First-player and second-player win

From Wikipedia, the free encyclopedia
Jump to navigation Jump to search
Diagram showing optimal strategy for tic-tac-toe. With perfect play, and from any initial move, both players can always force a draw.

In game theory, a two-player deterministic perfect information turn-based game is a first-player-win if with perfect play the first player to move can always force a win. Similarly, a game is second-player-win if with perfect play the second player to move can always force a win. With perfect play, if neither side can force a win, the game is a draw.

Some games with relatively small game trees have been proven to be first or second-player wins. For example, the game of nim with the classic 3–4–5 starting position is a first-player-win game. However, Nim with the 1-3-5-7 starting position is a second-player-win. The classic game of Connect Four has been mathematically proven to be first-player-win.

With perfect play, checkers has been determined to be a draw; neither player can force a win.[1] Another example of a game which leads to a draw with perfect play is tic-tac-toe, and this includes play from any opening move.

Significant theory has been completed in the effort to solve chess. It has been speculated that there may be first-move advantage which can be detected when the game is played imperfectly (such as with all humans and all current chess engines). However, with perfect play, it remains unsolved as to whether the game is a first-player win (White), a second player win (Black), or a forced draw.[2][3][4]


See also[edit]

References[edit]

  1. ^ "Checkers Is Solved". Science. 317: 1518–1522. doi:10.1126/science.1144079. Retrieved 2008-11-24.
  2. ^ J.W.H.M. Uiterwijk, H.J. van den Herik. "The Advantage of the Initiative". (August 1999).
  3. ^ Shannon, C. (March 1950). "Programming a Computer for Playing Chess". Philosophical Magazine. 7. 41 (314). Archived from the original (pdf) on 2010-03-15. Retrieved 2008-06-27.
  4. ^ Victor Allis (1994). "PhD thesis: Searching for Solutions in Games and Artificial Intelligence" (pdf). Department of Computer Science. University of Limburg. Retrieved 2012-07-14.