Is poker broken?
Maybe. My buddy came up w\ a way to break Scrabble. I'm almost certain there's a way to break Clue very, very quickly using some very rudimentary logic. Checkers just got broke. If they can do this to poker, it'll be impressive.
Diogenes of Sinope: "It is not that I am mad, it is only that my head is different from yours."
Arnold Judas Rimmer, BSC, SSC: "Better dead than smeg."
Arnold Judas Rimmer, BSC, SSC: "Better dead than smeg."
Actually, I want you to explain "break".
And one poker player does not make a good test (ok, two, but it was one at a time).
Enter this computer into a few tournaments and let's see how it finishes.
I actually saw that article a few days ago and thought about Skynet and how we're teaching all our PCs how to beat us in everything. Then I realized we could teach it everything about war and it still wouldn't know how to beat us. In a game that is limited, like checkers or chess (in which the best PC still can't consistently beat the best humans) or poker or whatever, the PC may have an advantage, but in war there is no advantage. There's no way to know all possible moves. There's no use in having history as a guide. In fact, it might be the weakness that we exploit as the AI would protect itself against the things it knows, which any plan could exploit.
P.S. My name is really John Connor.
And one poker player does not make a good test (ok, two, but it was one at a time).
Enter this computer into a few tournaments and let's see how it finishes.
I actually saw that article a few days ago and thought about Skynet and how we're teaching all our PCs how to beat us in everything. Then I realized we could teach it everything about war and it still wouldn't know how to beat us. In a game that is limited, like checkers or chess (in which the best PC still can't consistently beat the best humans) or poker or whatever, the PC may have an advantage, but in war there is no advantage. There's no way to know all possible moves. There's no use in having history as a guide. In fact, it might be the weakness that we exploit as the AI would protect itself against the things it knows, which any plan could exploit.
P.S. My name is really John Connor.
"Happy slaves are the worst enemies of freedom." - Marie Von Ebner
"It was always the women, and above all the young ones, who were the most bigoted adherents of the Party, the swallowers of slogans, the amateur spies..." - Orwell
"It was always the women, and above all the young ones, who were the most bigoted adherents of the Party, the swallowers of slogans, the amateur spies..." - Orwell
Two birds w\ one stone.TheCatt wrote:I'll bite, how do you break Scrabble?
By "broken," I mean that there's a mathematical way to prove one particular game strategy is optimal & can be automated. No one can ever outperform the method.
Tic-Tac-Toe is broken as well, for example.
My buddy broke Scrabble by writing some code that analyzes the board layout, your current letters, & digs thru a word list to determine what word in what position will score the most points.
Diogenes of Sinope: "It is not that I am mad, it is only that my head is different from yours."
Arnold Judas Rimmer, BSC, SSC: "Better dead than smeg."
Arnold Judas Rimmer, BSC, SSC: "Better dead than smeg."
I read that article a few days ago as well.
You can only have a perfect/optimal strategy when you know all of the details.
In Scrabble you know the maximum points you can get with the pieces you have, and you can know what words your opponent is likely to make out of your words.
In games like Hold'em you can't know all of the details and betting adds way too many details.
When you're playing heads up and someone bets aggressively 10 hands in a row, is it because they are overplaying the value of their cards, or did they just happen to get better than average cards?
Now, I *can* envision a computer examining tens of thousands of hands that a player has been in, comparing/contrasting/estimating his tendencies in certain situations, and making better than average estimates based on the actions of its opponent. After all, humans are creatures of habit and tend to repeat themselves.
But there's so much data, based on chip stacks, emotional responses to previous hands, the mental state of the players, the perception the players have of each other, the tiredness/irritability of the players etc. Even off-table stuff like having to quit the game in 10 minutes to pick up the kid at pre-school is a factor in how people play at any given moment.
Plus people change how they play based on how they thing their opponent is playing, and how they think their opponent will react.
Plus, when you read a poker book it'll say stuff like, "With X/X raise to 4 big blinds 70% of the time, raise to 3 big blinds 15% of the time, raise to 2 big blinds 10% of the time, and limp in 5% to keep people from reading you."
There are 1,326 different combinations of hold cards in Texas Hold'em (52 x 51 = 2652, divided by 2 is 1,326 [you divide by two because Kh/Qc is the same as Qc/Kh]). Who's to say that your raise to 3 big blinds was because you have X/X and bet with way 70% of the time. Or maybe it's because you have X/Y and this is the 10% of the time you make an over bet with that hand. Or maybe you have Y/Y and it's part of the 50% of the time you make that sized bet? Or maybe it's a lousy Z/Q because you've folded the last 10 hands due to lousy cards which makes you look conservative, so a steal attempt is likely to work. Or maybe it's the 15% of the time he makes that bet with R/S?
You can only have a perfect/optimal strategy when you know all of the details.
In Scrabble you know the maximum points you can get with the pieces you have, and you can know what words your opponent is likely to make out of your words.
In games like Hold'em you can't know all of the details and betting adds way too many details.
When you're playing heads up and someone bets aggressively 10 hands in a row, is it because they are overplaying the value of their cards, or did they just happen to get better than average cards?
Now, I *can* envision a computer examining tens of thousands of hands that a player has been in, comparing/contrasting/estimating his tendencies in certain situations, and making better than average estimates based on the actions of its opponent. After all, humans are creatures of habit and tend to repeat themselves.
But there's so much data, based on chip stacks, emotional responses to previous hands, the mental state of the players, the perception the players have of each other, the tiredness/irritability of the players etc. Even off-table stuff like having to quit the game in 10 minutes to pick up the kid at pre-school is a factor in how people play at any given moment.
Plus people change how they play based on how they thing their opponent is playing, and how they think their opponent will react.
Plus, when you read a poker book it'll say stuff like, "With X/X raise to 4 big blinds 70% of the time, raise to 3 big blinds 15% of the time, raise to 2 big blinds 10% of the time, and limp in 5% to keep people from reading you."
There are 1,326 different combinations of hold cards in Texas Hold'em (52 x 51 = 2652, divided by 2 is 1,326 [you divide by two because Kh/Qc is the same as Qc/Kh]). Who's to say that your raise to 3 big blinds was because you have X/X and bet with way 70% of the time. Or maybe it's because you have X/Y and this is the 10% of the time you make an over bet with that hand. Or maybe you have Y/Y and it's part of the 50% of the time you make that sized bet? Or maybe it's a lousy Z/Q because you've folded the last 10 hands due to lousy cards which makes you look conservative, so a steal attempt is likely to work. Or maybe it's the 15% of the time he makes that bet with R/S?
That is so not breaking Scrabble.Malcolm wrote:Two birds w\ one stone.TheCatt wrote:I'll bite, how do you break Scrabble?
By "broken," I mean that there's a mathematical way to prove one particular game strategy is optimal & can be automated. No one can ever outperform the method.
Tic-Tac-Toe is broken as well, for example.
My buddy broke Scrabble by writing some code that analyzes the board layout, your current letters, & digs thru a word list to determine what word in what position will score the most points.
Having the highest scoring possibility for a given turn does not necessarily maximize player welfare throughout the entire game: Does it open too many opportunities for your opponent? Does it maximize possibilities for a bingo later? Does it manage the rack effectively?
That does not, even by your own definition, mean broken.
It's not me, it's someone else.
I will accept Tic Tac Toe is broken, but we've all known that since childhood.
It's not me, it's someone else.