Bots and Poker

[TexasKowboy]

Here is a link to a MSN artical on bots.

http://www.msnbc.msn.com/id/6002298/

I would like to know what everyone thinks?

Tetsuo I didn't know where to put this. Would you please move it if it needs moved?

[Nortonesque]

This is probably the most revealing part:

The University of Alberta’s Computer Poker Research Group has developed an artificially intelligent automaton known as “Vex Bot,” capable of playing poker at the master level, though as yet it can only apply its gambling genius to two-player games.

Two players is one thing, but a full game is something else. You might be able to program something to win at the lowest limits, with straight ABC strategy, but higher limits and NL would be extremely difficult to program. The good players would catch on to its tendencies pretty fast.

[iceman5]

Bring on a predictable bot.

[tetsuo]

There are such things as ring game bots. The poki poker academy has a few types - sparbot, vexbot, and ringers with various 'traits' that you can tweak.

Whether they'd win or not at Party is another matter lol

[TexasKowboy]

[k3nt]

[WildBillHickok]

shades of Blue and Junior playing world champ Gary Kasparov...

but i don't think they'll develop a bot that would beat the best poker players...like someone mentioned, the live player would catch on and eventually trounce the bot

[tetsuo]

That time Blue won quickly, Kasparov basically announced to the world that he had been 'on tilt' (not his words, but that's what he meant) and didn't play his A game.

WildBill - the trouble with AI is it will eventually be able to tell when you have 'caught on' and change its strategy. As it is, the VexBot will undoubtedly have some game theory built into it to keep it unpredictable.

Think about how much computers have come on in the last few decades.. Think about what they will be capable of in another five, then, twenty years... If you aren't scared and/or thrilled, you aren't thinking far ahead enough

Moore's law is still pretty applicable. He said (paraphrasing) that computing power would double every year. This law has held for quite some time, but in the future we will be seeing scalability breakthroughs which uncouple us from the theoretical boundaries of the tehnology we use today.

By Moore's law (which I think is pessimistic, going forward), in 10 years you computer will have over a THOUSAND times more power that it does today. I'm quite convinced that this power will be adequate to beat humans at any/all games.

BTW Big Blue is not 'intelligent' as far as I know - it just 'brute forces' game scenarios, i.e. it calculates all possible combinations of moves to the finish from its current point and selects the best one.

This post turned out far longer than I had intended. Apologies.

[WildBillHickok]

...about bots changing its strategy after it sees you've caught on...i thought of that after i'd posted...
that's ok...i still got my 44, they'll never beat me!

[euri10]

the big difference between poker and chess or backgammon for instance is the non-complete state nature of the game.
bots are tremendously better in complete games where all the game is known by every player at each instant, this is the case for chess & backgammon. At each instant of time you can describe precisely what is the game. and each player has the same exact information.

however for poker, information is incomplete. IA made a lot of progress these past years developing game models to play backgammon. TD-gammon, now in its version 3 consistenly beats all world class players and he learns from itself, it's amazing to study this program/
research on games where information is incomplete is far more complicate and gives poor results.
university of Alberta indeed works on such IA problems, which are THE questions of the future but as far as I know u can beat very easily such a bot.