Introducing Tsuro - The game of the Path, my new strategy game, based on the board game.

https://scratch.mit.edu/projects/369811739/

This version can be played by a single player against up to three AI opponents, or by up to four players on the same screen. Here's a screenshot of a four player game:



Since it's mobile friendly, a great way to play is with a tablet laying on a table between you and your friends or family. But it's also fun to play solitaire.

Let me know what you think. I would appreciate all constructive feedback. Post here, or in the project comments.

I hope you enjoy Tsuro!

You can learn more about Tsuro at Wikipedia.

I've made a few changes to the game, based on your feedback:
1) Paths are drawn in bolder and lighter pen lines, to make them stand out against the background (thank you @colinmacc for pointing this out).
2) Iterations per path segment bezier curve have been halved, due some lag when the board gets near full.
3) The highscore is now based on tiles traversed (or path segments traveled). This was because @colinmacc quite quickly set the maximum “tiles placed” highscore possible. You can traverse the same tile several times.


There are still some issues I'm working on, based on your feedback. Among other things:
- Some affirmation when a player is eliminated (thank you @TheBurlyCoconut). (edit: I can't figure out a good way to do it. I'll think on it…)
- Improved in-game instructions (thank you @TheBurlyCoconut). (edit: the instructions have been updated)

If you notice any issues, please let me know. Thanks!

I've reworked the game to accommodate up to 3 AI opponents. I welcome any ideas on how the AI can be improved. As it is, the AI only identifies possible tile / rotation combinations which produce paths which doesn't lead straight to the edge of he board.

Are there strategies that could be implemented if there are several possible tile / rotation combinations?

• Should it try to stay away from the edge?
  
• Should it try to stay away from opponents?
  
• Should AI opponents work together?
  
• Should the AI try to save certain types of tiles?
  
• Should the AI strategy change depending on the stage of the game (early / late) or depending on what strategy the opponents appear to employ?
  

awesome

I played your project earlier, and I am not an expert at this game, but you should definitely add some music to make it get that Chinese feel, you know?

kriblo wrote:

I've reworked the game to accommodate up to 3 AI opponents. I welcome any ideas on how the AI can be improved. As it is, the AI only identifies possible tile / rotation combinations which produce paths which doesn't lead straight to the edge of he board.

Are there strategies that could be implemented if there are several possible tile / rotation combinations?

• Should it try to stay away from the edge?
  
• Should it try to stay away from opponents?
  
• Should AI opponents work together?
  
• Should the AI try to save certain types of tiles?
  
• Should the AI strategy change depending on the stage of the game (early / late) or depending on what strategy the opponents appear to employ?
  

I'm not exactly a master at this game, but I do have some strategies worked out, so to answer:

• Probably. Though a bigger thing would be trying to get to wider open spaces and avoiding places where there's just one empty tile, as it's easier to outlast your opponents in bigger spaces.
  
• Not really? Going on the offensive early on (at least for me) is a good way to corner your opponents before the game becomes one of survival and hoping you can get that one specific tile that you need in order to not hit a wall. They should probably avoid paths that would put them on the same tile as an opponent, though, since that tends to put you at their mercy.
  
• I'd say yeah. That'd probably be an easy way to up the difficulty for the human players. (though I have no idea how easy that would really be to implement.)
  
• I dunno? Though tiles that have a wide variety of path types are less likely to result in a loss for every rotation, so those might be good to have them save.
  
• In the late game, it becomes more important to try to reach larger open spaces than your opponents. And if an opponent appears to be playing more aggressively and heading directly towards an AI, the AI could run away from them and try to take the path that lands them farthest away from the opponent, forcing them to either give up or chase them and risk winding up on the same tile.
  

It's been a while since I've played against actual humans, so I don't have too many strategy tips, but maybe these will help?

StrixCattus wrote:

kriblo wrote:

I've reworked the game to accommodate up to 3 AI opponents. I welcome any ideas on how the AI can be improved. As it is, the AI only identifies possible tile / rotation combinations which produce paths which doesn't lead straight to the edge of he board.

Are there strategies that could be implemented if there are several possible tile / rotation combinations?

• Should it try to stay away from the edge?
  
• Should it try to stay away from opponents?
  
• Should AI opponents work together?
  
• Should the AI try to save certain types of tiles?
  
• Should the AI strategy change depending on the stage of the game (early / late) or depending on what strategy the opponents appear to employ?
  

I'm not exactly a master at this game, but I do have some strategies worked out, so to answer:

• Probably. Though a bigger thing would be trying to get to wider open spaces and avoiding places where there's just one empty tile, as it's easier to outlast your opponents in bigger spaces.
  
• Not really? Going on the offensive early on (at least for me) is a good way to corner your opponents before the game becomes one of survival and hoping you can get that one specific tile that you need in order to not hit a wall. They should probably avoid paths that would put them on the same tile as an opponent, though, since that tends to put you at their mercy.
  
• I'd say yeah. That'd probably be an easy way to up the difficulty for the human players. (though I have no idea how easy that would really be to implement.)
  
• I dunno? Though tiles that have a wide variety of path types are less likely to result in a loss for every rotation, so those might be good to have them save.
  
• In the late game, it becomes more important to try to reach larger open spaces than your opponents. And if an opponent appears to be playing more aggressively and heading directly towards an AI, the AI could run away from them and try to take the path that lands them farthest away from the opponent, forcing them to either give up or chase them and risk winding up on the same tile.
  

It's been a while since I've played against actual humans, so I don't have too many strategy tips, but maybe these will help?

Thank you for your insights! It's been over a year since I worked on this project, and I'm not sure when I'll get back to it. But I appreciate you taking the time!

Hello

First, great game!

I am also not a master but I have done some work on the longest path you can make in the game. That isn't really important in a multi-player game but it has lead to some thoughts. As @StrixCattus said, stay away from the edges. But I think more importantly is to try to move to the largest open area/fore your opponent into the smallest open area. When I played the bot I knew I had won by turn 3 since it had moved into a small area with no way out. The forcing the opponent is probably hard to program. But moving towards open areas is most likely doable. You may also want to evaluate “exits” from a region so the computer can determine the max number of moves or expected number of moves if it goes to a certain area.

I have a idea for the CPU. Make it have Easy, normal and hard (I am not a master this is just an idea).

kriblo wrote:

I've reworked the game to accommodate up to 3 AI opponents. I welcome any ideas on how the AI can be improved. As it is, the AI only identifies possible tile / rotation combinations which produce paths which doesn't lead straight to the edge of he board.

Are there strategies that could be implemented if there are several possible tile / rotation combinations?

• Should it try to stay away from the edge?
  
• Should it try to stay away from opponents?
  
• Should AI opponents work together?
  
• Should the AI try to save certain types of tiles?
  
• Should the AI strategy change depending on the stage of the game (early / late) or depending on what strategy the opponents appear to employ?
  

• The AI can try to stay away from the edge, but it can go to the edge to combat a strategy in which you never go more than 1 block from the edge.
  
• I have several wins of the interesting 3-move variety, which has several steps:
  

1. Select a 2-player game with player 1 and CPU.
   
2. Start on the top right square.
   
3. On your first move, select a tile where you exit on the opposite side of the tile that you started from.
   
4. The CPU moves to the square you just moved to.
   
5. Your move forces the CPU, but not you, off the board.
   

So, the CPU should not go to the tile you're on (putting it at your mercy), but it can go to a tile you could move to with your next to (putting you at its mercy).

• I have no opinion. The AI can be better, but humans should have some chance to win.
  
• Some tiles that are good (to me) to save are the tic-tac-toe board (4 straight lines), the puzzle piece (4 U's), and the plus (4 small right angles). The AI, though should not save too many tiles, as it should have some choices to work with.
  
• In the early game, the AI can gravitate towards tiles you could move to (as I mentioned earlier) but in the late game, the AI could try to go to large open spaces. But when is the boundary between early and late game?
  

MathMathMax wrote:

thanks to kriblo for tsuro, I wouldn't know about it without him