Poker Bots
Current Roll: $5166.
I had the worst losing streak I've experienced in a long time. I worked my way to up $150 over about an hour and a half. Then, I hit a ton of big hands and lost it all. Lost the whole $150 in 15 minutes. It's the most angry and closest I've been to tilting in a while, so I quit. Quitting itself is a success I am proud of, that I was able to quit when I knew emotions would prevent me from playing my A game.
Slashot covered the possibility of poker bots today.
The article itself is ok, but most of the comments in slashdot are terrible. A few common misconceptions: Making a poker bot to beat pp $2/4 wouldn't be difficult at all. Nor would the creator need to be a good poker player. Most people suck at poker because they are impatient and don't understand the odds. For most people, playing bad poker is far more fun that playing good poker.
For someone at my level of computer science, all the difficult problems have been solved and published by University of Alberta's game group: http://www.cs.ualberta.ca/~games/poker/ A lot of their work - and I've read it all - deals with heads up but there is plenty enough information to create a very good bot at a full table.
I've implemented some screen scraping and event reading for party poker. Combine some ANN's, game theory, hand strength evaluator, and you could have a very strong bot. If I worked on it full time, it would take me max 3 months to have a winning bot, maybe less. I can't think of anyway that party poker would ever be able to detect that I'm using a bot. There are almost certainly bots out there thought I doubt they are widespread. They are certainly not all strong nor winning.
Someone from MIT claims to have implemted a bot with friends.
Reading his post it sounds like they implemented an 'expert system' for play, which is certainly the most simple, but also the least pure and least adaptable system to implement. Expert systems are the most basic but can still be powerful. Basically, expert systems have some hardwired plays they always make. Good players would be able to exploit the predicitability of expert systems.
Here's my prediction - there will be a technological battle between sites and bots. This battle has had a few skirmishes (i.e. party poker and winholdem). (For those that don't know, winholdem is a commercial program that will play your cards for you.)
These skirmishes haven't been pretty. There have been innocent casulties, including some reports of people having their accounts closed for running microsoft's calculator in the background. Poker sites will work hard to improve their defense and eliminate false positives, but I think there will always be false positives. People who did nothing wrong will be accused of cheating and their accounts may become forfeit.
I've already considered some of the things that poker sites can do to combat bots. PokerStars is already a lot more protected that party poker. This is because they don't write each action into a text box that is easily readable by another program, as party does. That doens't make it impossible to get the information, it just makes it non-trivial. I havn't specifically dealt with some of the issues that come up with creating a bot on poker stars, but there are plenty of code libraries out there to help capable programmers.
In summary, there's not a lot the sites can do. Given time and energy, the bots will probably win out. I remain hopeful the sites will come up with some technological savior, but since I understand the technological challenges, I remain skeptical. We'll see, and online poker is certainly a field I am considering working in someday.
Legally, there is little to no recourse for the poker sites, as Loic Dachary points out. Unfortunately, the same laws that protect poker bots protect valid and important rights of software developers and consumers themselves. It may be possible to carefully construct laws to protect against bots but leave other rights in check, though enforcing these laws would be difficult since Loic is completely right about the impossibility of detecting bots. I think Loic trivializes some of the options poker sites have to make life difficult for bots, but otherwise he is completely on the mark.
The future looks bleak for honest poker players who want to compete against other humans. I'd like to end this post on an optimistic note, but I just can't think of one.
I had the worst losing streak I've experienced in a long time. I worked my way to up $150 over about an hour and a half. Then, I hit a ton of big hands and lost it all. Lost the whole $150 in 15 minutes. It's the most angry and closest I've been to tilting in a while, so I quit. Quitting itself is a success I am proud of, that I was able to quit when I knew emotions would prevent me from playing my A game.
Slashot covered the possibility of poker bots today.
The article itself is ok, but most of the comments in slashdot are terrible. A few common misconceptions: Making a poker bot to beat pp $2/4 wouldn't be difficult at all. Nor would the creator need to be a good poker player. Most people suck at poker because they are impatient and don't understand the odds. For most people, playing bad poker is far more fun that playing good poker.
For someone at my level of computer science, all the difficult problems have been solved and published by University of Alberta's game group: http://www.cs.ualberta.ca/~games/poker/ A lot of their work - and I've read it all - deals with heads up but there is plenty enough information to create a very good bot at a full table.
I've implemented some screen scraping and event reading for party poker. Combine some ANN's, game theory, hand strength evaluator, and you could have a very strong bot. If I worked on it full time, it would take me max 3 months to have a winning bot, maybe less. I can't think of anyway that party poker would ever be able to detect that I'm using a bot. There are almost certainly bots out there thought I doubt they are widespread. They are certainly not all strong nor winning.
Someone from MIT claims to have implemted a bot with friends.
Here's my prediction - there will be a technological battle between sites and bots. This battle has had a few skirmishes (i.e. party poker and winholdem). (For those that don't know, winholdem is a commercial program that will play your cards for you.)
These skirmishes haven't been pretty. There have been innocent casulties, including some reports of people having their accounts closed for running microsoft's calculator in the background. Poker sites will work hard to improve their defense and eliminate false positives, but I think there will always be false positives. People who did nothing wrong will be accused of cheating and their accounts may become forfeit.
I've already considered some of the things that poker sites can do to combat bots. PokerStars is already a lot more protected that party poker. This is because they don't write each action into a text box that is easily readable by another program, as party does. That doens't make it impossible to get the information, it just makes it non-trivial. I havn't specifically dealt with some of the issues that come up with creating a bot on poker stars, but there are plenty of code libraries out there to help capable programmers.
In summary, there's not a lot the sites can do. Given time and energy, the bots will probably win out. I remain hopeful the sites will come up with some technological savior, but since I understand the technological challenges, I remain skeptical. We'll see, and online poker is certainly a field I am considering working in someday.
Legally, there is little to no recourse for the poker sites, as Loic Dachary points out. Unfortunately, the same laws that protect poker bots protect valid and important rights of software developers and consumers themselves. It may be possible to carefully construct laws to protect against bots but leave other rights in check, though enforcing these laws would be difficult since Loic is completely right about the impossibility of detecting bots. I think Loic trivializes some of the options poker sites have to make life difficult for bots, but otherwise he is completely on the mark.
The future looks bleak for honest poker players who want to compete against other humans. I'd like to end this post on an optimistic note, but I just can't think of one.
posted by mike at 1:19 PM
0 Comments:
Post a Comment
<< Home