Techie Babble
Working on the BOT is a pain in the ass. I decided to go with C# because I love the language, however it is not great for subclassing. I have gotten a few subclassing algorithms to work but nothing seems stable. I am forced now to reconsider the method I use. I have two choices:
A) Send a GetWindowsText Message every few seconds and figure out where I am. I would need to read the entire string stored in the CHAT window and figure out the hands, etc.. It may turn out that this is the best option.
B) Read the files off the Hard Drive. It should work, and it also will give me a marker to the last thing I read. It has its limitations too. I am leaning towards this for two reason: 1. No API 2. All Managed Code (Which is good in C#). The other nice thing is I can grab a textfile or two from my machine and work on this anywhere. I do not need to be really playing.
Next step after figuring out that part is to again use Windows APIs to effect the BUTTONS in party. I will need to have a wait loop that figures out when the buttons appear, and then does a final read of the textfile and makes its move. Since the computer is much faster at doing this stuff I do not think there will be much delay.
So the boring stuff will be done and I can work on the interesting portion of the project. The AI. Since I am a sucky player and even worse AI designer v1 is going to use the KISS method of development: Keep it Simple Stupid! It is going to play the way I would play if I had percentage and pot odd information in my head. I have yet to master figuring out my odds so in this aspect the BOT will actually be superior to me. I am not going to go into details of exactly how the BOT will work, but since I am not extremely aggressive it will probably be a conservative BOT.
Once I have tested v1 of the BOT I will then work at making it a better player than me. The only real way to do this is to have it analyze other people’s play and use it against them. Unlike me it can remember every move that every player has made. It can figure out who the bluffers are and what people are likely to play. It probably will not be able to sense when it is beaten. However sometimes I think these hunches are overrated. I think at least in low limit games it has to be more playing the percentages perfectly. No deviation. Perfect for a computer. I guess this theory will be put to the test.
1 Comments:
did it work ?
6:12 PM
Post a Comment
Subscribe to Post Comments [Atom]
<< Home