Neverlose

[GameTheory]

I think at one point he was a bot. The issue was brought up to PS by a number of the 100/200 players. It was investigated, and NLose still plays on Stars quite frequently, however, it seems to me that he's not doing half as good as he used to.

[Diplomat]

...on that same thought though, if he was a bot, and assuming pokerstars cared, why would they let him/her/it keep playing, even if he agreed to not hook up the bot? Why not just ban the account? If true, this is actually more unsettling; it's not that they are stupid and do not pick up on the bot, but rather they just don't care.

It's easy to upgrade software. It's much tougher to change a mindset.

-Diplomat

[Turning Stone Pro]

Very intersting thread. I am curious if the 2+2 experts feel that poker is like chess, where you can build a program (like BigBlue in the chess world) that can beat all (or nearly all) humans?

In chess, I can see that, depending on the layout of the board, there can be only one "best" move at any one time.

BUT, is poker like this? Isn't it more player-specific, where there might not be a "best" play at any one time, depending on how well the player knows the opponent(s)?

Just curious. I think that for a 10 handed table, preflop, it might be possible to come up with some sort of "basic strategy", similar to blackjack, for each situation.

But for short handed or especially heads-up play, considering the blind-structure of modern holdem, I think that one plays the person much moreso than the cards. When I play someone heads up, I try to learn what they are going to do in certain situations. (For example, some players lead on the flop with nothing if an A flops and I didnt raise pre-flop, 100% of the time).

How could this be programmed into a computer, especially when your opponents on PP change tables so quickly? Games immediately break down and start up again, it's not like you know the computer will be playing with Kasparov for an extended period of time.

Just wondered what people's thoughts are.

TSP

[Evan]

[ QUOTE ]
But for short handed or especially heads-up play, considering the blind-structure of modern holdem, I think that one plays the person much moreso than the cards

[/ QUOTE ]
As I understand it, the only scenarios that bots have been able to destroy humans is heads up.

[ QUOTE ]
Games immediately break down and start up again

[/ QUOTE ]
Does the 100-200 game on Stars break and restart that often?

[David Steele]

[ QUOTE ]
I am curious if the 2+2 experts feel that poker is like chess, where you can build a program (like BigBlue in the chess world) that can beat all (or nearly all) humans?


[/ QUOTE ]
Yes. ( It was Deep Blue BTW )

[ QUOTE ]
BUT, is poker like this? Isn't it more player-specific, where there might not be a "best" play at any one time, depending on how well the player knows the opponent(s)?


[/ QUOTE ]

Poker can not be beaten with a chess program, true, but here is nothing to stop the programmers from statistically modeling the opposition and perhaps have a better model then human players are able to make. On the other hand some game theory type non-exploitive strategy might work well enough.


[ QUOTE ]
... short handed

[/ QUOTE ]
It has been reported that the short handed bots have been very strong already.

[ QUOTE ]
How could this be programmed into a computer, especially when your opponents on PP change tables so quickly?

[/ QUOTE ]

How can humans play either? I would think it would be easier for the computer to access its player data ( should any be available ) faster then humans. If none is available then the bot and the human will have to start modeling from scratch.

D.

[glen]

"How could this be programmed into a computer, especially when your opponents on PP change tables so quickly? Games immediately break down and start up again, it's not like you know the computer will be playing with Kasparov for an extended period of time."

When neverlose was first being accused, a bunch of player all sat in a room and talked about him. Thecount1729 said he thought that neverlose was the kind of bot that compiled all of the hand histories that it has played against you, and then learns from them. So, at first you might be able to hang with it, but then it would just get better and better against you, i.e. see how prone you are to cr the turn with picked up flush draws, notice if you always bet an ace high flop when he doesn't raise preflop etc. . . .

[NLfool]

I believe theCount or something like that actively worked and played against a bot called poki and said that there were big gains in regards to bot play in poker. He maybe the expert you're looking to talk to

[glen]

yeah, but this was not the case with neverlose. He would often have like 70k spread out over 3 tables, and clearly more in his account. so say one day he started playing 3 tables with 10k each, then a few hours later he would have 17k on one, 8500 on another, and 200 on a 3rd, but he would not reload even though he had plenty of time to just click the dealer button for more chips. . . so then he might double up once or twice by going all-in and have like 800 on the table, THEN he would reload. bizarre. . .

[turnipmonster]

I don't know much about this specific situation, but as a programmer here's my .02 cents.

- the rebuying thing is probably the most convincing piece of evidence, especially if rebuying requires you to do something graphically that a program would have a hard time doing. It would surprise me if someone smart enough to program a bot this good wouldn't be able to figure this out though.

- I am interested to know how good the pokibots (sparbot,vexbot etc) are at playing shorthanded hold 'em. they are very good at heads up.

- I would again be shocked if someone who could write a bot that good at HE could not implement something in O8, which is generally a simpler game (I may be wrong, it is at low limits though).

my guess? well, if I were going to have a bot on a site, I would definitely have it do some things that were very un-bot like. I would also probably play under the same account myself, which could account for lots of weird behavior.

anyone smart enough to write a bot that can beat those games, is also smart enough to make it pretty hard to figure out if the bot is actually a bot. so my guess is you won't come up with a definitive answer unless pokerstars does some serious data mining and even then it would be very difficult to tell.

one final thought. it would be weird to write a bot to target games of that size (100/200). the players there are going to be very tough, and unless it was extremely strong I would imagine that unleashing 100 bots on the party 15 would be far more profitable.

--turnipmonster

[Vaftrudner]

Yes, and both the players (and the site I assume) are much more aware (of bots, collusion and such) at that level. Another reason for "unleashing 100 bots on the party 'some-lower-level'". /v