"Bots" vs humans, get ready to be trounced
Dennis
players you might want to check this out
http://www.rci.rutgers.edu/~cfs/472_html/Intro/ChessContents.html
The strengths that poker players compaired to computers have, reading
other players, recognizing bluffing opportunities as they arise etc. are
greatly diminished online. On the other hand, the strengths of computers
are magnified online. They never tire, never make mistakes, instantly
calculate odds and more. The above link has some fascinating articles on
the subject.
_________________________________________________________________
Posted using RecPoker.com - http://www.recpoker.com
VooDoo Chile
competitive backgammon.
I was purchasing one of the neural-network base BG programs from him. The
neural network
BG programs are so good, you would swear they cheat with the dice. They
have literally changed the
championship level game of BG. Of course, the programs (or the programmers)
couldn't explain what
the programs did, being as they were neural networks. The person I was
speaking to mentioned that
they were going to try the same technique with poker.
I don't think it will work. The game of BG is totally static. There is
ALWAYS a best move for a
particular board position and dice roll, and it is ALWAYS the same
(excluding scoring considerations
in a match). This is totally untrue in poker. Championship level poker
demands two things which I
wonder whether the computer will be capable of: Varying your own play and
(far more difficult), being
aware of your opponents play. I think "bot" play will be beatable based on
the predicatability of play...
Dennis <dls9...@yahoo.com> wrote in message
news:3d334643$0$68544$9a6e...@news.newshosting.com...
Dennis
this subject comes up, what is missing in the conversation is that
computers are still in their infancy. We are light years ahead of where we
were 5 years ago and few people fully appreciate where we will be 5 years
hence.
James Martin, the inventor of voice recognition technology and a leading
thinker in the field predicts that in 5 years time IBM will have a
computer that will be indistinguishable from a very bright person when you
talk to it on the phone. It will be creative, flexible, adaptive, much
brighter than we are and it will have instant access to all information
via the web. Tireless, mistake free, it will answer the phone for you and
make business decisions better than you can. And that is just the begining.
If you are interested here is a link to an article about it;
http://www.freep.com/money/tech/mwend9_20010809.htm
regards, Dennis
VooDoo Chile wrote:
> I dunno about this. I was speaking recently to someone involved with
> competitive backgammon.
> I was purchasing one of the neural-network base BG programs from him. The
> neural network
> BG programs are so good, you would swear they cheat with the dice. They
> have literally changed the
> championship level game of BG. Of course, the programs (or the programmers)
> couldn't explain what
> the programs did, being as they were neural networks. The person I was
> speaking to mentioned that
> they were going to try the same technique with poker.
>
> I don't think it will work. The game of BG is totally static. There is
> ALWAYS a best move for a
> particular board position and dice roll, and it is ALWAYS the same
> (excluding scoring considerations
> in a match). This is totally untrue in poker. Championship level poker
> demands two things which I
> wonder whether the computer will be capable of: Varying your own play and
> (far more difficult), being
> aware of your opponents play. I think "bot" play will be beatable based on
> the predicatability of play...
_________________________________________________________________
Clfranck01
>Championship level poker
>demands two things which I
>wonder whether the computer will be capable of:
If a computer can beat the world's best chess player,
it can play damn good poker, I'm sure. In fact, any
average PC would be a good platform since poker is
not at all that computationally challanging compared
to chess. Deep Blue had special hardware to chew
through moves.
Once you move from online, however, I'd say it would
still do extremely well at limit play, but would probably
get its ass kicked in expert NL play.
Craig Franck
clfra...@aol.com
Cortland, NY
Tom Weideman
> If a computer can beat the world's best chess player,
> it can play damn good poker, I'm sure.
I can't imagine how you could conclude this. Computers are "naturals" when
it comes to games of complete information like chess, where they can crank
through the plies. Games of incomplete information like poker are a bit
tougher to treat. For one thing, one would need to decide whether to have
the program play exploitively or optimally, and both of these approaches
require immense amounts of additional baggage.
> In fact, any
> average PC would be a good platform since poker is
> not at all that computationally challanging compared
> to chess. Deep Blue had special hardware to chew
> through moves.
Do you have any idea how much "chewing" a computer would have to do in order
to play poker in a similarly exhaustive manner?
> Once you move from online, however, I'd say it would
> still do extremely well at limit play, but would probably
> get its ass kicked in expert NL play.
I'm always amazed when I see people assert this (as I have seen many times
on rgp), because it is exactly the opposite of what is true.
Tom Weideman
JT Autry
you are speaking of the type of program that allows the computer to "learn".
If there is always a best move for a particular roll with a particular board
position, which are finite in number, then what advantage would a neural-net
program have? There is nothing to learn.
JT
"VooDoo Chile" <notana...@nospam.com> wrote in message
news:q5JY8.45888$6r.14...@news4.srv.hcvlny.cv.net...
Nat Silver
> Clfranck01 wrote:
>
> > If a computer can beat the world's best chess player,
> > it can play damn good poker, I'm sure.
That has to be the case.
> I can't imagine how you could conclude this...
> Do you have any idea how much "chewing" a computer
> would have to do in order to play poker in a similarly
> exhaustive manner?
It doesn't have to be in an exhaustive manner.
There are only 7 cards visible in holdem and
only 9 in Omaha. IMHO, experts like,
say, Caro and Sklansky working together
could create programs that would win
consistently at these games in all but the toughest
competition. Note that in the instance of
randomizing ones play of particular holdings, the
computer could call a random number to determine
its action. Moreover, the program could have code
allowing it to adjust to the weaknesses of individual
players. But even without that, IMHO, it would win.
Note that bots have been winning on-line for years.
> > Once you move from online, however, I'd say it would
> > still do extremely well at limit play, but would probably
> > get its ass kicked in expert NL play.
NL would be cut-and-dried for Caro and Sklansky.
It's pot limit that might be the difficult game.
A. Prock
>Tom Weideman wrote:
>> Clfranck01 wrote:
>>
>> > If a computer can beat the world's best chess player,
>> > it can play damn good poker, I'm sure.
>
>That has to be the case.
Proof by assertion. My favorite.
>> I can't imagine how you could conclude this...
>> Do you have any idea how much "chewing" a computer
>> would have to do in order to play poker in a similarly
>> exhaustive manner?
>
[snip some speculative theories about bot capabilities]
>Note that bots have been winning on-line for years.
Really? How So?
- Andrew
Nat Silver
> Nat Silver wrote:
> >Tom Weideman wrote:
> >> Clfranck01 wrote:
> >> > If a computer can beat the world's best chess player,
> >> > it can play damn good poker, I'm sure.
> >That has to be the case.
> Proof by assertion. My favorite.
Chess is much more difficult and unforgiving
than poker. I am not pretending to have proof.
> >> I can't imagine how you could conclude this...
> >> Do you have any idea how much "chewing" a computer
> >> would have to do in order to play poker in a similarly
> >> exhaustive manner?
> [snip some speculative theories about bot capabilities]
Not really speculative theories.
> >Note that bots have been winning on-line for years.
The UNLV bot, r00lie, destroyed IRC players in all limit
games. I know that bots are playing for real money
now at several sites but do not have details. Since I
have no hard information to present, I ask you to
consider what experts like Caro and Sklansky
would say, if they want to air their views.
Nat Silver
> Nat Silver wrote:
> >> Clfranck01 wrote:
> >> > If a computer can beat the world's best chess player,
> >> > it can play damn good poker, I'm sure.
> >That has to be the case.
> Proof by assertion. My favorite.
> [snip some speculative theories about bot capabilities]
Consider these factors:
1. Playing poker (for long periods of time) is tedious and boring.
As a result, those who play too long eventually will lose
concentration, perspective and maybe their bankrolls.
2. Humans lose the most money (and are the most
vulnerable) when they go on tilt.
3. Most hands do not require creative play to be
profitable, just sound play.
4. In most games, all trouble hands can be folded
without significant losses.
Computer programs would be immune to 1 and 2
and consistent on 3 and 4.
wg bradley
"A. Prock" <jeffy...@yahoo.com> wrote in message
news:3d339a20$0$1056$8026...@spool.cs.wisc.edu...
> >Note that bots have been winning on-line for years.
>
> Really? How So?
Here's a chart of Pokibot's results on IRC:
http://spaz.ca/aaron/poker/index.html
Background is here:
http://www.cs.ualberta.ca/~games/poker/
T. Chan
wrote:
>Here's a chart of Pokibot's results on IRC:
>
>http://spaz.ca/aaron/poker/index.html
IIRC, Andrew was around IRC for quite a while, and I'm sure he knows
about the existence of Poki. While it's true that the IRC bots have
done well against play-money players, based on my (very, very) limited
experience with them, I think they would get killed in a typical
online 10/20. I wouldn't even back one in a 3/6.
--
Self-indulgent homepage warning:
http://www.sfu.ca/~tchand/
wg bradley
would be completely flabbergasted if Poki couldn't beat the 3-6 games on
Paradise. It would be an interesting experiment to be sure though.
"T. Chan" <terren...@telus.net> wrote in message
news:coe7ju075sfplsmme...@4ax.com...
Dennis
> Not sure about the 10-20's because I just haven't played them enough, but I
> would be completely flabbergasted if Poki couldn't beat the 3-6 games on
> Paradise. It would be an interesting experiment to be sure though.
In the near future I fear Poki will look like the old game Pong compared
to the new bots.
Octo the Genarian
really matter if someone designs a bunch of perfect bots and sticks them on
Paradise. Why? Because when you find yourself in a room full of perfect
poker players who do not make conversation and play is if they know what
each other has, you will realize that the game is not worth playing and
leave.
It's analogous to cheating in a casino. Sure, there may be a hidden camera
and your opponent may know what your cards are, but if you are a normal
person, you will quickly realize that this game is not for you. Did every
abandon professional baseball in droves when the world series was rigged?
No. You becomes a little more cautious, a little more cynical, and you move
on.
"Dennis" <dls9...@yahoo.com> wrote in message
news:3d334643$0$68544$9a6e...@news.newshosting.com...
Asha34
human. I'm thinking of this in a b&m poker room more than online. I'd suggest
a gas mask. There's no way that even a perceptive opponent can tell whether you
are in fact human when you're wearing the large, military gas mask, whether you
are breathing or just pretending to breath.
Ashley
Manny
> This I don't understand. I assume that when you speak of neural-networks
> you are speaking of the type of program that allows the computer to "learn".
> If there is always a best move for a particular roll with a particular board
> position, which are finite in number, then what advantage would a neural-net
> program have? There is nothing to learn.
>
> JT
Because instead of having to program the 8 trillion things the program
COULD do, a neural net selects the best things to do from the set
which does not contain the 7.9 trillion things that are clearly wrong.
It can place itself in that situation billions of times and "score"
the outcome from each one of the moves that it chooses.
That's at least the most basic idea behind NN's (give it a situation,
let it make it decision, rate the decision with an objective function,
and repeat till you're convinced it has seen every situation you could
care about). Problems arise in exactly how one chooses the things in
the set you're going to test. You can rarely be convinced that you
have found the optimal solution for a lot of problems they use neural
nets on. Some problems lend themselves much more nicely to it.
Nat Silver
> It would be interesting to see how "bots" could be disguised to appear to
be
> human. I'm thinking of this in a b&m poker room more than online. I'd
suggest
> a gas mask. There's no way that even a perceptive opponent can tell
whether you
> are in fact human when you're wearing the large, military gas mask,
whether you
> are breathing or just pretending to breath.
> >What annoys me about this "The bots are coming!" hysteria is that it
doesn't...
not hysteria
> >really matter if someone designs a bunch of perfect bots and sticks them
on
> >Paradise. Why? Because when you find yourself in a room full...
not a room full
> > of perfect poker players who do not make conversation...
Conversation would not be a problem for one or two bots at a table
but lack of conversation is irrelevant and not a necessarily indicative of a
bot.
> > and play is if they know what each other has, you will realize that the
game
> > is not worth playing and leave. ...
We're not discussing collusion here. Online players collude effectively now.
> >It's analogous to cheating in a casino. Sure, there may be a hidden
camera
> >and your opponent may know what your cards are, but if you are a normal
> >person, you will quickly realize that this game is not for you. Did every
fan
> >abandon professional baseball in droves when the World Series was rigged?
> >No. You becomes a little more cautious, a little more cynical, and you
move
> >on.
You are trying to minimize what effect the presence of bots would have in a
game that you played in. You seem upset and maybe a little hysterical over
the
prospect of playing against bots, really.
Nat Silver
> It would be interesting to see how "bots" could be disguised
> to appear to be human. I'm thinking of this in a b&m poker
> room more than online. I'd suggest a gas mask. There's no way
> that even a perceptive opponent can tell whether you are in fact
> human when you're wearing the large, military gas mask, whether you
> are breathing or just pretending to breath.
> >What annoys me about this "The bots are coming!" hysteria is...
not hysteria
> > that it doesn't really matter if someone designs a bunch of
> > perfect bots and sticks them on Paradise...
You're whistling inthe dark, here.
> > Why? Because when you find yourself in a room full...
not a room full
> > of perfect poker players who do not make conversation...
Conversation would not be a problem for one or two bots at a table
but lack of conversation is irrelevant and not a necessarily indicative of a
bot.
> > and play is if they know what each other has, you will realize that the
> > game is not worth playing and leave. ...
Aren't you confusing player collusion with bots? This is not about
collusion.
Online players collude effectively now.
> > It's analogous to cheating in a casino. Sure, there may be a hidden
> > camera and your opponent may know what your cards are, but if
> > you are a normal person, you will quickly realize that this game is
> > not for you. Did fans abandon professional baseball in droves when
> > the World Series was rigged? No. You become a little more cautious,
Steven Green
......if someone designs a bunch of perfect bots and sticks them on
Paradise
Can anyone explain to me how you can implement a bot on a poker site?
How will it "push the buttons"
Thanks,
Steven Green
"Octo the Genarian" <inv...@nomail.com> schreef in bericht
news:ah0kle$sb5$2...@geraldo.cc.utexas.edu...
Nat Silver
> ......if someone designs a bunch of perfect bots
> and sticks them on Paradise
> Can anyone explain to me how you
> can implement a bot on a poker site?
> How will it "push the buttons"
Ah, you have conceded the main point
and now are questioning the reach of
programming techniques.
Although not a programmer, I have come
upon some enlightening sources, which
are given below.
The template of a stationary card table and
its action buttons is not difficult to
duplicate. Hence a accomplished programmer
can provide mouse commands for various
locations on the screen using screen coordinates.
One source that refers to computers reading
templates and making autobets as a result is
in Steven Skiena's Calculated Bets:
www.amazon.com/exec/obidos/search-handle-form/104-0518258-9831907
an example of computers reading screens in real time poker play:
http://softwaredev.earthweb.com/sdtech/article/0,,12065_616221_1,00.html
Brad Singer
> Can anyone explain to me how you can implement a bot on a poker site?
> How will it "push the buttons"
That's not so difficult. You can write a program that finds the poker
program's window, matches bitmaps to find out what the cards are, does
some OCR (Optical Character Recognition) to read the bet amounts and
manipulates mouse events to push the buttons.
In order to prevent spammers from registering in AOL groups, AOL prints
a registration code on the screen that has been optically warped to
prevent OCR from recognizing it, so there's proof that spammers have
been using OCR this way.
If the poker program randomly shuffles the position of the buttons, the
bot can just read their labels off the screen, unless they have been
optically warped a la AOL. I know something about OCR and I believe
that AOL style text warping can be overcome. But at least that changes
what amounts to a two week software project to build a bot into a much
larger OCR problem that fewer people are qualified to solve.
Given that a bot has been used on IRC poker, I would be interested to
know what is running on online poker today. Someone might even be using
a program for all their computation and doing the input/output manually.
This is much simpler than a chess program by far. Chess programs also
randomize their moves, otherwise they would open the same way every
time. If opponents could count on a chess program always responding the
same way in the same situation, they could just run a huge amount of
offline (i.e., before the game) analysis to be prepared for it.
One of the advantages of a poker bot is that it could run much more
accurate probability calculations than are ever used by a human player
and never make a mistake or take an uncalculated risk. Also, the plays
that are referred to as "moves" (various kinds of bluffing, slow
playing, inducing bluffs, etc.) can be exactly quantified using
probabilistic game theory. Moves are the part of poker where one can
really use game theory, part of it much like the "he thinks that I think
that he thinks that I think, etc." found in scissors, paper, stone.
Add to this the extra knowledge that comes with mult-bot collusion and
any perceived human advantage (although I cannot imagine what that might
be) is overwhelmed by the odds. Trying to bet someone out of the pot
with a multiplayer bluff just becomes another "move".
Then again, what will happen when one group of bots comes up against
another group of bots? It might be a game of musical chairs to see
which side gets most seats at the table. :-)
I am sure there have been some hard feelings in Las Vegas occasionally
about who gets access to the suckers. I was actually invited one time
by a professional to take part in a pot limit hold-em game that was set
up specifically to sucker one tourist. There was no collusion of any
kind, but there was also very little action between the two pros and
myself. The sucker went back to the ATM machine several times and in a
period that couldn't have been more than an hour and a half we cleaned
him out of several thousand, all quite legitimately.
brad
A. Prock
>> Nat Silver wrote:
>> >Tom Weideman wrote:
>> >> Clfranck01 wrote:
>> >> > If a computer can beat the world's best chess player,
>> >> > it can play damn good poker, I'm sure.
>> >That has to be the case.
>> Proof by assertion. My favorite.
>
>Chess is much more difficult and unforgiving
>than poker. I am not pretending to have proof.
Really? I assert that poker is more difficult and
unforgiving than chess.
How's that for a parry?
>> >> I can't imagine how you could conclude this...
>> >> Do you have any idea how much "chewing" a computer
>> >> would have to do in order to play poker in a similarly
>> >> exhaustive manner?
>
>> [snip some speculative theories about bot capabilities]
>
>Not really speculative theories.
>
>> >Note that bots have been winning on-line for years.
>
>The UNLV bot, r00lie, destroyed IRC players in all limit
>games.
Really? That wasn't the impression that I got. r00lie
made some money from the 10/20 games, and made little
(or lost) in the 20 40 games.
>I know that bots are playing for real money
>now at several sites but do not have details.
I see.
>Since I
>have no hard information to present, I ask you to
>consider what experts like Caro and Sklansky
>would say, if they want to air their views.
Neither of these two really have much interesting
to say about programmed poker players. Sklansky
knows nothing about programming, and every time I've
prodded Caro for anything interesting, he come up
short.
- Andrew
Nat Silver
> Really? I assert that poker is more difficult and
> unforgiving than chess.
Okay. I guess we would have to define what
we mean when we say such a thing. An inferior
move usually ends a game of chess. There is more than
one way of playing a poker hand, leading to comparable
outcomes. As you know, the long run is is very important.
One can lose many hands and still show a profit.
> How's that for a parry?
Not bad.
> Really? That wasn't the impression that I got. r00lie
> made some money from the 10/20 games, and made little
> (or lost) in the 20 40 games.
If that is so, it wasn't in the long run. Since 20-40 games
required bankrolls usually large enough for pot limit, they
did not run all the time as did 10-20.
> > Since I have no hard information to present, I ask you
> > to consider what experts like Caro and Sklansky
> >would say, if they want to air their views.
> Neither of these two really have much interesting
> to say about programmed poker players. Sklansky
> knows nothing about programming, and every time I've
> prodded Caro for anything interesting, he come up
> short.
Well, the best Bridge programs are stronger than 99% of
all bridge players. Below are several outdated links:
http://www.jimloy.com/bridge/review.htm#strongest
http://www.cs.umd.edu/~nau/bridge/bridge.html
Anyway, I don't feel we are in much disagreement here.
If you think that there are few, if any, pokerbots winning,
probably you do realize that it would be a matter of time
before they do.
Scott Marks
> Steven Green wrote:
>
> an example of computers reading screens in real time poker play:
>
> http://softwaredev.earthweb.com/sdtech/article/0,,12065_616221_1,00.html
Just a small correction, we did not do any screen reading in the above
referenced URL. We exploited a flaw in PRNG that provided the seed for
the shuffling of the deck and were able to determine what the other
players' hands were and what the turn and river cards would be.
Scott...
Bing
<inv...@nomail.com> wrote:
> What annoys me about this "The bots are coming!" hysteria is that it doesn't
> really matter if someone designs a bunch of perfect bots and sticks them on
> Paradise. Why? Because when you find yourself in a room full of perfect
> poker players who do not make conversation and play is if they know what
> each other has, you will realize that the game is not worth playing and
> leave.
But, you see, the difference here is that perfect human
players do care if you catch on to them and leave. They
spent years working on their act and it will affect
them to be found out. If enough players wise up to the
perfect player, that player may have to pack up and
go elsewhere or change his ways.
A bot wouldn't care. It would be happy enough to just sit
there and wait until some other sucker comes along.
Similarly, that perfect player hopes to sit with the
victim for many hands and hours in a row, waiting for
the long run to hand him his profits. The bot is more
than happy to play just 3-4 hands with the victim.
The bot will be equally ready to play at 1 PM or 1 AM
and won't mind playing 300 hands or just 3. You can't
just wait for it to get bored and leave the casino.
-- Bing Monopoly Expansion Set
Visit us at http://www.paxentertainment.com
A. Prock
>
>> Really? That wasn't the impression that I got. r00lie
>> made some money from the 10/20 games, and made little
>> (or lost) in the 20 40 games.
>
>If that is so, it wasn't in the long run. Since 20-40 games
>required bankrolls usually large enough for pot limit, they
>did not run all the time as did 10-20.
In the "Good Ol' Days" (tm) 20/40 was pretty much a 24/7
game, and it was the 50/100 that had a hard time getting
off every day. But maybe I'm miss-remembering things.
Regardless, R00lbot got in about four years worth of
play. I don't know if those were 1000 hour years or 3000
hour years, but I'm sure that whatever it was rivals what
most recreationaal players might play in a lifetime. I'm
sure that a significant portion (25%?) was at the 20/40
level.
- Andrew
Bing
> ......if someone designs a bunch of perfect bots and sticks them on
> Paradise
> Can anyone explain to me how you can implement a bot on a poker site?
> How will it "push the buttons"
That's not too difficult. Recall that the screens and buttons
you see on your monitor are just part of the poker program
you downloaded. When you press a button, the program sends
a specific signal to the main computer server in Antigua. The
signal travels from the program to your operating system and
then through your modem and cables. Similarly, when that poker
server sends your poker program a signal describing your cards,
other players' actions, etc., that signal also travels across
those cables.
It would be easy to write a bot program that intercepts these
signals and acts upon them. So, if the poker server sends your
computer the signals for "four of spades" and "seven of hearts",
your bot program would call up a database and determine that the
best thing to do would be to fold. So, it then sends a signal
down the cable that is the same one that would have been sent
if you had pressed the "fold in turn" button.
Obviously, this is a very simple example. A good bot would have
a huge database of the best thing to do in any situation (probably
initialized by experts). It would also track these "best"
decisions and determine which ones made money and which didn't.
It would adjust future decisions.
It might also track the plays made by other players. Sophisticated
programs could be used to anticipate what a specific player might
do. The options are only limited by the level of effort that the
programmers want to put into the project.
Nat Silver
> Nat Silver wrote:
> > an example of computers reading screens in real time poker play:
> > http://softwaredev.earthweb.com/sdtech/article/0,,12065_616221_1,00.html
> Just a small correction, we did not do any screen reading in the above
> referenced URL. We exploited a flaw in PRNG that provided the seed for
> the shuffling of the deck and were able to determine what the other
> players' hands were and what the turn and river cards would be.
Scott, that was a big correction. I stand corrected.
Now that you have responded in this thread,
would you care to offer your opinion as to the
feasibility or current state of winning pokerbots?
Dennis
<thoughtful comments snipped>
> programs could be used to anticipate what a specific player might
> do. The options are only limited by the level of effort that the
> programmers want to put into the project.
Yes, and a sophisticated program that had access to a specific player's
hand history would be devastating to that player.
Scott Marks
>
> Scott, that was a big correction. I stand corrected.
> Now that you have responded in this thread,
> would you care to offer your opinion as to the
> feasibility or current state of winning pokerbots?
I'm not aware of anyone that is currently doing this. However, I
wouldn't be surprised if someone was using a bot to assist themselves or
play others outright.
I would be more worried about running into a well-designed bot on a
heads-up or 5-player max table, as opposed to a full 9 (or 10) handed
table because of the difficulty involved in programming a bot to play
well in a full game.
There's probably people using bots (or other player assistance tools) in
the online rooms now. I guess it's perhaps just another risk of playing
online, which I do, BTW.
Scott...
William Loughborough
> James Martin, the inventor of voice recognition technology and a leading
> thinker in the field predicts that in 5 years time IBM will have a
> computer that will be indistinguishable from a very bright person when you
> talk to it on the phone.
Voice recognition technology wasn't "invented" by James Martin and in fact
is older than I am (76 years). The "seers" at places like IBM have been
predicting such things as voice dictation recognition software for as long
as I can remember and it was always "6 months away" - nice to see that
they've gotten real enough to extend it to 5 years. In either case it's
vapo(u)rware of the worst kind. The Ray Kurzweils of this technology are
able to pontificate about this stuff, but it's still bullshit.
Love.
It's bad luck to be superstitious
William Loughborough
>
> Really? How So?
>
> - Andrew
Why they adopt names like Brock that rhyme with Spock and get in their
time machines and go forward to when the upcoming cards (which the dealing
algorithm didn't even know!) have been dealt and then come back in time,
remembering the outcomes, which they then change by their actions. Then
they spend all their time answering absurd paranoid postings about it
being impossible to beat online poker. They have the advantage over
someone like me, whose creeping senility prevents me from remembering what
cards are coming.
If they program it to properly select games, they don't need much of a
'bot because many spots online are so soft that it's pathetically easy to
win. If they play REALLY GOOD, you'd just avoid them - which is what you
do for really good human players if you don't want to waste your time when
there are so many really terrible players online.
Dennis
I am dictating this sentence using "ViaVoice", a software and hardware
package that costs less than $100 and is amazingly proficient at
understanding and dictating my voice. And, the more I use it, the better
it gets. Stop by any computer store to check it out.
As for your other totally unfounded, unsuported and untrue statements... I
just have to wonder what your motivation was.
William Loughborough wrote:
> Voice recognition technology wasn't "invented" by James Martin and in fact
> is older than I am (76 years). The "seers" at places like IBM have been
> predicting such things as voice dictation recognition software for as long
> as I can remember and it was always "6 months away" - nice to see that
> they've gotten real enough to extend it to 5 years. In either case it's
> vapo(u)rware of the worst kind. The Ray Kurzweils of this technology are
> able to pontificate about this stuff, but it's still bullshit.
_________________________________________________________________
Keith Ellul
particular site uses so that your bot doesn't have to go through the
UI?
Shuffling the positions of buttons is a BAD IDEA. I accidentally click
the wrong button enough as it is.
Keith
JT Autry
JT
"Keith Ellul" <kbe...@fe02.math.uwaterloo.ca> wrote in message
news:Pine.SOL.4.44.020716...@fe02.math.uwaterloo.ca...
Don shennum
> A. Prock wrote:..
>
> > Really? I assert that poker is more difficult and
> > unforgiving than chess.
>
> Okay. I guess we would have to define what
> we mean when we say such a thing. An inferior
> move usually ends a game of chess. There is more than
> one way of playing a poker hand, leading to comparable
> outcomes. As you know, the long run is is very important.
> One can lose many hands and still show a profit.
>
Competitively?
An inferior move does not end a game..it's the superior play of the
opponent to take advantage of the inferior play. And, there are many
more branches in a chess game than a poker game. The ebb and flow of
plans within a game is enormous. One plan can "fail," but you can
recover, play defensively, and hopefully wait for an opportunity to
regain at least equilibrium.
Had conversations with both Master-level Chess players and poker
players. The two lend themselves to each other. It's the belief that
you are in control of the situation with the information presented to
you, and thus exert your will on the situation...that's a trait shared
by master poker and chess players alike.
As for the topic of bots. Kasparov showed he was human in his loss to
Deep Blue, and played badly in the last, and tie-breaking game, of a
very strenuous match. To me it does not prove that the computer was
better, just that it won this seven game match 4-3. In Chess, Kasparov
had to play anti-computer, and not rely on playing "book" lines so
much because obviously the computer had a better memory of the "book"
lines. In the last game, Kasparov played a very bad "book" line, and
was never really in the game. In situations where Kasparov could play
more on intuition and feel, he more than held his own against the
computer. (Let's not get started on a debate as to whether it's fair
the computer had access to a database of all game Kasparov has ever
played WHILE playing against him, whereas Kasparov did not have access
to a comparable database during game play).
In poker, it's much more about feel and intuition than hard
calculating. I don't think a computer would ever be able to develop
that same 'feel' for opportunities that statistics don't show. In the
short run, a bot should hold his own. In the long term I'm not so
sure. My instinct tells me that a bot would fare better at heads-up or
short tables as compared to a full ring game, but I can't realy argue
why I think that would be the case.
Regards,
Don
Nat Silver
> Nat Silver wrote:
> > A. Prock wrote:..
> > > Really? I assert that poker is more difficult and
> > > unforgiving than chess.
> > Okay. I guess we would have to define what
> > we mean when we say such a thing. An inferior
> > move usually ends a game of chess. There is more than
> > one way of playing a poker hand, leading to comparable
> > outcomes. As you know, the long run is is very important.
> > One can lose many hands and still show a profit.
> Gawd, do you even play Chess?
> Competitively?
Not anymore.
> An inferior move does not end a game..it's the superior play of the
> opponent to take advantage of the inferior play...
We're talking about computer play here.
> And, there are many more branches in a chess game than a poker game.
> The ebb and flow of plans within a game is enormous. One plan can "fail,"
> but you can recover, play defensively, and hopefully wait for an
opportunity to
> regain at least equilibrium...
Through mistakes
> Had conversations with both Master-level Chess players and poker
> players.
Asa Hoffman and GM Walter Browne play both games.
They are from my generation.
> As for the topic of bots. Kasparov showed he was human in his loss to
> Deep Blue, and played badly in the last, and tie-breaking game, of a
> very strenuous match. To me it does not prove that the computer was
> better, just that it won this seven game match 4-3.
Deep Blue won. Kasparov drew a won position early on in the match.
Kasparov suffered psychological strains that lead to exhaustion,
something Deep Blue is immune to.
> In Chess, Kasparov had to play anti-computer, and not rely on playing
> "book" lines so much because obviously the computer had a better
> memory of the "book" lines.
That was Kasparov's strategy.
> In the last game, Kasparov played a very bad "book" line, and
> was never really in the game.
By the last game Kasparov had psyched himself out of the match.
> In situations where Kasparov could play more on intuition and feel,
> he more than held his own against the computer.
That's to be expected. He was World Champion.
> (Let's not get started on a debate as to whether it's fair
> the computer had access to a database of all game Kasparov
> has ever played WHILE playing against him, whereas Kasparov
> did not have access to a comparable database during game play).
Hey, should Deep Blue be rated at least 2600? Of course. Is Deep Blue one
of the top 10 chess players in the world? Of course.
> In poker, it's much more about feel and intuition than hard
> calculating. I don't think a computer would ever be able to develop
> that same 'feel' for opportunities that statistics don't show.
There is more than one way to skin a cat.
A computer's style can be just as intimidating or even more so.
Consider that Kasparov was a broken man at the end of his
ordeal with Deep Blue, really.
> In the short run, a bot should hold his own. In the long term I'm not so
> sure. My instinct tells me that a bot would fare better at heads-up or
> short tables as compared to a full ring game, but I can't realy argue
> why I think that would be the case.
Time will tell.
tadperry
"A. Prock" <jeffy...@yahoo.com> wrote in message
news:3d3448b0$0$1061$8026...@spool.cs.wisc.edu...
r00lie made money like clockwork on IRC. He may very well have struggled at
higher limits when the stronger limit players opened a game of 50-100, but
this was rare. Day in and day out it was almost entirely 10-20 and r00lie
went "up the slope, like the antelope."
tvp
DaveM
>I wonder why the least knowledgeable posts are often the most hostile?
>
>I am dictating this sentence using "ViaVoice", a software and hardware
>package that costs less than $100 and is amazingly proficient at
>understanding and dictating my voice. And, the more I use it, the better
>it gets. Stop by any computer store to check it out.
>
>As for your other totally unfounded, unsuported and untrue statements... I
>just have to wonder what your motivation was.
Well all the voice recognition systems I've seen for Radiology were
laughable. At least secretaries make phonetic errors that make sense but
are obviously wrong and often understandable. Voice recognition produces
pure gibberish when it has problems.
DaveM
Dennis
> Well all the voice recognition systems I've seen for Radiology were
> laughable. At least secretaries make phonetic errors that make sense but
> are obviously wrong and often understandable. Voice recognition produces
> pure gibberish when it has problems.
You were probably using "Dragon" or a derivative of it. Yeah, that was
pretty poor. IBM's "ViaVoice" is very impressive but you still have to
speak clearly and enunciate well. When it makes a mistake you can correct
it and it remembers your voice so it improves as you use it. It's a joy
for slow typists or disabled people. As yet it isn't as good as a fast
typist but that is only a matter of time.
Pokerbase
this since the 60s. It is still a long way off. I would be suprised
if we got anywhere close to this in my lifetime. Voice recognition
would be by *far* one of the easier parts of technology like you
describe.
Now, do I think it is possible to write a poker bot that can beat
moderately tough online poker games? Definitely. I'm not sure if
they are out there, and frankly, someone(s) that could do it would
probably have a better EV applying there skills to something else
(generally true for people beating live poker as well). If you were
building collusive bots, it would certainly be much easier, by the
way.
Don't believe everything you read.
"Dennis" <dls9...@yahoo.com> wrote in message news:<3d337016$0$68565$9a6e...@news.newshosting.com>...
> James Martin, the inventor of voice recognition technology and a leading
> thinker in the field predicts that in 5 years time IBM will have a
> computer that will be indistinguishable from a very bright person when you
> talk to it on the phone. It will be creative, flexible, adaptive, much
> brighter than we are and it will have instant access to all information
> via the web. Tireless, mistake free, it will answer the phone for you and
> make business decisions better than you can. And that is just the begining.
Joe Long
>That's not too difficult. Recall that the screens and buttons
>you see on your monitor are just part of the poker program
>you downloaded. ...
>It would be easy to write a bot program that intercepts these
>signals and acts upon them. So, if the poker server sends your
>computer the signals for "four of spades" and "seven of hearts",
>your bot program would call up a database and determine that the
>best thing to do would be to fold. So, it then sends a signal
>down the cable that is the same one that would have been sent
>if you had pressed the "fold in turn" button.
Actually, Bing, I doubt this is a workable method. I expect the poker
sites encrypt the data flowing between client and server (at least I
hope so). But it doesn't matter. So far at least, all of the
graphical interfaces are static: your cards appear in the same place,
ever time; the "fold check bet raise" buttons appear in the same
place, every time. The same bitmaps are used for the cards and
buttons, every time. It is not that difficult for a program to "read"
the screen and send the appropriate mouse actions.
In theory a site could make such a 'bot a lot more difficult to
program by making their GUI's dynamic, that is, move the locations
around and randomly change the bitmaps. But I doubt that the players
would accept that!
Dennis
> The Artifical Intelligence community has been making predictions like
> this since the 60s. It is still a long way off. I would be suprised
> if we got anywhere close to this in my lifetime.
There is a real possibility that you will be surprised.
> Don't believe everything you read.
Those are wise words that I take to heart. I also try not to discount what
I haven't read because it makes me uncomfortable or conflicts with my
world view.
Regards, Dennis
Keith Ellul
> The Artifical Intelligence community has been making predictions like
> this since the 60s. It is still a long way off. I would be suprised
> if we got anywhere close to this in my lifetime. Voice recognition
> would be by *far* one of the easier parts of technology like you
> describe.
>
> Now, do I think it is possible to write a poker bot that can beat
> moderately tough online poker games? Definitely. I'm not sure if
> they are out there, and frankly, someone(s) that could do it would
> probably have a better EV applying there skills to something else
> (generally true for people beating live poker as well).
Not really. If you have a +EV bot, that is just a license to print
money. Get one to sit at a few simultaneous online high-stakes
tables. Permanently. If you want to avoid being caught have it leave
and come back (with a different account) at random intervals.
Think about it. If your bot could win even a small bet per hour at
$10/$20, just have 10 of them going simultaneously and you are being
paid $100 per hour to sleep / eat / sit on your ass. $2400 per day of
additional income really adds up, especially when you are not doing
anything for it. It's kind of like running a casino, except that you
don't have to spend money on dealers, cashiers, security, comps, etc.
Keith
VooDoo Chile
mentioned.
The "neural networks" for these programs were trained by playing out
trillions of games.
Once "training" was complete (and the neural network was "frozen"), the
neural network
could then score every position in terms of percentage chances of various
end results
(wins/losses/gammons/backgammons). Then, the right play can be selected.
The key
to all of this, is that the end product is "frozen". It no longer is
updating its "program"
while it is playing. Even if a "poker" neural network were allowed to
"learn" while it
played live, those hands would be statistically insignificant compared to
the trillions
of hands required to train it originally (you could teach it more, but not
much). So,
a neural network based program would be just as unflexible as a "rule-based"
program
(you just wouldn't know what the rules were).
Manny <ceea...@yahoo.com> wrote in message
news:5e096d64.02071...@posting.google.com...
> "JT Autry" <tau...@satx.rr.com> wrote in message
news:<ah02k1$p6624$1...@ID-147725.news.dfncis.de>...
> > This I don't understand. I assume that when you speak of
neural-networks
> > you are speaking of the type of program that allows the computer to
"learn".
> > If there is always a best move for a particular roll with a particular
board
> > position, which are finite in number, then what advantage would a
neural-net
> > program have? There is nothing to learn.
> >
> > JT
>
> Because instead of having to program the 8 trillion things the program
> COULD do, a neural net selects the best things to do from the set
> which does not contain the 7.9 trillion things that are clearly wrong.
> It can place itself in that situation billions of times and "score"
> the outcome from each one of the moves that it chooses.
>
> That's at least the most basic idea behind NN's (give it a situation,
> let it make it decision, rate the decision with an objective function,
> and repeat till you're convinced it has seen every situation you could
> care about). Problems arise in exactly how one chooses the things in
> the set you're going to test. You can rarely be convinced that you
> have found the optimal solution for a lot of problems they use neural
> nets on. Some problems lend themselves much more nicely to it.
>
> >
> > "VooDoo Chile" <notana...@nospam.com> wrote in message
> > news:q5JY8.45888$6r.14...@news4.srv.hcvlny.cv.net...
> > > I dunno about this. I was speaking recently to someone involved with
> > > competitive backgammon.
> > > I was purchasing one of the neural-network base BG programs from him.
The
> > > neural network
> > > BG programs are so good, you would swear they cheat with the dice.
They
> > > have literally changed the
> > > championship level game of BG. Of course, the programs (or the
> > programmers)
> > > couldn't explain what
> > > the programs did, being as they were neural networks. The person I
was
> > > speaking to mentioned that
> > > they were going to try the same technique with poker.
> > >
> > > I don't think it will work. The game of BG is totally static. There
is
> > > ALWAYS a best move for a
> > > particular board position and dice roll, and it is ALWAYS the same
> > > (excluding scoring considerations
> > > in a match). This is totally untrue in poker. Championship level
poker
> > > demands two things which I
> > > wonder whether the computer will be capable of: Varying your own play
and
> > > (far more difficult), being
> > > aware of your opponents play. I think "bot" play will be beatable
based
> > on
> > > the predicatability of play...
> > >
> > > Dennis <dls9...@yahoo.com> wrote in message
> > > news:3d334643$0$68544$9a6e...@news.newshosting.com...
> > > > If you are skeptical that computer bots are not a threat to online
poker
> > > > players you might want to check this out
> > > >
> > > > http://www.rci.rutgers.edu/~cfs/472_html/Intro/ChessContents.html
> > > >
> > > > The strengths that poker players compaired to computers have,
reading
> > > > other players, recognizing bluffing opportunities as they arise etc.
are
> >
> > > > greatly diminished online. On the other hand, the strengths of
computers
> > > > are magnified online. They never tire, never make mistakes,
instantly
> > > > calculate odds and more. The above link has some fascinating
articles on
> > > > the subject.
Bad Bob
Good to hear from you! You really ought to post here more than once every
two years or so. RGP can use all the "mature" insight it can get ;o)
"William Loughborough" <lov...@gorge.net> wrote in message
news:3d3469d3$0$76674$9a6e...@news.newshosting.com...
Pokerbase
you think. I would be willing to come up with the poker logic part of
the bot for 1 million. I will guarantee that it can beat 5-10 online
games for $5 an hour, or your money back.
Keith Ellul <kbe...@fe02.math.uwaterloo.ca> wrote in message news:<Pine.SOL.4.44.020717...@fe02.math.uwaterloo.ca>...
Bing
> >It would be easy to write a bot program that intercepts these
> >signals and acts upon them.
> Actually, Bing, I doubt this is a workable method. I expect the poker
> sites encrypt the data flowing between client and server.
Ok, so it may not be easy for the average hack to write a
bot from scratch, but a good programmer shouldn't have a
problem. If the data stream is encrypted, then the programmer
would need to decompile the poker client and mimic the
encryption/decryption schemes found inside the client.
Then, a bot could do all the things I described, despite
any encryption.
> So far at least, all of the graphical interfaces are
> static: ... It is not that difficult for a program to "read"
> the screen and send the appropriate mouse actions.
That is another approach. The benefits of your approach
is that the bot could probably be adapted to any graphical
poker program, new or old.
The downside would be that you would have to dedicate your
computer to the task of playing poker. If you tried to use
the computer to read your email, surf the web, or post to
newsgroups, the "graphics reader" would quickly get lost.
My proposal for a bot would be able to run unattended in
the background while you did other things. You could
probably run it 24 hours a day if you knew it would be
making a profit.
> In theory a site could make such a 'bot a lot more difficult to
> program by making their GUI's dynamic, that is, move the locations
> around and randomly change the bitmaps. But I doubt that the players
> would accept that!
No they wouldn't. And it would only defeat a few kinds of bots.
But, that leads into another question. Would online poker
sites make any effort to stop people from using bots? It
seems that a bot fills a seat and generates a steady rake.
JT Autry
days a year.
JT
"Pokerbase" <poke...@yahoo.com> wrote in message
news:bccd445d.02071...@posting.google.com...
Keith Ellul
damn, I left my million in my other jeans. I'll have to get back to
you. Do you take Canadian Tire money? (trust me, if you were from
Canada you would find that last bit funny)
Keith
jw_steve
"Dennis" <dls9...@yahoo.com> wrote in message
news:3d346d5e$0$76692$9a6e...@news.newshosting.com...
> I wonder why the least knowledgeable posts are often the most hostile?
>
> I am dictating this sentence using "ViaVoice", a software and hardware
> package that costs less than $100 and is amazingly proficient at
> understanding and dictating my voice. And, the more I use it, the better
> it gets. Stop by any computer store to check it out.
>
> As for your other totally unfounded, unsuported and untrue statements... I
> just have to wonder what your motivation was.
> <snip>
If only it was amazingly proficient at knowing where
to place a comma and a quotation mark.
jw steve
hapster
> > I am dictating this sentence using "ViaVoice", a software and hardware
> > package that costs less than $100 and is amazingly proficient at
> > understanding and dictating my voice. And, the more I use it, the better
> > it gets. Stop by any computer store to check it out.
And how slow did you have to talk to get it to do what it did? I'm a court
reporter who used to worry about programs like this, but I'm not going to
worry about it until it can accurately transcribe a Pakistani neurologist
speaking/mumbling at 225 words a minute LOL.....
Verbally Yours....
Octo the Genarian
that I don't think the existence of even very successful bots would be that
big of deal. Just like collusion in casinos. It's one strike against
casinos, but it's no reason to swear off them for life. If every starts
losing online, everyone will stop playing online. (Well, except the players
who would have lost regardless).
"Bing" <pax...@yahoo.com> wrote in message
news:3d344fa6...@news.fu-berlin.de...
> On Mon, 15 Jul 2002 20:39:34 -0500, "Octo the Genarian"
> <inv...@nomail.com> wrote:
>
> > What annoys me about this "The bots are coming!" hysteria is that it
doesn't
> > really matter if someone designs a bunch of perfect bots and sticks them
on
> > Paradise. Why? Because when you find yourself in a room full of perfect
> > poker players who do not make conversation and play is if they know what
> > each other has, you will realize that the game is not worth playing and
> > leave.
>
> But, you see, the difference here is that perfect human
> players do care if you catch on to them and leave. They
> spent years working on their act and it will affect
> them to be found out. If enough players wise up to the
> perfect player, that player may have to pack up and
> go elsewhere or change his ways.
>
> A bot wouldn't care. It would be happy enough to just sit
> there and wait until some other sucker comes along.
>
> Similarly, that perfect player hopes to sit with the
> victim for many hands and hours in a row, waiting for
> the long run to hand him his profits. The bot is more
> than happy to play just 3-4 hands with the victim.
>
> The bot will be equally ready to play at 1 PM or 1 AM
> and won't mind playing 300 hands or just 3. You can't
> just wait for it to get bored and leave the casino.
Dennis
> If only it was amazingly proficient at knowing where
> to place a comma and a quotation mark.
>
> jw steve
Funny thing is, I backed up and added those myself, thinking they were
needed. The program got it right the first time...
OrangeSFO
Dennis
Probably 10% slower than normal and I do have to enunciate carefully.
jw_steve
> jw_steve wrote:
> > If only it was amazingly proficient at knowing where
> > to place a comma and a quotation mark.
> >
> > jw steve
>
> Funny thing is, I backed up and added those myself, thinking they were
> needed. The program got it right the first time...
>
Yeah, that is "funny." Just to be clear, what I really
meant by my post was, "I don't believe you."
jw steve
Dennis
> > Funny thing is, I backed up and added those myself, thinking they were
> > needed. The program got it right the first time...
> >
>
> Yeah, that is "funny." Just to be clear, what I really
> meant by my post was, "I don't believe you."
>
> jw steve
No need to be insulting. And no need to believe me. If you are interested
go to a computer store that will demonstrate it for you. Call around,
shouldn't be hard to find.
Dennis
tadperry
"jw_steve" <ssny...@nyc.rr.com> wrote in message
news:KypZ8.98266$QD2.27...@twister.nyc.rr.com...
The coolest aspect of voice recognition software to me is the ability to
control computer command sequences.
As for text entry, I'm hooked on emacs. I LOVE emacs. I use it under Windows
and there has never been (and apparently never will be) a "word processor"
that handles text editing duties like emacs does.
I'd much rather get the text all finished and perfect and "flow" it into a
template than mess around trying to type something into a piece of bloatware
like Word for Windows.
I type very quickly, but the ability to do absolutely everything without my
hands leaving the keyboard is a huge plus.
I'd say any translator that's been working for several years can type about
as fast as I can, but when it comes down to me versus some guy typing into
Word, I'll produce twice as many finished pages in the same number of hours.
Another favorite technology is programmable keyboards.
Between programmable keyboards and voice recognition, you'd hardly have to
touch anything. You can assign one or two keys to go anywhere and open any
app and even control that app once inside and use the voice recognition for
typing.
Mice are a real drag. :-)
tvp
The Beet Man
<mat...@worldnet.att.net> wrote:
>A. Prock wrote:..
>
>> Really? That wasn't the impression that I got. r00lie
>> made some money from the 10/20 games, and made little
>> (or lost) in the 20 40 games.
>
>If that is so, it wasn't in the long run. Since 20-40 games
>required bankrolls usually large enough for pot limit, they
>did not run all the time as did 10-20.
I didn't play on IRC all that much but I did play from time to time
many years ago, and I recall the 20-40 always went in the evenings.
r00lbot was ok enough to beat play money players but it basically
played a weak-tight, "by the book" strategy. Any decent player could
kill it. As I recall, opening with A8o on the button for a raise got
you a huge negative rating from r00lie, which is a routine play for a
decent player.
--
This post brought to you courtesy of the Beet Man!
The Beet Man
wrote:
>On Mon, 15 Jul 2002 23:12:12 -0700, "wg bradley" <wgb...@yahoo.com>
>wrote:
>
>>Here's a chart of Pokibot's results on IRC:
>>
>>http://spaz.ca/aaron/poker/index.html
>
>IIRC, Andrew was around IRC for quite a while, and I'm sure he knows
>about the existence of Poki. While it's true that the IRC bots have
>done well against play-money players, based on my (very, very) limited
>experience with them, I think they would get killed in a typical
>online 10/20. I wouldn't even back one in a 3/6.
Did any of those IRC bots play Omaha/8? From what I saw of the IRC
bots, I wouldn't back one in an online 3/6 Hold'em game either, but I
suspect it would be rather simple to write a bot to beat an online 3/6
game of Omaha/8.
The Beet Man
<kbe...@fe02.math.uwaterloo.ca> wrote:
>On 16 Jul 2002, Pokerbase wrote:
>> Now, do I think it is possible to write a poker bot that can beat
>> moderately tough online poker games? Definitely. I'm not sure if
>> they are out there, and frankly, someone(s) that could do it would
>> probably have a better EV applying there skills to something else
>> (generally true for people beating live poker as well).
>
>Not really. If you have a +EV bot, that is just a license to print
>money.
My EV would go through the roof if I just had a bot that could
automatically fold unplayable hands for me and alert me when I had a
potentially playable hand, because I could be constantly doing
something else while playing poker, like washing the dishes, cooking
dinner, shaving my cat, etc. As it is I need to stop playing poker if
I want to do these things.
Bill Sicaras
to program a bot that would look at your cards and a given board and
action ahead of you and make a decision, ala B&M. Shit i can write that
in a day. This is like what the Chess bot did- although instead of 5
moves to choose from (check bet raise call fold), it had to consider
every piece and its possible moves when weighing its decision, and
counter moves and counter counter moves, etc. Not that much compute
power these days is needed. Go to radio shack and see some great
computer opponents for $5.99.
But poker would require a very very VERY complicated learning system, so
it would adjust to an infinite amount of styles. Shit, any player can
change his/her style mid session, new players in and out, etc.
Even still, it DOESNT give an edge much greater than an expert player.
Maybe in the future, but we are far from that point yet. Today bots
may exists that calculate odds and determine the best move based on
that... which is nothing a good player cant do on the spot. A bot has
the disadvantage of not adjusting to style. I love seeing people post
"first hand knowledge" of a bot that does A,B,C, D, etc..... they watch
too much star trek. Its not reality. Creating the screen capture and
mouse control is baby stuff... ill bet there are bots using this right now.
Take it from a Computer Science M.S. and Computer Engineering B.S.... a
successful bot would take a GENIOUS of a programmer, and a computer
capable of crunching billions of calculations lighning fast... nothing a
Pentium 4 could do (which I designed circuits for) for sure.... maybe a
few parallel Sun Blade workstations at $20k a pop. Remember it has to
decide in a split second for online.
Everyone relax.
:)
Bill
Bill Sicaras
Bing posted :
Ok, so it may not be easy for the average hack to write a
bot from scratch, but a good programmer shouldn't have a
problem. If the data stream is encrypted, then the programmer
would need to decompile the poker client and mimic the
encryption/decryption schemes found inside the client.
Now thats funny. "all the programmer would need to do is decompile and
mimic the encryption..." If that were possible, dont you think I would
be breaking into banks and CIA sites?!!? Thats like saying, "all we
need to do to cure cancer is crack the genetic code".
128bit encryption, over SSL, is almost impossible to intercept, even
with your OWN client. We arent talking average hacks or expert
programmers- no one. Thats what encryption is all about. If it was so
easy to crack as you say we'd all be in a load of shit wouldnt we!
Decompiling would give you hundreds of millions of lines of machine
language code. Assuming you could do the impossible and interpret it
(its not like you get source C code with comments when you decompile!),
you still couldnt break encryption as it is dynamic, even during one SSL
transaction the symmetric keys change and handshaking is continuous.
Profe
oh, how I miss the olden days...
-LL
"The Beet Man" <i-eat-beets-an...@juno.com> wrote in message
news:j6dtjukpshnvlbp22...@4ax.com...
Paul Phillips
Bill Sicaras <bsic...@dal.asp.ti.com> wrote:
>But poker would require a very very VERY complicated learning system, so
>it would adjust to an infinite amount of styles. Shit, any player can
>change his/her style mid session, new players in and out, etc.
Playing poker well does not require this. Playing close to optimally
might, but that's not necessary to win money even in tough games, let
alone the cakewalks available online.
>I love seeing people post
>"first hand knowledge" of a bot that does A,B,C, D, etc..... they watch
>too much star trek. Its not reality.
What specifically are A,B,C,D, etc? I do have first hand knowledge of
bots that do E,F,G but I'm not sure how that compares.
>Take it from a Computer Science M.S. and Computer Engineering B.S.... a
>successful bot would take a GENIOUS of a programmer, and a computer
>capable of crunching billions of calculations lighning fast...
Now you're completely high. Successful bots will be prevalent online
within a few years at most if they are not already. A marginally +EV
bot would require minimal work for any competent programmer, and less
CPU than is in my toaster. They don't need to be the greatest poker
players ever to win money (but I do expect bots to be as good as all but
perhaps the very best humans within a fairly short period, say a decade.)
And actively colluding bots will be better than any human that isn't
cheating as well.
--
Paul Phillips | The Cynics regarded everyone as equally corrupt.
Future Perfect | The Idealists regarded everyone as equally corrupt,
Empiricist | except themselves. -- raw
pal, i pill push |----------* http://www.improving.org/paulp/ *----------
The Beet Man
wrote:
>r00lbot: "You called KdQc for two bets cold? BWAHAHAHAHAHA!"
>
>oh, how I miss the olden days...
I always liked "Up the slope goes the antelope!" and "I don't
understand this game!"
Keith Ellul
> One more thing-
>
> Bing posted :
>
> Ok, so it may not be easy for the average hack to write a
> bot from scratch, but a good programmer shouldn't have a
> problem. If the data stream is encrypted, then the programmer
> would need to decompile the poker client and mimic the
> encryption/decryption schemes found inside the client.
>
>
> Now thats funny. "all the programmer would need to do is decompile and
> mimic the encryption..." If that were possible, dont you think I would
> be breaking into banks and CIA sites?!!? Thats like saying, "all we
> need to do to cure cancer is crack the genetic code".
> 128bit encryption, over SSL, is almost impossible to intercept, even
> with your OWN client. We arent talking average hacks or expert
> programmers- no one. Thats what encryption is all about. If it was so
> easy to crack as you say we'd all be in a load of shit wouldnt we!
Not really. You have to remember that they are sending your client the
key to decrypt this stuff. This is obviously not the same as reading
someone else's encrypted data. You are not trying to crack the
encryption.
> Decompiling would give you hundreds of millions of lines of machine
> language code. Assuming you could do the impossible and interpret it
> (its not like you get source C code with comments when you decompile!),
> you still couldnt break encryption as it is dynamic, even during one SSL
> transaction the symmetric keys change and handshaking is continuous.
You could easily mimic this. All that you have to is re-write the
client. Not as hard as you think (not a 1-day job, either, but
definitely doable). Remember, you are NOT breaking the encryption.
You don't need to break it... you are the one who is authorized to
decrypt it. It is just a matter of writing your client to go through
all the same steps (key exchange, etc) that the actual client does.
Once you know what their protocol is, the job becomes easy (you just
have to write a client program that follows that protocol... this part
IS a 1- or 2-day job, especially since you are not concerned about pretty
graphics) The hard part is reverse-engineering their protocol. And,
when I say "hard", what I really mean is "boring".
Keith
Joe Long
<kbe...@fe02.math.uwaterloo.ca> wrote:
>Not really. You have to remember that they are sending your client the
>key to decrypt this stuff. This is obviously not the same as reading
>someone else's encrypted data. You are not trying to crack the
>encryption.
OK, gang, let me say up front that I am not an encryption expert. But
from what I do know about it, there seems to be some fundamental
misunderstanding here. Let me share a few points, AIUI:
1. with modern public-key encryption, knowing the program code is
useless to decoding the traffic. One of the most widely used
programs, PGP (stands for Pretty Good Privacy), has its source code
freely distributed.
2. intercepting the keys exchanged between the client and server is
useless ... in fact, they are called "public keys" because they are
public, that is, not secret at all. The public keys from each side
are part of a private-public key pair (two keys, one private, one
public). The pairs are generated randomly, together, but the program
keeps the private key secret and shares the public key with the other
end (or the world). Traffic is encrypted by using your private key
and the other fellow's public key; only the combination of your public
key and the other fellow's private key can decrypt it. In this way,
by the simple exchange of PUBLIC information, only the server can
decrypt traffic from the client and only the client can decrypt
traffic from the server, because no one else knows their private keys.
3. that is why knowing the poker client's code is useless in
decrypting the traffic coming to the client. You would have to know
the client's private key, which is generated randomly every session
(or more often). It is theoretically possible to crack public-key
encryption, but at 128 bits even the most powerful supercomputers
couldn't do it in time to be of any use in a poker session.
But, could you not edit and re-compile a client program and have it
log in to the server? If the client is sloppily written, yes. But a
properly written client uses a similar key exchange based on
computations against the compiled code to authenticate itself; a
modified program would be instantly recognizable to the server as a
forgery. In fact, it is common practice to include "signature" keys
with downloadable software, so that the person downloading and
installing it can be certain that he is getting the real program and
not some hacked "trojan horse."
Keith Ellul
Everything so far is good. I don't think that anyone was suggesting
cracking the encryption. The fact is, you don't need to. Like I said,
YOU are the one who is authorized to decrypt it. You don't need to
crack it. They are sending you the keys.
FWIW, one small thing in what you wrote... in practice, public-key
encryption is generally only used for key exchange. This is why I say
that they are sending you the keys. Private-key stuff tends to use
shorter keys and is faster to encrypt-decrypt. So a reasonable way of
doing things is for a client to send it's public key to the server, and
the server to reply by sending the client a private key (encrypted
using the public key). The client receives this key, decrypts it, and
uses that private key to encrypt traffic for a certain amount of time
until the public key "expires" and the whole key exchange thing is done
again. There are specific protocols to handle this "key management"
task, but the details aren't really relevant here. The only reason
that I mentioned it is in case someone gets some misconceptions from
what you wrote about public-key encryption...
> But, could you not edit and re-compile a client program and have it
> log in to the server? If the client is sloppily written, yes. But a
> properly written client uses a similar key exchange based on
> computations against the compiled code to authenticate itself; a
> modified program would be instantly recognizable to the server as a
> forgery.
This is easily hacked. All that you have to do is compute the hash of
the original compiled program and send that out (obviously you don't
send the hash of YOUR program... that would be stupid.
> In fact, it is common practice to include "signature" keys
> with downloadable software, so that the person downloading and
> installing it can be certain that he is getting the real program and
> not some hacked "trojan horse."
This is a little different. In this case, the person who is
downloading the program has access to the actual program so they can
compute the hash and see if it checks out. On the other hand, in the
poker client/server example, the server does not have access to the
program that you are running so it can't tell if you are sending the
hash of the program that you are running, or just a hash of the
original client.
So, doing this is kind of pointless... it doesn't really add any
security. Anyone who is going to be able to get far enough to write a
working "fake client" could get around this with no effort at all.
Keith
Joe Long
<kbe...@fe02.math.uwaterloo.ca> wrote:
>Everything so far is good. I don't think that anyone was suggesting
>cracking the encryption. The fact is, you don't need to. Like I said,
>YOU are the one who is authorized to decrypt it. You don't need to
>crack it. They are sending you the keys.
Not quite: YOU are not, your copy of the client you downloaded is.
Having the keys the server sends the client, without having the
client's private key, is not going to let you encrypt/decrypt the
packets.
...
>> But, could you not edit and re-compile a client program and have it
>> log in to the server? If the client is sloppily written, yes. But a
>> properly written client uses a similar key exchange based on
>> computations against the compiled code to authenticate itself; a
>> modified program would be instantly recognizable to the server as a
>> forgery.
>This is easily hacked. All that you have to do is compute the hash of
>the original compiled program and send that out (obviously you don't
>send the hash of YOUR program... that would be stupid.
If the client is simply sending a hash of the entire binary ... but
there are more sophisticated ways to use keys for authentication, no?
Probably online poker clients today are not taking full
countermeasures against someone forging the client to use with a 'bot
-- but if 'bots become a problem, I expect they would do so.
...
>So, doing this is kind of pointless... it doesn't really add any
>security. Anyone who is going to be able to get far enough to write a
>working "fake client" could get around this with no effort at all.
A simple hash, yes. However, the use of keys for authentication
(proving that you are who you say you are, and not a forgery) is well
developed. I may be mistaken, as I said I'm not an encryption expert,
but AIUI if the client and server are properly done
reverse-engineering the client would not allow you to produce a
working forgery. The site could provide you with the source code and
it still wouldn't help.
Keith Ellul
I have never really heard of anyone using any type of security like
this. That main problem is that it is not really secure. This key has
to be in the client somewhere. If it is there, it can be found. There
is no security here... perhaps some obfuscation (ie, it might be
difficult to find in the client's executable) but this is certainly not
secure.
Anyway, like I said, I have never heard of anyone trying anything like
this. There are certain schemes that some companies use to prevent the
copying of executables (ie, copy-protection). These generally don't
work for the same reason.
> >So, doing this is kind of pointless... it doesn't really add any
> >security. Anyone who is going to be able to get far enough to write a
> >working "fake client" could get around this with no effort at all.
>
> A simple hash, yes. However, the use of keys for authentication
> (proving that you are who you say you are, and not a forgery) is well
> developed.
For sure. But, in order to do this, the keys must be kept private. If
they are embedded somewhere in the executable they are not private.
> I may be mistaken, as I said I'm not an encryption expert,
> but AIUI if the client and server are properly done
> reverse-engineering the client would not allow you to produce a
> working forgery. The site could provide you with the source code and
> it still wouldn't help.
This is certainly not true. If the source code is going to be provided
then I could find the keys in there and take them out. If the source
code is not provided, this just obfuscates things somewhat... they keys
can still be found.
Basically, what is comes down to is this. It is impossible for a
downloaded client to have access to information that I do not have. If
it is in there, I can find it. If the client can extract it, so can
I. The concept of signing something (to prevent forgeries) requires
that you "sign" it with your private key. If I have your private key I
can do the "signing" and pretend to be you. If the "private" key is
embedded in the client then I can decompile the client and find the key.
There is no security here.
Keith
Bad Bob
<kbe...@fe02.math.uwaterloo.ca> wrote:
zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz!
Bing
<kbe...@fe02.math.uwaterloo.ca> wrote:
> Basically, what is comes down to is this. It is impossible for a
> downloaded client to have access to information that I do not have. If
> it is in there, I can find it. If the client can extract it, so can
> I. There is no security here.
Exactly correct. Your posts on the subject are right on
the money.
Manny
> In article <3D3EC98A...@dal.asp.ti.com>,
> Bill Sicaras <bsic...@dal.asp.ti.com> wrote:
> >But poker would require a very very VERY complicated learning system, so
> >it would adjust to an infinite amount of styles. Shit, any player can
> >change his/her style mid session, new players in and out, etc.
>
> Playing poker well does not require this. Playing close to optimally
> might, but that's not necessary to win money even in tough games, let
> alone the cakewalks available online.
>
> >I love seeing people post
> >"first hand knowledge" of a bot that does A,B,C, D, etc..... they watch
> >too much star trek. Its not reality.
>
> What specifically are A,B,C,D, etc? I do have first hand knowledge of
> bots that do E,F,G but I'm not sure how that compares.
>
> >Take it from a Computer Science M.S. and Computer Engineering B.S.... a
> >successful bot would take a GENIOUS of a programmer, and a computer
> >capable of crunching billions of calculations lighning fast...
>
> Now you're completely high. Successful bots will be prevalent online
> within a few years at most if they are not already. A marginally +EV
> bot would require minimal work for any competent programmer, and less
> CPU than is in my toaster. They don't need to be the greatest poker
> players ever to win money (but I do expect bots to be as good as all but
> perhaps the very best humans within a fairly short period, say a decade.)
>
> And actively colluding bots will be better than any human that isn't
> cheating as well.
A marginally +EV bot would require MINIMAL work for ANY COMPETENT
programmer?
Come on.
First of all designing the bot would require a better understanding of
probability and statistics than 95% of all the competent programmers I
know.
Then, the programmers I know who understand the mathematics well
enough would have a difficult time interfacing with the online
software. I think maybe <1% of programmers could design a marginally
+EV bot in a non-raked game, and it would not require a MINIMAL work.
I suspect the bots that are out there right now (and will be there in
the future) will be designed via a neurel net, and that is never a
MINIMAL work.
Go have a look at Poki. They had a team of computer scientists working
on this and there are tens of thousands of lines of code for that
thing. Hardly minimal.
And do you mean marginally +EV with a rake?
I don't know, maybe your competent programmer is my expert programmer
and your minimal amount of work is my 9 months of coding, testing,
implementation and tweaking.
I'd be really interested in hearing if you or anyone else has such a
bot going. I'd even be intrigued by anonymous claims of it.
I do agree that it would not take a lot of computing power once made.
Also, if you're referring solely to a heads-up bot, thats a different
story. It still wouldn't be easy, but it would be far easier to make
one +EV.
Keith Ellul
What does +EV mean? Especially heads up. +EV again whom? Against
some guy who can't play? Sure. Easy. Against a pro? Good fucking
luck.
All this talk has gotten me more and more interested. I think that I
may spend some time trying to do this. Looks like a fun project that I
will undoubtedly fail at.
Keith
Paul Phillips
Manny <ceea...@yahoo.com> wrote:
>A marginally +EV bot would require MINIMAL work for ANY COMPETENT
>programmer?
That might be a misuse of "minimal" because from scratch it would
take a while, but my point is that it's not hard. I have no evidence
for this other than lots of experience, but correct starting hand
selection and utterly unimaginative post-flop play are enough to beat
many, perhaps most, low-limit games.
That, observing bots which collect statistics on everyone online, and
a sensible game selection algorithm are enough to create a winning online
poker bot. I almost wish I had a decent financial motivation to prove
this because I'm so sure it's true and super surprised that a lot of people
seem to disagree, but I don't so I'm mostly happy to wait for someone
else to demonstrate this. (Though that may not happen soon because anyone
with a winning bot has little incentive to advertise its existence and
much to hide it.)
>And do you mean marginally +EV with a rake?
Yes, actually. Call me crazy.
>I don't know, maybe your competent programmer is my expert programmer
>and your minimal amount of work is my 9 months of coding, testing,
>implementation and tweaking.
As to the former, very possibly: my idea of a competent programmer is
one who has done (solo) those things that you have to do to get a CS
degree from a good college. A lot of people working as programmers today
learned their skills from "teach yourself java in 21 days" and whatnot,
which I am not knocking, but often doesn't make you much of a programmer.
As to the latter, yes as well: given that a winning bot is essentially
a license to print money, nine months sounds like very little work.
>I'd be really interested in hearing if you or anyone else has such a
>bot going. I'd even be intrigued by anonymous claims of it.
I'm toying with financing the creation of one. I think I know exactly
how to go about it at every level of its creation, but I'm too lazy and
unmotivated these days (at least in a programming sense) to actually
do it myself.
>Also, if you're referring solely to a heads-up bot, thats a different
>story. It still wouldn't be easy, but it would be far easier to make
>one +EV.
Our bot would have the choice of all the games at paradise, so that would
be among its choices.
--
Paul Phillips | Name something a blind person might use - A sword
Stickler | -- heard on Family Feud
Empiricist |
all hip pupils! |----------* http://www.improving.org/paulp/ *----------
Gary Carson
<kbe...@fe01.math.uwaterloo.ca> wrote:
>>
>> Go have a look at Poki. They had a team of computer scientists
working
Yeah, but they're Canadian, it would take a team.
Gary Carson
http:// garycarson.home.mindspring.com
Keith Ellul
> On Thu, 25 Jul 2002 13:41:28 -0400, Keith Ellul
> <kbe...@fe01.math.uwaterloo.ca> wrote:
>
> >>
> >> Go have a look at Poki. They had a team of computer scientists
> working
I did not. I don't even know what Poki is....
Keith
Gary Carson
Then you should learn something about a subject before you start
pontificating about it.
Jacob Johannsen
I forget the web site, but ask google for "Computer Poker" and click
"I'm feeling lucky".
--
/Jacob Johannsen aka CNN
--------------------------------------------------------------------------
You're in a phrase of mythic riddle messages, all aligned.
--------------------------------------------------------------------------