If Connect6 uses an infinite board, both
state-space and
game-tree complexities are infinite as well. Instead, assume that a Go board is used. The game-tree complexities for it are still much higher than those in
Gomoku and
Renju, since many more moves are possible placing two stones than one—specifically
n(
n−1)/2 moves are possible, where
n is the number of unoccupied spaces before a move. However, the state-space complexity is largely unchanged, since any legal position in one game will also be legal in the other. Based on the standard in Herik, Huntjens, and Rijswijck, the
state-space complexity of Connect(19,19,6,2,1) is 10172, about the same as that in Go or Gomoku. If a larger board is used, the complexity is much higher, since the number of moves increases exponentially with board size; it should still be the same as the other two games on the same size board. Now, let us investigate the
game tree complexity. Assume that the averaged game length is still 30, the same as the estimation for Gomoku (Allis 1994). Then, the number of grids chosen to put one stone is about 300, and the number of choices of one move is about \left(\frac{300\times300}{2}\right) or 45,000. Thus, the game-tree complexity is about \left(\frac{300\times 300}{2}\right)^{30} ≈ 10140, much higher than that for Gomoku. Alternatively, if one assumes that the total number of stones placed (instead of the number of moves) is the same as that for Gomoku, that leaves us with an average game length of roughly 15. Then the game-tree complexity is roughly \left(\frac{300\times300}{2}\right)^{15} ≈ 1070, the same order of magnitude as that for Gomoku given in Allis 1994. Again, if a larger board is used, this complexity becomes much higher. ==History==