RNG's at Full Tilt.

156 views
Skip to first unread message

veeRob

unread,
Apr 18, 2005, 4:12:07 AM4/18/05
to
Interesting transcript from Howard Lederer in a razz S&G with him tonight
regarding the RNG and card shuffling at Full Tilt. Figured some people
would be interested in reading...

Caitlin215 (Observer): i understand that FTP uses
action cards placed at random intervals to speed
up play is that true in ring games as well as
tourneys or are they just used in tourneys
Howard Lederer: our deal si totally random
Howard Lederer: action flops would be hard to
program
Caitlin215 (Observer): well i had the opportunity to
meet chris recently and i was talking to him about
ftp someone over heard and asked him if ftp used
action cards and he said yes placed at totally
random intervals but i wasnt sure if he was just
talking about toruuneys
esfd283: how would that be possible
Howard Lederer: you misheard
Shangri La (Observer): what are action cards?
Howard Lederer: he wwas either joking
Caitlin215 (Observer): i guess i did then
Howard Lederer: or you misheard
Howard Lederer: the notion of action cards is
funny to us
Shangri La (Observer): to ftp?
Howard Lederer: because we know how hard it
would be to pull off
Dakon (Observer): strange thing to joke about
Howard Lederer: and the risk of discovery would
be huge
Howard Lederer: our deal is rnadom on three levels
Howard Lederer: first we used an audited rng
Howard Lederer: but we also use a deck queue
Howard Lederer: basically each deck on the site is
put in a queue after it is used
Howard Lederer: then when we need a deck we
pick one at random
Howard Lederer: then we run it through the rng
Howard Lederer: then after each round has been
dealt
Howard Lederer: the remaining cards are shuffled
during the action
Caitlin215 (Observer): wait the deck that s in play
is shuffled during the action
Howard Lederer: there is no way to know what is
coming because trhe cardes are not set
Howard Lederer: right cait
Caitlin215 (Observer): now is that fair to shuffle the
deck once its in play?
Howard Lederer: since the deck is not set, you
can't cheat
Caitlin215 (Observer): ok thta makes sense
Howard Lederer: we areonly suffling the cards that
remain
Caitlin215 (Observer): but how would u be able to
cheat unless u found a way to ping the deck off
the server
Howard Lederer: if some how you could crack the
deck, it wouldn't help you at our site
Howard Lederer: I am not suggesting that it would
be easy

--
veeRob's poker blog :: http://www.veerob.com
Full Tilt Poker :: http://www.fulltiltpoker.com/?aff=394&code=

------- 
looking for a better newsgroup-reader? - www.recgroups.com


andreas kinell

unread,
Apr 18, 2005, 6:50:53 AM4/18/05
to

"veeRob" <a2e...@webnntp.invalid> schrieb im Newsbeitrag
news:nn6cj2x...@recgroups.com...

the good news:
a nice idea to shuffle the deck during play, it makes it way
harder to cheat.
the bad news:
your play may actually change the next card to come.
it doesn't really matter theoretically, but i feel uncomfortable
when i know that, if i had bet one second later, i probably
would not have been outdrawn and stuff like this.

andreas


UKPaul

unread,
Apr 18, 2005, 7:06:58 AM4/18/05
to

"andreas kinell" <ne...@kinell.ch> wrote in message
news:4263911c$1...@news1.ethz.ch...

> the bad news:
> your play may actually change the next card to come.
> it doesn't really matter theoretically, but i feel uncomfortable
> when i know that, if i had bet one second later, i probably
> would not have been outdrawn and stuff like this.
>
> andreas
>

Agreed, but it also means that next time you chuck 72o and the flop is 772
you dont feel quite so bad.


N. Ziember

unread,
Apr 18, 2005, 7:30:06 AM4/18/05
to

"veeRob" <a2e...@webnntp.invalid> skrev i meddelandet
news:nn6cj2x...@recgroups.com...

> Caitlin215 (Observer): well i had the opportunity to
> meet chris recently and i was talking to him about
> ftp someone over heard and asked him if ftp used
> action cards and he said yes placed at totally
> random intervals but i wasnt sure if he was just
> talking about toruuneys


I was railbirding too at the table during the bustout bonanza where this was
discussed.

This Caitlin215 asked if there were action cards dealt, to which Chris
Ferguson replied something like "Yes, at the correctly expected rates".

So, what are the expected rates?

well, 52:1


It's funny how Caitlin misunderstood the joke.


PS. The bad english is only mine and not a correct transcript, it was more
fun when it happened, but you had to be there =)

/NZ


HARTH G

unread,
Apr 18, 2005, 9:14:36 AM4/18/05
to

Reminds me of a time i was at a table IRL with Juanda.  Some guy took a pot of
him, so Juanda asks the guy.. " u play on FTP?''' Guy says yes...  Juanda says
whats your username?... so the guy tells him... Juanda then pretends to use his
cell phone and says he is callin the programmers to rig his account.

hahahah funny at the time, but really not that funny either.

_______________________________________________________________
* New Release: RecPoker.com v2.2 - http://www.recpoker.com

snuix114

unread,
Apr 18, 2005, 11:36:48 AM4/18/05
to

Chris is hilarious!  I saw on the U.S. Poker Championships on ESPN in the early
rounds:

In an allin pot one guy had KK, another had Kx with a K on the board.  Chris
says: "You know if a king comes there's a split pot."  Everyone looked at him
confused and someone asked why?  He replies: "That would be 5 kings and they'd
have to split the pot!"

Off subject but shows Chris' sense of humor.

_______________________________________________________________
Block Lists, Favorites, and more - http://www.recpoker.com

ScotchRox

unread,
Apr 19, 2005, 7:17:09 PM4/19/05
to
"veeRob" <a2e...@webnntp.invalid> wrote in news:nn6cj2xubj.ln2
@recgroups.com:


Eeeenteresting. Thanks for posting this, VeeRob. I've wondered if sites
do this. It makes a lot of sense for security reasons. Though I do
wonder how much they tax their entropy pool by constantly reshuffling.

-ScotchRox

Perry Friedman

unread,
Apr 19, 2005, 7:39:57 PM4/19/05
to
In article <Xns963DA5AB629...@140.99.99.130>,

ScotchRox <ScotchRox***NOSPAM***@***NOSPAM***gmail.com> wrote:
>Eeeenteresting. Thanks for posting this, VeeRob. I've wondered if sites
>do this. It makes a lot of sense for security reasons. Though I do
>wonder how much they tax their entropy pool by constantly reshuffling.
>
>-ScotchRox

Actually, the way things are done, it actually is less taxing on the
"entropy pool" since cards are only requested as needed. Hence, the
entire deck need not be set before a hand is dealt. If there are only
5 people preflop, only 10 cards are requested initially. If there is
no flop, no further cards are requested.

The actual process is that basically there are banks of decks.
There is a 52 card deck, a 51 card deck, etc. More correctly,
there are banks of unique indices, 0..51, 0..50, etc.

As some examples, when the down cards are ready to be dealt to a full table,
the table would ask for 18 cards from a 52 card deck.
18 random cards are selected using a Knuth algorithm from a 52 card deck.
The actual random numbers are selected from a hardware TRNG (not a PRNG)
meaning it is TRULY random on a quantum physics level. This source
is XOR'ed against an PRNG as a failsafe against hardware malfunction.
The process of obtaining the 18 cards from the 52 card deck caused
the deck itself to be reordered.

Hence, the next request for cards from the 52 card deck would start
with a different initial state than the previous request.

The request for cards returns 18 unique indices between 0 and 51.

Now, when it comes time to deal the flop, the table asks for 3 cards
from a 34 card deck. The table knows what cards are left in it's own
deck. All the RNG returns is 3 indices between 0 and 33, which are
used to pick from the remaining cards in the table's deck.
Again, the process of choosing these new indices causes the 34 card deck
(or more correct, the set of indices from 0 to 33) to be reordered.

This means that all tables are basically shuffling decks for each other,
and that the timing of the hand affects the outcome (in two ways... first,
the TRNG itself will return different results depending on when it
is queried, because it is not a PRNG and is stateless... it's is truly
random... second, even if the TRNG returned the same numbers, other tables
could have modified the state of the indices before you made your
request).

And yes, this effectively makes rabbit hunting useless. I actually like
this. It removes any guilt about folding a hand. You can't say
"I would have won had I called preflop" because the board would have
been different.

Perry

ScotchRox

unread,
Apr 19, 2005, 8:41:25 PM4/19/05
to
frie...@xenon.Stanford.EDU (Perry Friedman) wrote in
news:d444sd$kk4$1...@xenon.Stanford.EDU:

And thank YOU, Perry. Very informative. I wish FTP posted something
about this on their website. I couldn't find anything on there about the
RNG or the shuffle algorithm the last time I looked for it.

-Brian "ScotchRox" Peterson
(ScotchRox on GamesGrid, Scotch Rocks on PS and FTP)

Perry Friedman

unread,
Apr 19, 2005, 8:47:02 PM4/19/05
to
In article <Xns963DB46C857...@140.99.99.130>,

You are quite welcome. I know it's been on the todo list for some time
to get a detailed explanation about the RNG as well as the auditing
we've had done on it.

I will remind the powers that be about this :-)

Perry


Quiet Lion

unread,
Apr 23, 2005, 10:28:58 AM4/23/05
to
Ferguson's joke was very subtle.

"Yes, we use action cards, but at totally random intervals." That is
isomorphic to not using action cards at all. I.e., an action card has an
equal chance of coming off the deck as any other random card.

Richard Brodie
www.liontales.com

Reply all
Reply to author
Forward
0 new messages