I’m dubious of the idea that I should be training mentally in an area that a computer program can already trounce all humans in. Playing optimal poker is computationally solvable and not demanding, except for figuring out the biases of the human players.
I assume that a human can’t do better than a computer when playing against a player with no biases. At the professional level, play should converge closer and closer to the unbiased ideal, so that poker experience and human insight should become less and less valuable.
Do professional poker players make most of their money playing against rubes, or against other professionals? Is any of the money counted as their earnings prize money contributed by sponsors, or advertising sponsorships?
What are the arguments that a human can outplay a computer at poker? How is poker more useful than practicing multiplying large numbers together?
Bots only win at 1v1 limit poker. No bot can play professional no-limit poker, especially at a full table.
Again, the best humans are much, much better at poker than the best bots. The idea that optimal NL poker is computationally solvable and not demanding is just wrong. No one has solved it yet.
Yup. There are also many more situations in limit poker that have a clearly optimal play than no-limit poker.
In limit poker, you have the choice of check/call/raise/fold, where in no-limit poker the raising is fully continuous and you almost definitely don’t have the complete information to make the actual optimal play across the full possible range of bets.
Phil, if you wanted to read the best literature on this, The University of Alberta Poker group (run by the guy who weakly solved checkers, I think?), made a bot years ago that wins 1v1 limit poker against professional players and they keep writing about it while probably winning millions of dollars secretly on the internet. Or possibly they are too true of academics to actually run bots.
I’m dubious of the idea that I should be training mentally in an area that a computer
program can already trounce all humans in. Playing optimal poker is computationally
solvable and not demanding, except for figuring out the biases of the human players.
You obviously have very little knowledge of the topic you’re presenting yourself as an expert on.
Do professional poker players make most of their money playing against
rubes, or against other professionals?
As is a logical necessity, most make their money off of non-professionals. The very best probably make more off of other professionals, though there are some very bad players even at the highest stakes (e.g. billionaires who just like poker).
Is any of the money counted as their earnings prize money contributed
by sponsors, or advertising sponsorships?
There isn’t good public data on sponsorship deals. When one hears how much a particular professional makes, they are seldom included.
What are the arguments that a human can outplay a computer at poker?
The fact that you don’t see computers beating the best humans (except in some somewhat marginal forms of poker, and even there it’s debatable), and in most forms of poker, not even the semi-good players.
This isn’t a matter of “argument”, but a matter of observing the facts.
How is poker more useful than practicing multiplying large numbers together?
You don’t usually get money for simple multiplication.
You obviously have very little knowledge of the topic you’re presenting yourself as an expert on.
If I thought I were an expert, I would be answering questions instead of asking questions.
What are the arguments that a human can outplay a computer at poker?
The fact that you don’t see computers beating the best humans (except in some somewhat marginal forms of poker, and even there it’s debatable), and in most forms of poker, not even the semi-good players.
That isn’t an argument unless the best humans frequently play against computers. Do they?
A human could be better than a computer at beating another human. In a game with one computer and four humans, I can easily believe that one human might win more than the computer did.
In a game with four well-programmed computers and one human, I predict the computers will trounce the human regularly. I’m not an expert at poker; but I am an expert at computation, so I feel pretty confident about this prediction.
(A game with 3 well-programmed computers, one human, and one poorly-programmed computer would count as a game with 3 computers and 2 humans.)
I’m dubious of the idea that I should be training mentally in an area that a computer program can already trounce all humans in. Playing optimal poker is computationally solvable and not demanding, except for figuring out the biases of the human players.
I assume that a human can’t do better than a computer when playing against a player with no biases. At the professional level, play should converge closer and closer to the unbiased ideal, so that poker experience and human insight should become less and less valuable.
Do professional poker players make most of their money playing against rubes, or against other professionals? Is any of the money counted as their earnings prize money contributed by sponsors, or advertising sponsorships?
What are the arguments that a human can outplay a computer at poker? How is poker more useful than practicing multiplying large numbers together?
Bots only win at 1v1 limit poker. No bot can play professional no-limit poker, especially at a full table.
Again, the best humans are much, much better at poker than the best bots. The idea that optimal NL poker is computationally solvable and not demanding is just wrong. No one has solved it yet.
This doesn’t make sense to me. Why would no-limit be much harder than limit?
I’m no expert, but I expect it’s because the game tree is sparser in limit than in no-limit.
Yup. There are also many more situations in limit poker that have a clearly optimal play than no-limit poker.
In limit poker, you have the choice of check/call/raise/fold, where in no-limit poker the raising is fully continuous and you almost definitely don’t have the complete information to make the actual optimal play across the full possible range of bets.
Phil, if you wanted to read the best literature on this, The University of Alberta Poker group (run by the guy who weakly solved checkers, I think?), made a bot years ago that wins 1v1 limit poker against professional players and they keep writing about it while probably winning millions of dollars secretly on the internet. Or possibly they are too true of academics to actually run bots.
They’re working on no-limit poker now though, and I’d be surprised if bots haven’t passed humans within 5 years. For now, humans still dominate. http://poker.cs.ualberta.ca/publications.html
I’m no expert, but I expect it’s because the game tree in limit is helpfully pruned by the betting rules relative to the game tree of no-limit.
I believe that they can win LIMIT poker at a full table; does not have to be 1-1 in that case.
Out of curiousity, if computers were better to become poker players than humans (this is highly likely, in the long run), what would you say then?
Stop playing poker online unless you have a bot to play for you or a software tool to enhance and regulate your play.
You obviously have very little knowledge of the topic you’re presenting yourself as an expert on.
As is a logical necessity, most make their money off of non-professionals. The very best probably make more off of other professionals, though there are some very bad players even at the highest stakes (e.g. billionaires who just like poker).
There isn’t good public data on sponsorship deals. When one hears how much a particular professional makes, they are seldom included.
The fact that you don’t see computers beating the best humans (except in some somewhat marginal forms of poker, and even there it’s debatable), and in most forms of poker, not even the semi-good players.
This isn’t a matter of “argument”, but a matter of observing the facts.
You don’t usually get money for simple multiplication.
If I thought I were an expert, I would be answering questions instead of asking questions.
That isn’t an argument unless the best humans frequently play against computers. Do they?
A human could be better than a computer at beating another human. In a game with one computer and four humans, I can easily believe that one human might win more than the computer did.
In a game with four well-programmed computers and one human, I predict the computers will trounce the human regularly. I’m not an expert at poker; but I am an expert at computation, so I feel pretty confident about this prediction.
(A game with 3 well-programmed computers, one human, and one poorly-programmed computer would count as a game with 3 computers and 2 humans.)
“Well-programmed computer” sounds like “sufficiently smart compiler” to me :).