Tic-Tac-Toe Duel

Why You'll Like Tic-Tac-Toe Duel

Tic-tac-toe is a solved game in the strict mathematical sense: with optimal play from both sides, the result is always a draw. That fact has been understood since the 19th century, but it took until 1952 for someone to actually demonstrate it on a computer — A.S. Douglas wrote OXO for the EDSAC at Cambridge as part of his PhD thesis on human-computer interaction, and it is now widely cited as the oldest video game with a graphical display, predating Tennis for Two by six years. Douglas’s program played perfect tic-tac-toe against humans on a 35×16 cathode-ray tube, and we know it existed because the EDSAC’s machine code was preserved on punched paper tape.

The game’s solution is small enough to enumerate by hand. Without symmetry reductions there are 255,168 distinct game sequences; account for the eight-way board symmetry (four rotations and two reflections) and that drops to 26,830 unique games. Of the terminal positions, the first player can force a win in roughly 91 distinct lines, the second player in 44, and 3 are forced draws — but those win counts only matter if both players play badly. Optimal play guarantees a draw because the first player can always force at least a draw and the second player can always avoid a loss with the standard defensive responses.

This browser build implements perfect AI using minimax with alpha-beta pruning. The ‘O’ opponent computes the full game tree, scores wins as +10, losses as -10, and draws as 0, then picks the move with the optimal value. There is no difficulty slider — every move from the AI is provably the best response. That has a specific consequence: against this AI, you can never win. The best outcome you can force is a draw, and that requires you to claim the centre on move one, mirror the AI’s threats correctly, and avoid forking attacks. If you lose to the AI you played a suboptimal move; if you draw you played perfectly. Local two-player mode is the only path to a non-trivial win — and that is exactly where the strategic interest lives, because two humans can both play badly enough to produce one of the 135 non-drawing terminal positions.

Tic-Tac-Toe Duel works best when you slow down enough to see how one move changes the next few moves. The answer is rarely just the first legal action. Good puzzle play reduces uncertainty, opens new paths, and avoids moves that make the board look cleaner while secretly removing flexibility.

For players arriving from search, the practical question is how to make the next attempt better. In Tic-Tac-Toe Duel, that usually means focusing on state reading, move order, and preserving future options. The more you understand that core loop, the less the game feels random and the more each restart becomes useful practice.

Strategy notes

Before acting, ask what new option the move creates. If two moves both seem correct, choose the one that leaves more exits, reveals more information, or keeps the board easier to repair.

How to Play

• The Setup: A 3×3 grid, X moves first. Click any empty square to place your mark.
• Win: Be the first to line up three of your marks horizontally, vertically, or diagonally.
• Draw (Cat’s Game): If all 9 squares are filled with no three-in-a-row, the game ends in a tie.
• Vs AI: Toggle ‘vs AI’ on to face the perfect minimax opponent. The best you can achieve is a draw — every loss against this AI is on you.
• Vs Friend: Toggle ‘vs AI’ off and X / O hot-seat on the same display. This is the only mode where wins are actually possible.

Tips and Strategy

• Scan for forced moves before experimenting.
• Prefer moves that reveal information or open future options.
• Avoid tidy-looking moves that trap important pieces or spaces.
• When stuck, work backward from the goal state.
• Use failed attempts to identify which move first caused the board to lock up.