Very interesting thread on 2p2 about NL bots at Full Tilt

[Zoozie]

You do not have permission to view link Log in or register now.

Only read the first post since the thread is growing rapidly and nothing really interesting comes up.

Zoozie

 

[mikepipe]

Very intersting indeed, thx Zoozie!

 

[catrina m]

I found this thread on full tilt, after I wen't in and found one of the bots from the 2+2 thread, 0_Drunkenboxer playing at a $1/$2 no limit table. I watched them play for a while. He took a long time to make minor poker descisions but it may be because he is playing multiple tables. Or he may be a bot. Its all very interesting.

Outdated URL (Invalid)

 

[Rusty]

What surprised me about many of the responses to that thread were the amount of supposedly decent poker players that stated they would easily exploit a poker bot.
I am certain many of them have lost money to bots and will never know it.
Poker is the perfect game to operate a bot because they ruthlessly exploit the Egos of their human opponents.
The only difficult part to get right when programming a bot is how to spot a bluff and when to make one the rest is basic maths.
All that needs to be done then is to input the data from players stats and a bot can come up with the right decission on bluffs more often than not.
It is also easy to put in a subroutine so that the bot plays differently on occasions and you could also make it adaptable to live data.
My guess is there are some very sophisticated bots out there perhaps many and that they would be very difficult if not impossible to detect.
It is 10 Years since Deep Blue beat Kasparov and as I remember his Ego would not allow him to believe he had been beaten fairly by a computer.
Now what does that remind me of?

 

[brutaldeluxe]

There is a fair amount of evidence to suggest he wasn't beaten fairly though.
Also he did beat Deeper Blue in a game so the computer is not infallible or unbeatable anyway. There's a good documentary about it all which tells you a few things that were never mentioned at the time.

 

[Rusty]

There is a fair amount of evidence to suggest he wasn't beaten fairly though.
Also he did beat Deeper Blue in a game so the computer is not infallible or unbeatable anyway. There's a good documentary about it all which tells you a few things that were never mentioned at the time.

Yes I saw that documentary it was fascinating.
What Kasparov felt was that he was being followed and/or spied on and felt intimidated.
He also felt there was some human intervention that was helping Deep Blue but he had no evidence of this and I think that it was just ego that felt bruised.
I know IBM had a lot riding on Deep Blue being successful and this has helped fuel the conspiracy theories.
The point I was trying to make is that desktop PC's are far more powerful than Deep blue ever was and chess programs so sophisticated that they are virtualy unbeatable by human players now.
For me it is obvious that a very sophisticated Poker program could and would outplay good players.
It would have no ego never get upset never lose concentration and never make a mistake on % plays. As the only way you can read an opponent online is by referencing their past play a bot could do this better than human too.
Damn, I might even program one myself

In

 

[brutaldeluxe]

Kasparov has a huge ego, no doubt about that. But it is justifiably huge you might say. As you say IBM shares sky rocketed after the result of that match.
Now why was Kasparov denied the computer logs from deeper blue ?
Also deeper blue made a move which allowed Kasparov to draw by a perpetual check (which he missed). Seems very odd that a computer with that much calculating power would let that happen. In fact it shouldn't happen as computers don't make errors like that. Again odd.
Why deny the logs if no human intervention ? How could a computer make a mistake like that - it shouldn't.
Also come the end they dismantled deeper blue rather than actually continue the project. Why was that ? Plus there were also other factors such as an entire team of chess experts and programmers studying all of Kasparov's games beforehand whilst Kasparov was not denied any access to any of the Deeper Blue program or any of it's games.

Chess computers now are indeed very good but they aren't unbeatable just yet. The thing is now you need the very very best humans to be able to even have a chance against them. Before even a reasonable chess player could beat most computers.

It'll be interesting to see how good poker bots can get. The thing is we may not always know as people may try to cover their existence up a little so they can get their bots playing. It does have some interesting factors that a chess program won't. How do you go about successfully bluffing a bot ? How will the bots be programmned to bluff ? If you know you are up against a bot can you employ an "anti-bot" strategy the same way people employ anti-computer chess strategy by playing closed positions that need positional play rather than raw calculating.

 

[ace4suited]

in terms of the chess related discussion I was reminded of an episode of Star Trek in which Spock is able to beat the computer at chess when he claims he should only be able to draw at best because computers are infallable.

The problem is that this is just not the case. A computer is only as infallable as the programers and can only go at the speed that its processor allows it to go at.

In the strictest sense the computer is not really thinking it is going through all the possibilities.

More crutially modern computers have a HUGE amount of data stored in them predominantly about openings but also i suspect a little with end game theories.

As such it is like playing an opponent with a library of books at the table. No human can keep 1000's of openings in their head to the degree that a computer can and as such it is not a fair test.

In terms of poker I am sure that bots do exist but equally I suspect it has more to do with bums on seats than anything more sinister. That and a few people who have been conned into buying dodgy software.

There is far more going on in terms of online poker than people give it credit for. There is lots that can be read into chat, pauses etc and even egos. It is 90% about reading people and this is very hard for any programer to accurately code. It would be fairly easy i am sure to program a computer to play basic solid poker and it would be possible in ring games to probably have +ve EV (given the number of mistakes that people make )from doing this but this is totally different from becoming a master at the game.

 

[Rusty]

It is a fascinating subject and I guess what we have to decide is, just that, how we decide.
As I said online there no tells as such excluding the time it takes someone to bet therefore the only way we can decide how strong or weak an opponents hand is;
Position
stack size
bet size
other bets
play history
etc
A computer program can calculate all these more efficiently than a human player so if you accept that it is not that difficult to program a very sound poker stratergy then it follows a poker bot could be very good indeed.
Also bear in mind a decent program could recall every play any and every player had made a feat beyond human players.

As for Deep Blue and the conspiracy theory I certainly would not rule it out but Kasparov's ego is so huge it is hard to know the true facts although watching that documentary one could not help but feel a little sorry for him.

 

[Simmo!]

You do not have permission to view link Log in or register now.

The video above gives some background & analysis on the Kasparov vs Deep Blue games for those who weren't aware. It's rough-cut for the first minute or two but worth sticking with.

 

[Rusty]

You do not have permission to view link Log in or register now.

The video above gives some background & analysis on the Kasparov vs Deep Blue games for those who weren't aware. It's rough-cut for the first minute or two but worth sticking with.

Yeah that is some choice cuts from the said documentary.
Some interesting comments on the subject there as well.
Thanks for that Simmo