands apart: our board is proven to be a first-player win, and nobody knows the winning line.
None of this makes our opponents perfect. The engines on this site search a few plies and guess the rest with an evaluation function; a solved game means a perfect player exists, not that the one you are facing is it.