Pokerbots Making Online Players Sad

Anonymous Coward wrote to mention a Wired article about the rise of Pokerbots in online gaming venues. From the article: "Smart, skilled players are rewarded in the long run, especially online, where there are plenty of beginners who would never have the nerve to sit down at a real table. But WinHoldEm isn't just smart, it's a machine. Set it to run on autopilot and it wins real money while you sleep. Flick on Team mode and you can collude with other humans running WinHoldEm at the table. For years, there has been chatter among online players about the coming poker bot infestation. WinHoldEm is turning those rumors into reality, and that is a serious problem for the online gambling business."

Comments

Is Zonk an editor bot?

[Anonymous Coward]

That would certainly explain a lot, especially if Taco wrote it.

Re:Is Zonk an editor bot?

[Baricom]

<?php

for ($submissions as $s) {
  if (($s->news_for_nerds && $s->stuff_that_matters && $s->dupe_count <= 1) || ($s->category == 'Google') || (rand(1,10) > 5)) {
    $s->make_inflammatory();
    $s->add_random_speling_errors();
    $s->post();
    sleep(86400);
    $s->post();
  }
}

?>

Poker Cheaters

[mfh]

My initial thought is that anyone who would run a pokerbot is evil. Then my attention turns to Las Vegas and the enormous rooms of metalic robots who are all fixed to win and win big, suck the life, time (24/7 baby), and money, out of would-be regular people. Then I don't feel as bad. I still don't like cheaters, tho. The answer? Play free online poker. Save your money for BYOB -- real games with your friends. We play Texas Holdem from time to time at the cottage and it's a hoot. Games should be fun -- not business, IMHO.

Re:Poker Cheaters

[Travoltus]

[Games should be fun -- not business, IMHO.]

Anything that involves real money is, or becomes, business.

Darwin never sleeps.

Re:Poker Cheaters

[melikamp]

Here's a thought: TFA says that the online poker biz already makes $1.4 billion annualy. Now this cash will pay for R&D of The Perfect CAPTCHA. This will be interesting to watch.

Re:Poker Cheaters

[sydb]

Darwin never sleeps.

Must be a bug, OSX sleeps; what version are you running?

Re:Poker Cheaters

[melikamp]

They work really badly anywhere except at (1) long-hand (2) fixed limit (3) low stakes (4) loose (5) passive tables where the winning strategy is clear and mostly consists of waiting for a good hand and then playing pot odds correctly. If anyone is so bored that they are willing to babysit a bot which makes less than $1 per hour, more power to them.

It is all about complete automation. Without it these bots are useless because it is simply not fun to play the long-hand fixed limit poker correctly. Take it from an avid hold'em player.

Re:Poker Cheaters

[jonadab]

> > [Games should be fun -- not business, IMHO.]
> Anything that involves real money is, or becomes, business.

This is why we only play for M&Ms. We assign values to the different colors, and use them as chips. Everybody brings a pack of M&Ms to the game. Eating your profits is explicitely allowed. Somehow, mysteriously, we always seem to run out of our M&M poker chips, and then the game's over. Funny how that works.

I've heard of playing for homemade cookies, as a form of higher-stakes game, but to me it always sounded a bit too much like a way to put on weight.

Poker

[Descalzo]

Well, I'm sorry but I don't lose any sleep over people who lose money gambling, or who feel it is unfair. It's gambling! Who do you think pays for all those lights in Vegas? The losers!

Re:Poker

[Hedonist23]

I understand where this comment is coming from, but what people don't understand is that poker is actually a much different game than other forms of gambling. It's a game in which you play against other players, not the house. As far as the house is concerned, you can win every hand and become a millionaire, as long as they get their rake (the percentage a house takes out of every pot).

Poker is a skill game, that's why people can become pro's at it. That's why even semi-pro's like me can make a decent living off of it, especially now with the boom in popularity of the game.

Re:Poker

[adam31]

You've just described real poker, where you're physically at a casino.

That misses the point entirely. There are 2 separate, distinct issues that the pokerbot addresses that are unique to online play:

1) Any player could be using this program to evaluate the current live hand in an off-line fashion. Attempting to weed them out by chatting is useless. As far as using a bot is cheating, this would be cheating (and many players "seeking to understand the game better" would deem it as excusable!)

2) Outright collusion. This can be done by two humans using the same on-line poker forum. No bots are neccessary. That bots also do it is irrelevant. The reason the bots can collude is because the program author thought that people need to be aware of the issue!

Re:Poker

[jonadab]

> the other 95% is comprised of statisical knowledge, human psychology,
> knowing *how* to bet, and the 5% is just a touch of luck/karma/whatever

That's actually a pretty good description of gambling.

find a flaw

[nihaopaul]

all you have to do is find a flaw in the poker bot and then exploit it, they always have one!

Re:find a flaw

[ikkonoishi]

Generally a torpedo down the exhaust port works.

If that fails try throwing an old man shooting lightning from his fingers into a conveniently placed pit to the energy core.

Re:find a flaw

[Flyboy Connor]

all you have to do is find a flaw in the poker bot and then exploit it, they always have one!

This is a pretty apt comment. I think professional poker players would love to play against a bot. It gives them a considerable advantage, because if they studied the bot they can predict what it will do.

There are, basically, two possibilities. Either the bot plays purely statistically. If that is the case, it may win against dumb players, but can break even at best against good players. Or, the bot tries to model its opponents and tries to take those models into account when playing. If that is the case, as soon as a good player recognises that a bot is playing, he can ensure that the bot will have the wrong model of him, and then exploit that.

And, of course, as the parent says, it is possible that the bot contains an exploitable flaw. The bot creator goes to sleep, someone on the net recognises the flaw and posts about it in a newsgroup, and by the time the bot creator awakes he is broke. I would not sleep soundly with a bot playing with my money.

Re:find a flaw

[randyest]

True, but that's not how the bots work. No offense -- everyone posting here seems confused. Google winholdem and you can learn that it simply plays ideal poker. No player profiles, no fanciness -- just simple statistically ideal poker.

That's enough to win more than lose against average to good players (even greart players, over the long run) but pair a few winholdem bots in the same game cooperating and you can rake it in.

Re:find a flaw

[randyest]

Of course they play statistically and, surprise, computers are better at playing accurately based on statistics than even the best human players. WinHoldEm doesn't try to profile or model players. It just plays perfect poker (statistically.) And against most players, that is a sure win over time. Even against great players, it doesn't lose over time (think Las Vegas house.)

The point you're missing is that several accounts, all playing WinHoldEm bots which are communicating with one another can rape even the best players over time. It's cheating at poker, and the gambling sites can't seem to control it yet.

Re:find a flaw

[Jah-Wren Ryel]

And, of course, as the parent says, it is possible that the bot contains an exploitable flaw. The bot creator goes to sleep, someone on the net recognises the flaw and posts about it in a newsgroup, and by the time the bot creator awakes he is broke. I would not sleep soundly with a bot playing with my money.

I doubt that would happen. It's the online gambling equivalent of posting a misprice to fatwallet.com. Except that online merchants are big slow and stupid and most still haven't figured out how detect hordes of people taking advantage of a misprice. Most still aren't smart enough to page a human when there is an abnormal spike in the sales numbers for an item.

For a poker-bot, it is simple to prevent large scale exploitation of a flaw - give the bot a sanity check. If it loses more than $X, then it stops playing and pages a real human to check things out. There will probably be false positives due to the nature of gaussian distributions but experience ought to indicate what a good enough value for $X is to minimize those false positives and still make automated playing profitable.

Re:find a flaw

[Illserve]

Even against great players, it doesn't lose over time (think Las Vegas house.)

The "house" doesn't make money from statistics. It makes money from the rake, a small percentage of each pot which go to the establishment. Just like party poker.

WinHoldEm bots which are communicating with one another can rape even the best players over time.

Collusion of this sort doesn't give you a very huge advantage. You have a bit more information about the statistics of card distribution by knowing the other players' hole cards, but it's not a terribly big deal.

Re:find a flaw

[Richthofen80]

all playing WinHoldEm bots which are communicating with one another

This is the important thing, collaboration. In all scenarios, casinos, both real and online, factor in the odds or frequency of the player winning. For every percentage over 49% in favor of the player, the casino adjusts accordingly. It just doesn't make any sense for the casino to win less than 51% of the time. int he case of these poker sites, they take a certain percentage of the 'take' in any hand. In blackjack, the odds are in favor of the dealer about 51% of the time. Casinos have unlimited money and can continue to play, knowing in the long run statistically they'll win.

What scares Casinos is collusion. To any one player in blackjack, he has a 49% chance. However, multiple players sharing information changes those odds, in favor of the group over the casino . (this only applies to 'house' games, like blackjack) If you read Ben Mezrich's 'Bringing down the house', a group of students at MIT figured this out. They were able to play statistically and when they found a table whose odds leaned into the players, they called in a big fish who would bet more, knowing that the odds had swung.

The same collusion applies to Poker, except against other players, not the house. If I am dealt two Aces, and I collude with another player who indicates that he got one ace, I can tell two things... One, that no one else can match my aces, since there's a single ace somewhere else, and second, the other player can drop out, minimizing the loss of the teams.

The great thing about card games is that there's a finite number of cards dealt, and therefore statistical rules apply... the chances of drawing an Ace from a deck of cards increase for every non-Ace you draw. Since robots can keep track of every card dealt, they have an excellent chance to quickly calculate poker and blackjack situations. Collusion allows even more input to be gathered and for computers to make even more informed decisions

Re:find a flaw

[tsotha]

It just plays perfect poker (statistically.) And against most players, that is a sure win over time. Even against great players, it doesn't lose over time (think Las Vegas house.)

This is clearly wrong. There isn't any such thing as "perfect poker (statistically.)" The best players tailor their game to other players around the table, both in live games and on sites like PartyPoker.com.

Statistical poker isn't that hard to play. Most books on the subject include handy tables to figure out how likely you are to make your hand based on the number of "outs" you have. Throw in a little hand groupings for preflop play and you're all set to play "statistically". You should be ready to have your ass handed to you, though, since every professional poker player in the world can beat a "statistical" player.

From a professional's point of view, the poker bots will ruin the game, though. Not because the pros can't beat the bots. The problem is the bots will drive out the truly bad players, which is where most of the money comes from. A pro playing bots all the time will make a small profit, but it's a lot of work for the amount of effort. I predict if this bot is good enough it will simply drive out the fish, and the sharks will move back into the local card clubs where they were making a comfortable living two or three years ago.

Re:find a flaw

[SlowEmotionReplay]

"You see, pokerbots have a preset win limit. Knowing their weakness, I sent wave after wave of my own dollars at them, until they reached their limit and shut down." Zapp Brannigan

Re:find a flaw

[Psyonic]

I don't mean to be rude... but clearly you don't understand what a royal flush is in poker. A Royal flush is the highest of the straight flushes, meaning it is a straight and a flush. Or in other words, the cards rankings are in order, and are all of the same suit. In your example, all 4 players would have a straight, but only one of them could have a flush, and therefore a royal flush.

Re:find a flaw

[Znork]

Yes, but the guy with the five aces wins anyway.

Why not do what casinos do?

[Limburgher]

If you're caught cheating, you're escorted out.

Sometimes, to a shallow ditch in the desert.

Hard to implement online, actually. Nevermind. :)

Re:Why not do what casinos do?

[slavemowgli]

Not *that* hard, though. Yes, it's possible to remain quite anonymous online, but if you're playing for real money, then you will have to identify yourself in some way so that the money can be sent to you - and I don't think you'd be able to get away with creating a hundred PayPal accounts.

The only ones who'll be able to get around that are the organised crime people, and I wouldn't be surprised if these really turn out to be a hard problem in the end. But casual players that just want to make money fast... not really, or at least not in the long run.

Re:Why not do what casinos do?

[Hott of the World]

Caught Cheating? Hell, sometimes, if you're caught WINNING, you're escorted out. Or worse.

Story, Here.

The end game....

[pieterh]

No more unassisted human players, but networks of bots competing against each other, ultimately controlled by individuals, and creating a larger and more interesting game... Bots are just another tool, after all.

Where's the problem?

[lightspawn]

If they're allowed to play, there's no problem. Humans should deal (heh), or retreat to humans-only venues.

If they're not allowed to play (why not?), but still do, there are two problems. The social one of people running them (I'm assuming the bots don't decide to play by themselves) which probably can't be solved - some people are inherently dishonest. Then there's the technical problem - how do you let humans play while shutting out bots? There really isn't a feasible solution, especially if humans decide to play physically but let a bot decide their moves for them. But of course some will still try to implement a partial solution. Discuss.

Re:Where's the problem?

[drdink]

Or you could put a captcha at the start of every few hands. Not every hand, since that'd be annoying. Instead, just do it every few random hands, such as every fifth or sixth. It won't solve it, but it'll cut down on the problem.

Re:Where's the problem?

[ae]

Why wouldn't that hypothetical Indian just run the bot himself?

Re:Where's the problem?

[izomiac]

How about incorporating captcha into showing the cards? That way a bot couldn't read the cards but a human (theoretically) could. Such a measure would cut down on bots, but it still wouldn't stop someone from physically playing, but still consulting a program to see what they should do. But, I doubt there's a way to stop people from doing that.

Re:Where's the problem?

[Mal-2]

Oh god no... don't go putting up barriers or the players will stay away in droves. The fact is, folding 4 out of 5 hands you're dealt (or more) leads directly to playing multiple tables so as to actually have something to DO once in a while. If you interrupt that flow with captchas all the time, it will make it incredibly annoying to play those multiple tables and people won't want to play at all. I know some people are content to chug away at one table, but spreading out is essential to increasing profits while keeping a rein on variance. You will take much smaller swings from bad luck playing four .10/.25 tables at once than you will from playing a single .50/$1 table.

If you start modifying the cards in a captcha-like way, that too will be extremely annoying for people playing at multiple tables. You expect the ace of spades to look the same no matter where or when it appears. Same with any other card. If you just start juggling filenames, that won't affect the human players at all... but it won't affect screen-scraping bots either, only ones that depend on constant filenames. Those that do would quickly be replaced by those that can "read" directly from the data in video RAM. And I really can't see a way to stop people from running the bot on a second computer which controls the first over a VNC connection.

Even chat queries are spotty. Playing four tables at once (as I often do) doesn't leave a whole lot of time for witty banter. I keep the dealer in Verbose mode where every action of every player is announced, allowing me to look over the history of that hand quickly when it's my turn to act at a given table. Unfortunately, this has the side effect of scrolling chat text off the screen very quickly, and I've pretty much given up on trying to be sociable while playing. If you want to talk to me, that's what IRC is for. Then I don't have to worry about tables stealing focus while I'm trying to type.

For those of you that don't play (or don't play multiple tables), any table will steal focus when it is your turn to act. It doesn't matter if you are typing into another table's chat box, you are forcibly taken to the one where it is your turn. This often leads to half-formed thoughts being uttered at the wrong table and the first half never being said at all. I find this focus jumping to be worse than any other aspect of trying to chat with other players. The tables will steal focus from ANY app, not just other poker tables, but somehow they don't steal the input devices. If you keep typing, your text will still go where you intended (provided it's not another table). This means I'll happily blather away in IRC while playing poker, but won't talk to the other players much if at all.

Mal-2

From Someone Who Makes His LIving Playing

[Hedonist23]

I make my living playing poker. I used to play mainly online, and now split my time fairly evenly between brick and mortar (B&M) play and the online realm.

I can tell you that the bots are not a big deal yet. First of all, I'll be amazed when they ever come up with the technology to play no limit hold 'em. That would be a miracle program. Poker is much more than just betting and raising, and the occasional bluff. Just as important are reading your opponent, making bets that damage others pot odds, and playing your position in relation to the blinds. Plus, there's just a certain amount of feel needed in the game. Even Doyle Brunson claims ESP is important in Super System.

Limit ring games are a different ballgame, and a bot does have some chance of success. However, that chance is at best only at the low level games, where a program could actually outplay the players. Any mid to high stakes game has players who will quickly figure out the way a bot plays, and milk it for all it's worth.

As it is now, winholdem is a pretty bad program. I don't know of anyone who has made a profit with it, and I do know a couple of people who have at least taken a look at it. If you're worried about something in online poker, be much more worried about collusion, with multiple people at the same table sharing their hands with each other. But, even that doesn't give a huge advantage against a good player, unless there are upwards of six or seven people in a room sharing information against the rest. Poker is, and always will be a skill game, and none of these cheating methods can change that.

hed.

Re:From Someone Who Makes His LIving Playing

[Hedonist23]

There's a lot more involved, although that information is of course valuable. You would also have to program it to understand that good players vary their betting based on their position related to the blinds, as well as if the game is currently aggressive/passive, etc. Knowing every possible hand is really only a first step.

A friend of mine was actually working on the same thing, for an artificial intelligence class. The program failed miserably, but he did get an A, so he at least got a start :)

hed.

Re:From Someone Who Makes His LIving Playing

[demachina]

Maybe thats because its impossible to catch the people running the servers cheating unless they are completely clueless or you have a regulatory body going in and scrutinizing their software and operations which I doubt there is especially for off shore sites.

The house software controls the deal. They can write software where they press a button and the program finds the undealt cards they need for the house player to win a pot they want, or assuming its well written software tell them it can't be done this hand without looking like its cheating. If you have your pick of the undealt cards you can arrange to win most hands.

It boggles the mind anyone would think they could catch a minimally well written piece of software cheating for the people who control the server and all its software.

Re:From Someone Who Makes His LIving Playing

[demachina]

First the caveat I wouldn't touch internet poker with a ten foot pole so I know not of what I speak, but I find it hard to believe that anyone would let anyone save every card dealt, so I assume all you can save is what you saw when you played right? If you could see all the cards you could analyze the secret behavior of players which would be bad. If you are only seeing the cards that were shown in the game I would think your ability to do statistical analysis would be weak.

I can really see no possible way you could tell if the deal was altered as long as they decide to cheat selectively and with some randomness. I can see you not wanting to stack the deck at the beginning because you might be able to detect that but every so often hitting a button that says house wants to win this hand, and letting an algorithm pick any random undealt card that would make that so would be nearly impossible to detect.

The challenging AI algorithm in this business is figuring out the optimal strategy for setting the hook (i.e. you want the sucke...player to win a lot early so they get hooked, then slowly turn them in to losers, and mix in an occasional win to keep them from giving up as you slowly break them.

I would agree there probably isn't much motivation for big reputable sites to cheat assuming they are making a killing already. What mix do they use to make their money and how do you know how much they make and how. As for a third party audit of the software, that is useless unless they have the power to secretly watch the server from inside, constantly checksum the binaries, watch everything on every server etc. Sounds like a tactic to make the sucke...players feel good about it without actually insuring its up and up.

I just can't see anything that would stop a site that was in a financial crunch or if owners decide they want to make a killing and cash out from cheating.

Poker is one of the few worth while gambling games, if you are good at it, because you are playing against other people instead of a house with the odds stacked in its favor. It would seem to me if you do it online you give up that assurance and are back to a game where the house can set the percentage they take.

Re:From Someone Who Makes His LIving Playing

[TexVex]

I would suggest you read some of David Sklansky's books on poker. He is big on his Fundamental Theorem of Poker, and the utilization of Game Theory.

The FTOP states that you profit every time you play a hand exactly as you would if you could see all the cards, and you profit every time an opponent makes the wrong play assuming he could see all the cards. Making a "mistake" in this context means giving your opponent favorable odds to chase a draw, calling when you don't have favorable odds, failing to value bet a winning hand, calling with a losing hand, etc.. Sklansky uses Game Theory to propose ways in which you bet, bluff, call, and fold with the correct frequency to give your opponent the most opportunities to make mistakes and make as few mistakes as possible yourself.

Actually applying what Sklansky writes takes a lot of knowledge of the game. You have to be able to recognize betting patterns, calculate pot odds on the fly, accurately estimate your implied odds, put your opponents on ranges of hands, and many other things. All in real-time.

Some of those are things computers are good at. Many of them are not. :)

Re:From Someone Who Makes His LIving Playing

[ramblin billy]

I'm amazed (but probably shouldn't be) that so many posts say so much shit without any evidence of research...especially considering the amount of Google worship on /. As it turns out, pokerbots are much more complex and varied than any of these posts consider. You are right to consider the bots more as AIs than simple if/then engines. Take a look at the WinHoldEm website. You'll notice there are different models available. There is also information regarding casinos detection schemes and how to defeat them. The forum lists topics like Teams, hacks to the main bot code and play formulas. Bots are designed to be customized for the individual owners taste. It's much more evolved than most folks imagine.

The real value of teaming is not in the knowledge of the cards in play, it is in the freedom to chase a bad hand when a partner's hand can cover. It's easier to hang around trying to fill an inside straight when your partner has a pair of aces.

Look at the bright side - it might finally be possible to make some REAL money by defeating the Turing Test. You could code a bot who explained playing 20 hours a day by pretending to be a crankhead.

billy - how about a nice game of chess?

Automated

[Klar]

I'd think that the same sort of approach could be taken as has been done in the past with macro's in mmorpg's. Track behaviour, and if there is suspicion, have an admin personal message the player, asking them a question a bot wouldn't have difficulty answering. Also,

I know a few people who play high stakes online games(2k+ buyin tables), are people trusting the bots at high stakes?

Re:Automated

[painkillr]

the problem isn't pokerbots running 24/7, it's ppl using pokerbots on their PC while they're playing. anyone stupid enough to run a pokerbot 24/7 will get caught, it's the people who sit at their pc and solely handle the clicking after consulting the pokerbot that can get away with it.

Re:Automated

[XMyth]

There isn't a whole lot a poker bot can do for you that most skilled players can't do in their head anyways. The ONLY advantage to a bot is the automation. Being able to consult one while you're playing isn't an edge against a good player.

All it is is a crutch weak players can use against other weak players.

Also, if you're consulting a 'bot' for all of your play then your play isn't going to change and you'll get swallowed whole by the first decent player you sit at a table long enough with.

And if you can change up your play while using the bot/app then you're probably going to see that it's not that hard to figure up the pot odds and other things a bot would do for you on your own.

Re:Automated

[phriedom]

A poker bot can play 10 tables at the same time and keep track of every statistic about all 90 other players at the tables.

Of course, playing ten tables at a time is a good way to get yourself noticed, but you could probably get away with 5 or 6 tables at a time. My brother-in-law plays 5 tables live, without a bot. He does, however, use Pokertracker, which helps him keep statistics on everyone he plays with, which in my opinion isn't cheating, it is just automating something that you could do manually. Having seen his statistics for the average 3/6 player on PartyPoker, I have no doubt that a bot could make money there. Maybe it wouldn't make 2.2 big-bets per hour the way the best human limit players do, but I have no doubt that 1 BB/hr would be easy. Play 5 bots at 5/10 and that is $50/hour. Run it 4-6 hours a day to avoid getting noticed by the admins and it wouldn't make you a millionaire, but it would be a nice chunk of change.

Re:Automated

[ArcSecond]

He does, however, use Pokertracker, which helps him keep statistics on everyone he plays with, which in my opinion isn't cheating, it is just automating something that you could do manually.

Would you include counting cards in Blackjack in that analysis of what is and isn't cheating? I believe counting cards in your head at a real casino is okay. But just try "automating" it and then see what happens.

Re:Automated

[AvitarX]

Try doing it in your effectivly and see what happens.

Re:Automated

[CastrTroy]

You are technically allowed to count cards. They are also technically allowed to kick you out of their casino, and technically allowed to blacklist you from all the casinos in the state. They don't like it when you win. There's no laws against cheating, but then again, there's no laws against kicking you out of the casino either.

don't be afraid of poker bots

[Anonymous Coward]

I am not aftraid of poker bots. I play in a casino 20 hours/week and online 15 hours a week.

A good player adjusts to his players in a very human way.

Artificial intelligence of a high variety would be need to emulate this adaptive behaviour in a robot/software program.

There are two benefits that I see from poker bots:

1. They will provide self-funded development in artificial intelligence, just like the stock market provided advancement in certain aspects of physics, statistics, and probability.

2. Good poker players will detect bots and 'trap' them to make money from them.

Implied odds (http://rds.yahoo.com/S=2766679/K=implied+odds/v=2 /SID=e/TID=F588_121/l=WS1/R=1/IPC=us/SHE=0/H=3/;_y lt=AuwgGElWrfwm0QKAD7BMCEdXNyoA/SIG=12dj0jprq/EXP= 1125273382/*-http%3A//www.cs.ualberta.ca/~jonathan /Grad/papp/node22.html) provide disproportionate benefit to folks that explot low probability/high payoff situations.

The poker bots do provide more information to pople, and they give them an edge, in that respect they are simply a tool.

The poker bots provide a better mathematical approach to poker, which means statistically, they will beat bad players.

Regardless of how it happens, a fool and his money are soon parted.

Should be obvious

[keraneuology]

Who is going to set up the first bot only online poker site? Let people compete by setting their bots against other bots.

Of course, won't be long until really good poker players start cheating by pretending to be bots...

on-line poker is for marks

[frovingslosh]

Come on, on-line poker is for chumps.

Do you really believe that the operators of these on-line "casinos" are above playing poker against you while they can watch your hands, or when they can tell the computer what to deal next? And while dealing themselves the good cards too often might be caught by statistical analysis of the decks (if you can afford to loose enough to gather maeningful data), their watching and knowing your hands would only look like skillful play on their part.

Another form of cheating that I know is going on (because I know someone who admits to doing it) is to play multiple hands in the same game against another player and share information about your hands. This is a great way to part the fools from their money, since having lots more information about the deck than non-cheating players geatly improves your odds. You know, for example, if the chance of drawing that fourth king is very high because it hasn't been dealt to the other hands you know about, or zero because it has. And when one of the positions you control has a particularly good hand you can drive up the pot by having the other hands you control place small raises when they would otherwise drop.

If you like on-line poker, let me introduce you to three card monty. Some people confuse it with a game of chance too, but it's just a very expensive private magic show.

Re:on-line poker is for marks

[XMyth]

Why would they waste their time cheating?? The rake alone is massive profits.

And besides, if they EVER get caught cheating (former employee rats them out or something) then people will simply stop playing there and now they've lost what was their big money maker...the rake.

A B&M casino could cheat you just as well.

Re:on-line poker is for marks

[mosch]

Come on, on-line poker is for chumps.

statistically speaking... about 90% of them are chumps, yes. The other 10% win.

Do you really believe that the operators of these on-line "casinos" are above playing poker against you while they can watch your hands, or when they can tell the computer what to deal next?

Yes. Personally I have about a million hands logged in a postgres database. Any statistical analysis I've ever done regarding: 'how often should this scenario happen, versus how often DID it happen' has shown that it was on the level.

I personally generate thousands of dollars in rake each month, by playing winning poker. They could rip me off one-time for a few grand... or have all that money, every month, for as long as they exist and spread a fair game. There's no reason to kill the golden egg.

Another form of cheating that I know is going on (because I know someone who admits to doing it) is to play multiple hands in the same game against another player and share information about your hands.

Unfortunately, colluders exist. Fortunately, they're relatively easy to detect, and the information isn't "extremely valuable", it's "usually worthless" and generally "statistically insignificant". Sometimes these douchebags try to trap guys for extra bets, or run squeeze plays on them... though these are extremely exploitable strategies that will only work against the worst players. (and are easily detected by a review of the hands by the casino, should a player file a complaint)

In conclusion: stop talkin about things you don't understand, kiddo. I have no doubt that you lose at online poker, but the problem isn't that you're getting cheated.

Re:on-line poker is for marks

[mosch]

Bullshit. Nope. 1 Million complete hands from one single site and version would take aprox 9 years to gather at 24 hours per day of play. What sites even give you the entire contents of each hand including folds. Um none? Yes, most of the hands are not played to completion, but that doesn't make them statistically worthless. a million hands is relatively easy to gather. play 4 short-handed tables at a time, you'll be logging about 300 hands per hour. Do that a few hours per day for a few years. Now find a friend who has been playing every day and swap histories with him. There's your million hands. If I got aggressive with hand history trading, I could probably put together a ten million hand db using *only* people I've actually met in real-life, not just purely online poker geeks. If you want to be cheezy you can also "data-mine" where you just observe hands. It's possible to do this at 16-tables at a time, and gather between 800 and 1200 hands per hour, depending on what type of game you're watching. I didn't get any of mine through this technique. Fukin idiot. Yes, you are.

Simple way to judge

[zymano]

Most online poker sites have a CHAT going.

Just ask them a question. Could be an idea for the poker software programmers.

Just send me the $$$ and you don't have to pay any copyright fees.

Cheating?

[Mathonwy]

I keep seeing comments talking about "those durned cheat'n bots". But it makes me wonder, what exactly is "cheating" in this case?

In most cases, I'd say that "cheating" is doing anything that is against the rules of the game that gives you an unfair advantage. But what here is the bot doing?

As far as I can tell, none of the things this bot does are things that acutal human players couldn't do, if they wanted to bother. So then at that point is it still cheating?

The one exception to this is the collusion. That's clearly against the rules of poker. But I predict that that will be a self-correcting problem. Since after all, it won't be long before someone makes an alternate version of the BOT that feeds incorrect data to the other BOTs so that you're more likely to win money from them. (Since game-theory wise, if you're the only cheater in a room full of honest people, you have an advantage.) And shortly after that, other bots will start to do the same, until the "collusion" a bot gets cannot be trusted, and is no longer a worthwhile channel of information.

(Heck, a whole war of bots trying to "Read" other bots based on their (possibly erronious) collusion information could start. That could actually be kind of fun to watch. From a distance.)

Anyway, as long as the BOTs aren't actually hacking the system, or forcing other peoples' clients to crash, etc, then I think you could make a good argument that it's not really cheating.

(And no, I'm not a bot user or apologist. All of my online poker playing is restricted to free sites, anyway...)

Re:Cheating?

[Sparr0]

You seem to misunderstand the process. Bots are not "sharing" information in the sense you seem to think. This is not a case of 3 guys all running bots and sharing. It is a case of one guy running 3 different bots that work together. There is no chance of misinformation entering the picture, and knowing the location of 4-10 extra cards in play gives a SERIOUS advantage, far more than the cost of the blinds or ante for the extra bots.

Of course there's gambling bots, spam bots, etc...

[danielDamage]

I'm no gambler myself, but I do understand that part of what actually makes gambling "fun" for people is the risk and potential reward. For many people it is a mixed professional and entertainment pursuit.

Granted, I'm not very good, and so I just have NEVER found enjoyment from pitting my wits against people in this way. It always just seems like luck whether I get a good hand or not. But for people WITH this skill, it's very enjoyable and exciting. As I understand it, the strategy of some poker variants like T#xas H#ld Em is pretty deep.

But as models of real money get pushed to the online universe (MMORPG's, pagerank, etc.) people are going to try and use them as automated moneymaking avenues. It just goes hand in hand with putting ANY kind of real value online. If people can find an algorythm to exploit the particular system, then they're going to.

REALLY good humans have advantages over bots in poker (and perhaps still in chess as well), but it's the above average casual players who are going to get raked.

So it's just time for the online gambling industry to mature a little, just like the MMORPG market, blogging, or any online universe that's had to combat bots and keep them from raping all the possible value from those systems.

Either they can find viable ways to combat bots and make play work for human players, or they will not be able to remain competitive.

Re:Who cares?

[The Evil Couch]

there's even less reason to care than you think.

The bot does nothing crazy. it knows the odds on everything. which means that it plays no better or worse than a human who knows the odds of each hand.

sure, it can tap into other bots playing so that it develops a huge advantage over other players, but so can two humans with an instant message or voice chat program.

unfair? you bet.

cheating when it's used stand-alone? hell no.

Look to the westerns for the answer.

[Helios292]

This can only be solved in one way. Impliment a networked FPS engine for knocking over the server and pulling out virtual Derringers when cheating is suspected.

It's coming:

[ImaLamer]

We don't deal to droids in here!

Bots should be easy to deal with.

[btarval]

I honestly don't see what the problem is here. This is simply a technology-vs-technology situation, and the Casinos always have the upper hand.

Just put in a challenge-response mechanism which stays one step ahead of the bots. When a bot is detected, skew the results against them, in favor of the humans. Eventually the bot accounts will go broke.

This basically ends up encouraging the bot accounts to go elsewhere, while retaining and attracting players who don't want to play against bots.

Ultimately, it ends up rewarding the Casino's which can stay on the technology edge, while forcing the other ones to be primarily attractive to bots. Those will undoubtedly end up going broke.

The only real problem that I see here is one of lazyness on the part of the Casino operators.

Ok, sorry I got another one...

[ImaLamer]

"I fold"

"Fold? I can't let you do that Dave"

Stay away from the cash tables.

[PotatoHead]

The tournament games are where it's at. After a time, you get to know the regular players. New faces (avatars?) are easy to spot and collusion becomes very difficult because the players are sprinkled about the tables.

If you are a good player and have a bankroll large enough to handle the variance, you have a very good chance at winning some nice cash. I personally found this too expensive and risky to enjoy the whole experience.

I like the idea of asking questions and including turing tests, BTW. These two things, applied to the cash tables, would go a long way toward thwarting the bot problem. Collusion will remain an issue however.

The problem with the cash tables, and to a lesser degree on the sit 'n go games, is the ability for players (bot or human) to communicate outside the game environment. I'm not sure we are going to be able to solve that. --Stay away from the cash games, unless they are very high stakes. (Even then variance is several times your buyin cost --be ready and beware!)

The wife and I play regularly --it's a lot of fun when you've got some good players online. We started out playing cash games and sit and go contests. However, variance was just huge compared to real table action. Ended up losing a fair amount, despite solid play.

After doing some analysis and research, we decided to give it another go and stick to tournament play. --Much better experience. We've got our losses back and are now profit taking while slowly building the bankroll.

Coupla things I've noted:

- the cards often appear balanced for high action. Almost every hand sees flops that are difficult for players to let go of. It's our perception that bad beats on the river are far more common online than seen at the meatspace tables. (Undecided if this is just due to more hands being played however...)

- be aware of the overall game speed. Long rounds allow time to play cards that matter, short ones don't. Speed games are very profitable for the house, putting pressure on skilled players. Avoid those at higher buy in's.

- rebuy games often generate very good payouts in relation to the intial buy-in. avoid the temptation to rebuy however, unless it's very early in the game. Extra chips won't matter to a skilled player and you just pay a lot more in relation to your potential winnings. Rebuy speed games are pure evil at higher buy-ins, but can be fun and very profitable at lower ones. (Given you don't mind the greater chance factor.)

- the large sites are more difficult to manage than the smaller ones are. When considering online poker, pay close attention to the tournament games offered. This will tell you a lot about the site and what players they are looking for. Number of initial chips, buy ins offered and round length are key.

I'm posting this out of self-interest as well. (Like any solid poker would!) The more players in the game, the bigger the winnings are for everyone involved. Just thought that disclaimer was appropriate to make everything clear.

Want to play where my wife and I do and save yourself the trouble of learning what we have? Shoot me an e-mail and I'll trade our learning in return for a signup referral. (Referrals generate points and some small dollars, we use to play more tournament poker.) I'm not a sales shill by the way. Google me and you will find nothing of the kind. I simply enjoy the game and have been winning enough to continue playing to learn, earn and the occasional nice dinner with winnings :-P If you see some success, do exactly what I have done here and lower your overall game cost. (Do it with some tact though.)

One important rule, passed on to me during our last trip to play at Binyons: Play as cheap as you can and as often as you can. Keeping play overhead low helps to manage player variance and thus overall profit.

Re:Stay away from the cash tables.

[mosch]

the cards often appear balanced for high action. Almost every hand sees flops that are difficult for players to let go of. It's our perception that bad beats on the river are far more common online than seen at the meatspace tables. (Undecided if this is just due to more hands being played however...)

Action flops are a myth. Pull any reasonably sized hand database into pokertracker's postgres version, and then do queries on the hand histories. Look for things like "odds that I am going to flop a set" or "odds that I will be dealt X versus Y" and check it. You'll see it's right in range. Obviously some questions require very large sample sizes to know for sure... but as my sample size has increased, everything has always come out in line.

Whenever I see these hands happen in the casino I joke that the casino is rigged for action, and that's why I only play online. (I say this because the old coots who play daytime poker near me all claim online is rigged...)

be aware of the overall game speed. Long rounds allow time to play cards that matter, short ones don't. Speed games are very profitable for the house, putting pressure on skilled players. Avoid those at higher buy in's.

Good tournament advice. You just need different skills as the blind/stack ratios change. "Solid" play just because BAD play once the blinds get big in relation to the stacks.

If you're serious about tournament poker, buy harrington's books, and read both of them... volume 2 is NLHE tournament gold.

rebuy games often generate very good payouts in relation to the intial buy-in. avoid the temptation to rebuy however, unless it's very early in the game. Extra chips won't matter to a skilled player and you just pay a lot more in relation to your potential winnings. Rebuy speed games are pure evil at higher buy-ins, but can be fun and very profitable at lower ones. (Given you don't mind the greater chance factor.)

There's no reason to avoid the rebuy. If you're a better player, it's foolish not to (especially since rebuys generally don't have any fees attached).

Imagine for a second that you were offered entry into a tournament where everybody else bought in for $400, and you can enter for $200. However, you will start with half the chips of everybody else. You'd be a fool NOT to take this offer! Given equivalent skill, a half-sized stack is MORE than half as likely to win. If you have an edge, then it's even better.

Fast rebuy tournaments (as are common in casinos) are high gamble, but they offer an excellent ROI for the expert player. MTTs require an absurdly large bankroll, though, if you're routinely playing in large fields. (I'm not honestly sure since I'm not an MTT specialist, but I'd guess you'd need 100x the buy-in to get down to a 1% risk of ruin).

the large sites are more difficult to manage than the smaller ones are. When considering online poker, pay close attention to the tournament games offered. This will tell you a lot about the site and what players they are looking for. Number of initial chips, buy ins offered and round length are key.

If you're serious about poker, use neteller. Keep some money stashed there, and use it when you have the urge to play a different site or try something out. This also makes it easy to have cash on hand if you like to take advantage of deposit bonuses.

Doesn't matter

[Salvarus]

I play online poker (mildly seccessfully) and a poker bot would win at limit poker, no bot could win at No Limit. Especially if it judges by math only. In NL Poker the point is to throw your opponents odds off. Robot or Human, I'll still take their money! ^_^

Not a very good bot

[Propaganda13]

$5 table and only $30 after letting it run all night? $.50/$1 tables of Texas Hold'em generally have a pot average between $5 and $10 and cost $.75 for small blind/big blind.
I'm more worried about collusion at a table and there's no way to stop this, whether they're using this bot, Teamspeak, or sitting next to each other.
The bot does do the hardest thing for a real person to do which is to sit and not play. Fold junk hands for an hour, and you're willing to bet on anything that's playable.

Don't Fear

[herk]

(a) Winholdem is terrible. I'd sit at a table full of win holdem bots (provided they're not sharing information) any day of the weak for any stakes I could statistically afford.

(b) It's a little frustrating to see the endless stream of people spouting off about how online poker sites are surely rigged when they know absolutely nothing about it. I've been playing online poker for a year and have turned $20 into a small fortune. I beat the game for more than I make as a developer on the good months now, and I've successfully withdrawn a little less than half my profits. Poker sites take a rake from each pot that's played. Some of the larger sites have 80k concurrent users at peak times. Start taking a little piece of every pot with that many people online and you're earning a small fortune every freaking minute. They have very little overhead, computers, bandwidth, support, the rest is pure profit, and there's plenty for them. Why the hell would they risk this by cheating their players? If it was impossible to beat online poker, how do many of us do so consistently?

Truth be told I suspect these comments are coming from people who've never played online, or are influenced by the same stereotypes of poker being a game for cheats and hustlers. Bad players who try online poker and can't seem to win tend to enjoy spouting off that the sites are rigged, when in reality weird things happen in poker everywhere. Knowing how to bankroll yourself for what limits to sustain the unavoidable statistical downswings is the key.

Don't worry about the foolishness spouted in this tread. Win holdem is no threat, nor are any other bots at this point in time. And any of the major poker sites are plenty reputable, I was wary at first too but I've seen their business practices for a solid 12 months now. Online poker is booming right now and there's plenty fun to be had and money for the taking if you're half intelligent and can learn the required discipline.

Its all FUD

[litewoheat]

Even an intermediate player who can spot a bot can bust that bot. Bots aren't smart, they follow set logic based on hand strength, player betting patterns, and general statistics. If you play them "by the book" you will loose. If you play them by varying your play, playing overly aggressive and not playing crap hands, even in late position, you will always beat them. They are predicable.

Oh and you can always get up and find another table if they make you uncomfortable, thats be beauty of online poker, there's lots of tables out there so you don't have to be stuck on one.

Its also just a matter of time until the poker sites develop bot detection and destruction software. Its already in some clients.

it's taken this long?

[danny]

I can remember people getting unhappy about bot-clients in Netrek, way back in 1994...

Brilliant solution

[The Monster]

How about incorporating captcha into showing the cards?

Like this:

For game 13814397 today, the face-up cards are
<img src="card/20050828/13814397/1.gif> through
<img src="card/20050828/13814397/52.gif>

The numbers in the image URLs would have no fixed correlation to the card values (1.gif is NOT always the Ace of spades). The browser will unfortunately have to download all 52 images every game, but not every hand. Maybe a player could be allowed to keep a 'deck' for several games, but the basic idea would make it difficult for a bot to play. It would keep visually-impaired humans out, too.

Finally real hope for AI!

[logicnazi]

First of all I have to say I entierly support the people writing/running the online poker bots. They have found a clever way of winning and I find the intellectual competition between various bot writers far more stimulating than people just playing poker (I also think all those rules in F1 racing to reduce the benefits from creative engineering are horrible especially as this undermines the supposed purpose of racing as fueling research).

While some people might argue that using poker bots is wrong because it is a violation of some user agreement with the casino consitancy requires us to give no more credance to the casino liscensce agreement than any other clickwrap. Even if the Casino has a EULA type agreement preventing bots how is this really any different than a hypothetical clause in the MS EULA requiring that you won't dual boot linux or use OSS in general? In both cases the company is demanding you not use your own computers in the way you choose so as to protect their profits. We should treat both cases exactly the same.

Of course this isn't to say that the online casinos shouldn't do what they can to detect and evict bots (though seizing their money goes too far...you can ban whoever you like from your sight but stealing their money is a whole other matter). This brings me to the title of this post. No, I don't think that this poker bot or even poker playing in general will create significant advances in AI.

However, the battle between the automated bots and those trying to detect (or even take advantage of) the bots does offer real promise. Online casinos are only the tip of the iceberg, I fully expect the war between bots and anti-bots to only get more ferocious and spread from casinos and MMPOGs to more and more online activities. Finally their will be competitive pressure to develop incrimentally more and more sophisticated AIs dealing with more and more types of situations.

Unfortunatly, the academic community is particularly ill-suited toward developing integrated human like AI. We know from brain research and evolution that incrementally equipping and improving a system gradually with pragmatic hacks and adding specialized functional subunits can create human-like intelligence (it made us). Moreover, the continued difficulties faced by AI research suggests that no simple elegant algorithm will serve. If you want a computer to do all the things a person does you you need to program that computer with all hundreds of specialized sub-functions our brains posses. In short no clever idea will allow us to circumvent the fact that human like AI will take a massive number of lines of code.

Unfortunatly, while the academic community is very apt at creating algorithms for well specified functions, e.g., computer vision, it is very poor at creating massive integrated applications. While the core concepts and algorithms for the OS, database and the like have often come from academia it isn't a coincedence that the complex fully featured non-reasearch versions are either commercial or open-source. Quite simply academia rewards novelty and creativity not dilligence and the quality of the final product. As a CS prof you are far better off (and have more fun) testing your own pet idea or at best creating a demonstration app with a few other research group members than incrimentally contributing small features to a massive code base.

Only the commercial software companies and open source communities have the sort of reward structures suitable to creating usefull AI. In these communities it is overall product usefullness that is rewarded and many people are happy to make incrimental improvements even if they won't make for a good paper. I just hope these bot wars provide the begining baby steps necessery to get some of these projects rolling.

Old news in the poker world

[Rufford]

The problem is that playing against a team of players in Hold Em is quite different than everyone for themselves. Several people are getting this wrong.

The team can decide who has the strongest hand and the weak hands can stagger their bets to lure an opponent into betting more or getting out of the hand. Not only a statistical advantage with knowledge but also a finer level of control with the game.

The surprise? This happens in real life in real casinos. Regulars will team up on new people and then fight for the money between themselves. They even make crappy TV shows about it.

Players can get the same edge

[John Cage]

Using real-time odds calculating software players can get the same information the robots use to determine their own betting, without cheating. For example, Pokulator software tells a player their overall odds of winning as well as chances to make a particular hand. This is the kind of information the robots are programmed to use, but armed with this information and a brain a real player could beat out a robot.

Re:if you can make so much money

[stonecypher]

For the same reason the people that make $70,000 a month part time with their home business have infomercials: because what they're really making money from is convincing people to buy into some scheme of theirs.

It's invariably the case that even when these systems really do work (a few of them, particularly the real estate redevelopment models, actually do work,) selling them to millions of people is significantly more profitable than actually running them.

It's called franchising. Pizza Hut doesn't own all its stores; neither do McDonalds, Wendys, and so on. Why? The stores are immensely profitable.

The answer is simple: reaping margin on people investing their own money is far more profitable than doing it all yourself, because the growth rate is far larger, and you step back from all of the personal risk. If they're profitable, they break you off a proper chunk. If they fail, well, it's just not your problem.

Be honest: would you rather own five fast food resteraunts, or reap a percentage on two hundred? The franchise model is so successful that better granchises, like fast food, often have smaller companies subfranchising. Next time you go to a Burger King or whatever, take a look at the drive through window, where they have the complaint phone number. Surprisingly often, it'll list a smaller company you've never heard of, but which owns a few hundred local fast food places from a variety of name brands. In Pittsburgh, most of the fast food in the college district, regardless of who the sign reads as, is owned by either of two companies (to the tune of some 100 or so storefronts covering a good two dozen brands.)

In short, selling a scalable business model is usually more profitable than using it.