anyone wanna talk texas hold em?
anyone wanna chat texas hold em?
hi.
i'm writing an article about online poker, and why people love it so much. are you worried about bots? how much do you play? what is the appeal? there's a poker group at the University of Alberta that are designing the perfect poker program - that is expected to beat the top human player in a very short time. i'm looking for people to interview, either by a quick email, or by phone. it won't take long. if you're interested, be in touch soon. i would really appreciate it.
darcat_sweetness@yahoo.com
cheers, darc
hi.
i'm writing an article about online poker, and why people love it so much. are you worried about bots? how much do you play? what is the appeal? there's a poker group at the University of Alberta that are designing the perfect poker program - that is expected to beat the top human player in a very short time. i'm looking for people to interview, either by a quick email, or by phone. it won't take long. if you're interested, be in touch soon. i would really appreciate it.
darcat_sweetness@yahoo.com
cheers, darc
Comments
There is an easy way for sites to defeat bots. They just need to put up a prompt every few minutes (or every hand if they want!) and force players to respond within a time limit or be logged off. Something simple like the registration screens a lot of sites use to defeat bots today.
Question is: do the poker sites have the will to beat the bots? Or do they like bots? Is it possible some of them are using bots? Hmmmm... I've always thought stars was rigged!
Until a true AI is developed (ie terminator style) a poker program will never beat 'the best player in the world' or anyone even decent. I am sure these programs could beat low limit games...pure math and counting pot odds. The reason they can do this with chess is because it is a game of perfect information, while poker is not. Many low limit and some middle limit games can be beat in an ABC mathematical method, but not a chance beyond that. In reality, the math is quite simple to any high stakes player, even on the fly. The computer edge at this level from a math perspective is limited. Do you think a computer's math abilites applied to poker are better enough than Gus Hansen's (who is considered a math genius) that it would outweigh the unpredictableness and ability to change gears that Gus has? No way.
Id love to see the developer bankroll the program and have it 'sit down' in the 'big cash game' - HORSE with Hansen, Brunson, Ivey, Seed, Greenstein, Chau, Reese, Negreanu etc
Ill give the developers 5:1 odds that if their program sat in this game with 50bb they would be to the felt within 10 hours.
Cheers
Magi
An individual bot has little value BUT a network of bots running 24 hours a day could generate tens of thousands of profit per day. What is the value of that??? Of course, once low limit players get wind of it, it pretty much kills the opportunity.
As to the earlier post about beating a big game - why bother? There aren't that many big games but there are literally millions of players at low stakes to beat up. A bot with a bit of AI built-in (or a database of players' styles, which is not that complicated and in fact already exists to a certain extent...) could easily clean up. Once the program is written, it can be replicated thousands of times and run on tables all the way down to .01/.02. After all, how much does it cost to run a computer all day? Something like this could win 2-3BB/100 hands and do it on hundreds of tables simultaneously. And there is absolutely nothing illegal or immoral about it. Sweeeeet! Anyone want to invest a couple hundred grand in this to help me get it off the ground?
A successful high stakes player will change his style during the session. That alone will help to defeat a bot. Can a bot do the same thing? Sure! You can add randomness to the play to keep others guessing.
Please discuss.
You contradict yourself -
"Of course, once low limit players get wind of it, it pretty much kills the opportunity."
Then why would we do it? Online sites have a vested interest in protecting their players. They could ruin the best laid plans (and $$ invested in it) with one simple algorithm or prompt like you mention in your first post.
Imagine a hedge fund that did this.
It would be stupid for a place like Party to run bots. The risk long-term from damage to their reputation is greater than the short-term upside. But what if another, nameless site, put up at least 1 bot on each table to simulate action and get real players to join in? House players as it were. Only as long as they didn't play too well! But aren't house players fairly good players? Not too aggressive, maybe but they don't lose much either.
So take it out of the house. Say a team of guys got together to sponsor the creation of a bot that could beat medium-level players on low stakes games. Together they could open a few hundred accounts and play 24 hours per day. As long as they didn't get too greedy, they could probably generate 5000BB per day, 150,000BB per month, 1.8M BB per year. Hmmm. Sounds like a good business. Do you think maybe someone is already doing it? Kinda makes you wonder. And what are the big sites doing about it? I haven't heard a peep...
Maybe I should put together a business plan???
I'm still new to this whole game, and not sure if I have ever played a bot, but by the sounds of it, maybe I must have by now. Are there really alot out there to worry about? And do they primarily haunt the low limit tables? If so, whats the best way to protect myself against them?
No. For one, I play online for recreation and practice with the math of poker, so I don't put up much money - I save that for live games. And reason two, a bot is simply the ultimate expression of a book player, who plays according to a particular algorithm based on cards and position; such play addresses less than half of what's going on in a poker game, and I'm reasonably sure I'm better in the rest of the game.
Online, not much - a few hours a week at most. Live, three or four times that. I'm not a poker pro, just a casual player who shows a profit in low limit games.
Money won is twice as sweet as money earned, and it's nice to have a hobby that consistently gives you money.
Hmm. I've heard of Poki, who had such trouble playing even average players in a regular 10-handed game that the U of A developers decided to focus purely on headsup play. In that limited universe, they still believe they're years away from being competitive. Since each additional player added to the game increases the complexity geometrically, it sounds to me like, barring something like a quantum computing breakthrough, your prediction is a bit of a pipe dream.
One of the interesting things about Poker - as opposed to something like Chess or Backgammon, where computer players are at least competitive with the top echelon players - is that very few top level Poker players use a strictly game-theoretical approach. Of course they all know their Sklansky inside out, but they don't follow his methods particularly closely, and Sklansky himself often has trouble in top level games (though I understand he did well in the recent Headsup Tournament). This makes it much harder to analyze what a top level player is doing to win, and so harder to build an algorithm that can beat the game.
If they DO exist, I would think they'd have to haunt the low limit tables at least to start. The inventor would want to prove it first before putting up big cash. Unless he had a major backer. Plus they could be less sophisticated and still make a terrific profit.
If they DON'T exist yet, they will soon. Put a good Windows programmer together with a decent poker player in a room for 6 months. I don't think it would be that hard to distill Sklansky's starting hands charts and TJ's or Miller's post flop play with a your opponent's history to put together a winning strategy. Most of this stuff already exists (PokerTracker, GameTime+, odds calculators, various books). In it's simplest form, it should be able to beat most low limit players since it would be patient, find the best tables, know how its opponents play and tailor its play to maximize profits. The kind of stuff the really good players would do. Plus, it can do it 24 hours per day! As an added benefit, it would be a great testing ground to see which strategies work best at which sites, stakes, times of day, etc.
How can you protect yourself against a really good player today? Go to another table. If every table has a bot what do you do? Go to another site. If every site has bots - go to B&M. (Of course, then there's collusion to worry about.) It is really up to the sites to come up with strategies to defeat them. Not that hard to do and they should be public about it as well. To my mind, being up front about this gives them more credibility vs the sites that pretend there is no threat.
How do you know you're up against a bot? If they build them right, you don't! Sounds a bit like 'The Twilight Zone'.
LOL -- I'm too busy in my wacky thread to notice this response. The very best bot out there, is built by the UofA mathmeticians led by Darse Billings. The first Bot was called Loki, and then they created Poki. I bought their program and the bots are amazingly good. You can buy the program as well ( http://www.poker-academy.com/ ) It's pretty awesome software to learn with. However, I trully believe the bot can be beat by pretty much most winning players. I believe the bots would mostly feed on losing players, so they wouldn't /shouldn't notice anything different. The winning players would continue to win, and they would take money from the bots. That's why I said, as long as the bots put up real money (ie. not house bots), I don't mind.
Now, if someone were to use some sort of teaming with the bots, I think it would be quite difficult to win against. As you mention in a later post, I believe this can only be worth several million dollars at best. I think people would notice that the money had dried up and you'd see far fewer players, and rake!! This is a billion dollar problem for the house and I think they will figure out how to combat the effective bots. So, I'm not that worried. I just won't play if the money dries up.
Cheers
Magi
Does anyone know how relatively advanced the bots are in Turbo Texas Holdem?
Agreed. I was being too terse. My worry comes from the house running the bot. They could get access to hole cards etc. which would make it very difficult to beat.
Not sure about Turbo TH, but the poki bot won a gold medal in AI design. I think they (poki) are the gold standard.
Cheers
Magi
Artificial Intelligence: Man vs. Machine
The industrial revolution ushered in a new era of human history where civilization advanced by inventing machinery that could do a job faster and with more accuracy than its human counterpart. Since that time machines have made significant strides, first in matching and exceeding human physical labor and now trying to match human intelligence.
In May 1997 the World Chess Champion, Garry Kasparov, a Grand Master and considered the best chess player of his time was defeated by Deep Blue; a machine made of silicon, plastic, and metal. The machine triumphed not once, but twice. This highly publicized match brought game theory to the attention of the masses and questioned whether man had usurped his claim as the most intelligent being with his own invention.
Three years earlier, researchers at the University of Alberta had studied machine learning as it applied to game theory by developing a checkers playing program named Chinook. Chinook became the best checkers playing entity on the planet, eventually winning the checkers world championship by defeating competitors in qualifying tournaments and leaving a trail of stunned human players in its wake. Recognized by the Guinness Book of World Records, the University team wanted to build on this success and apply their knowledge and programming skills to other game arenas. In the early stages of development at the time of Chinook’s checker victory was a program that played Texas Hold'em Poker.
Play the Cards and the Man
" [It] can be unnerving for some of the better human players, who often rely on unbridled aggression to win. The machines don't feel challenged as humans do; they simply crunch more numbers to decide the proper response."
- New York Times, July 10, 2003
The next natural focus for the GAMES group was opponent modeling. Neural networks could predict to satisfactory degree human responses based on a set of stimuli; and so Poki was built on the shoulders of its predecessor, Loki. Though deriving its name from humbling beginnings (Poki was indicative of how long the early versions would take to run through modeling scenarios), Poki soon proved itself to be a significant step forward in computer poker technology. The GAMES group took its testing to the next level by hosting a server that would allow human opponents to come and pit their skills against Poki with play money. The more people that came to try and beat the poker robot, the more information Poki had to work with and model against. Soon Poki was taking all comers and playing a solid, profitable poker game. Now that Poki had proven itself in full handed table games, the researchers turned to another area of poker research.
One-On-One
The movie A Beautiful Mind introduced the public to a concept long known by mathematicians; Nobel laureate John Nash’s concept of economic game theory. Nash’s idea was applied to heads-up poker to find an approximation of an equilibrium strategy—which is optimal play by both players. This new program was named Sparbot. Sparbot had a simpler and more elegant approach to poker than Poki; it would not play to win but merely to not lose. In heads-up poker, unlike other games of skill, this can be a profitable strategy. If your opponent plays well, a Nash-optimal strategy will break even over the long run; but if the opponent makes any serious mistakes (such as weak betting, or folding too much), they will lose money. Sparbot forces its opponent to accept that breaking even may be their best scenario.
Future of Artificial Poker
“ The artificial intelligence researchers have moved one step closer to creating an unbeatable computer poker program. An account of their most recent program, called PsOpti - for pseudo-optimal poker program - will receive the top paper award at the world's premier AI meeting in August.”
- National Sciences and Engineering Research Council of Canada, June 2003
Now that poker has exploded onto the stage of the mainstream media, the research group at the University of Alberta has risen to the forefront of game theory technology, and they show no signs of slowing down. They have recently been interviewed by the New York Times, and had a feature piece filmed by Discovery Channel Canada. Edmonton has been thrust into the limelight by their Games research, and with the guidance of poker theorist Darse Billings, the development of poker game theory can only grow in this "Vegas of the North".
The newest development by the University of Alberta games group is Vexbot, which is an AI system based completely on opponent modeling. This is a radical departure and very different from the original rule-based Loki. It has been successful thus far in proving that opponent modeling is critical when playing poker at a world class level. Vexbot forces an opponent to continually change strategies and adapt their play, as it will attempt to exploit any and all weakness or predictability it finds in their playing style. Players facing Vexbot find themselves asking "how can I win against myself?"
In the future, poker enthusiasts can look forward to a poker playing program that incorporates all the best aspects of Poki, Sparbot and Vexbot, resulting in the ultimate training tool for aspiring poker students dreaming of victory at the final table.
What Can I Do?
Fortunately, the GAMES group at the University of Alberta is not keeping this technology locked up in academia. Poker enthusiasts and aspiring and veteran players can now play Poki, Sparbot, and Vexbot in the comfort of their own home. Poki's Poker Academy brings poker research technology to your desktop, and enables you to practice and hone your skills against an opponent who will always play a competitive game. Poki will never go on tilt, never throw his chips at you, nor will he grow tired and bored of the game. Poki will always play a game based on sound poker strategy. It will force you to learn, adapt, and improve your poker abilities. If you can learn how to be profitable in Poki's Poker Academy, you will be a formidable foe in the online card rooms and real life casinos.
One day, the world's best poker champions may be seated across from an opponent with the coldest and most lifeless eyes they have ever seen. They will stare into a pixilated screen which belies the sophisticated modeling taking place in a cage of steel and plastic, utterly insensitive to its opponent’s searching stares.
Some suspect computer programs have infiltrated Internet games
By Mike Brunker
Reporter
MSNBC
Updated: 11:35 p.m. ET Sept. 21, 2004
http://www.msnbc.msn.com/id/6002298/
The solution, in the case of the Vex Bot, was adding a layer of artificial intelligence over its ability to calculate probabilities.
“It will show you things that no human player has ever shown you before,” Billings said of the latest incarnation of the bot, which also has the ability to model its opponent’s behavior. “… One of the biggest advantages that programs have is that they have no fear, no shame. Humans can be intimidated. They will back off in the face of a very aggressive player. A bot will not. It has no compunction about doing whatever it will take to win. It will raise you with any two cards if it thinks that it has a very slight advantage based on your history. And it can induce a lot of anger and emotional upset. These things are ‘tilt monsters.’”
:eek:
It was about the most recent developments in Robotics and AI. I only caught the last 10 minutes, unfortunately.
But the program ended by mentioning that "Honda had unveiled a top secret program 2002".
The feasibility study was created in 1970 but they had to wait 10 years for technology to progress to the point where they could actually build what they wanted.
It took 10 years to build ( finishing in 1990 ) and they waited a further 10 years to unveil it - a robot that's optical recognition systems were so advanced that it could walk around and up and down stairs.
It looked impressive and clunky at the same time.
It's probably the first time I've seen a robot walk up stairs and actually look human-esque ( rather than looking like it belonged in the Special Olympics ).
Kind of scary at the time.
In this book, based on Moore's law ( processing power doubling every 18 months ) Kurzweil says that by 2020, AI will be comparible with the Human brain.
On the downside - the US government spends 1/3 of a trillion dollars per year on the military ( 25 billion per year going into black ops ). They probably have some nice robots.
nice... killer... robots...