Can a bot settle the luck/skill argument?

[TreyWilly]

There's an article here (http://www.edmontonsun.com/News/World/2007/06/11/4251177-sun.html) about how a bot programmed by the University of Alberta will play Phil Laak and Ali Eslami for 50K over 2,000 hands.

Kind of disturbing. Hope Laak and Eslami crush it. Whatever.

But the article got me thinking about what a successful bot could do for the luck/skill argument. If programmers could successfully build a computer that could play optimally and crush the best human opponents, wouldn't that prove that poker was, in fact, a game where skill dominates?

Such a development would be bittersweet, of course. I don't want such a bot to exist. But the exhibition could possibly lead the way for more states to accept the luck/skill argument, and thus create more B&M options.

Just a thought ...

[DrewOnTilt]

I don't think that we need to bring bots into the legal debate at this point, for obvious reasons. The whole issue of "luck vs skill" is irrelevant to the fundamentalists, anyway.

[nc001]

For bad players, poker is LUCK. They are gambling.

For good players, poker is skill. Predatory in nature and based on one's knowledge of stats and human behavior.

Evangalists most likely dislike that bad players are gambling and being victimized.

[J_B]

[ QUOTE ]

Evangalists most likely dislike that bad players are gambling and being victimized.

[/ QUOTE ]

If poker players are spending time/money on poker that is less $$$ for them.

[Dire]

The top roshambo bots would beat any humans who tried to 'play against' them. Doesn't mean anybody's going to be convinced paper rock scissors is a game of skill all the sudden.

[TreyWilly]

[ QUOTE ]
The top roshambo bots would beat any humans who tried to 'play against' them. Doesn't mean anybody's going to be convinced paper rock scissors is a game of skill all the sudden.

[/ QUOTE ]

Why not?

To put it another away, if a bot can crush poker over the long run, that pretty much proves poker IS NOT a game of luck, doesn't it?

I guess this holds true for PRS as well, but so what? If one was sharp enough to spot his opponents' tendencies, and those opponents continued to play him for money, it is plausible one could become a pro at roshambo. It's silly, but that doesn't mean it isn't valid.

Bots can't exploit luck.

[hotbacon]

[ QUOTE ]


Bots can't exploit luck.

[/ QUOTE ]

But humans can?

[TreyWilly]

[ QUOTE ]
[ QUOTE ]


Bots can't exploit luck.

[/ QUOTE ]

But humans can?

[/ QUOTE ]

Damn.

[TreyWilly]

OK. What I'm suggesting (still hashing it out) is that the possibility of programming a computer to beat a game proves that game can be beat.

Are humans who beat the game also proof? Yes. But a computer could prove to be a better model for showing this. You can't program a computer to beat craps or roulette. If the deck is shuffled after each hand, you can't program one to beat blackjack.

You can program a computer to beat chess, poker and -- yes -- roshambo. Comparing data of games a computer can beat to data of games a computer can't beat could help convince some that poker is a skill game.

Put simply: If a computer can beat it, tangible skill is a factor. If a computer can't (such is the case with lotteries) then it must be subject mostly to chance.

Forgive me if I'm repeating myself here, but I'm on little sleep and this idea has arrived and left about a dozen times.

[govman6767]

Ask Mr Gatorade he "thinks" he is the bot master

[theflyingcow]

[ QUOTE ]
You can program a computer to beat chess, poker and -- yes -- roshambo.

[/ QUOTE ]

A little off topic, but seriously? Roshambo? I'm a little skeptical because this implies there's an optimal way to play roshambo. I'm assuming if this is actually the case that there should be a reasoning based on game theory to prove it, but what assumptions must one make?

[TreyWilly]

[ QUOTE ]
[ QUOTE ]
You can program a computer to beat chess, poker and -- yes -- roshambo.

[/ QUOTE ]

A little off topic, but seriously? Roshambo? I'm a little skeptical because this implies there's an optimal way to play roshambo. I'm assuming if this is actually the case that there should be a reasoning based on game theory to prove it, but what assumptions must one make?

[/ QUOTE ]

I'm guessing the skill element is derived from tendencies, patterns of execution, etc.

[govman6767]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
You can program a computer to beat chess, poker and -- yes -- roshambo.

[/ QUOTE ]

A little off topic, but seriously? Roshambo? I'm a little skeptical because this implies there's an optimal way to play roshambo. I'm assuming if this is actually the case that there should be a reasoning based on game theory to prove it, but what assumptions must one make?

[/ QUOTE ]

I'm guessing the skill element is derived from tendencies, patterns of execution, etc.

[/ QUOTE ]

They proved in the mathmatics of poker that the optimum strategy for Roshambo is being totally random in your picking (Dice or something)

In this game you can only exploit your opponents tendacys. But by playing a bot who is being totally random with his picks it would be impossible to beat him without LUCK.

In turn while being random with his picks he can be aquiring data from your picks and get an idea of what your tendacys are giving the computer the ability to adapt to your play to exploit it.

[w_alloy]

The fact that they are only playing for 2k hands and they are acting like this means somethings shows how little poker knowledge/experience this team has.

Obviously they have some practical concerns about the humans not wanting to play for that long, but short handed limit has arguably the highest variance of any holdem game.

2k hands is pretty much a coinflip between any expert players.

[TreyWilly]

[ QUOTE ]
The fact that they are only playing for 2k hands and they are acting like this means somethings shows how little poker knowledge/experience this team has.

Obviously they have some practical concerns about the humans not wanting to play for that long, but short handed limit has arguably the highest variance of any holdem game.

2k hands is pretty much a coinflip between any expert players.

[/ QUOTE ]

I'm speaking more to the principle, not the example.

[Quanah Parker]

I'm getting pretty sick of the "skill vs. luck" angle as a rationale for why online poker should be legal.
It reminds me of the stoners touting all the stupid benefits of hemp, when bottomline, it's about getting high.

It's all about the right of a person to do what they want to do, as long as they're not hurting anybody.
IMO, the issue is one of personal freedom. I believe I have the right to spend my money how I see fit. If I want to gamble on the lottery or play poker, it's my business. If I want to put $1k on long odds for a crappy payout, it's my business.

Smoke weed, or play online poker . It's all the same to me. Embrace the argument head on, and have the balls to stand up for what's right. Getting all caught up in "luck vs. skill" is just playing by the rules that some conservative politician crafted to protect us from ourselves. I've already got one mother, I don't need the government acting as my 2nd mommy.

[Skallagrim]

As a matter of philosophy, I agree with Quannah.

As a matter of practicality, he is just plain wrong. If we dont fight this battle on the skill v. luck level, we will have just as much chance of seeing legal poker in my lifetime as we do in seeing legal marijuana. The majority of americans are not libertarians, most think "there oughta be a law" any time they see something they dont like or fear.

With poker, the best way to end the fear is to separate poker from the types of gambling that really do bother people (slots and other house-edge games).

In other words, precisely because online poker is a skill game like chess or golf, it is less of a danger than online slots.

Do I wish we lived in a country that valued personal liberty and freedom above all else? Yes. Do we? No.

Skallagrim

[Quanah Parker]

Durn it, Skallagrim, you're probably right.
I hate to admit it, but I'd be a starry eyed fool not concede your well written points.

Personally, I don't have any interest in blackjack or sports betting, so it shouldn't matter to me. I guess I just needed to vent, everyone seems to think the "skill vs. luck" issue has merit, and it does.
Maybe we don't need to push one angle on why online poker should be legal. We've got multiple legitimate reasons that it would be beneficial to legalize. We need to state our case to the public and the politicians, and keep the debate running. The issues of taxing the proceeds of online gaming, Biblical support of gambling, personal freedom, and the fact that poker is a game of skill should all be highlighted.

[meleader2]

[ QUOTE ]

Do I wish we lived in a country that valued personal liberty and freedom above all else? Yes. Do we? No.

Skallagrim

[/ QUOTE ]

do any exist?

[Artsemis]

[ QUOTE ]
[ QUOTE ]

Do I wish we lived in a country that valued personal liberty and freedom above all else? Yes. Do we? No.

Skallagrim

[/ QUOTE ]

do any exist?

[/ QUOTE ]

No, that's why we're trying to push closer to it.

[Skallagrim]

[ QUOTE ]
[ QUOTE ]

Do I wish we lived in a country that valued personal liberty and freedom above all else? Yes. Do we? No.

Skallagrim

[/ QUOTE ]

do any exist?

[/ QUOTE ]

In my personal experience, the Netherlands comes closest (at least Amsterdam). But even they have their nanny-staters (just a lot less moralists).

But we will keep fighting, maybe we get there in one of you younger folks' lifetime /images/graemlins/wink.gif.

Skallagrim

PS, I also agree that skill v. luck is only ONE argument on our side and that ALL arguments with merit need to be pushed.

[MiltonFriedman]

"Evangalists most likely dislike that bad players are gambling and being victimized. '

No, they mostly dislike that millions of people are enjoying themselves in their Gor-given, unalienable right to the "pursuit of Happiness".

These Un-American religous fanatics defy the Declaration of Independence which deemed such rights to be dviniely granted:

"We hold these truths to be self-evident, that all men are created equal, that they are endowed by their Creator with certain unalienable Rights, that among these are Life, Liberty and the pursuit of Happiness."

[Dire]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
You can program a computer to beat chess, poker and -- yes -- roshambo.

[/ QUOTE ]

A little off topic, but seriously? Roshambo? I'm a little skeptical because this implies there's an optimal way to play roshambo. I'm assuming if this is actually the case that there should be a reasoning based on game theory to prove it, but what assumptions must one make?

[/ QUOTE ]

I'm guessing the skill element is derived from tendencies, patterns of execution, etc.

[/ QUOTE ]

They proved in the mathmatics of poker that the optimum strategy for Roshambo is being totally random in your picking (Dice or something)

In this game you can only exploit your opponents tendacys. But by playing a bot who is being totally random with his picks it would be impossible to beat him without LUCK.

In turn while being random with his picks he can be aquiring data from your picks and get an idea of what your tendacys are giving the computer the ability to adapt to your play to exploit it.

[/ QUOTE ]

Very off topic here, but randomness is not the 'optimum' strategy for roshambo in reality. Randomness is nonexploitable, but similarly incapable of exploitation. In reality, anybody who is 'playing' the game is going to be playing exploitably and so long as you are the 'superior' player, randomness is going to be far from optimal. And even if you do run into an opponent playing 100% randomly then your expectation is the same no matter what strategy you use.

So any exploitative strategy is obviously exploitatable making randomness the mathematically 'correct' play, but in reality - it's not so easy to figure out how to exploit exploitative play.

So no, the top bots in roshambo do not play random except as a fail safe when it seems that they are being succesfully exploited.

[TheEngineer]

[ QUOTE ]
So no, the top bots in roshambo do not play random except as a fail safe when it seems that they are being succesfully exploited.

[/ QUOTE ]

http://chappie.stanford.edu/cgi-bin/roshambot

[coxquinn]

[ QUOTE ]
The top roshambo bots would beat any humans who tried to 'play against' them. Doesn't mean anybody's going to be convinced paper rock scissors is a game of skill all the sudden.

[/ QUOTE ]

it is a game of skill

[coxquinn]

[ QUOTE ]
[ QUOTE ]
So no, the top bots in roshambo do not play random except as a fail safe when it seems that they are being succesfully exploited.

[/ QUOTE ]

http://chappie.stanford.edu/cgi-bin/roshambot

[/ QUOTE ]

woops missed this--- ironically exactly how a poker bot would be programmed, when it thought it was getting outplayed it would default to always making all of your choices equal EV

[shdw01]

I think a stronger argument would be to design a bot that plays absolutely horrible. A game based only on luck, bad play is not penalized. Design a bot that will fold AA, go all-in while playing the board, etc. That will prove how skill does play a factor

[Dire]

[ QUOTE ]
I think a stronger argument would be to design a bot that plays absolutely horrible. A game based only on luck, bad play is not penalized. Design a bot that will fold AA, go all-in while playing the board, etc. That will prove how skill does play a factor

[/ QUOTE ]

The argument isn't about whether poker is exclusively luck based or not, it's about whether luck is the 'predominate' (however you want to define that) factor in the game. Like with roshambo, I think it's obvious that there is some skill in the game - but it's still not a game of 'skill' per say, since luck is the predominate factor.

[TreyWilly]

[ QUOTE ]
I think a stronger argument would be to design a bot that plays absolutely horrible. A game based only on luck, bad play is not penalized. Design a bot that will fold AA, go all-in while playing the board, etc. That will prove how skill does play a factor

[/ QUOTE ]

I think the bot should be programmed to do both. The designers could have it play 200k hands trying to win and 200k hands playing like a dolt. Then the results of each could be compared against the distribution of cards -- which should be somewhat the same after this sample -- and show that skill is THE PREDOMINATE factor.

You could take it even further and have a whole spectrum of skill levels for the bot, to show exactly how each improvement affects the results.

Of course, there is no bot capable of this at the moment (I don't think) but I don't see how the principle could be challenged.

[ike]

[ QUOTE ]
If programmers could successfully build a computer that could play optimally and crush the best human opponents, wouldn't that prove that poker was, in fact, a game where skill dominates?


[/ QUOTE ]

There are already lots of people who play very well who have consistently crushed their opposition over large numbers of hands. Bots that can do it too don't prove anything new about the game.

The endless skill vs luck argument is so frustrating. Its obvious to anyone intelligent and open minded that poker involves both luck and skill. The proportions can be easily quantified with numbers like ROI, BB/100, stddev, etc. You can say things like "Given player X's past results we can be 95% confident that he will be up money over the next N hands."

The problem is that the "mostly skill" criterion is vague and simple-minded enough that people who want to count poker as "mostly luck" can define it such that they are right. "Any given hand, the guy who gets the best cards wins" is infantile, worthless logic and trying to argue against it is futile. None of the clever tricks to show that its mostly skill that get discussed on these boards ad nauseum will make any difference. The facts of the matter should speak for themselves, but they don't, because the people making the decisions are not especially interested in facts.