Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

How A.I. Conquered Poker

nytimes.com

Post reply on HN
Print view View original

181–188 of 188 posts

Re: How A.I. Conquered Poker

#181

Post by misja111 »

bluecalm wrote: post #160
misja111 wrote: post #151

Earlier quoted context omitted.

> You can only win in poker if you recognize how your opponent deviates from the optimal strategy and then play a strategy to exploit him. This is not true. If you play optimal strategy, you will win against any opponent except one that plays optimal as well, in which case you'll break even. But, of course you'll win a lot more if you are able to adapt your strategy to exploit any weaknesses that you have detected. A…

This is almost correct but not quite. Optimal strategy wins against many other strategies and breaks even against others. The ones it breaks even against are not necessarily optimal themselves, they just don't make mistakes vs the optimal one but might be exploitable themselves. To be more precise: you only need to replicate not mixed (pure) plays of the optimal strategy to not lose against it. Your frequencies for m…

In theory you are right but when talking specifically about poker, I can't think of any example of imperfect play that isn't exploited (to a certain extent) by optimal play. Yes, playing the opponent would win more, but optimal play makes a (smaller) profit as well. Or would you know an example?

Re: How A.I. Conquered Poker

#182

Post by kuboble »

3pt14159 wrote: post #173
kuboble wrote: post #161

Earlier quoted context omitted.

Poker isn't like this. Think rock, pair, scissors, crap where rules are that crap always loses. Now uniform strategy between rps will yield profit vs an opponent who plays crap sometimes. It's very easy to play crap in poker

Right, but so what? If the optimal strategy is a 50-50 split against someone else playing the optimal strategy and at least a 50% win against someone who is not, it doesn't follow that if they other player plays crap every game that the optimal strategy is best against them. Identification of the crap player means that you increase your bets vs what you would normally bet for perfect play. Perfect play is bullet proo…

The point being that in poker and in particular Texas Holdem - unlike in RPS - the optimal strategy will win against good players and completely obliterate weak players. Yes, the good human can win even more against bad human but make no mistake - it's extremely hard to not lose a lot vs optimal strategies.

>Identification of the crap player means that you increase your bets.

That's not true. There are a lot of different ways to exploit bad players depending on how bad they are. Also a lot of weak players these days have much more subtle leaks (e.g. never bluff-raise the river) and it takes some skill to spot and exploit them.

Re: How A.I. Conquered Poker

#183

Post by HALtheWise »

hashberry wrote: post #120
HALtheWise wrote: post #8

Earlier quoted context omitted.

The taxonomy I personally use is: * Every computer program has algorithms, it's an extremely general term for "the idea behind how the computer will solve the problem". Advanced algorithms are typically those that took a lot of human effort to come up with. * Machine Learning refers to a specific class of algorithms where the computer automatically figures out (part of) what it should do based on data. * Deep learnin…

I like your taxonomy. But just to clarify: is the cruise control on my car "artificial intelligence?" Because a lot of dumb 15-year-old teenagers don't know they have to increase power while going up a hill, while my cruise control algorithm does.

My category for AI is based on "popular perception", not necessarily an actual comparison to what humans can or can't do. Most humans don't know how to multiply complex numbers, but that doesn't mean doing so demonstrates human-level intelligence and decision making.

In the specific case of cruise control, I don't think maintaining speed accurately is typically viewed as "difficult" for a computer today, so it would be deceptive to refer to as AI. 100 years ago that may have been different.

Re: How A.I. Conquered Poker

#184

Post by HALtheWise »

JacobThreeThree wrote: post #45
HALtheWise wrote: post #8

Earlier quoted context omitted.

The taxonomy I personally use is: * Every computer program has algorithms, it's an extremely general term for "the idea behind how the computer will solve the problem". Advanced algorithms are typically those that took a lot of human effort to come up with. * Machine Learning refers to a specific class of algorithms where the computer automatically figures out (part of) what it should do based on data. * Deep learnin…

I like your framework. Would you put statistical analysis "algorithms" in the same category as "Machine Learning"?

Sure, "classical" ML of various forms (knn, decision trees, etc.) are Machine Learning, but aren't Deep Learning. That's exactly why the distinction between those phrases is useful.

Re: How A.I. Conquered Poker

#185

Post by smallerfish »

MaxfordAndSons wrote: post #146
conductr wrote: post #32

Earlier quoted context omitted.

I'm actually working on this as a side project. I've been dragging my feet as I wasn't sure anyone would want to "code to play" and had no real marketing plan. It's a fun project though, especially trying to design in a way that proves I'm not manipulating the deck behind the scenes. Email in profile if anyone is up for being an alpha player.

I've been thinking about a similar project. Do you intend to have it be only bots? I think it would be interesting to also allow human players to play against the bots (perhaps incentivized by getting preferential rakeback deals compared to the Bot players?). Many players fret about bots on other sites; it would be cool to enable those of us who want the challenge play them knowingly and with a slight financial handi…

I sketched out plans for the idea a while back (and have been too busy to do anything about it.) I'd have bot only games, and mixed games. I'd also have tools built into the game to let you pull up full stats on any player (there's add on software right now for major platforms that gives asynchronous advantage to players who use it, so to mitigate that it'd need to be adopted.) I think if the stats were good enough it'd be relatively easy to spot bots, which might make human-only tables possible also. The system would be punitive against bots which didn't declare themselves as such, but it would fully supportive of bots being part of the game on the online sphere.

Re: How A.I. Conquered Poker

#186

Post by bluecalm »

misja111 wrote: post #181
bluecalm wrote: post #160

Earlier quoted context omitted.

This is almost correct but not quite. Optimal strategy wins against many other strategies and breaks even against others. The ones it breaks even against are not necessarily optimal themselves, they just don't make mistakes vs the optimal one but might be exploitable themselves. To be more precise: you only need to replicate not mixed (pure) plays of the optimal strategy to not lose against it. Your frequencies for m…

In theory you are right but when talking specifically about poker, I can't think of any example of imperfect play that isn't exploited (to a certain extent) by optimal play. Yes, playing the opponent would win more, but optimal play makes a (smaller) profit as well. Or would you know an example?

The imperfect play is in the mixing. Imagine you play like optimal but every time you have a mixed action of bluff or check you bluff. You won't be exploited by an optimal strategy but you will be very vulnerable vs someone who bluff catches a lot.

Re: How A.I. Conquered Poker

#187

Post by 3pt14159 »

kuboble wrote: post #182
3pt14159 wrote: post #173

Earlier quoted context omitted.

Right, but so what? If the optimal strategy is a 50-50 split against someone else playing the optimal strategy and at least a 50% win against someone who is not, it doesn't follow that if they other player plays crap every game that the optimal strategy is best against them. Identification of the crap player means that you increase your bets vs what you would normally bet for perfect play. Perfect play is bullet proo…

The point being that in poker and in particular Texas Holdem - unlike in RPS - the optimal strategy will win against good players and completely obliterate weak players. Yes, the good human can win even more against bad human but make no mistake - it's extremely hard to not lose a lot vs optimal strategies. >Identification of the crap player means that you increase your bets. That's not true. There are a lot of diffe…

I reread the entire chain and realized I misinterpreted the context around what you said and thus came to the wrong conclusion of your understanding and meaning.

> > You can only win in poker if you recognize how your opponent deviates from the optimal strategy and then play a strategy to exploit him. > This is not true.

You were right here when you said "This is not true." I still think there is more nuance than the discussion as a whole credits, but overall I think you were more right than wrong in our discussion and I was more wrong than right. Thanks for the chat.

Re: How A.I. Conquered Poker

#188

Post by conductr »

MaxfordAndSons wrote: post #146
conductr wrote: post #32

Earlier quoted context omitted.

I'm actually working on this as a side project. I've been dragging my feet as I wasn't sure anyone would want to "code to play" and had no real marketing plan. It's a fun project though, especially trying to design in a way that proves I'm not manipulating the deck behind the scenes. Email in profile if anyone is up for being an alpha player.

I've been thinking about a similar project. Do you intend to have it be only bots? I think it would be interesting to also allow human players to play against the bots (perhaps incentivized by getting preferential rakeback deals compared to the Bot players?). Many players fret about bots on other sites; it would be cool to enable those of us who want the challenge play them knowingly and with a slight financial handi…

Both could work. I’m particularly interested in the all bot variety because the pace can be very accelerated. For example, an idea I’m kicking around is having timeout be under 1 second.

I see it like HFT. You may have multiple bots running multiple playing strategies and an entire tournament could occur in minutes/hours. A large weekly tournament might even be like a sporting event for hackers.

The human vs bot scenario you’re interested in would be possible on the same platform with a longer timeout period.

Post reply on HN