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

No limit: AI poker bot is first to beat professionals at multiplayer game

nature.com

Post reply on HN
Print view View original

321–330 of 407 posts

Re: No limit: AI poker bot is first to beat professionals at multiplayer game

#321

Post by bjoli »

diggan wrote: post #298
jacquesm wrote: post #296

Earlier quoted context omitted.

Not to belittle the man's other achievements but speaking four languages is pretty normal in Europe, except when you're from the UK.

> speaking four languages is pretty normal in Europe Clearly we have different experiences (swedish person living in spain currently) but I haven't met that many people who speak four languages and are from a european country (but have yet to been in eastern europe). That finns speak swedish is a special case though, as AFAIK, they learn swedish in school and being finn-swedish is a thing too.

I am a classical musician, and in my profession it is quite common. I speak a lowly 3 languages, but many colleagues speak 4+. It is a very international market, and if you leave your home country to study, it is not uncommon to work in yet another country before returning home.

Our solo flute speaks a whopping 6 languages well, and I suspect our harp player knows even more.

Re: No limit: AI poker bot is first to beat professionals at multiplayer game

#322

Post by Cybiote »

agent008t wrote: post #320
tialaramex wrote: post #308

Earlier quoted context omitted.

Player tagging just makes you exploitable. I play one way now, you tag me "Haha, fool bet-folds way too much" and then I change it up to exploit you, "Huh, I keep trying to fold him out with worse and he doesn't bite even though my notes say he will". The theoretically optimal play just skips that meta and meta-meta play and performs optimally anyway. Because poker involves chance the optimal play will be stochastic…

I think we need to make a distinction between two kinds/styles of play: 1. Coming up with an unexploitable strategy, then scaling it up by playing as many hands as you can, earning the slim expected value each time. 2. Picking a good table / card room / 'scene', and then trying to extract as much value from it as possible. You most often see 1 online, and 2 live, for obvious reasons. A skilled human would be a lot mo…

Note that 1) is only true if your opponent is also not making many mistakes. Which fails to be true for most humans, where the combination of randomization and calculating state appropriate ranges is very difficult. This means that weak players can still lose heavily from mistakes/poor play within a reasonable number of hands, it need not be slim.

Furthermore, you can kind of account for such players by including more random or aggressive profiles in the inference/search stage.

Re: No limit: AI poker bot is first to beat professionals at multiplayer game

#323

Post by Phillipharryt »

pbhjpbhj wrote: post #309
Phillipharryt wrote: post #285

Earlier quoted context omitted.

The same argument of money being on the line applies to the detection. Poker software is already pretty damn impressive with its tracking. The online casinos actually stand to lose more money than the bot creators could make, so the detection has a greater incentive, and is likely to triumph.

They only lose if there are less plays, surely? I assume they take a cut of all winnings, they're not putting up stakes.

Yes, I'm assuming that if bots work their way into everyday online poker that people will stop using it, so there would be less players.

Re: No limit: AI poker bot is first to beat professionals at multiplayer game

#324

Post by albedoa »

maehwasu wrote: post #310
albedoa wrote: post #132

Earlier quoted context omitted.

> when it bets with a weak hand, the opponent (another copy of itself) sometimes folds and it makes more money than if it hadn't bet. This makes no sense. If I am betting for thin value with a weak hand, then I make less money when my opponent folds. Does the bot not know whether it is bluffing or value betting?

“Value betting” and “bluffing” are human heuristics to simplify complicated situations. The bot doesn’t “know” whether it’s value betting or bluffing—it’s not a relevant question. The relevant question is whether to bet, and what amount, in order to maximize value of the particular hand it has, with reference to the board and opponent actions taken.

Right, we agree on that, but the above comment lumps all of what you describe (“betting with a weak hand”) under “bluffing” and says the bot learns that it makes more money when its opponent folds.

Re: No limit: AI poker bot is first to beat professionals at multiplayer game

#325

Post by ryanlol »

loup-vaillant wrote: post #301
jacquesm wrote: post #296

Earlier quoted context omitted.

Not to belittle the man's other achievements but speaking four languages is pretty normal in Europe, except when you're from the UK.

> speaking four languages is pretty normal in Europe Northern Europe, maybe. French people for instance tend to suck at foreign languages. We rarely go beyond 3 languages (French, English, then German or Spanish. The last two are often forgotten after school.) I suspect Spain and Italy are similar.

For example Spanish, Catalan, English and French would hardly be an unusual combination.

Re: No limit: AI poker bot is first to beat professionals at multiplayer game

#326

Post by noambrown »

estomagordo wrote: post #319
noambrown wrote: post #6

I'm one of the authors of the bot, AMA

Do you expect the human players to play at the best of their ability when they're not playing for actual real money?

There was real money at stake in this experiment. The pros were guaranteed $0.40 per hand just for participating, but that could increase to $1.60 per hand depending on how well they did.

To answer your question, no, I don't think human players would play at their best when not playing for actual money.

Re: No limit: AI poker bot is first to beat professionals at multiplayer game

#327

Post by chongli »

disgruntledphd2 wrote: post #286
chongli wrote: post #226

Earlier quoted context omitted.

No, I think your earlier example of a swarm of just-above-break-even bots would be much more difficult to combat. Even if they can be detected, the anti-detection countermeasures can evolve, turning it into an arms race. Anything you can model in your bot detection algorithm, the bot-maker can model too. Reaction times ought to be one of the easiest things to fake. All it would take is a bunch of monitoring of large…

Not normally distributed, as negative reaction times are unlikely. You could use log-normal, but I believe that a mixture of exponential and gamma tends to be used by reaction time researchers (search ExpGamma).

negative reaction times are unlikely

Oh, right. I was thinking along the lines of 100m dash, where people often do have negative reaction times (which we penalize as false starts).

In poker we don't have much of an incentive to react instantly to any play.

Re: No limit: AI poker bot is first to beat professionals at multiplayer game

#328

Post by noambrown »

cambaceres wrote: post #293
noambrown wrote: post #6

I'm one of the authors of the bot, AMA

Are all the hands available to the public?

The hand logs from the 5 humans + 1 AI experiment are included in the supplementary material of the Science paper.

Re: No limit: AI poker bot is first to beat professionals at multiplayer game

#329

Post by hdkrgr »

Traster wrote: post #297
csa wrote: post #279

Earlier quoted context omitted.

You clearly didn’t read the additional links they posted. They mentioned why they chose 10k (AIVAT), and it goes far beyond any of the variables you mentioned. For any number of hands, my money is on the bot.

That really doesn't address the point that was raised. It's not that the bot wins through luck and that 10k is too small a sample, it's that a good professional poker player isn't good over 10k hands, they're good over 5 years. Any good player will have their play analyzed and responded to, so there's a feedback loop there - any good player will have their play analyzed, exploited and will have to re-adjust their str…

This will be interesting to see.

Brown and Sandholm's algorithm aims to play a Nash Equilibrium which by deifnition _cannot_ be exploited by a single opponent player as long as all players are playing the equilibrium strategy. As they note in the paper this gives you a strong optimality guarantee in the 2-player setting. It was unclear whether this would transfer to real-world winnings in the multi-player case, and while it looks like it does for now (for current strategy-profiles of human players) humans might be able to adapt to the strategy played by the bot. Given the fact that the bot wins against current human strategy-profiles in the n-player setting, it's likely (but not a sure thing) that human players will have to team-up against the bot to exploit it. That seems rather unlikely to me.

Re: No limit: AI poker bot is first to beat professionals at multiplayer game

#330

Post by crazypyro »

gexla wrote: post #273

It's easy to "take away" too much information from this. The focus is that an AI poker bot "did this" and not get too much into other adjacent subjects. But what's the fun in that? 10,000 hands in an interesting number. If you search the poker forums, you'll see this is the number you'll see people throw out there for how many hands you need to see before you can analyze your play. You then make adjustments and see a…

>Remember that one of the most exciting events in online poker was the run of isildur1. That run was put to rest when he went bust against players who had studied thousands of his hand histories.

Perhaps more famously, Jungleman compiled hand histories from many different people while he was playing Tom Dwan in the 'durrrr' challenge (which I guess technically isn't over....)

Post reply on HN