- Created Sunday, November 8th 2015 @ 10:21:43
If two bots always play in the same way, there is no sense to play more than two games between those bots because the start position is always the same.
So the proposal is to start with two stones already in place, put at random positions. For fairness, the random stone of the beginning player should not be in the middle, otherwise the first player's advantage would be too high.
Then we have 42 possible start positions, which should be enough. Also this way we prevent bots entering with a precomputed perfect solution that always wins.
DeveloperCreated Monday, November 9th 2015 @ 08:50:22
That might be a nice idea :)
- Created Tuesday, November 10th 2015 @ 07:29:09
Or, at least, play in pairs. So one with red, and one with yellow.
- Created Tuesday, November 10th 2015 @ 09:06:12
+1 for the two starting stones.
- Created Tuesday, December 8th 2015 @ 14:39:16
I think pre-defined starting position is great idea, or at least auto re-match - because starting player can always win, if playing perfectly.
for all battles between perfect bots it will be a draw (1:1), and for any other after two games it will be really visible, whose bot is better.
- Updated Thursday, December 10th 2015 @ 09:53:33
I think two random stones wont help a lot. Don't forget, if player1 doesn't play D1 with first move it will lose, if player2 plays perfectly. So any starting position gives you predefined result. You will only decrease the game tree.
- Updated Tuesday, April 19th 2016 @ 08:48:31
Isn't the premise of the competition that it's a beginner contest?
Random starting stones will only make it a bit harder to define perfect play. Changing the nature of the game is the way to go if we want to make it hard. E.g.:
- After each place_move there's a 10% chance of a stone being removed from the board, or
- When you place_move you give two places the first will be there with 60% likelihood, and for the other 40% chance the second move is chosen -- and you don't get told where the stones actually land. Each player only gets told what the place_moves are, not the actual board state. Or
- You see the board game state one move behind the actual state (so you don't know what move they made). Or,
- Moves are played simultaneously (i.e. a random order of the two player's moves each turn much like in warlight2).