Online Poker Question
RzItUp
Over this time I am + $2755 mostly from 3-6 and 5-10, and some 10-20.
Sporadically over this time I have noticed what I would consider peculiar
hands. I see strange plays every day, and I am not quick to judge online poker
as unfair.
My question is this: Having known that the shuffling algorithm was broken a
little while back on Planet Poker, are we as players to assume that at no point
in time are we safe from having the same thing happen again??
I know that maybe you can make it more secure, but I would reason that it can
always be broken. Am i correct??
And if it was broken again, would Planet Poker or Paradise Poker know that it
had been??
Some of the hands I have seen recently on Paradise Poker seem suspicious to me
because i have replayed the hand using the mind set of the player had he
actually seen the all of the cards after the deal. Example: Player in midlle
position calls a cap before the flop with 67o. He sees his opponents have AA
and AK. Flop comes 48K. The player with 67o capped the flop. He called the
turn bet and raise, then raised the river when an innocent 5 appeared. Happens
all the time right? I know that, and I could have used a more interesting hand
here, but the mindset of the player with 67o, is that he will get his money in
on the flop (he knows with AA and AK that it will be capped), then caps it on
the flop to get value, only calls the turn so as not to appear suspicious, then
gets the raise in on the end.
I will keep playing online poker so long as I continue to make money. But some
of the hands I see will always arouse my suspicion unless someone can verify to
me that the shuffling algorithm cannot be broken.
I know that people will say, "why would someone with the intelligence to hack a
computer program for poker, waste his time if he could hack into a bank or
something else worth lost of $$?". If I could do it, I wouldnt mind sitting at
home making $500 - $1000 a day without arousing the suspicion of the FBI, or
some other federal agency.
Just my thoughts and no one elses.
Mike O'Malley
Dark Scorpion
"RzItUp" <rzi...@aol.com> wrote in message
news:20000128181224...@ng-ch1.aol.com...
> Some of the hands I have seen recently on Paradise Poker seem suspicious
to me
> because i have replayed the hand using the mind set of the player had he
> actually seen the all of the cards after the deal.
Coming from the other side of the coin, here is an example from a 10-20 hand
I played recently on Paradise poker. I will repost again when I can get
onto Paradise poker to request a hand history, so I can show you the hands
in question. Keep the following things in mind:
1- I am *NOT* a cheater. While I am a programmer, I choose, purposefully
not to write so much as a robotic program, to allow myself to "play all
day". A robot program (which I see others are in the process of writing)
isn't a program that can see anything other than what you or I could as you
play. All it does is stand in and play for you while you are not there.
Why would I do this? Mainly because I think that most of the people I play
against on there are so easy, that all I need to do is write a programming
bot with the way I think, and be done with it. Perhaps I am being
arrogant...but after going up $750 in less than one week of play...all too
easy.
2- I am from California. As most posters from this group would
agree..."California players are completely unpredictable". What is not
realized by most people that make the statement...poker in CA is "prove it"
poker. If I smell a rat, or think I'm being "bullied"...I will call all the
way to the end with a tiny pair...just to see the look on the guy's face
when he gets AKs beaten.
Okay, here we go with the example. 10-20 Holdem on Paradise poker. 5
players at the table...all with move than $500. I buy in for $200.
- At this point, I should note...while $200 is more than acceptable to sit
at in most 10-20 HE games...the opposite is true online...if all things were
equal (that is...I had it in my roll), I wouldn't really sit down with less
than $1000. I tried sitting with $20 at 2-4...HA, that was a joke! I got
my ass re-raised out so fast, it wasn't funny. When I went to the table
with $100, though...things were just fine, thank you.
I decide to post right away...so no one would leave the table (since there
are only four others besides me). Now, I have noticed that...no matter what
table I sit at, or for what buy in, if I post prior to BB...the hand is
ALWAYS raised. Irregardless...if there is so much as one player posting
into the game that is off the BB...that hand is raised.
My post hand: J4s (clubs I think). Normally, I wouldn't call this hand. I
would fold it up...whether I posted or not. But, sure as there is a
sun...the pot was raised, and I semi-tilted. I said to myself: "I'm sick of
*that* crap". I call the raise.
flop (K4s)-8...ZERO in my suit. Checked to the raiser, who bets out. At
this point I say to myself that he *might* have a K, or Ax of the board's
suit...maybe. But I am on semi-tilt, and call with a pair of fours. All
others before and after me fold.
Turn (K4s)-8-8: Now I have two pair. Check - bet. I think now the chances
are a little better that he has a K...but now I am in for 2 big bets and a
small...so *prove it*. I am still thinking I am being bullied. I call.
River (K4s)-8-8-4: FULL BLAST! Check - bet. I am thinking
that...maybe...just MAYBE...he was bullying me...with K8o, and bullied me
into his full house. I call.
Raiser turns over pocket rockets, for the losing hand. Naturally, there are
comments about how poorly I played (and the comments are correct), and got
lucky on the river (because I did). But I knew NOTHING in advance other
than I was semi-tilted, and I got lucky.
A few hands later on the same table...original guy is gone, and I am left
with a new player (who got raised out when he posted) and two people that
commented on how poorly I played my hand (and they were right). Since I
don't remember the details of the hand, I will make it short. I got a case
of the runs to an AQ flush against a player that didn't showdown. Again...I
called because I thought I was being bullied, not because I knew anything in
advance.
Yes...you do see a lot more *magic* hands online. In my opinion, there is
only one reason...because you simply see more hands.
When I play in the club...what's the thing that takes the most
time...shuffling! In some 20-40 games, more time is spent shuffling than
playing. When a computer shuffles the cards, it takes a scant few moments
to complete. Almost no time at all, and the cards are in the air for the
next hand...and yes, you are bound to see more magic hands.
It's to the point where, even though I personally believe in the integrity
of Paradise Poker's shuffling algorithm (based on the number of possible
decks in poker...that puts me in awe...), I will NOT play Omaha on there
server. The magic there is overwhelming to me. I can't catch a single hand
playing there, and everyone always wins and I always lose, irregardless of
the hand. Do I think that the source has been cracked, and the omaha
players are all out to get me? LOL HELL NO! If someone was gonna crack
the interface, they'd be dumb to do it in a 3-6 omaha room, where there are
5 or 6 10-20 holdem tables all the time! You'd HAVE to get some of that
money....simply out of greed. Another game I can NEVER win is the 1 on 1
Holdem. I *NEVER* win at it. I always get beaten, cracked, defeated, no
matter what. I have an example of that, too, on the play table, but the one
time I played it live...same story. Bullied and beaten...and I won't play
it.
More hands means more *magic* hands. That simple. If you look like you are
being a bully, I'll probably call you down...so I am not gonna give my
Paradise screen name because I don't feel like being hunted...LOL.
Besides...I get bullied enough already. :-)
All the best in all you do,
Dark Scorpion
Andrew Prock
>
>1- I am *NOT* a cheater. While I am a programmer, I choose, purposefully
>not to write so much as a robotic program, to allow myself to "play all
>day". A robot program (which I see others are in the process of writing)
>isn't a program that can see anything other than what you or I could as you
>play.
Just out of curiosity, how are the robots supposed to interface with
the server? It's not like there are open standards. Of course it
can be done, but it really does require a pretty darn high level of
programming competence. This is actually one of the biggest problems
online poker is going to face in the future.
Automated poker players which can pass this form of "Turing test"
are going to be hard to spot, and even harder to beat. That is
if someone has the sophistocation to "break" the communication
standard, they'll be able to write a winning bot which can pass
the Turing test well enough to get away with this for quite a while.
- Andrew
TheBigDenominations
won't happen inside the next 5 years.
Here are a few points explaining why...
1) If the client app is written correctly it will be
running in secured/encrypted memory. (Application Layer
data can be encrypted any way you wish. This means that the
data is encrypted through the whole OSI model and cannot be
intercepted prior to entering encrypted memory)
2) Point(1)means that any robot type program would need to
hook into the video output and interpret that correctly.
3) Simple countermeasure - Remove client control of video
output and move the video output around randomly.
4) Any robot type program would need to pass an advanced
Turing test anyway...
"Hi, this is 'online-cardroom' support. We are worried
about robot players ruining our games. Please answer the
following questions to reassure us that you're a real
person. How many stone would you say Vanessa Feltz weighs?
How many stone would you say Kate Moss weighs? Please reply
immediately or your account will be put on hold"
5) My guess is that if Microsoft put the Windows 2000 team
on it for a year, then they would still be nowhere near...
So, with only a little disrespect to Dark Scorpion...
I smell a big pile of horse poo !!
* Sent from AltaVista http://www.altavista.com Where you can also find related Web Pages, Images, Audios, Videos, News, and Shopping. Smart is Beautiful
Isai
communications/interface aspects. But are there any programs in existence
that can consistently beat reasonably good human players?
Isai
Tony L. Svanstrom <to...@svanstrom.com> wrote in message
news:1e574kj.1itlywp1pqs662N%to...@svanstrom.com...
> TheBigDenominations <thebigdenomin...@hotmail.com.invalid>
> wrote:
>
> > In my humble opinion, it (a poker robot for online play)
> > won't happen inside the next 5 years.
> >
> > Here are a few points explaining why...
> >
> > 1) If the client app is written correctly it will be running in
> > secured/encrypted memory. (Application Layer data can be encrypted any
way
> > you wish. This means that the data is encrypted through the whole OSI
> > model and cannot be intercepted prior to entering encrypted memory)
> >
> > 2) Point(1)means that any robot type program would need to hook into the
> > video output and interpret that correctly.
>
> That's not hard to do, besides... point 1 is just bs, no one bothers to
> do anything like that... and even if they did you can still use a
> debugger and stop things at any moment to see what's going on, or you
> simply tear their program a part and takes a look at how it's put
> together.
>
> > 3) Simple countermeasure - Remove client control of video output and
move
> > the video output around randomly.
>
> Not that hard to overcome, in worse case you'll simply run their program
> within a "virtual computer", meaning that everything they try to do is
> caught by your software. The easiest way to do this would be to use an
> existing Win-emulator and write your bot for another platform. Then you
> have your Mac-app controlling their PC-app without them being able to do
> a thing to avoid you getting the information.
>
> > 4) Any robot type program would need to pass an advanced Turing test
> > anyway...
> >
> > "Hi, this is 'online-cardroom' support. We are worried
> > about robot players ruining our games. Please answer the
> > following questions to reassure us that you're a real
> > person. How many stone would you say Vanessa Feltz weighs?
> > How many stone would you say Kate Moss weighs? Please reply
> > immediately or your account will be put on hold"
>
> That's just stupid, _I_ wouldn't know the answer to that question. Why?
> Simply because I'm not form a country where we use anything like that...
> Ever heard of the metric system? it's what the rest of the world's
> using...
>
> > 5) My guess is that if Microsoft put the Windows 2000 team
> > on it for a year, then they would still be nowhere near...
>
> That's because they're M$.
>
>
> Within 3 months I could have a "robot" controlling the computer just
> like a human would... That would require me to bring in some people that
> want money for their work, but I'm sure there's people here with big
> enough bankroll to be able to afford it if they want to...
>
>
> /Tony
> --
> /\___/\ Who would you like to read your messages today? /\___/\
> \_@ @_/ Protect your privacy: <http://www.pgpi.com/> \_@ @_/
> --oOO-(_)-OOo---------------------------------------------oOO-(_)-OOo--
> DSS: 0x9363F1DB, Fp: 6EA2 618F 6D21 91D3 2D82 78A6 647F F247 9363 F1DB
> ---ôôô---ôôô-----------------------------------------------ôôô---ôôô---
> \O/ \O/ ©1999 <http://www.svanstrom.com/?ref=news> \O/ \O/
Thats_hersh_foll...@tc.umn.edu
Isai <is...@pathcom.com> wrote:
>You are right, of course, that it is not too difficult to handle all the
>communications/interface aspects. But are there any programs in existence
>that can consistently beat reasonably good human players?
The answer to this is no (IMHO of course). The closest I have seen is
Pokibot on IRC. Pokibot is not bad but I would welcome him at my
table. Poki probably could beat the 3/6 online, but not for so much
money that I would hire a team of programmers to write the interface.
A decent programmer could make much more money elsewhere.
TheBigDenominations
Why can't you have a little respect for other peoples
opinions?
OK, Here we go.
1) Point 1 is not bullshit - Lots of people are doing just
that right now. Think online banking, think PGP, think just
about any client-side application that takes "security"
seriously.
2) Your points about debugging and using windows emulators
are complete bullshit...
If the client app is running in encrypted memory YOU CANNOT
ACCESS ANY DATA EXCEPT FOR THE VIDEO OUTPUT.
3) Ditto for decompiling the program.
4) The Turing Type question was (surprise!) JUST AN
EXAMPLE, anyway let me know when you solve that one between
supper and bedtime.
5) Just like in "real" life, the good guys hold all the
cards. An online site could wipe hundreds of 000's of
development $'s simply by making some minor changes to the
client software. As they say on Paradise Poker "You will
not be able to play until you have updated your software -
click here to download"
6) You're the worst type of computer geek. Arrogant,
opinionated and showing no grasp of reality whatsoever.
7) It's people like you that give our industry a bad image
in the "real world" (Yes Tony, there is a world outside of
computers)
PS. I had a look at your website, it's littered with bad
links - very sloppy work. "If it's not worth publishing
properly, it's not worth publishing at all"
<end of communication>
HrgSmes
begin to play l on l with someone I can't see.
It is daunting to think someone could design a robot program. Maybe I am a
robot program and I don't know it. Coujld account for a lot of other things
going wrong. . Bah.
Best. HR GREENBERG ENDit
JW Steve
Svanstrom) wrote:
<snip>
>
>Within 3 months I could have a "robot" controlling the computer just
>like a human would... That would require me to bring in some people that
>want money for their work, but I'm sure there's people here with big
>enough bankroll to be able to afford it if they want to...
>
It's a very interesting situation and one I hadn't considered. From a
technical standpoint, I agree that it could be done in the time frame
you describe, or maybe a little longer, but certainly not 5 years.
An interesting question is... how easy is it to program the AI so that
it always wins? Something even more dangerous than the purely
computer player would be an online assistant. One that takes in all
of the information presented and displays to you the best play. If
someone could develop this and they were smart, they wouldn't try to
have it play by itself. They would simply have it tell them the
correct play and they do it. I see no defense to this. Hell,
someone in the know could even package and sell an online assistant
that works with certain sites (although, the bots playing each other
would be interesting..). Or worse yet, develop and give it out for
free, virtually ruining the site.
jw steve
JW Steve
Svanstrom) wrote:
>JW Steve <jw_s...@hotmail.com> wrote:
>
>> On Sun, 30 Jan 2000 01:32:35 +0100, to...@svanstrom.com (Tony L.
>> Svanstrom) wrote:
>
>> >Within 3 months I could have a "robot" controlling the computer just like
>> >a human would... That would require me to bring in some people that want
>> >money for their work, but I'm sure there's people here with big enough
>> >bankroll to be able to afford it if they want to...
>
>> It's a very interesting situation and one I hadn't considered. From a
>> technical standpoint, I agree that it could be done in the time frame you
>> describe, or maybe a little longer, but certainly not 5 years.
>
>It will be more or less the same as runing the client inside an
>emulator, it just sounds so much more undetectable that I thought I
>should mention it; just to prevent a useless "it can be done" "no" "it
>can be done" "no"-thing.
>
>> An interesting question is... how easy is it to program the AI so that
>> it always wins?
>
>Here you have two things to consider; the first one is how hard it is to
>figure out the correct odds and the other one is how hard it is to make
>it act human. The first one is easily done, it's what computers do best,
>but what about the second one? Well, that's a question about what you
When you say, correct odds, are you considering any human element? By
saying it's "easily done" it leads me to believe that you are
considering some *basic strategy* for poker. I guess this means that
you would plug in some sort of calculation of EV and let the computer
run. The problem is that EV relies on what you think the other
players are going to in the future. I don't believe this is easy and
as someone pointed out an experienced player would gladly play with
pokibot which seems to have been developed with a decent amount of
research and insight.
It seems to me that a good poker player benefits from intangible
player reads, intuition, etc. moreso than almost any other game. I
think you are oversimplifying the concept of creating a winning bot.
The simpler it is, the easier it will be to beat, and once someone
figures out your bots nuances, they will be able to clean your clock.
jw steve
<snip>
Andrew Prock
But in the end, it *is* a possibility. In fact,
I know people who certainly could handle any kind
of encryption except the most paranoid.
Tools *do* exist which allow you to do almost
anything to running programs, including rewrite
them while they are running (www.cs.wisc.edu/projects/dyninstAPI).
These tools are very sophistocated and have been
used to do things like rewrite kernels while they
are running, and break liscence server protection
for software.
So...
Technically, it is possible to make the interface.
Pokibot (www.cs.ualberta.ca/~games/poker/)
also shows that it is possible to write a
bot which can make a *significant* amount of money
playing against good players.
On the other hand, as Doug points out, it is almost
always going to more profitable for someone with
such skills to use their skills honestly to make
money.
On the other hand, 16-year olds with too much time
on their hands have been known to do the darndest
things.
- Andrew
Thats_hersh_foll...@tc.umn.edu
>also shows that it is possible to write a
>bot which can make a *significant* amount of money
>playing against good players.
Perhaps I need to back off from my original statements a bit. When I
last played against pokibot (a few months ago) he had a pretty low win
rate. He was ahead something like .02 sb/hand. I don't think that is
enough to overcome the rake. However, I just checked back and he is
pulling in a whopping .18 sb/hand over 4000 hands. This is a pretty good
win rate. If most of this is in the 20/40 game then I am impressed.
I have not played against poki lately. Perhaps I would not be so
confident of my ability to beat him if I did.
>On the other hand, as Doug points out, it is almost
>always going to more profitable for someone with
>such skills to use their skills honestly to make
>money.
One thing that I object to is the idea that it would be dishonest to
have a computer playing poker. This is sort of like saying that it
would be dishonest for a professional poker player to play low limit,
or dishonest to take notes or use a calculator to figure pot odds. I
would not even mind if the casino themselves put in a few computer
props to bolster a game that was starting to sag. I would rather have
to play against a couple of computer players in the lineup than have
the game break up. I am just not afraid of a couple of tight solid
players at my table. There is usually enough loose money to go around.
Note: Some people would object to the casino having props because they
would be afraid that the dealers would cheat in favor of the props.
This is also a common concern in live poker. But, I figure if you can't
trust the casino not to dish off to the props then you also can't trust
them not to dish off to friends of the casino. At least a prop will be
labeled as a prop and you can watch them accordingly.
Andrew Prock
>
>>On the other hand, as Doug points out, it is almost
>>always going to more profitable for someone with
>>such skills to use their skills honestly to make
>>money.
>
>One thing that I object to is the idea that it would be dishonest to
>have a computer playing poker.
In general, it would only be dishonest if you tried to
convince the world that it *wasn't* a computer player,
something which I think would be necessary if you used
an online play for money situation.
- Andrew
JW Steve
<Thats_hersh_foll...@tc.umn.edu> wrote:
>In article <877gmf$4...@spool.cs.wisc.edu>,
>Andrew Prock <jeffy...@yahoo.com> wrote:
>>Both Tony and Doug bring up some good points.
>>
>>Pokibot (www.cs.ualberta.ca/~games/poker/)
>>also shows that it is possible to write a
>>bot which can make a *significant* amount of money
>>playing against good players.
>
>Perhaps I need to back off from my original statements a bit. When I
>last played against pokibot (a few months ago) he had a pretty low win
>rate. He was ahead something like .02 sb/hand. I don't think that is
>enough to overcome the rake. However, I just checked back and he is
>pulling in a whopping .18 sb/hand over 4000 hands. This is a pretty good
>win rate. If most of this is in the 20/40 game then I am impressed.
I think Poki plays in the 10/20 game almost always (at least since
I've been playing). This is the same game that seems to attract
people starting with their token $1000 and trying to get the $3000 to
get to pot limit ASAP. I was in a game today that where several
people were betting like crazy raising the whole time. I had flopped
a four flush and had the nut flush after the turn and was able to cap
the turn and the river. The one guy who was betting folded on the
last $20 call, pot odds were 34 to 1. Playing completely
nonsensically. Several hands later I played a pair of twos and
flopped a boat. Bets were capped until the river where someone turned
over a 95o which meant they were playing the board pretty much. It
was idiotic. At that point, someone said something like, I don't we
all want to get to pot limit? Then let's bet every hand. At that
point I left the game.
The point is obvious, poki's competition isn't all that good.
>I have not played against poki lately. Perhaps I would not be so
>confident of my ability to beat him if I did.
Not sure, but you are also talking about a presumably pretty complex
bot. Is it learning from experience or did they tweak it, or did the
game just get softer?
<snip>
jw steve
Linda Sherman
>
> Both Tony and Doug bring up some good points.
>
> I know people who certainly could handle any kind
> of encryption except the most paranoid.
>
> Tools *do* exist which allow you to do almost
> anything to running programs, including rewrite
> them while they are running (www.cs.wisc.edu/projects/dyninstAPI).
>
> These tools are very sophistocated and have been
> used to do things like rewrite kernels while they
> are running, and break liscence server protection
> for software.
Except there's a cheap, simple, and elegant defense against this sort of
hacking - just change your software periodically. It doesn't take a big
change - a few lines of code will throw all the addresses off and
totally break the hack.
The online outfits could make you download a new client every so often.
I'm surprised they don't, frankly.
Lin
--
Linda K Sherman <lin...@concentric.net>
lins...@gte.net www.cti-pro.com www.dalati.com
Andrew Prock
>Andrew Prock wrote:
>>
>> Both Tony and Doug bring up some good points.
>>
>> But in the end, it *is* a possibility. In fact,
>> I know people who certainly could handle any kind
>> of encryption except the most paranoid.
>>
>> Tools *do* exist which allow you to do almost
>> anything to running programs, including rewrite
>> them while they are running (www.cs.wisc.edu/projects/dyninstAPI).
>>
>> These tools are very sophistocated and have been
>> used to do things like rewrite kernels while they
>> are running, and break liscence server protection
>> for software.
>
>Except there's a cheap, simple, and elegant defense against this sort of
>hacking - just change your software periodically. It doesn't take a big
>change - a few lines of code will throw all the addresses off and
>totally break the hack.
Actually, they would pretty much have to redesign the
entire communication layer of their program to make this
kind of fix effective.
The sophistocated tools that are used today *don't* operate
on the address level, they operate on the function call level.
If you just changed the execuatble a bit, tools like dyninst
are still going to be able to trap the same function calls
and manipulate them accordingly.
If you just change the name of the function call, all you
have to do is find the new name. I was talking to someone
who works on the project yesterday and he said that unless
there was some kind of hardware level encryption he should
be able to "hack" the program as long as he can get it
running properly in the fisrt place.
- Andrew
Andrew Prock
>The point is obvious, poki's competition isn't all that good.
Don't look at how poki does absolutly, look at how he does
relativly. Few players earn as much as it does, and the ones
that do are expert players.
>Not sure, but you are also talking about a presumably pretty complex
>bot. Is it learning from experience or did they tweak it, or did the
>game just get softer?
You really should read the papers. Poki does opponent modeling.
He uses how you've played your hands in the past to set up an
internal simulation which is used to select check/bet/fold actions.
Regardless of how soft the game is Poki is going to be one of the
top money earners.
- Andrew
JW Steve
I am not sure if "read the papers" is sarcasm or you mean read the
research papers. If you mean the latter, I will probably check them
out eventually. For now I was hoping that someone who had could
answer a simple question. How hard is it, how much development went
into poki? Obviously if poki has to lose in order to learn, this
might not be a good option. I've also noticed that poki seems to only
play with full tables and plays all day long. If it took them 2 years
of development to get him to win, etc.. it might not be worth it. It
goes back to my first question that no one really answered. How hard
is it?
Anyway, I don't doubt that poki is good, but the results of many
institute research projects are. If the people who created poki
weren't willing to help defraud everyone else and help out, how hard
is it for someone else to create a winning bot?
Do you know? If not, who does?
jw steve
Andrew Prock
>On 2 Feb 2000 15:56:17 GMT, jeffy...@yahoo.com (Andrew Prock) wrote:
>>You really should read the papers. Poki does opponent modeling.
>>He uses how you've played your hands in the past to set up an
>>internal simulation which is used to select check/bet/fold actions.
>>Regardless of how soft the game is Poki is going to be one of the
>>top money earners.
>
>I am not sure if "read the papers" is sarcasm or you mean read the
>research papers. If you mean the latter, I will probably check them
>out eventually.
I meant read the research papers. Most of the questions you
ask are addressed in the papers.
>For now I was hoping that someone who had could
>answer a simple question. How hard is it, how much development went
>into poki?
Ph.D. researchers are working on it. They've been working on the
research for about three years I think.
>Obviously if poki has to lose in order to learn, this
>might not be a good option.
See the papers.
>I've also noticed that poki seems to only
>play with full tables and plays all day long.
Full table play and shorthanded are very different. Poki has
been "trained" to play full table games.
>If it took them 2 years
>of development to get him to win, etc.. it might not be worth it. It
>goes back to my first question that no one really answered. How hard
>is it?
I could implement the papers in a month or two of solid coding.
Is that hard? It really depends on what you mean by hard.
>Anyway, I don't doubt that poki is good, but the results of many
>institute research projects are. If the people who created poki
>weren't willing to help defraud everyone else and help out, how hard
>is it for someone else to create a winning bot?
Using the UA work as the basis for creating a winning bot wouldn't
be astronomically difficult for most AI researchers. Creating a
stable interface through which the bot could play wouldn't be
terribly difficult for most OS researchers.
The hardest thing is to get people with the two skill sets
together and convince them that writing an automated poker
player that plays online for real money is worth their time.
- Andrew
JW Steve
>According to JW Steve <jw_s...@hotmail.com>:
>>On 2 Feb 2000 15:56:17 GMT, jeffy...@yahoo.com (Andrew Prock) wrote:
<snip>
>>Anyway, I don't doubt that poki is good, but the results of many
>>institute research projects are. If the people who created poki
>>weren't willing to help defraud everyone else and help out, how hard
>>is it for someone else to create a winning bot?
>
>Using the UA work as the basis for creating a winning bot wouldn't
>be astronomically difficult for most AI researchers. Creating a
>stable interface through which the bot could play wouldn't be
>terribly difficult for most OS researchers.
>
>The hardest thing is to get people with the two skill sets
>together and convince them that writing an automated poker
>player that plays online for real money is worth their time.
Not to mention fraudulent and probably illegal.
From what you've told me, I am going to classify the task as *very
difficult*. The fact that poki with several years of research behind
it can't adjust to non-full tables would make it next to useless in
trying to deceive a casino. It leads me to believe you would have to
put in a great deal of effort in the AI end, a decent amount of effort
and manipulation on interface end. All of this to do something
fraudulent in the hopes that you won't be detected and that the
maintenance costs won't be prohibitive.
Sounds like a bad bet to me. Let me ask you this... what do you think
of Tony's 10K to put together the whole package? I'd say he's
probably off by a factor of 10 at least.
jw steve
>- Andrew
>
>
>
>
Andrew Prock
>>The hardest thing is to get people with the two skill sets
>>together and convince them that writing an automated poker
>>player that plays online for real money is worth their time.
>
>Not to mention fraudulent and probably illegal.
The legality of such a program is a very gray area. Not
only because online wagering in a gray area, but also because
there is only an implied assumption that all opponents are
human. You certainly aren't going to go to jail for creating
such a program.
>From what you've told me, I am going to classify the task as *very
>difficult*.
For the average person certainly. For a certain class of
experts, not really. The biggest problem, as I said, would
be convincing them it was worth their time.
>Sounds like a bad bet to me. Let me ask you this... what do you think
>of Tony's 10K to put together the whole package? I'd say he's
>probably off by a factor of 10 at least.
Probably not. My guess is that you could probably find
capable people who could pull off the contract in a month
for less than $20K. Unfortunatly, this wouldn't include
maintenance costs (as the sites try and keep the bots off
you'll have to come up with more counter-measures).
- Andrew
JW Steve
>According to JW Steve <jw_s...@hotmail.com>:
>
>>>The hardest thing is to get people with the two skill sets
>>>together and convince them that writing an automated poker
>>>player that plays online for real money is worth their time.
>>
>>Not to mention fraudulent and probably illegal.
>
>The legality of such a program is a very gray area. Not
>only because online wagering in a gray area, but also because
>there is only an implied assumption that all opponents are
>human. You certainly aren't going to go to jail for creating
>such a program.
No.. but there is an assumption by me that most academic types have
loftier goals than cheating some rubes out of money.
>>From what you've told me, I am going to classify the task as *very
>>difficult*.
>
>For the average person certainly. For a certain class of
>experts, not really. The biggest problem, as I said, would
>be convincing them it was worth their time.
Ahh. yeah... for classes of experts very difficult things aren't
difficult, that's why they are experts.
>>Sounds like a bad bet to me. Let me ask you this... what do you think
>>of Tony's 10K to put together the whole package? I'd say he's
>>probably off by a factor of 10 at least.
>
>Probably not. My guess is that you could probably find
>capable people who could pull off the contract in a month
>for less than $20K. Unfortunatly, this wouldn't include
>maintenance costs (as the sites try and keep the bots off
>you'll have to come up with more counter-measures).
That's an interesting amount. In Boston doing a little project
management for an Internet startup and I have a hard time finding
someone who can create a web directoy of maybe 10-15 pages with a
modest amount of cgi scripting for under 10K. You'd have a really
tough time convincing me you can find two experts in two different
fields to do this job for 20K.
jw steve
JW Steve
Svanstrom) wrote:
>Andrew Prock <jeffy...@yahoo.com> wrote:
>
>> According to JW Steve <jw_s...@hotmail.com>:
>
>> >Let me ask you this... what do you think of Tony's 10K to put together
>> >the whole package? I'd say he's probably off by a factor of 10 at least.
>>
>> Probably not. My guess is that you could probably find capable people who
>> could pull off the contract in a month for less than $20K. Unfortunatly,
>> this wouldn't include maintenance costs (as the sites try and keep the
>> bots off you'll have to come up with more counter-measures).
>
>I think I said something about if the right people are available... The
>main guy I was thinking about is teaching advanced programming at
>university-level; so that's pretty "capable".
>
I find most people teach either because it's easier than applying the
subject matter and they can't hack it in the real world, teaching and
consulting is lucrative and flexible and lower stress, or they think
they are making a difference and are well intentioned.
The first type wouldn't be very good, the second would cost you an arm
and a leg and the third wouldn't be part of such a scheme. Which type
is the guy you are thinking of?
jw steve
Gary Carson
JW Steve wrote in message <38999b92....@news.psn.net>...
>On 2 Feb 2000 19:45:34 GMT, jeffy...@yahoo.com (Andrew Prock) wrote:
>
>>According to JW Steve <jw_s...@hotmail.com>:
>>
>
>No.. but there is an assumption by me that most academic types have
>loftier goals than cheating some rubes out of money.
I think you'd be wrong. Most academic types become academic types becuase
it's much easier than getting a real job. At least that was my obervation
when I was an academic type. Many of them would think find the prospects of
cheating some rubes out of some money an attractive and entertianing
diversion. But, maybe my sample is biased somewhat by the self-selection of
only hanging out with co-workers that shared my attitudes back when I was an
academic type.
Gary Carson
>
>>>From what you've told me, I am going to classify the task as *very
>>>difficult*.
>>
>>For the average person certainly. For a certain class of
>>experts, not really. The biggest problem, as I said, would
>>be convincing them it was worth their time.
>
>Ahh. yeah... for classes of experts very difficult things aren't
>difficult, that's why they are experts.
>
>>>Sounds like a bad bet to me. Let me ask you this... what do you think
>>>of Tony's 10K to put together the whole package? I'd say he's
>>>probably off by a factor of 10 at least.
>>
>>Probably not. My guess is that you could probably find
>>capable people who could pull off the contract in a month
>>for less than $20K. Unfortunatly, this wouldn't include
>>maintenance costs (as the sites try and keep the bots off
>>you'll have to come up with more counter-measures).
>
JW Steve
<garyc...@mindspring.com> wrote:
>
>
>JW Steve wrote in message <38999b92....@news.psn.net>...
>>On 2 Feb 2000 19:45:34 GMT, jeffy...@yahoo.com (Andrew Prock) wrote:
>>
>>>According to JW Steve <jw_s...@hotmail.com>:
>>>
>>
>>No.. but there is an assumption by me that most academic types have
>>loftier goals than cheating some rubes out of money.
>
>I think you'd be wrong. Most academic types become academic types becuase
>it's much easier than getting a real job.
Our experiences differ greatly. My experiences in technical fields is
that the most qualified people are there because they can't get a
green card and work, so they are stuck doing post-doc work. Another
subset of people are in graduate school simply to get the PhD and
teach. Others are looking for the edge an MS or PhD brings in the job
market. Still others are truly interested in the subject matter and
are trying to make some difference in the world. Can't say I met many
of your type.
> At least that was my obervation
>when I was an academic type. Many of them would think find the prospects of
>cheating some rubes out of some money an attractive and entertianing
>diversion.
>But, maybe my sample is biased somewhat by the self-selection of
>only hanging out with co-workers that shared my attitudes back when I was an
>academic type.
I hope you understand I wasn't talking about just any student. The
discussion was about experts in their field who were also in academia.
Was that you and your friends? The graduate school I attended, people
worked harder than anyone with a 9-5 job, and it was quite
competitive, in fact some people paid to attend graduate school, might
explain why nobody was there for the reasons you describe.
jw steve
Andrew Prock
>>Probably not. My guess is that you could probably find
>>capable people who could pull off the contract in a month
>>for less than $20K. Unfortunatly, this wouldn't include
>>maintenance costs (as the sites try and keep the bots off
>>you'll have to come up with more counter-measures).
>
>That's an interesting amount. In Boston...
Well that's the coasts for you. They are expensive.
>You'd have a really
>tough time convincing me you can find two experts in two different
>fields to do this job for 20K.
Your best chance for this would be to hire
CS graduate students with solid resumes, in the
appropriate sub-fields, who are looking for something
to do over the summer. Most of them would be happy to
work for a month and get $10K out of the deal. Heck,
for that price you might even get two months of work out
of them, as long as they were planning on returning to
grad school after the summer.
Going rate for someone with a masters in computer science
is about $60K - as a lower bound. Most every CS grad student
will work at this level for a summer.
- Andrew
JW Steve
Svanstrom) wrote:
>JW Steve <jw_s...@hotmail.com> wrote:
<snip>
>Now does the 10k sound enough?
>
Sure.. but if I was footing the bill it would be contingent on the
fact that the thing was going to;
a) work undetected
b) win regularly until I get my money back
Problem with CS is that people want to work for time and expense with
no gaurantee of any results.
jw steve
Gary Carson
JW Steve wrote in message <3898a8fc....@news.psn.net>...
>On Wed, 2 Feb 2000 15:24:00 -0600, "Gary Carson"
><garyc...@mindspring.com> wrote:
>
>>
>>
>>JW Steve wrote in message <38999b92....@news.psn.net>...
>>>On 2 Feb 2000 19:45:34 GMT, jeffy...@yahoo.com (Andrew Prock) wrote:
>>>
>>>>According to JW Steve <jw_s...@hotmail.com>:
>>>>
>>>
I consider graduate students to be graduate students. By academic types I
mean faculty members or those graduate students who aspire to become facutly
members.
As far as paying to go to graduate school, I never knew many of those. I
can only think of one offhand. Professional programs like law school or MBA
programs are differently, most of those I've known where paying. But, not
in academic programs. I sure would have never attended a graduate program
that expected me to pay.
How many hours you put in isn't a very good measure of the challange or risk
of the job, btw. The competivite nature of academic success is very
different from real world success.
One of my professors (when I was a grad student in engineering) said it best
I think, "In the real world you are expected to solve problems that might
not have nice solutions. In academia I get to define a problem that I
already know how to solve, then solve it, and I'm considered a success". He
was a professor of IE/MS at Nothwestern then, he's a professor of CE at
Princeton now. I think he's an academic success.
Gary Carson
Gary Carson
JW Steve wrote in message <3899adc4....@news.psn.net>...
The problem with you is that you want the benifits of ownership but want
some wage earner to assume ownership risks for you.
Gary Carson
>
>jw steve
>
>
JW Steve
Svanstrom) wrote:
>JW Steve <jw_s...@hotmail.com> wrote:
>
>> I never meant to imply that a) I was seriously offering anyone 10K
>
>Yeah, right, I have a whole NG that can back me up... C ya in court,
>pal. ;-)
>Damn, I wish I was living in the US, then I'd instantly pick up the
>national hobby of suing people; hey, you only have to get lucky once so
>bring on the hot water... *LOL*
>
It's a tough hobby.. you have to go to law school which usually
involves going into debt around 60K depending on how good of a school
you want to attend.
jw steve
Chris Downs
> >difficult*.
>
> For the average person certainly. For a certain class of
> experts, not really. The biggest problem, as I said, would
> be convincing them it was worth their time.
Consider how much time and effort went in to creating TTH. Would that
level of play be good enough for your bot? If so, you are talking a
tremendous amount of effort/manpower.
TheBigDenominations
permutations calculable.
Poker is a "Closed" game - Any game where one side has
information that the other side does not.
After about 20 years and millions of $'s (Mainly IBM's) a
computer program better than the best human player was
finally created.
The reason it took so long for brute computing power to
overcome humanity at chess was the fact that human
intelligence can "discard" so many permutations and options
that a computer would have to process...
The Chess Player would say "Well, I won't move my Knight
cause that clearly won't help" while the computer has to
process the millions of permutations arising from a move of
the Knight to see if anything good happens.
This human advantage is more clearly evident in closed
games because simple analsis along the lines of "He's
called under the gun and I KNOW THIS PLAYER DOESNT EVER
CALL A7 LET ALONE UTG Therefore ...." is so much harder for
computers.
To mention another closed game, Bridge.
Again, at least Five Million $ have been spent on
development and only now do we have a computer program that
can play at "club" (Translation for Poker and non-bridge
Players - "Totally Bonkers Fish") level.
So (IMHO) to program a computer to play seriously quality
poker is going to be very hard.....
Regards,
The Big Denominations
William Loughborough
program better than the best human player was finally created."
I know it won but the playing parameters were rigged in its favor. The
commercial showing David Robinson dribbling rings around it was a
metaphor for how lame it actually is.
"So (IMHO) to program a computer to play seriously quality poker is
going to be very hard....."
That's the real truth and in this case "very hard" is very close to
impossible. The people who would sell you their services are always
misguided hucksters. Estimates of $20k to do a meaningful poker bot are
fantasies by people who have no clue as to how absurd their claims are.
It will always take longer, cost more, and not fulfill promises. One of
the ultimate oxymorons is "artificial intelligence" because what is so
touted is actually neither artificial nor intelligent.
As to "cracking the shuffle/deal algorithm": it would be fairly simple
to make a machine that shuffled and dealt a bunch of decks and no part
of the program would know the card sequence until the deal. No seeds, no
RNG, just shuffle, cut and deal.
--
Love.
ACCESSIBILITY IS RIGHT - NOT PRIVILEGE
http://dicomp.pair.com
William Loughborough
spare time that could beat a typical LA 3-6 Omaha game."
WL: I was responding to a post that dealt with a program that beat the
world's champion chess player. Other posts have been about beating
*real* games, however that's defined. From what I've seen of online
poker I will take your bet if the last phrase is "paradisepoker 10-20 HE
game." No need to automate it, just follow its instructions as you
perform the interface chores. It's a lot easier to type the words "write
a bot" than to do it. Lots of people who are in fact ace programmers
have tried and so far the results are at best spotty. This whole thread
is populated by the usual confusion between "simple" and "easy". It's
like speech recognition or OCR, the claims look good in Popular Science
but when you buy the program it usually sits on the shelf after you find
it's more hype than help.
Jerrod Ankenman
: That's the real truth and in this case "very hard" is very close to
: impossible. The people who would sell you their services are always
: misguided hucksters. Estimates of $20k to do a meaningful poker bot are
: fantasies by people who have no clue as to how absurd their claims are.
: It will always take longer, cost more, and not fulfill promises. One of
: the ultimate oxymorons is "artificial intelligence" because what is so
: touted is actually neither artificial nor intelligent.
Define 'meaningful.'
I'm not an ace programmer, but I bet I could write a bot in my spare time
that could beat a typical LA 3-6 Omaha game.
It wouldn't be artificially intelligent. It would just play very
selectively, and only bet the nuts or draws thereto.
I don't even think it would be difficult.
Will it beat the omaha section of the HOE tournament at ESCARGOT?
No.
I think mayhaps you underestimate the absurd weakness of very many many
poker players.
Jerrod Ankenman
jer...@crl.com
Matt Hill
I've been following this thread and I think I have something valid
to add. If you have ever played any of Wilson's Hold Em software, like
"Turbo Texas Hold Em", then it should be obvious that Pokibot is not the
only poker robot out there. Turbo Texas Hold Em has some tough opponents,
which, while beatable, are probably better than 90% of the people in the
2-4 game I play in at Paradise Poker.
The authors of that software could easily write an interface
between their existing software and an online cardroom. The AI part is
already there, they would simply need to modify it...
=]
matt hill
Boulder, CO
USA
> On Thu, 3 Feb 2000 02:50:01 +0100,
to...@svanstrom.com (Tony L. > Svanstrom) wrote:
>
> >JW Steve <jw_s...@hotmail.com> wrote:
> >
> >> On 3 Feb 2000 00:04:31 GMT, jeffy...@yahoo.com (Andrew Prock) wrote:
> >>
> >> >According to JW Steve <jw_s...@hotmail.com>:
> >> >>
> >> >>It's not a risk for the wage earner if he knows he can do the job.
> >> >>Kind of like an engineer who accepts the task of designing a new
> >> >>building. I didn't see many designs that said "this may or may not
> >> >>work".
> >> >
> >> >Well sort-of. You also never see Engineers/Architects that work
> >> >for employers whose contracts are:
> >> >
> >> >"we'll only pay you if the building you design is profitable."
> >>
> >> Cute, but not a very good analogy. A successful building design is
> >> one that stays up, a successful pokerbot design is one that wins. The
> >> latter equates directly to earning money, the former doesn't. If an
> >> engineer stands behind his design (which he has no choice) he is
> >> assuring it will stand up, if a pokerbot designer stands behind his
> >> design (up to him), he should be willing to assure it will win.
> >
> >No, a successful building design is one that works as intended; you go
> >to an architect and you say "this is what I want and how it should work"
> >and they try to design/build it.
> >Later on you discover that the elevators should have been 30 meters to
> >the right, that it should have had one less elevator and that the
> >material used for the floor isn't hard enough...
>
> Please tell me about forensic structural engineering.. LOL.. I did it
> for years. Please.. tell me about the relationship between architect,
> engineer (you seem to think they are the same), project managers, GC's
> (means..), sub-contractors, owners. Tell me about the contractual
> obligations between the groups involved. Tell me about bonding and
> how that enters into the picture. Then elaborate on how/why a dispute
> might be resolved.
>
> Your description above demonstrates your ignorance on the subject.
> Later on you don't "discover the elevators should have been 30 meters
> to the right". If you "discovered" that, the owner sure as hell isn't
> going to pay for it. Someone made a mistake and someone is going to
> pay for it. In structural engineering, the cost for repair and
> retrofit is often 10 times what it cost to do something in the first
> place. Mistakes are made, but liability is assessed and someone pays.
> If you make too many mistakes, you go out of business. Simple as
> that.
>
> >Same thing with a bot
> >playing poker; you design it for a certain situation, and then you might
> >find that the situation to use it in isn't as planned... Then you
> >upgrade it.
>
> Of course and if I were seriously entertaining the thought of hiring
> someone to do something like this, I wouldn't put such rigid
> constraints on.
>
> >If you think that 10'000 is enough to create a bot that goes into a
> >2000/4000 game and beats the crap out of them then you just don't
> >understand how things work IRL.
>
> Now Tony, that's called strawman. I've never said anything remotely
> like that. In fact, the 10K figure was your number, not mine. I said
> it was probably a factor of ten more. What I did say is that IF I
> were to pony up the money I would put on tight restrictions. The
> reason (which you and Andy both seem to have missed) is not some
> ignorance of wage economics, it's because I don't believe you can do
> it for 10K. Get it? If someone severely underbids a job, I want to
> put extra requirements on it to insure that this person is not just
> talking out of their ass.
>
> jw steve
>
>
>