Why Chess’s Game Tree Grows So Quickly
The Shannon number estimates possible move sequences, not the number of different board positions. Two games can reach the same arrangement of pieces through different moves, and each sequence still counts as a separate game.
The estimate comes from a simple multiplication repeated many times. At a typical position, a player may have around 30 legal moves. After each choice, the opponent has more choices, and then the player has more again. If a game lasts about 40 moves per side, a rough tree of possibilities is on the order of 30 choices repeated across roughly 80 individual turns. The true number varies because some positions offer fewer moves, games end at different lengths, and the rules can stop repetition or other endless cycles.
Why Computers Don’t Try Every Game
Even a fast chess engine cannot simply follow every branch to the end. Instead, it searches a limited number of moves ahead, evaluates promising positions, and skips lines that are unlikely to change its decision. This is called pruning: cutting away branches of the search tree while keeping the ones most relevant to the move being chosen.
That strategy works because a useful move often can be judged without calculating every possible continuation. But it also means an engine’s search is not an exhaustive proof of what will happen in every game. The branching structure is the reason chess can feel both orderly and inexhaustible: a player makes one move, and the number of futures worth considering expands again.
