D
dgroes
Visionary
- Joined
- Mar 6, 2019
- Total posts
- 692
What did Jerod Ankerman meant by solving Poker as a game?
Solving poker means the same as it does for solving any other game. Formally a 'solved game' is one where given the state of the game at any point one can accurately predict the ending assuming player play perfectly. Perfectly meaning that they always make the best move for themselves.
Let's explore what that means by starting small.
Tic Tac Toe (or Naughts and Crosses ) is a simple game that you have probably already solved. So what does it mean to have solved TicTacToe? It means that you know every line of play that could happen, and you take the move that is most likely to lead to your winning. There are only 138 possible final boards for a game of tic tac toe, and so it's easy to imagine the ones that exist for the state of the game at any point, and make the move that leads to you winning in more of them.
Moving up in difficulty, we come to chess, which the human mind has not solved due to the Ultra High complexity of the game. - There are about 4.4x10^46 legal states of a chess game. In recent years, chess has been solved by a super computer that can do in chess, what you can do in TicTacToe. That is, see all finished game states, and make a move that will lead to winning in most of them.
Stepping up again, look at the game GO. Go has NOT been solved. There are simply too many board variations for our computing power at this time. There are computers that are amazing at it, but they play as we do, imagine a number of steps ahead, and plan according. It's just that computers can imagine a lot further and a lot faster than we can.
Finally we move on to poker. Image a NLHE SnG, with 6 people. 1500 starting ships, 50/100 blinds, and a deck of 52 cards. How many combinations can exist for just the hole cards? about 1.4x10^11 --- that's just hole cards before anyone bets... then multiply by preflop bet combinations for UTG.. fold, 1BB, 2BB, 3BB, 4BB, etc, then, multiply by the response possibilities from seat 2, times seat 3, etc...
You easily surpass all the number of atoms in the known universe trying to figure how many possible positions there are of a 6 player sit and go at any one time.
Poker can not be solved by any means that we know or theorize about. Some say quantum computing could hold an answer like that. But then, remember that every player has to play perfectly, which players don't do.
That being said, you can make a machine that can play like we do, given the proper inputs of player ranges, win percentages, bet/response lines, pot odds, etc. But such a machine should be easy to bluff. You would need a generalized Artificial Intelligence to really beat the game with a computer.
Again, that's not to say that bots don't make money sometimes. If you or I had the patience of a bot to play perfect percentages all the time, we'd usually come out ahead too.
Anyway, I hope that answers your question. You can read more about solved games on wikipedia here.
Solving poker means the same as it does for solving any other game. Formally a 'solved game' is one where given the state of the game at any point one can accurately predict the ending assuming player play perfectly. Perfectly meaning that they always make the best move for themselves.
Let's explore what that means by starting small.
Tic Tac Toe (or Naughts and Crosses ) is a simple game that you have probably already solved. So what does it mean to have solved TicTacToe? It means that you know every line of play that could happen, and you take the move that is most likely to lead to your winning. There are only 138 possible final boards for a game of tic tac toe, and so it's easy to imagine the ones that exist for the state of the game at any point, and make the move that leads to you winning in more of them.
Moving up in difficulty, we come to chess, which the human mind has not solved due to the Ultra High complexity of the game. - There are about 4.4x10^46 legal states of a chess game. In recent years, chess has been solved by a super computer that can do in chess, what you can do in TicTacToe. That is, see all finished game states, and make a move that will lead to winning in most of them.
Stepping up again, look at the game GO. Go has NOT been solved. There are simply too many board variations for our computing power at this time. There are computers that are amazing at it, but they play as we do, imagine a number of steps ahead, and plan according. It's just that computers can imagine a lot further and a lot faster than we can.
Finally we move on to poker. Image a NLHE SnG, with 6 people. 1500 starting ships, 50/100 blinds, and a deck of 52 cards. How many combinations can exist for just the hole cards? about 1.4x10^11 --- that's just hole cards before anyone bets... then multiply by preflop bet combinations for UTG.. fold, 1BB, 2BB, 3BB, 4BB, etc, then, multiply by the response possibilities from seat 2, times seat 3, etc...
You easily surpass all the number of atoms in the known universe trying to figure how many possible positions there are of a 6 player sit and go at any one time.
Poker can not be solved by any means that we know or theorize about. Some say quantum computing could hold an answer like that. But then, remember that every player has to play perfectly, which players don't do.
That being said, you can make a machine that can play like we do, given the proper inputs of player ranges, win percentages, bet/response lines, pot odds, etc. But such a machine should be easy to bluff. You would need a generalized Artificial Intelligence to really beat the game with a computer.
Again, that's not to say that bots don't make money sometimes. If you or I had the patience of a bot to play perfect percentages all the time, we'd usually come out ahead too.
Anyway, I hope that answers your question. You can read more about solved games on wikipedia here.
Solving poker means the same as it does for solving any other game.