Back to General discussions forum

Hi Friends! as most problems I have created recently were buggy :) I suppose we can try different approach. The new problem is available but not listed yet, so that it can be tried by enthusiasts and any necessary amends could be made before it is finalized (it can't be successfully solved in its current state).

Trick-Taking Game

For some time I wished to have some traditional card game exercise (particularly because I don't well understand how they are programmed in the sense of "computer player" algorithm). Here comes simple game with very primitive "computer opponent" (you'll see it can less or more logically defend but it attacks randomly). Several simultaneous games are to be played - and I guess we need to set some threshold of games won to judge the problem as solved. For example 10 of 15 games should be won (draw doesn't mean victory!) or so.

Thus the questions, besides spotting some bugs or description flaws, are:

• does it make sense to try making machine attack more clever than random (and what logic you suggest) - or it is enough as is?
• what amount of games to play and threshold value to win you suggest?
• any other suggestions (deck size? I feel it is better to have odd number of ranks)

Thanks in advance!

P.S. interactive problems now should work with JSON format also, using the Content-Type:application/json header, e.g.:

curl http://codeabbey-games.atwebpages.com/tricks.php \
    -HContent-Type:application/json 
    -d '{"token":"dhXR6/3xWYBMM+5KiXFvbjQw","move":"Y4 W5 X0"}'

{"states":"+X4 >Y2+X1 -Y3>X4","scores":"0:1 0:1 1:0"}