Poker: Human versus computer programme

[TightEnd]

Heard an interesting discussion last night

one person asked if we could ever foresee the day when a piece of computer software would be able to beat all the best poker players in the world

this person is a skilled backgammon player, and pointed out that computational programmes can now match and beat all but a very small number of players

previously we have seen "Deep Blue", developed by IBM, beat Kasparov at Chess

Similar has been seen in Bridge

I think the key constraint to developing a cast iron method of computer software beating humans at poker is two fold

a) the psychological aspect of poker ( it might be right to make a play, doesn't mean we do it)
 
b) poker is a game of partial information and deduction, not merely programmable logic...this is a key difference from backgammon for example where all the stones are visible and information is therefore more complete and one can do probability regressions on likely throws of the dice etc

and yet I know many programmes are being worked on, (above the existence of "bots" already) and if successful could change the nature of internet poker for ever

Your thoughts please

[RED-DOG]

I think I would have an advantage over a computer because it can't pull My plug out

[TightEnd]

trust you 

[Nose]

I think the key to a succesful Bot will be its ability to gather information on other players and use it to its advantage.

For example, if a Bot wants to beat the 15/30 game on party, it can 'watch' every table for a month, gathering a huge amount of information on the regular players and their habits, betting patterns,indicators that they are on tilt etc etc. Once it has this info, a lot more of the hidden pieces are now in full view of the bot. It will just take some programmers with a keen eye for the most significant information to design something that can use this information to be a winning player.

[ifm]

The whole bot thing will keep cropping up.
The fact is that yes computers will at some point be able to win at poker against humans. They said it couldn't happen at chess because of the unpredictable nature of the human (making unconventional moves) but the computer became more advanced.
The same principle applies here, eventually nobody will be playing on the net it will be bot v bot.

[bundle]

I think the key to a succesful Bot will be its ability to gather information on other players and use it to its advantage.

For example, if a Bot wants to beat the 15/30 game on party, it can 'watch' every table for a month, gathering a huge amount of information on the regular players and their habits, betting patterns,indicators that they are on tilt etc etc. Once it has this info, a lot more of the hidden pieces are now in full view of the bot. It will just take some programmers with a keen eye for the most significant information to design something that can use this information to be a winning player.

The way a successful bot wins is in the formula that is programmed into it. Generally by a geek and a very good poker player.

 It’s then tested thought playing at least 10,000 hands, and then they look for any weakness.
  These formulas are tweaked and tweaked until it is running at a successful win rate, it really doesn’t matter about watching a table for days or weeks before, it all boils down to probability.  A machine will grind without needing food, sleep, and will never go on tilt after a bad beat. But with that said, I think if you know it’s a bot you could beat it without too much trouble, you can figure it’s programmed and will soon get an idea of how it plays, the trouble lies in playing more than one at the table, now your in all sorts of trouble

[Karabiner]

I reckon some kind of "superbot" could well be nigh unbeatable at fixed limit.

Not so sure about big bet poker. IMHO

[The Baron]

One of the pros in Vegas claimed he used a bot or bot software to win 7 main event seats. Judging by the size of games I saw him play I dont think he needed to lie about it.

[thetank]

My concern over bots killing the fixed limit game influenced my desicion to switch to tournaments.

That and collusion being less of a problem in tournaments.

[Nem]

My concern over bots killing the fixed limit game influenced my desicion to switch to tournaments.

That and collusion being less of a problem in tournaments.

You hit the nail on the head.

[thetank]

I hope computer software is close to human ability coz I won the WSOP on Wilson software's tournament texas holdem for windows the other day (after 57 attempts, 3rd final table)

I think in real life it might take me closer to 5700 attempts as a lot of the big boys could read me like a book. That and fatigue setting in late in the day leading to poor desicions on my part.

If I give up smoking there may be a greater chance of me living to the ripe old age of 5723 and taking home a bracelet. (If the sun hasn't exploded by then)
 

[ifm]

some bot programs can be linked, then you will know others hole cards on tables (without knowing the person using it) and then maximise winnings.
I have said it before but i got a bot specifically written for pacific poker, more curiosity than anything, anyway it got creamed!!!
Though you can tamper with it's settings, haven't tried it since though.

[SupaMonkey]

I think bots could beat players in low stakes NLH but the higher up you go it just won't be possible. The reason that it can beat chess and backgammon players is because there are a limited number of moves but this is not so in poker. Also, good players will figure out what the bot thinks of their own playing style and be very willing to show specific betting tells at it sometimes just to wipe the floor with it at other times.

No good poker player plays mechanically.

[ifm]

You think chess has a limited number of moves??

[thetank]

I think he means that chess is a game of complete information, whereas poker is a game of partial information.