| |
 |
|
| Science Forum Index » Mathematics Forum » Online poker RNG... |
|
Page 1 of 2 Goto page 1, 2 Next |
|
| Author |
Message |
| Gerry |
 Posted: Sat Feb 18, 2006 7:49 pm |
|
|
|
Guest
|
I'm hoping someone(s) on here will be kind enough to help ease my mind
(or, for that matter, to verify my reasons for concern!) And I should
say that, while I'm a good poker player, I'm only a marginal student of
mathematics and computing, so keep that in mind, please, as you answer
this question, but:
Is it possible that, as I'm playing online, any of the other players
are cheating by way of predicting the order of the cards in the
"shuffled" deck?
A few (some math, some poker) truths to keep in mind as you answer:
1. Many sites use an RNG algorithm based on 2^32. Isn't that pretty
easily susceptible to brute-force attacks from someone who has a
program relying on distributed computing?
2. In the most popular type of online poker (Texas Hold 'Em), there are
four rounds of betting, and the most important three rounds come AFTER
(a) each player gets two (face-down) cards and (b) three face-up cards
are dealt on the table -- SO, if the cheateer is, say, the third player
to be dealt cards at a table of ten players, then he will know the 3rd
and 13th cards (i.e. the two cards he's been dealt) as well as the
21st, 22nd and 23rd cards (the three dealt face-up immediately
following the two distributed to each player).
So, I guess my question is: how easily would someone be able -- either
by brute-force run-through of all possible combinations of "deck
arrangements" with those 5 cards in those specific spots, or by some
sort of reverse-engineering, or whatever -- to conclude the sequence of
cards in the deck (which, of course, would give him knowledge of
everyone's cards, as well as the two cards left to be dealt)?
And, for that matter, a related question: there's a piece of software
currently on the market (Check out pokerrng.com.) which claims that, if
a player enters into the software the details of a few thousands
recently dealt hands, the software will "synch" with the poker system's
RNG well enough to predict three of the five cards (not suits, but
values, which is enough to be dangerous!) left to be dealt! Is this
even plausible? Do I need to be afraid of this type of thing?
Anything online has inherent pitfalls, and cheating is -- both online
and offline -- a fact of life, but there's a certain level of security
I'd like, and I wonder if I'm jumping at shadows or rightly
concerned...
Any thoughts would be appreciated!
Thanks,
Gerry |
|
|
| Back to top |
|
|
|
| richard.blankman@gmail.co |
| Posted: Sat Feb 18, 2006 11:31 pm |
|
|
|
Guest
|
Gerry wrote:
[quote:f19630bc7f]I'm hoping someone(s) on here will be kind enough to help ease my mind
(or, for that matter, to verify my reasons for concern!) And I should
say that, while I'm a good poker player, I'm only a marginal student of
mathematics and computing, so keep that in mind, please, as you answer
this question, but:
Is it possible that, as I'm playing online, any of the other players
are cheating by way of predicting the order of the cards in the
"shuffled" deck?
[/quote:f19630bc7f]
Unless the programmers are complete idiots, no.
[quote:f19630bc7f]A few (some math, some poker) truths to keep in mind as you answer:
1. Many sites use an RNG algorithm based on 2^32. Isn't that pretty
easily susceptible to brute-force attacks from someone who has a
program relying on distributed computing?
[/quote:f19630bc7f]
An "algorithm based on 2^32?" So they're using one of infinitely many
algorithms based on the number 4,294,967,296? What if I told you I'm
picking a number using an algorithm based on the number 5? Would you
have _any_ idea at all what my number is? Now imagine it's based on 4
billion.... Even if someone wrote a program only searching through
algorithms conceivably based on 2^32, it would take years for it to
even find one that accurately matched the cards dealt during several
games of poker. And there's still no guarantee that'd be the right one.
[quote:f19630bc7f]2. In the most popular type of online poker (Texas Hold 'Em), there are
four rounds of betting, and the most important three rounds come AFTER
(a) each player gets two (face-down) cards and (b) three face-up cards
are dealt on the table -- SO, if the cheateer is, say, the third player
to be dealt cards at a table of ten players, then he will know the 3rd
and 13th cards (i.e. the two cards he's been dealt) as well as the
21st, 22nd and 23rd cards (the three dealt face-up immediately
following the two distributed to each player).
[/quote:f19630bc7f]
See above. I was assuming they were searching for an algorithm where
every card matched what was dealt (which is of course only rarely
revealed). If you're searching for algorithms with unknown cards, not
only would it still take quite a while, but there's absolutely no good
reason to assume any "found algorithm" comes anywhere near accurately
predicting what will come.
[quote:f19630bc7f]So, I guess my question is: how easily would someone be able -- either
by brute-force run-through of all possible combinations of "deck
arrangements" with those 5 cards in those specific spots, or by some
sort of reverse-engineering, or whatever -- to conclude the sequence of
cards in the deck (which, of course, would give him knowledge of
everyone's cards, as well as the two cards left to be dealt)?
[/quote:f19630bc7f]
Not easily at all. Realistically impossible given the speed of modern
day computers and infinite possible algorithms used by the programmers.
[quote:f19630bc7f]And, for that matter, a related question: there's a piece of software
currently on the market (Check out pokerrng.com.) which claims that, if
a player enters into the software the details of a few thousands
recently dealt hands, the software will "synch" with the poker system's
RNG well enough to predict three of the five cards (not suits, but
values, which is enough to be dangerous!) left to be dealt! Is this
even plausible? Do I need to be afraid of this type of thing?
[/quote:f19630bc7f]
Might work for extremely unprofessional poker sites where you have high
school students and extremely amateur programmers making poker games.
For any site where you can actually win money, or the programmers are
reputable (Yahoo, etc.), no, it's essentially implausible.
[quote:f19630bc7f]Anything online has inherent pitfalls, and cheating is -- both online
and offline -- a fact of life, but there's a certain level of security
I'd like, and I wonder if I'm jumping at shadows or rightly
concerned...
Any thoughts would be appreciated!
Thanks,
Gerry
[/quote:f19630bc7f]
Stay out of those shadows, Gerry. Let probability be probability...
sometimes you just can't win. |
|
|
| Back to top |
|
|
|
| Gerry |
| Posted: Sun Feb 19, 2006 1:08 am |
|
|
|
Guest
|
Perhaps my lack of knowledge of math led me to write the phrase
"algorithm based on 2^32" when what I actually meant -- and precision
is key to your science, so I'll ask for forgiveness in having misstated
this important premise -- was "a 32-bit seed space (2^32 possible
seeds)."
This, as per an article published six years ago
(http://www.developer.com/java/other/article.php/616221), can be
exhaustively searched (and perfectly determined) in just over a day,
using what at that time was a relatively state-of-the-art Pentium 400.
(And, given what I know of averages, the "average" time to find the
correct seed would therefore be about 12 hours, if the time to look
through all possibilities is a day.)
Given a Pentium 4 at 3GHz, or, better yet, a few of them distributing
the processing, is it REALLY unlikely that a guy somewhere with a few
lines of codes and a couple of chips strung together is using a
brute-force attack to find the seed within the approximately 2 or 3
minutes (depending on the game) necessary for him to make the critical
decisions? |
|
|
| Back to top |
|
|
|
| Teslia |
| Posted: Sun Feb 19, 2006 1:51 am |
|
|
|
Guest
|
"Gerry" <gerald_helmling@yahoo.com> wrote in message
news:1140329333.108945.172750@g43g2000cwa.googlegroups.com...
[quote:fda1f95958]Perhaps my lack of knowledge of math led me to write the phrase
"algorithm based on 2^32" when what I actually meant -- and precision
is key to your science, so I'll ask for forgiveness in having misstated
this important premise -- was "a 32-bit seed space (2^32 possible
seeds)."
This, as per an article published six years ago
(http://www.developer.com/java/other/article.php/616221), can be
exhaustively searched (and perfectly determined) in just over a day,
using what at that time was a relatively state-of-the-art Pentium 400.
(And, given what I know of averages, the "average" time to find the
correct seed would therefore be about 12 hours, if the time to look
through all possibilities is a day.)
Given a Pentium 4 at 3GHz, or, better yet, a few of them distributing
the processing, is it REALLY unlikely that a guy somewhere with a few
lines of codes and a couple of chips strung together is using a
brute-force attack to find the seed within the approximately 2 or 3
minutes (depending on the game) necessary for him to make the critical
decisions?
[/quote:fda1f95958]
If he knows the algorithm, and where he is in the sequence,
32 bit seed space is too small, that is only 4 bytes!
Searching is futile, as the next card can be any of the not-yet-played
cards.
I would be more concerned that the site could be fixed so it seems that
other players win more often, perhaps one of their computer generated
players, and they do not have to win that much more to bring in $$ to a
site. |
|
|
| Back to top |
|
|
|
| Gerry |
| Posted: Sun Feb 19, 2006 2:56 am |
|
|
|
Guest
|
But the next card couldn't -- online, at least -- be "any of the
not-yet-played cards." I mean, theoretically, it could, but, in
practice, the fact is that the entire virtual "deck" is "shuffled"
prior to the dealing. In other words, all cards are placed in a certain
order, supposedly "randomly," based on the PRNG, etc. But, as the
article I referenced points out, a 32 bit seed space can be subject to
a brute force attack in which, given the exact placement of 5 specific
cards, a Pentium 400 takes a day -- and a distributed network only
minutes -- to "find" the correct shuffled deck of the, what, 4 billion
possible permutations. So, with the distributed network, and given that
the poker room (as some still do!) uses a 32 bit seed space, my
opponent's computer could figure out the configuration of the deck, and
thereby "know" the cards to come...Unless I'm missing some key
ingredient. I guess my underlying question is: what's changed since the
1999 article I referenced, insofar as, at that time, the authors were
able to "sync" to the PRNG? |
|
|
| Back to top |
|
|
|
| David C. Ullrich |
| Posted: Sun Feb 19, 2006 8:15 am |
|
|
|
Guest
|
On 18 Feb 2006 16:49:23 -0800, "Gerry" <gerald_helmling@yahoo.com>
wrote:
Two points.
Yes, _if_ someone knows the _exactly_ what algorithm is being
used, and it has a 32-bit seed, then it's theoretically
possible to predict the next card after you've seen six
cards, and possible to narrow things down a bit after seeing
five. But that requires knowing exactly what algorithm is being
used; just knowing that it's a 32-bit seed is not enough.
Second, if I were playing online poker for money this
would be the least of my worries, because there are much
easier ways to cheat. For example collusion - if two
players agree to support each other's plays that gives
them an advantage (because it effectively means their
betting limit is twice that of the other players'.)
It wouldn't be too hard for one person to arrange to
be playing two hands (presumably the site would notice
two players at the same IP address so it would maybe take
a little trickiness.)
[quote:1dc75bd87c]I'm hoping someone(s) on here will be kind enough to help ease my mind
(or, for that matter, to verify my reasons for concern!) And I should
say that, while I'm a good poker player, I'm only a marginal student of
mathematics and computing, so keep that in mind, please, as you answer
this question, but:
Is it possible that, as I'm playing online, any of the other players
are cheating by way of predicting the order of the cards in the
"shuffled" deck?
A few (some math, some poker) truths to keep in mind as you answer:
1. Many sites use an RNG algorithm based on 2^32. Isn't that pretty
easily susceptible to brute-force attacks from someone who has a
program relying on distributed computing?
2. In the most popular type of online poker (Texas Hold 'Em), there are
four rounds of betting, and the most important three rounds come AFTER
(a) each player gets two (face-down) cards and (b) three face-up cards
are dealt on the table -- SO, if the cheateer is, say, the third player
to be dealt cards at a table of ten players, then he will know the 3rd
and 13th cards (i.e. the two cards he's been dealt) as well as the
21st, 22nd and 23rd cards (the three dealt face-up immediately
following the two distributed to each player).
So, I guess my question is: how easily would someone be able -- either
by brute-force run-through of all possible combinations of "deck
arrangements" with those 5 cards in those specific spots, or by some
sort of reverse-engineering, or whatever -- to conclude the sequence of
cards in the deck (which, of course, would give him knowledge of
everyone's cards, as well as the two cards left to be dealt)?
And, for that matter, a related question: there's a piece of software
currently on the market (Check out pokerrng.com.) which claims that, if
a player enters into the software the details of a few thousands
recently dealt hands, the software will "synch" with the poker system's
RNG well enough to predict three of the five cards (not suits, but
values, which is enough to be dangerous!) left to be dealt! Is this
even plausible? Do I need to be afraid of this type of thing?
Anything online has inherent pitfalls, and cheating is -- both online
and offline -- a fact of life, but there's a certain level of security
I'd like, and I wonder if I'm jumping at shadows or rightly
concerned...
Any thoughts would be appreciated!
Thanks,
Gerry
[/quote:1dc75bd87c]
************************
David C. Ullrich |
|
|
| Back to top |
|
|
|
| Richard Henry |
| Posted: Sun Feb 19, 2006 11:37 am |
|
|
|
Guest
|
"David C. Ullrich" <ullrich@math.okstate.edu> wrote in message
news:djrgv1ptomoj0rma2hjj9bl3bhpn6dn37l@4ax.com...
[quote:6fd275e5e8]
Second, if I were playing online poker for money this
would be the least of my worries, because there are much
easier ways to cheat. For example collusion - if two
players agree to support each other's plays that gives
them an advantage (because it effectively means their
betting limit is twice that of the other players'.)
It wouldn't be too hard for one person to arrange to
be playing two hands (presumably the site would notice
two players at the same IP address so it would maybe take
a little trickiness.)
[/quote:6fd275e5e8]
You mean like a telephone? |
|
|
| Back to top |
|
|
|
| Henry |
| Posted: Sun Feb 19, 2006 12:30 pm |
|
|
|
Guest
|
Gerry wrote:
[quote:bd0ac5daea]I'm hoping someone(s) on here will be kind enough to help ease my mind
(or, for that matter, to verify my reasons for concern!) And I should
say that, while I'm a good poker player, I'm only a marginal student
of mathematics and computing, so keep that in mind, please, as you
answer this question, but:
Is it possible that, as I'm playing online, any of the other players
are cheating by way of predicting the order of the cards in the
"shuffled" deck?
A few (some math, some poker) truths to keep in mind as you answer:
1. Many sites use an RNG algorithm based on 2^32. Isn't that pretty
easily susceptible to brute-force attacks from someone who has a
program relying on distributed computing?
2. In the most popular type of online poker (Texas Hold 'Em), there
are four rounds of betting, and the most important three rounds come
AFTER (a) each player gets two (face-down) cards and (b) three
face-up cards are dealt on the table -- SO, if the cheateer is, say,
the third player to be dealt cards at a table of ten players, then he
will know the 3rd and 13th cards (i.e. the two cards he's been dealt)
as well as the 21st, 22nd and 23rd cards (the three dealt face-up
immediately following the two distributed to each player).
So, I guess my question is: how easily would someone be able -- either
by brute-force run-through of all possible combinations of "deck
arrangements" with those 5 cards in those specific spots, or by some
sort of reverse-engineering, or whatever -- to conclude the sequence
of cards in the deck (which, of course, would give him knowledge of
everyone's cards, as well as the two cards left to be dealt)?
And, for that matter, a related question: there's a piece of software
currently on the market (Check out pokerrng.com.) which claims that,
if a player enters into the software the details of a few thousands
recently dealt hands, the software will "synch" with the poker
system's RNG well enough to predict three of the five cards (not
suits, but values, which is enough to be dangerous!) left to be
dealt! Is this even plausible? Do I need to be afraid of this type of
thing?
Anything online has inherent pitfalls, and cheating is -- both online
and offline -- a fact of life, but there's a certain level of security
I'd like, and I wonder if I'm jumping at shadows or rightly
concerned...
Any thoughts would be appreciated!
[/quote:bd0ac5daea]
FWIW, here is an opinion from a professional poker player:
Yeah, there was a problem with Planet Poker's algorithm way back in the day.
Pretty much killed the site, although it's still around and hanging out
despite doing very little in the way of advertising or trying to attract new
players.
Online poker rooms get their motivation to keep their games honest from the
fact that honest games attract tons of players and make tons of money.
Party Poker's annual *profit* is in the hundreds of millions, and perhaps
could exceed a billion this year. So a site can risk cheating players
for a few extra bucks at the expense of being exposed and losing their
ability to essentially print money as it is.
Now, this doesn't mean they're honest. Enron and Worldcom and such have
proved that whenever there's money involved there's always going to be
someone dishonest. But there are millions of poker players out there who've
played billions of hands and the majority of them seem to think most card
rooms are honest. In my opinion, they're honest "enough," meaning that if
they're cheating, it certainly isn't keeping me from making money as it is.
That's larger sites, like Party or PokerStars or UltimateBet. Smaller
cardrooms have the motivation to keep their games honest in order to become
larger cardrooms and make tons of money like the larger sites. However,
there have definitely been some smaller cardrooms that have been found to be
dishonest and/or have cheated customers out of money. These have been
quickly discovered and quickly blacklisted. With pokertracker tracking
every hand dealt, and thousands of players convening at online poker message
boards, it doesn't take long for the slightest perceived slight by a poker
site to be reported to anyone and everyone. So cardrooms with dishonest
shuffles or keeping players money will not last long. Believe me, every
time Party Poker takes more than 24 hours to reply to someone's email, or
someone loses a hand where they were a 95% favorite, I know about it.
Anyways, because of the potential problems with newer sites, I put no money
on any site that isn't well established and fairly big with a decently long
history without problems.
As for collusion, all the big sites have extensive collusion detection built
into their software. People have been banned from sites with their money
confiscated (and, in the case of pokerstars, that money is distributed to
their opponents) when caught. So while I'm sure collusion exists,
1) It's very hard to do profitably as it is
2) Colluders are often caught anyways, and this I believe deters most
colluders
So I don't worry about it. I've played hundreds of thousands of hands at
thousands of tables online and perhaps twice have I ever thought that maybe
I'm at a table with some colluders--and I was far from certain at that.
Each time I simply left the game, since there are thousands of other ones
going on at any one time.
But yeah, ultimately there's no way to know for sure. Sometimes I think
one cardroom screws with the odds a bit. Not a whole lot, just a bit. So
if
someone is a 60/40 favorite, they're really maybe 55/45. 90/10 becomes
85/15, etc. This would be hardly detectable, and keep bad players in the
game playing longer, making the site more popular, and keeping it able to
collect more rake. But who knows, that's probably just my own paranoia. At
the end of the day I continue to play as do so many thousands of others
because it's been many years and many billions of hands dealt out online and
people are satisfied with the way it's been going. |
|
|
| Back to top |
|
|
|
| David C. Ullrich |
| Posted: Sun Feb 19, 2006 3:07 pm |
|
|
|
Guest
|
On Sun, 19 Feb 2006 12:30:56 -0500, "Henry"
<henry432xRemove@comcast.net> wrote:
[quote:6db1b2402b]Gerry wrote:
[...]
FWIW, here is an opinion from a professional poker player:
[...]
As for collusion, all the big sites have extensive collusion detection built
into their software. People have been banned from sites with their money
confiscated (and, in the case of pokerstars, that money is distributed to
their opponents) when caught. So while I'm sure collusion exists,
1) It's very hard to do profitably as it is
2) Colluders are often caught anyways, and this I believe deters most
colluders
[/quote:6db1b2402b]
Surely they can catch and punish blatant collusion. But I don't see
how they're going to catch more subtle collusion.
[quote:6db1b2402b]So I don't worry about it. I've played hundreds of thousands of hands at
thousands of tables online and perhaps twice have I ever thought that maybe
I'm at a table with some colluders--and I was far from certain at that.
Each time I simply left the game, since there are thousands of other ones
going on at any one time.
[/quote:6db1b2402b]
If they were good at it you wouldn't think it was going on...
[quote:6db1b2402b]But yeah, ultimately there's no way to know for sure. Sometimes I think
one cardroom screws with the odds a bit. Not a whole lot, just a bit. So
if
someone is a 60/40 favorite, they're really maybe 55/45. 90/10 becomes
85/15, etc. This would be hardly detectable, and keep bad players in the
game playing longer, making the site more popular, and keeping it able to
collect more rake. But who knows, that's probably just my own paranoia. At
the end of the day I continue to play as do so many thousands of others
because it's been many years and many billions of hands dealt out online and
people are satisfied with the way it's been going.
[/quote:6db1b2402b]
************************
David C. Ullrich |
|
|
| Back to top |
|
|
|
| Gerry |
| Posted: Sun Feb 19, 2006 3:36 pm |
|
|
|
Guest
|
For me, David, collusion-detection is part scientific (i.e. I own
software which tracks the number of times particular players are
involved in hands together, and looks for "markers" of collusion --
raises and reraises in an effort to isolate a "mark" and drive him to
fold, etc., etc.) and part from-the-gut (based on years of playing B&M
and online poker). Certainly, incontrovertibly, your statement (I'll
paraphrase it as: "Good colluders know, as part of their colluding, how
to minimize the risk of detecion.") is true. But it isn't my particular
worry, wrong or right.
My particular worry is that, as was shown possible in 1999
(http://www.cigital.com/papers/abstracts/developer_gambling.html),
someone can "know" my cards...
Again, the authors of pokerrng (at pokerrng.com) make this claim, and I
have no way of knowing whether it's true that their product is snake
oil or that my fear of it is well-grounded.
None of you (and I mean this as a complimentary term, and one of
respect!) math geeks are curious? |
|
|
| Back to top |
|
|
|
| David C. Ullrich |
| Posted: Mon Feb 20, 2006 7:25 am |
|
|
|
Guest
|
On 19 Feb 2006 12:36:20 -0800, "Gerry" <gerald_helmling@yahoo.com>
wrote:
[quote:9b1e180864]For me, David, collusion-detection is part scientific (i.e. I own
software which tracks the number of times particular players are
involved in hands together, and looks for "markers" of collusion --
raises and reraises in an effort to isolate a "mark" and drive him to
fold, etc., etc.) and part from-the-gut (based on years of playing B&M
and online poker). Certainly, incontrovertibly, your statement (I'll
paraphrase it as: "Good colluders know, as part of their colluding, how
to minimize the risk of detecion.") is true. But it isn't my particular
worry, wrong or right.
My particular worry is that, as was shown possible in 1999
(http://www.cigital.com/papers/abstracts/developer_gambling.html),
someone can "know" my cards...
Again, the authors of pokerrng (at pokerrng.com) make this claim, and I
have no way of knowing whether it's true that their product is snake
oil or that my fear of it is well-grounded.
None of you (and I mean this as a complimentary term, and one of
respect!) math geeks are curious?
[/quote:9b1e180864]
Could be just that nobody knows anything about the question.
I've already said that it seems to me that yes, a person could
predict things after a certain number of cards, _if_ the person
knew exactly what algorithm was being used. Reverse-engineering
the algorithm from records of past hands seems like it would
take a _lot_ of hands - do they really publish records of
_billions_ of previous hands?
Looking at that url for the first time I see that sure enough,
the problem started when a site published the shuffling algorithm -
that seems like a bad idea to me. I haven't looked at the pdf -
if I think of it when I get to the office I will.
************************
David C. Ullrich |
|
|
| Back to top |
|
|
|
| rich burge |
| Posted: Mon Feb 20, 2006 11:55 am |
|
|
|
Joined: 04 Jan 2005
Posts: 95
|
Gerry wrote:
[quote:31e101c950]My particular worry is that, as was shown possible in 1999
(http://www.cigital.com/papers/abstracts/developer_gambling.html),
someone can "know" my cards...
Again, the authors of pokerrng (at pokerrng.com) make this claim, and I
have no way of knowing whether it's true that their product is snake
oil or that my fear of it is well-grounded.
[/quote:31e101c950]
Why would anyone suspect "100% legal poker cheat software" of being
snakeoil?
Rich |
|
|
| Back to top |
|
|
|
| Gerry |
| Posted: Mon Feb 20, 2006 7:26 pm |
|
|
|
Guest
|
True...Though, Occam's Razor suggests that the better reason may have
to do with your earlier post; the author of the "lizard with no legs"
might do well to get while the gettin's good... |
|
|
| Back to top |
|
|
|
| N. Silver |
| Posted: Mon Feb 20, 2006 9:55 pm |
|
|
|
Guest
|
Gerald Helmling wrote:
[quote:1a7ab047d9]Is it possible that, as I'm playing online, any of the other players
are cheating by way of predicting the order of the cards in the
"shuffled" deck?
[/quote:1a7ab047d9]
It's highly unlikely, given the wake-up call potential investors
received years ago from Paradise. Methods now are secret.
And, at, at least, one site, that I know of, the code is checked
and certified as generating pseudorandom deals by paid, qualified,
outside agencies.
[quote:1a7ab047d9]A few (some math, some poker) truths to keep in mind as you answer:
1. Many sites use an RNG algorithm based on 2^32. Isn't that pretty
easily susceptible to brute-force attacks from someone who has a
program relying on distributed computing?
[/quote:1a7ab047d9]
This aspect of running a site is crucial. Realize that in brick-and-mortar
games,
the (hand-shuffled) deals are almost always biased (if the table is not
using a
Shufflemaster). In fact, true random deals are much more likely to be
encountered
on-line. So, the casino player will "feel" a difference. Hence, it is
critically important
for on-line casinos to get it right, since it will be different.
[quote:1a7ab047d9]2. In the most popular type of online poker (Texas Hold 'Em), there are
four rounds of betting, and the most important three rounds come AFTER
(a) each player gets two (face-down) cards and (b) three face-up cards
are dealt on the table -- SO, if the cheateer is, say, the third player
to be dealt cards at a table of ten players, then he will know the 3rd
and 13th cards (i.e. the two cards he's been dealt) as well as the
21st, 22nd and 23rd cards (the three dealt face-up immediately
following the two distributed to each player).
[/quote:1a7ab047d9]
Give it a rest. There is always the possibility of cheating by two or
more players, which is more likely in a brick-and-mortar setting,
since casinos don't give much of a darn about anything except
the rake. Collusion does not immediately affect the bottom line.
Someone raised the specter of cheaters who are so good,
their actions would be undetectable at the table. Well an on-line
site such as Poker Stars, has ways of detecting such collusion
forensically, which is impossible in a B&M setting.
Another inhibiting factor, keeping good players from cheating,
is that, if exposed, they become outcasts in the community. Few
partnerships last forever. In my experience, by the way, in
Atlantic City, there was collusion.
[quote:1a7ab047d9]So, I guess my question is: how easily would someone be able -- either
by brute-force run-through of all possible combinations of "deck
arrangements" with those 5 cards in those specific spots, or by some
sort of reverse-engineering, or whatever -- to conclude the sequence of
cards in the deck (which, of course, would give him knowledge of
everyone's cards, as well as the two cards left to be dealt)?
And, for that matter, a related question: there's a piece of software
currently on the market (Check out pokerrng.com.) which claims that, if
a player enters into the software the details of a few thousands
recently dealt hands, the software will "synch" with the poker system's
RNG well enough to predict three of the five cards (not suits, but
values, which is enough to be dangerous!) left to be dealt! Is this
even plausible? Do I need to be afraid of this type of thing?
Anything online has inherent pitfalls, and cheating is -- both online
and offline -- a fact of life, but there's a certain level of security
I'd like, and I wonder if I'm jumping at shadows or rightly
concerned...
[/quote:1a7ab047d9]
You are jumping at shadows. |
|
|
| Back to top |
|
|
|
| David C. Ullrich |
| Posted: Tue Feb 21, 2006 6:54 am |
|
|
|
Guest
|
On Mon, 20 Feb 2006 06:25:00 -0600, David C. Ullrich
<ullrich@math.okstate.edu> wrote:
[quote:23e566e32c]On 19 Feb 2006 12:36:20 -0800, "Gerry" <gerald_helmling@yahoo.com
wrote:
For me, David, collusion-detection is part scientific (i.e. I own
software which tracks the number of times particular players are
involved in hands together, and looks for "markers" of collusion --
raises and reraises in an effort to isolate a "mark" and drive him to
fold, etc., etc.) and part from-the-gut (based on years of playing B&M
and online poker). Certainly, incontrovertibly, your statement (I'll
paraphrase it as: "Good colluders know, as part of their colluding, how
to minimize the risk of detecion.") is true. But it isn't my particular
worry, wrong or right.
My particular worry is that, as was shown possible in 1999
(http://www.cigital.com/papers/abstracts/developer_gambling.html),
someone can "know" my cards...
Again, the authors of pokerrng (at pokerrng.com) make this claim, and I
have no way of knowing whether it's true that their product is snake
oil or that my fear of it is well-grounded.
None of you (and I mean this as a complimentary term, and one of
respect!) math geeks are curious?
Could be just that nobody knows anything about the question.
I've already said that it seems to me that yes, a person could
predict things after a certain number of cards, _if_ the person
knew exactly what algorithm was being used. Reverse-engineering
the algorithm from records of past hands seems like it would
take a _lot_ of hands - do they really publish records of
_billions_ of previous hands?
Looking at that url for the first time I see that sure enough,
the problem started when a site published the shuffling algorithm -
that seems like a bad idea to me. I haven't looked at the pdf -
if I think of it when I get to the office I will.
[/quote:23e566e32c]
For some reason I thought it was a large pdf.
I have to say the first part, about the algorithm that
site was using, was just hilarious. They mention two
big errors, one of which I spotted immediately - the
shuffling algorithm simply does not give a uniform
shuffle, even if we assume a perfect rng. There's
another error that was obvious to me that the authors
don't mention explicitly: the code calls randomize
much too often! People get the idea from the name
of the function that calling randomize as often
as possible will give more random results, but
in fact the opposite is true.
You've said a few things here that I've been skeptical
about - the paper does not support those assertions,
as far as I can see. In particular, the attack they
describe is _not_ based on reverse-engineering the
shuffling algorithm, it depends on _knowing_ what
algorithm is being used! The idea that an attack
as they describe could work if the attacker knows
the details of the algorithm is something I haven't
disputed. (They give a few versions of the attack;
the extremely efficient one depends on the fact
that the seed is based on the current time, and
we can determine what the current time was within
a few seconds. Note that if the algorithm didn't
call randomize so often then this aspect would
not work, we'd be back to hundreds of millions
of seeds to check instead of thousands.)
Anyway. They describe well-known things that a poker
site could do to prevent this problem. You seem to
want someone here to tell you whether these
attacks are possible on currect poker sites. It's
impossible for anyone here to answer that: the
answer depends on whether the site is doing things
right, and that's not a mathematical question.
************************
David C. Ullrich |
|
|
| Back to top |
|
|
|
|
|
All times are GMT - 5 Hours
The time now is Sun Nov 08, 2009 1:42 pm
|
|