I guess it doesn't take a super nerd to see that computers are better than everyone at chess, and better than almost everyone (and probably everyone by now) at HU LHE. Given the last couple years trend of NL bots, I think it's generally accepted that they'll crush almost everyone soon as well.

The fact that programmers must attack each game in very different ways is immaterial to the person using computer assistance. What in the world are you trying to say?

I really love this MIT, thanks a lot for the lectures and the uploads on youtube! I really enjoyed them so much and the quality is super high for the most part.

chill. the sky is not falling.

Do you really think there weren't MIT/Harvard/Yale/Princeton etc people playing poker during the good times?

I hate when people correlate academic intelligence with poker skill. Although there is some correlation, it's not the be all and end all.

I used to play chess at a fairly high level and your comparison between chess and poker doesn't hold water. Chess is a game of complete knowledge, poker isn't, the two cannot be compared. All it takes is computing power to calculate every possible move in chess, there are no variables or unknowns. Once you calculate all the moves chess becomes tic-tac-toe. The same cannot be said for poker which has both unknowns and variables. Even with infinite computing power one cannot "solve" poker like you can chess.

Hi everyone, Kevin here (founder of the MIT Pokerbots Competition). There's some clear misconceptions about the competition and I just wanted to clarify a few points:

• We are NOT a class that teaches you how to write bots to play on online sites. We explicitly make it clear to our students that this is illegal and not tolerated. None of the (quite modest) curriculum goes anywhere near porting student's bots for playing online.
• Rather, we are an AI competition that uses the game of poker (this year, it's NLHE 3-handed cash) as a medium to teach concepts relevant to trading. Because our competition is bot-vs-bot, our class is not geared towards beating humans; in fact, we thoroughly cover why that would be a huge mistake for our tournaments.
• Will's class is completely separate from ours - his is simply a great poker player teaching you how to play great poker.

Feel free to leave any questions or further concerns here, I'd be happy to answer them.

well the bots at ipoker can already beat 5/10 6max cash games and have taken a million+ out of the poker economy already.

sky keeps falling

well what happens when the top student designs the best bot and realizes that he can run it on the shady poker sites where they dont care about detecting bots and rips the poker community for 100k+. hes not gonna take the moral high ground stance if i know anything about human nature

HU LHE has something like 10^18 game states. With infinite computing power it most certainly could be solved.

no bot will ever be able to play profitably in deep stacked plo high stakes games. So there's always that, there are too many variables in a big plo game for any program to win, so many of the situations in plo are counterintuitive and requires complex thought patterns that a computer is not capable of. I could see hold'em dying because of a bot, but plo is as much an art as it is a science so I'd say that game is safe.

lol MIT poker players are all probably ******ed.. last i heard was you actually had to be competent at poker to make a competent bot.

I've had a hard time believing we can shoot a missle down an air exhaust shaft in a house, but we can't shoot sandwiches at hungry people with similar technology.

I also have a hard time believing that they will develop technology able to make the computations in real time to beat NLHE pervasively (we all know some bots are out there making cheddar, but we don't have computers yet able to solve NLHE games completely), and yet the poker sites won't have the technology needed to ban this type of technology from their sites.

It's not in their business interests to allow the most popular game worldwide to completely degrade online. If they can stop me from accidentally leaving an ICM calculator open when I open PokerStars then they can figure out a way to block or ban those who use bots (or remote viewing, computer to computer hook-ups, KVM switch hook-ups, collusion, etc.). For every way to cheat online there is a way to block it...it's just a matter of cost versus benefits for the companies.

Unfortunately, the sites will likely wait until these things become a real problem before acting on them.

I think we have a few years before bots are 6bing with napkins pre based on reads and info from past hands on a certain player.

FTW.

Most of the MIT folks inclined to take poker seriously will quickly learn the math of the game to a very high degree. And of those so inclined, some will have computer programming skills, the ability to develop the skills, or know a classmate who has the skills, to create bots superior to those existing today.

Sure, there have been bright people who have developed bots to date. But MIT is the nuts in terms of intelligent young people. These classes are not positive EV for most players IMO.

Surprised everyone is so negative about this! I think its really exciting to advance the theory of the game. It will never be beaten. As Baluga Whale said in a Bart Hanson interview, on a macro level, poker is just a big game of rock/paper/scissors. Every strategy has a counter strategy. The trick is to be ahead of what everyone else is doing.

The idea of a world wide bot tournament is way overdue. We'd learn a lot from it.

I agree with this 100% after playing with it for a little bit you should be able to pick up on its patterns. Its the same as playing video games. We all have played video games growing up and we would get better every time we play If your like me and love to beat video games then this bot shouldnt be any different.

This MIT hysteria is laughable. It is just absurd the notion that nobody in the history of mankind has any remote idea how to program a bot but now omgz nooooez MIT realise that poker exists and computers exist so now will start producing world crushing bots! I mean, Uni of Alberta have been working on bots for like 15 years but what do these dirty foreigners know? The MIT kids will crush Polaris / Hyperborean / whatever it is these days with 2 hours of coding after a night out drinking just for laughs in between their class projects of sending a man to mars and curing cancer!!! We are all doomed!!!

FWIW, I think an MIT poker botting seminar is great, knowledge-wise. It'll almost certainly add nothing to current botters' arsenal, and even if it does - fighting the acquisition or spread of knowledge is a pointless endeavor.

I'm posting here because people seem to be worried about potential MIT botters when the real problem is already gigantic and (IMO) unstoppable. The problem is very similar to chess, providing people stop bringing up silly strawmen ITT (no chess is not gambling, yes the programming task is entirely different, no it doesn't matter if poker or chess is provable).

Thing only thing that matters: when you have bots that play better than most people (and we already do), eventually nobody will give real-money action online to unknowns that could be copying a bot's play move for move. Don't believe that'll happen? Look at HU LHE right now; extrapolate.

People said this same thing about NLHE several years ago. Now, nobody would make that claim. Don't kid yourself, any decision you can logically make in a poker game can theoretically be deduced by computers, and the practical limit to what bots can do is always increasing. Number of variables??? Computers can keep track of every variable.

A super advanced bot would reference all prior hands in it's history, check every HUD stat, analyze every time you took a similar action, factor for the type of player you were playing against, consult your recent history and tilt probability, throw in selective randomization, and perfectly keep in mind stack sizes and commitment. All in seconds, every hand it plays, ever, plus much more. It would make profitable moves no human would ever dream up, given the sheer amount of data available to it in an instant. I agree we have many years, but the sky is indeed inching downwards. Still, I think action will dry up from mediocre bots and non-existant fish before this uber-bot exists.

Just to reiterate, CardRunners has nothing to do with the actual course, aside from providing some prizes to the top finishers in their home game.

what?

WTF does MIT have to say about this? lol

See:

Then STFU.

Kevin,

If any of the students want to take the next step, please encourage them to enter the annual Computer Poker Competition:

http://www.computerpokercompetition.org/

level or extremely naiive? im guessing the latter.

decision trees are shorter. bots often occupy 2 seats per table in msnl games. your art/science bs is absolute rubbish and generally the view of a degen with an overly romantic view on card games. if the bot programmers get involved in plo (and theres no reason why they wouldnt), it would take them far less time to program effectively.

Because plo is generally a game of trading small equity pieces, human error is far higher, bots in plo would crush at higher winrates than nl. I would assume the main reason plo 6m bots have not been designed is because the programmers would be rightly worried about being exposed and dealing with colluding regs bot hunting.

PLO bots have been playing on known sites for a long time already.

What am I missing here? Poker bots have been around for quite some time, and their existence is common knowledge to a lot of people who don't even play the game. Both of my grandmothers have warned me about poker bots, and neither of them know how to play the game.

Call me naive, but the fact that a reputable institution is finally showing the science behind hold em seems like it will separate the game from it's "dumb luck" stigma. A little wishful thinking tells me that might help us get it back sooner in the US.