Neverlose

[schroedy]

I am not sure that your friend is right about backgammon.

I am not a player, but the doubling cube is as much a function of mathematics as anything else in backgammon.

[AviD]

Do you know of anyone that actually paid $100 for that program?

I'm lost as to how it uses "neural networking" which requires a brain, no?

This program a racket or actually plays well, good enough to pattern and beat most players?

[Evan]

[ QUOTE ]
I'm lost as to how it uses "neural networking" which requires a brain, no?


[/ QUOTE ]
No, neural networking is a term for artificial intelligence mediums that can "learn" form experience. Kinda like when the help thing pops up in Word and you keep closing it, eventually it will ask if you want it to stop popping up; its like that only a lot mroe advanced.

[schroedy]

Microsoft needs to program in an option that asks "Mr. Clippy" to perform an anatomically impossible function.

[Kevin J]

Thanks. I will check it out. I wasn't at all taking any offense. I just find this whole subject of bots very interesting.

When Paradise poker first came out, the first thing I thought of was to create a bot that could play 24/7. After seeing how bad some of the players were, I had visions of making gazillions! Unfortunately, my friend who is a computer programmer knew nothing about poker. I know nothing about computers. It quickly became apparent that it would be too daunting of a task for just the two of us and I gave up on the idea. Thanks again for the info.

[AviD]

Yes I was just perusing their site some more and saw that Poki applies this learning over time and will get better against you over time.

Some of the info on there is quite disturbing, considering they are condoning (some openly, some quietly) using Poki to play against humans for a profit. How is that legal? And how do they stay in business?

Keep in mind, I can be pretty dense at times...I'd imagine this is one of those times.

[vector2]

I've had the commercial version of Poki since it first came out, and its ring games are actually pretty good if you set the AI of each player to the right mix of tightness and aggressiveness. I probably have a winrate on the ring games of about .3sb/h over about 10,000 hands against it. I absolutely crushed it when I first started playing it, but now it's gotten so good that I just have to look at the hand history after I fold to see if a bot had the nuts or just overcards after it three-bet me.

As for the HU players in Poki, they have one bot who just attempts to play tricky and really aggressive, and one player who supposedly adapts to your style of play to exploit you. I'm about .6sb/h against both the HU bots, I don't have nearly enough hands against them HU to paint an accurate picture. I have, however, started new sessions where I played against the adaptive bot with the cards face-up, and I must admit that it can make some pretty impressive call-downs with weak holdings if it has me pegged on a bluff. However, I think at the same time that that's where it's strongest weakness lies is paying off too many hands. Whatever the case, I definitely think it's worth the purchase.

[Kevin J]

I also am not a backgammon player and don't understand why it is so much harder for a computer to beat a good player at that game than chess. But my friend is an excellent player of both games. I'm not sure if he's world class at backgammon, but he's damn good having traveled extensively to tournaments, etc. and winning more than his share. He's also made lots of money in the side backgammon games around the world. He is also an excellent poker player. If he says backgammon is a harder game to program a computer for, then I believe him.

[turnipmonster]

I paid for the program. it's definitely not a racket, it came directly out of U of Alberta's games research group:
http://www.cs.ualberta.ca/~games/poker/

there are a lot of good white papers on that site about opponent modelling for hold 'em. in addition, they did some research involving one of the best short players in the world (thecount on stars).


--turnipmonster

[scrub]

[ QUOTE ]

I haven't given poker bots much thought, despite being a programmer myself...but many of these "signs" indicating potential bot usage are not that difficult to overcome...

[/ QUOTE ]

[ QUOTE ]
I'm lost as to how it uses "neural networking" which requires a brain, no?

[/ QUOTE ]

Something here doesn't add up... Can you see why?

scrub