Online Poker Bots Becoming Problematic? 613
scumbucket writes "MSNBC has a story about how poker bots have started to appear on internet gambling sites and the implications. It also talks about how a 'master level' poker-playing bot already exists. Could this proliferation of poker-playing bots undermine the almost $1 billion online gambling industry?"
Good? (Score:2, Interesting)
A business that has millions, but really gives you nothing for your money in return, deserves a good stinging pinch like this.
Poker won't be the only one to have bots. Blackjack can be easily played via pre-defined rules. In fact, it's a lot easier to make a bot for Blackjack than for poker, since Poker is more affected by bluffing, and human interaction. That's why most poker players wear sunglasses, and show little emotion. Hence the phrase "poker face".
I guess in summary: "Online poker bots becoming problematic?"
Sure, but problematic for whom? The casinos are notorious for putting a winning blackjack player on a blacklist, and not letting them in. Why? Did they cheat? No, they just won, and casinos hate to lose money.
Bots are problematic for casinos, but is that something to cry about?
Of course, if you don't like casinos, don't go to them. They only exist
to fulfill a need/desire that many people have, and unless that goes away, they'll continue to exist.
Re:Good? (Score:5, Interesting)
So, yes bots are problematic in poker because if I play, I want to play against another human, not some computer simulation that can calculate the odds down to the decimal. Granted some humans exist like that, but not many.
Why a concern? (Score:2, Interesting)
Why should it really be a concern? I mean, poker companies will still be making money, and as long as nobody's cheating, what gives if the next person is a human or a bot?
Machines are no intellectual match for humans (at least for now). At least not for most humans. Given the fact that poker is a game of chance (unlike say chess, in which randomness has no play), a bot can only be as good as the expert that has created it.
Re:"I equate gambling... (Score:5, Interesting)
Online poker works by taking in a 'rake' from each pot, usually around 10% or so. Therefore if a pokerbot can clean out the 9 other people at the table, the casino will still earn the same profit as if those 9 people had just passed around money amoung themselves.
Re:Maple Leaf Forever! (Score:4, Interesting)
undermine the industry?? (Score:5, Interesting)
It is about time (Score:2, Interesting)
Go Bots !!!
Re:Good? (Score:3, Interesting)
Yeah. Unfortunately, this explanation also fits Microsoft. The lusers (in all levels, from home user to PHBs) will continue to drive their profits and existance, not technical merit, innovation or quality.
I just wonder whether /.ers will ever get this.
I hope Internet gambling goes down in flames (Score:5, Interesting)
I called the bank and while I was on the phone with the bank rep, more weird charges were coming in! We were both watching someone gamble away all my money in real time. So he red flagged them all and gave me a claim code.
The next day the phone rings. "Hello, this is Planet Poker..." and without thinking I say "No thank you" and hang up. The phone rings again a few minutes later. "Planet Poker..." and I say "please take me off your list" and hang up, still thinking it's a telemarketing call. Which sounds stupid given the withdrawls the day before, but I didn't put two and two together. (It was Planet Poker calling me to welcome me as a new degenerate gambler / customer.)
The phone rings again. "Don't hang up we think someone used your credit card!" she says really fast. I said, oh yeah, I reported those charges to the bank yesterday.
Then she sounds sullen. "Well... I guess we'll be getting the chargebacks then..."
I said, "yeah, I guess so!"
Don't know if the guy was using a program to help him cheat, but he played really badly.
Who cares about bots? Why trust the casino at all? (Score:2, Interesting)
There is no way for any user to know if an on-line "casino" is shaving points or that any of the other "players" are working for the casino. Unless you are betting on something that can be verified publicly and can't be rigged easily (such as major league sports) on-line gambling is too much of a gamble.
Re:not quite so hard... (Score:5, Interesting)
Umm... (Score:3, Interesting)
well considering that my friend and i implemented a system that does just that (over a year ago), i call bullshit on this one. what we did was capture an example of each card off the screen (we use ParadisePoker). then we came up with a best-match algorithm that compares each card to our reference cards to decide which card it is.
it's the "react in real-time" part that is the bitch of it all... the "read" from the screen is, while i won't say trivial, doable. it took me and my friend less than a week to implement using java's "robot" class (to read pixel values from the screen).
some guys we know were just trying to put together something that calulated odds for 7-stud on the fly, but found that inputting the cards by hand took too long. so they asked us if we could capture the cards from the screen. we didnt know if we could or couldnt, but we looked into it and it wasnt so hard.
i also know of more than one person back at school (MIT) who is working on a fully-functioning bot. i dont think they are "years" off from being finished either.
Old news at MIT (Score:2, Interesting)
Some of my buddies at MIT and I have been operating a program like this for over a year now. We wrote the control algorithms ourselves based on our preferred play styles and fit them into a screen scanning program. The program is no better than any of us, but it has the advantage of being cold and methodical. We included code for an emergency stop so that if something went wrong we wouldn't lose all of our money, though it hasn't been an issue after the initial testing phase. I'm still uncomfortable about using the program for higher stakes games, but on a 50 cent/1 dollar Hold 'Em table at peak hours (discovered through trial and error) the program consistently makes about $20 an hour.
Re:Good? (Score:2, Interesting)
Someone correct me if I'm wrong here but doesn't winning at Blackjack rely on card counting? If your bot is playing a computerized version you effectively have a new deck every time so you won't be counting cards.
Actually no (Score:3, Interesting)
"Really, if you want to make some money at poker you'd be better off learning to do the odds in your head and going to a real casino to find a table of marks."
ha,ha,ha thats pretty funny..and niave.
When you go to a casino, 4 out 5 people at a table are regulars, and the pretty much just sand bag until mister, 'I know the odds' shows up, then the fleece him.
However, one time I did make a semi-pro player come over the table at me... ahhh that was great.
Re:not quite so hard... (Score:4, Interesting)
Not nice, but thankfully it's about as immoral as I get.
Dont play poker online for money (Score:5, Interesting)
Foor the cool factor (Yay! its GNU!) consider "GNU Backgammon", the program uses 3 neural nets and humongous move databases. Backgammon compares well to poker btw, BG is ruled by dice and skill, Poker is ruled by chances and skill too. It is quite likely the strongest BG playing, ehm, thing, in the world.
Gnu BG plays an astounding 2200 rating on Fibs, if not higher if you get high end hardware, and give the bot a few secs between moves.
1800 is considered a worldclass human player, 1900 and above are grandmasters.
Friends, dont play backgammon online for money, and certainly not Poker. Instead if you must, visit tournies in the flesh.
Or get the bots, and a few spare comps... You will NEVER rob the casino thou, you will rob other suck^D^D^D^Dplayers.
"/Dread"
Re:not quite so hard... (Score:3, Interesting)
They already are here. (Score:1, Interesting)
I did this 3 years ago, and have had my sessions running for almost that whole time. I am neither a world-class poker player nor a world-class programmer. So far, they have earned me about $130,000.
The article got hung up on how well poker AI could compete against good players, but that's not where they are most effective. Bots can grind at the lower stakes ($1/$2 and $2/$4) far better than any human, and can beat the level of human play found there far easier than at the high stakes.
Some notes from experience:
-The sites don't give a shit. I've taken some measures to disguise my bots, but a monkey examining the casino records could figure it out. They collect their rake, and as long as I don't scare players away, they make more money with my being there.
-As much as I've tried, it's far from maintenance-free. I've put over 2000 hours into the whole mess so far.
-What I've got is not working nearly as well as it used to, and I'm on the verge of scrapping it and walking away. The upkeep is almost not worth my time, and unless I figure out how to regain the performance of 2003, I'm outie.
Re:not quite so hard... (Score:4, Interesting)
A few facts (Score:5, Interesting)
In poker you play against other player ; in blackjack you play against the house.
Blackjack used to be a game of chance with odds that could be slightly turned against the house in favour of the player if he played "perfectly". For this he needed to calculate odds given the number of cards left in the dealer's shoe and bet accordingly. If the game was played this way online, it would be a disaster for casinos as bots would rule the game. Online however, the deck is shuffled after each hand and there is no way of calculating the odds. Therefore, online blackjack has become a pure game of luck which is why bots are useless and why you shouldn't play blackjack.
On the other hand, bots can help you calculate odds in online poker. But that's only part of the game and they are largely ineffective against any decent players as they cannot understand human psychology as well as they can calculate odds.
Unlike chess - where bots are very effective - poker is not an information complete game. Therefore a player's skill depends strongly on his ability to "read" and bluff other players. Which is why poker bots will probably remain useless for a very long time. Probably until we reach hard AI.
Re:not quite so hard... (Score:5, Interesting)
Knowing how many hands you went the distance on, your winning percentage, and your betting habits during each betting round are the things that a really good player knows about every other player at the table. Those are the things you see people thinking about while deciding to make a call on a bluff/non-bluff. Those are also the exact things a program could help track.
Re:Old news at MIT (Score:2, Interesting)
The online casino's are going to have to go to a periodic human input check like you see on registering free email accounts. This will cut their rakes by reducing players abilities to play 6 tables at once, but will be required once average players start leaving in droves because of rampant cheating.
Stopping human based computer assisted collusion on higher limit tables is a much harder task. Despite their claims to be having security measures in place, this is already becoming epidemic. Bottom line: Online poker will be dead in 3 years. Congrats on getting yours while the getting is good. I am too, in fact, i've got 4 bots running at 2 different sites as i type this. +94 dollars since 7:30am CST.
Makes alot of sense (Score:1, Interesting)
Since the way the game works is that top player takes 50%, second takes 30% and third takes 20% of the total incoming pot (out of 10 players minus an entry fee), it's not really very difficult to come out of a game ahead. The problem is that you have to be extremely attentive and play very conservatively. This is horribly time consuming. You could spend 2 hours winning 10 bucks in the lowest pot games.
One way to get around it, play high stakes games. The problem is that, the high stakes players are very attentive and play conservatively and you don't have the easy win like you would in a 100 dollar game.
That's where the bots come in. They'd give an obvious statistical advantage if you could run a few hundred at the same time in low money games (ie where the newbies play). Even if you get third place, that's still a double your money on the entry fee. Sure some bots will lose, but if they play based on poker statistics, and bet according to the odds, they'll always be in the black over the long run.
Re:not quite so hard... (Score:2, Interesting)
This is the case for all of the top 10 online poker rooms: PartyPoker, PokerStars, Paradise, UltimateBet, PrimaPoker, PokerRoom, PacificPoker, CryptoLogic, Ladbrokes, Apex Poker Network.
They take security and integrity very serious and writing the clients from the bottom up assures them of that.
I question your authority (Score:3, Interesting)
Saying saying that blackjack "played at [sic] it's best... will translate into certain losses" is incorrect.
What's the point of saying you have a relatively "informed" view, when your view is wrong?
Most people play like bots? No, that's the problem, "most people" can't calculate perfect statistics on the fly, or do any of the other things bots can do in microseconds.
The 'human element' you talk about is present in face-to-face games, but in online games, you can't even see the other players! That's why bots can compete and disrupt the game. Maybe you can comment about playing poker in home/casino games, but I think your relatively informed view does not apply to this thread.
I suggest modding parent down.
Re:Is This So Wrong? (Score:1, Interesting)
And every one of these thousands of poker bots has to deposit money to play.
Hmmm, Suddenly doesn't sound like such a good idea.
Re:I hope Internet gambling goes down in flames (Score:2, Interesting)
Why? Here's Why: (Score:5, Interesting)
Consider the following case in Hold 'em:
You are up against only one other player and are in last position. The river has come out and there is a queen (from the flop), a jack, and then 3 low garbage cards. You have Ace-Jack. If that player checks to you, and has checked the previous bets, you will probably bet that even though you only have second pair. If he bets high, and has been betting the whole time, you may want to REALLY think before calling, and especially before raising.
This is an example of how BETS come into consideration.
And of course a good bot, like a good player, would not be able to just take all bets at face value.
Re:Good? (Score:3, Interesting)
The greater the risk, the greater the pay off. Playing by the odds, it is not a good idea to chase flushes or straights in Hold'em because most of the time it won't pan out for you. However, given some circumstances a human player might take the chance and get burned, or win big. Is the bot a worse poker player or better for not taking the chance?
Re:Old news at MIT (Score:3, Interesting)
NOTHING AT ALL!
That's the problem. The casino software can't possibly know that you're not using IM to chat with other players. At least in an offline cardroom they would likely get caught trying to signal.
Even if players don't collude, they have an opportunity to use aids to calculate pot odds and engage in other cheating that they'd have no chance of doing in a real cardroom.
Setting things straight (Score:5, Interesting)
I'm relatively sure that all of the online gambling sites use either Flash or Java applets to display cards and such. I wouldn't think they'd make it so easy as to give easy access to card names.
You do not actually need to break into the program in order to use some form of bot. Graphics recognition has advanced to the point where a hand can be analyzed on the fly by a concurrently running program. See Poker Office [pokeroffice.com]. Such programs can then immediately provide feedback based on the information they glean.
Could this proliferation of poker-playing bots undermine the almost $1 billion online gambling industry?
The end of the industry is not likely at hand. Poker is just one part of this industry, and the industry will continue EVEN IF bots are the only ones playing. The casino will just take the same percentage of each pot.
if a pokerbot can clean out the 9 other people at the table
Quite frankly it is ridiculous to think that a bot with perfect play can clean out any table. Good poker play results in a slow accumulation of profit at a faster rate than losses. A perfect bot will certainly not be playing more than 1 in 5 hands to begin with, and not win more than one in 3 of those. Good players can't just make the right cards appear, no matter what you saw in Maverick. They get the same crappy cards as everyone else, it's how they play them that differs.
you don't trust other players to not be using bots, but you trust the house to not add their own player to every game and fix the host software to guarantee that the house's player wins?
Yes, very much so. Contrary to popular opinion, most people are not complete retards. It's not difficult to tell when someone is consistently winning - certainly there are hot streaks, but any whiff of foul play will result in a huge exodus of players from any poker site. They have no reason to do such a thing, as profits are huge from both the rake AND the interest they are collecting on your bankrolled money.
_______
Any current bot is very likely for Limit poker - this is the 'easiest' style to play purely by the numbers. The state space required for a bot to make decisions in No Limit poker is absolutely huge- one poorly written part can get your bot cleaned out regularly.
Personally I would LOVE to be at a table where I have positively ID'ed a player as a bot because I could then run circles around it. There are a number of tactics that would play merry hell with a bot that plays the straight numbers, and even a bot that adjusts to my own play style is not difficult to take advantage of.
I play regularly online and I do not fear the bot. What I fear most is the bad player that will put all their money on a 20% draw, where any good player (or bot) would fold- because sometimes they hit, and that hurts.
Once they find a cure for bad players though, that's the end of poker, but I am content that that time is far in the future.
Temporary Solution (Score:3, Interesting)
Bots don't mean crap (Score:2, Interesting)
The only true way to cheat is what my brother and his friends do. They enter into the same tournament or play at the same table and then call each other on the phone or use a chat program. Then they tell each other what they have. They have a great system of not getting caught because the have serveral accounts and change the rooms they play in and what not. They have been doing this for months and win enough to cover their expenses. That is the biggest problem, some people get too greeding doing this and try to win millions and get caught.
But then again, what are the consequenses of getting caught..... nothing. This isn't Vegas where they can throw you in jail. They have no proof that you are really doing anything wrong, just logs telling them how you played and what good is that. All you have to tell them is that that is how you play. They can't call you a liar because they don't know you.
I think that people make this a bigger deal then what it is.
Re:Casino 101 (Score:3, Interesting)
Not true in blackjack, as alluded to by the previous poster. Card counting is NOT illegal if you are only using your brain. You are only using skill to beat a game that is beatable. If they suspect you of card counting, which again is not cheating, they can and will blacklist you....at least in Vegas. I do believe that they can't bounce you in Atlantic City...but, they will start to shuffle on you every hand or so if they suspect you of counting...
Some notes on the discussion... (Score:5, Interesting)
(Side note: If anyone is interested in playing some online poker and wants a bonus on their first deposit, drop a reply to this with your name and email address, and I'll send a referral out. We both get a bonus from this.)
Re:not quite so hard... (Score:2, Interesting)
The other players can still collude against you, even if the casino doesn't.
And the online casinos can't really stop them.
Re:Dont play poker online for money (Score:2, Interesting)
Re:Old news at MIT (Score:3, Interesting)
Who cares about them either? (Score:3, Interesting)
As far sa I'm concernd a spread of bots would be just like a spread of really good players - and even if they became the majority, why should it matter if online poker playing turns into a contest for besk poke developer? In fact if there was a widespread bot released I can think of no greater fun than developing the counter-bot to trigger bugs in the software and make all the clueless bot users fork over money to me!
IDNHAWSOPB but... (Score:1, Interesting)
1) Online Poker sites *do* try to prevent colluding. They actively monitor IP addresses (easy to get around, I know), but also who is playing with who. If you always end up at the same table with your buddies, they will find out and boot you and your pals.
2) Online Poker sites *do* try to prevent bots. They do this in a couple different ways, most of which are hackable of course. If you can write a bot to do OCR on the screen, it will never be detectable. Bot "play patterns" may still be detectable, but the client software will never be able to detect your bot.
3) The reason Online Poker sites care about collusion and cheating is NOT because they are losing money! At least not yet... They are worried about the online player community's confidence in the integrity of the online game. If players think others are cheating or using good bots, they simply will stop playing (eventually).
4) Creating a bot that can play No Limit is not very viable. No Limit games are MUCH harder to play properly than Limit. In No Limit games, being able to "read" the other players is much more important. Bluffing and reading bluffs is crucial. This is not nearly as important in Limit games.
5) Bots *can* easily make money in Limit games. The truth of the matter is that there is a ton of HORRIBLE players online. Players that re-raise with A6o, for example... Players that will call you down no matter what just because they have Ax and there was an Ace on the flop. If you play tight (think Dan Harrigton, minus his crazy WSOP bluffs), you can clean up. It's slow, but it works. Writing a bot that can play tight is not hard at all. I could easily code all the logic in a day. Of course, the hard part of the bot is the OCR (or decrypting/encrypting the network traffic if you want to hack it at that level).
Re:Be afraid of collusion, not bots. (Score:3, Interesting)
They track players who play together too much.
They most likely key off of credit card info.
You'd probably need to get a lot of real people in on it, and play decoy games away from each other and try to maintain a churn of new people.
Those online poker places could totally run a googleesque challenge by taking their logs, anonymizing the names/info and running a contest to see who can come up with the best method of flagging suspicious behavior. I wonder just how good the people they have doing it now are. I wonder how the various sites compare to each other along those lines.
Re:not quite so hard... (Score:3, Interesting)
What I don't get is this: why are groups of 5-6 people not getting together, playing at medium to high stakes tables, and just dominating the games since there's nothing stopping them from discussing what cards they have over some IRC channel or a VOIP conference call or something
Re:Original twoplustwo article (Loic Dachary) (Score:3, Interesting)
The bot owed its success to the fact that it remembered all the cards that were discarded (which in online stud often go by so fast it is very difficult to see). It then played about a million hands from the current position with a Monte Carlo simulation, taking into account all the discards and all the possible outcomes, to establish a base win probability.
Armed with a few heuristics on when to fold early, the bot did quite well.
Obviously it would be defeated by anyone who
a) knew he was playing against a bot
b) was a decent poker player
But that's not the point, is it? My plan was to improve the bot to start making judgments about particular players, put it in a corner, and then auto-haunt the low stakes tables 24 x 7, but then I got busy and put the project aside.
I claim a bot of this type could make money. Not tons, but enough to make it worthwhile.
Most bots suck at some part of the game. (Score:1, Interesting)
they could get the play money from the poor quality of players. They
mostly played just the premium hands. There was one that insulted
the other players play in chat, it probably made more money by getting people angry and putting them on tilt, than by any strategy.
But in a 3 or 4 handed game, the bots would lose. Short handed games
require more skill, and adapting to others play. That said, one on one games
may be easy for bots to beat-more game theory probabilities.
I used to see a player on pokerstars recently, I thought might be a bot. He was always there, didn't chat, just waiting for a one on one game to start. He played one on one strong and fast, but after a few more people entered the
game he would slow down and start to lose.
I am not worried about the bots yet. I am concerned about the poor quality
of shuffles. You definitely do not get 52! combinations. They obviously
are getting the seed for the randomness off the clock, and when you play quick
short handed games you often see similar cards come up a lot. Not to mention
the suspicion that many good players have that the deal is biased to favor
poorer players (it keeps the money in play longer, and moving across the rake more).
But cheating at cards has always been an issue, in the real world as well,
just like cheating at elections.