Jump to content

Man Vs Machine Update...


Recommended Posts

I understand that argument, and I'm not saying it's wrong, but I don't completely agree either. Analysing betting patterns and finding good places to bluff is not impossible to write into a computer program. Making marginal bets and calls after pinning down your opponent's hand ranges while playing mathematically perfect is not impossible to do for a computer program. Something that probably is very difficult to incorporate are psychological factors, but most other aspects of poker can be done, imo. A program doesn't need to get everything right to have an edge against most players.Also, I'm not good at chess, but it is a game where creativity and intelligence play a role, it is not just about making "the right move" and thinking a couple moves ahead.I guess my point is: You don't need a real AI for a poker program to be better at that game than a vast majority of humans, that it is just a matter of programming (and not even a very difficult one).Humans playing poker do that too!
That's an interesting point, unfortunately, just about all experts on programming and game theory disagree with you. I'm too lazy to find citations, but I believe there are some interviews with leading researchers on the topic in the book Positively Fifth Street (which admittedly is out of date info).
Link to post
Share on other sites

What would be necessary for a consistently winning pokerbot: the ability to detect the opponents' betting patterns and adjust to them, the ability to change gears, if you will. Humans spot patterns on an unconscious level. So far as I know, no computer can do that. Pattern recognition still must be explicitly taught to a computer.If the science of neural networking ever achieves its full flower, it will be possible to build a program that will adapt and adjust to players' patterns and the shifts therein. Will the computer be able to be creative and "switch gears" on its own when the time is right? Maybe, maybe not. Perhaps a computer that plays really well and changes gears at random would be good enough to beat the best.

Link to post
Share on other sites
Do you assume the range of your opponents? Or do you teach it to gather intelligence every time there is a show down?Phil Ivey starting range is different to that of Phil Laak. Do you feed the software with pre collated data on each player? But doing this you are not letting the computer 'play' but instead take educated guesses at what it should do.
wouldn't they just target tight starting hands and then adjust accordingly to the person's VPIP and showdowns?
Link to post
Share on other sites

One thing that's important to remember is that this study was only on heads up limit hold'em ... probably the easiest of all poker variants to program a bot for.Granted, the humans barely squeeked out a win, but I'd like to see a bot take on a full ring game, or a NL game. That would be a much more difficult task.

Link to post
Share on other sites
he got lucky this round
You do realise that the same hands are played in reversed roles in the other match?ie In one room Phil plays one hand against the computer in the other room Ali plays the same hand against the computer but with the opposite cards.To say 'he got lucky' is a jopke
Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...