Help Finding dice generator
tony
I would like to use another dice generator in Jellyfish 3.5, can anyone
help me find a .dll file on the net??
THANX
TONY
Max-X
Use plastic dice and roll them yourself. That would be the ultimate.
amni
Hi,
After an additional long search in the NET I found a site which
supplies non biased rolls for JellyFish 3.5.
The setting for JF3.5 can take rolls from a text file, in which the
rolls are written as pairs of numbers in each line. For example, the
file which includes the following three lines
3 4
6 2
4 4
will be interpreted as the rolls "3 and 4", "6 and 2" , "4 and 4"
(in that order).
Now the site WWW.RANDOM.ORG supplies (free) files with such lines.
You have to fill a simple "form" in which you are asked to tell, how
many numbers you need (maximum 10,000 numbers each download)
the lowest number (fill 1) and the highest number (fill 6),
the nubmer of colums (fill 2).
The rolls are claimed to be very un biased because they are created
fro a "natural data" (atmospheric noise or something like that).
amni
On Tue, 12 Oct 1999 11:54:09 +0100, tony <antonio...@cgi.it>
Alan Webb
amni schrieb in Nachricht <3811e73d...@news.infomall.co.il>...
>
>Hi,
>
>After an additional long search in the NET I found a site which
>supplies non biased rolls for JellyFish 3.5.
>
>The setting for JF3.5 can take rolls from a text file, in which the
>rolls are written as pairs of numbers in each line. For example, the
>file which includes the following three lines
>
> 3 4
> 6 2
> 4 4
>
>will be interpreted as the rolls "3 and 4", "6 and 2" , "4 and 4"
>(in that order).
>
>Now the site WWW.RANDOM.ORG supplies (free) files with such lines.
>You have to fill a simple "form" in which you are asked to tell, how
>many numbers you need (maximum 10,000 numbers each download)
>the lowest number (fill 1) and the highest number (fill 6),
>the nubmer of colums (fill 2).
>
>The rolls are claimed to be very un biased because they are created
>fro a "natural data" (atmospheric noise or something like that).
>
>
>amni
>
Excellent news! Now the "JF cheats clan" can get whopped with dice they
can't whinge about lol. Well done amni for finding that! I'm perfectly happy
with JF's dice however.
Webby's Backgammon Site
http://marina.fortunecity.com/frog/303/BGHome.htm
Michael Strato
^____^
>
> Oh Really? I hover around 40% win rate against JF in 5 point matches
> with 312 matches played,(since my last reset), using jellyfishes dice.
>
> In a 5 point closed situation with JF on the bar, it definately gets out
> on the first roll over 50% of the time, and never dances more than 3
> times. How many times have any of you danced more than 3 times on a 5
> point board? Why doesn't JF? And getting out on the first roll over 50%
> of the time is well over the 33% that it should.
>
> A statistical fluke? Maybe, as I suspect it would take thousands of
> matches to produce meaningful data, but I have read a large enough
> quantity of comments from others saying the same thing to make me
> suspect that my experience is not a fluke.
>
> And, since you are so certain that those of us who feel JF has a problem
> are so easily "whopped", how bout an 11 point match, me and you, on GG?
>
> Steve
>
>
> Sent via Deja.com http://www.deja.com/
> Before you buy.
stev...@my-deja.com
> Excellent news! Now the "JF cheats clan" can get whopped with dice
they
> can't whinge about lol. Well done amni for finding that! I'm perfectly
happy
> with JF's dice however.
>
> Webby's Backgammon Site
> http://marina.fortunecity.com/frog/303/BGHome.htm
>
>
Oh Really? I hover around 40% win rate against JF in 5 point matches
Rodger Poppleton
40 games using j's dice, and the stat's were - jellyfish 170, mine 114.
I'm used to the vagaries of dice, and in my opinion, jf cheats.
--
roger
<stev...@my-deja.com> wrote in message news:7v7ioa$ct7$1...@nnrp1.deja.com...
VSG
Rodger Poppleton <ro...@rpoppleton.freeserve.co.uk> schrieb in im Newsbeitrag:
7v83sc$hv8$1...@news5.svr.pol.co.uk...
> Of interest to those jellyfish cheats fans, I recently checked doubles over
> 40 games using j's dice, and the stat's were - jellyfish 170, mine 114.
> I'm used to the vagaries of dice, and in my opinion, jf cheats.
>
Doubles can be good and bad. There are very many situations where doubles can be
the last thing you wan't. Coming in against a back game; high doubles whilst on
the bar; blocked doubles due to duplication etc. etc. You shouldn't judge a bot
on whether it cheats by how many doubles it/you roll imo.
Alan Webb
Webby's Backgammon Site
www.isg-vsg.de/backgammon/BGHome.htm
gammon1
sp...@mail.com (amni) wrote:
>
> Hi,
>
> After an additional long search in the NET I found a site which
> supplies non biased rolls for JellyFish 3.5.
>
> The setting for JF3.5 can take rolls from a text file, in which the
> rolls are written as pairs of numbers in each line. For example, the
> file which includes the following three lines
>
> 3 4
> 6 2
> 4 4
>
> will be interpreted as the rolls "3 and 4", "6 and 2" , "4 and 4"
> (in that order).
>
> Now the site WWW.RANDOM.ORG supplies (free) files with such lines.
I had to laugh. I asked random.org for 100 pairs of numbers and in
the first 13 pairs were 3 sets of double-sixes!.........
Daniel Murphy
<ro...@rpoppleton.freeserve.co.uk> wrote:
>Of interest to those jellyfish cheats fans, I recently checked doubles over
>40 games using j's dice, and the stat's were - jellyfish 170, mine 114.
>I'm used to the vagaries of dice, and in my opinion, jf cheats.
Rodger, in all fairness it's not possible to conclude that your count
means anything without more information.
Did you count all the rolls in these 40 games? For example, let's
suppose that each player rolled 852 times in those 40 games. An
average number of doubles in 852 rolls is 142. Which would mean that
you rolled 38 doubles fewer than average and JellyFish rolled 38 more.
Could you explain why you think that would be unusual -- not just
unusual, but proof of cheating??? Because to me it's completely
unremarkable.
You might collect some more data by checking all the dice rolls in all
your games against JellyFish. You might check after every 40 games.
Count the rolls for each player in those 40 games. Count the doubles
for each player, too. Count carefuly, and please let us know the
results!
________________________________________________
Daniel Murphy www.cityraccoon.com/
Humlebæk Backgammon Klub www.hbgk.dk/
Raccoon on FIBS www.fibs.com/
Raccoon on GamesGrid too
Peppe
You are just so lucky, I only had 7 66's in the first 500 pairs :-)
Peppe
--
"but if we are lucky, maybe __/-\__ Preben Guldberg
another jesus comes and we can (o o) c92...@student.dtu.dk
count from the beginning again" oOOo (_) oOOo---------------------
--Winner for most original solution for the year 2000 problem
stev...@my-deja.com
My challenge was aimed at the person who said, "Now the "JF cheats clan"
can get whopped with dice they can't whinge about lol.", who I believe
was Alan Webb. My apaologies if I am mistaken.
That being said, I am always up for a good match, email me private and
we can set up a time.
Steve
> ^____^
>
> >
> > Oh Really? I hover around 40% win rate against JF in 5 point matches
> > with 312 matches played,(since my last reset), using jellyfishes
dice.
> >
> > In a 5 point closed situation with JF on the bar, it definately gets
out
> > on the first roll over 50% of the time, and never dances more than 3
> > times. How many times have any of you danced more than 3 times on a
5
> > point board? Why doesn't JF? And getting out on the first roll over
50%
> > of the time is well over the 33% that it should.
> >
> > A statistical fluke? Maybe, as I suspect it would take thousands of
> > matches to produce meaningful data, but I have read a large enough
> > quantity of comments from others saying the same thing to make me
> > suspect that my experience is not a fluke.
> >
> > And, since you are so certain that those of us who feel JF has a
problem
> > are so easily "whopped", how bout an 11 point match, me and you, on
GG?
> >
> > Steve
> >
> >
JP White
> Thus wrote gammon1 <gam...@my-deja.com>
> > In article <3811e73d...@news.infomall.co.il>,
> > sp...@mail.com (amni) wrote:
> > > Now the site WWW.RANDOM.ORG supplies (free) files with such lines.
>
> > I had to laugh. I asked random.org for 100 pairs of numbers and in
> > the first 13 pairs were 3 sets of double-sixes!.........
>
> You are just so lucky, I only had 7 66's in the first 500 pairs :-)
>
> Peppe
Since we are comparing random.org dice rolls, I did the following analysis a
few days ago of 500 dice pairs from random.org.
Roll Total
1-1 14
1-2 25
1-3 24
1-4 32
1-5 28
1-6 22
2-2 10
2-3 33
2-4 32
2-5 33
2-6 31
3-3 15
3-4 30
3-5 24
3-6 25
4-4 11
4-5 24
4-6 31
5-5 19
5-6 27
6-6 10
Total 500
Not being a statistician I can't shed any light onto the randomness or
otherwise of this. But at first glance I see nothing that looks like it is
awry. As one would expect, the doubles appear half as often as than the non
doublets. Looks random to my untrained eye. Maybe you are just unlucky
Peppe.
--
JP White
Mailto:jp.w...@nashville.com
Peppe
> > > I had to laugh. I asked random.org for 100 pairs of numbers and in
> > > the first 13 pairs were 3 sets of double-sixes!.........
> > You are just so lucky, I only had 7 66's in the first 500 pairs :-)
> Since we are comparing random.org dice rolls, I did the following analysis a
> few days ago of 500 dice pairs from random.org.
[--- snip --- list of numbers ---]
> Not being a statistician I can't shed any light onto the randomness or
> otherwise of this.
Nah, I am not going to bother. They seemed quite reasonable, but the
sample is too small to really tell how well the distibution actually
is.
> But at first glance I see nothing that looks like it is
> awry. As one would expect, the doubles appear half as often as than the non
> doublets. Looks random to my untrained eye. Maybe you are just unlucky
> Peppe.
That was actually my point, though the smiley didn't stress that much.
I might have misread the "........"'s to imply that the numbers could
not be trusted, so I had to give a counter example.
People whine a lot about dice, too much I think, especially people
tend to notice the doubles. But the rolls are after all just two
numbers and you must play accordingly.
I have seen a good deal of lucky sequences (with and without doubles,
of course), but they are only lucky under the given circumstances.
Tonight I was trapped behind a three point board and rolled four
doubles that were blocked, leading to a crunched home. Now, had my two
back men been one point further back, I would have escaped and rallied
home. Being just a friendly game, I was chuckling at the third and
laughing at the fourth roll - it was just an amazing sequence.
Peppe
--
Fight fiercely, Harvard, fight, fight, fight! __/-\__ Preben Guldberg
Impress them with our prowess, do! (o o) c92...@student.dtu.dk
Oh, fellas, do not let the crimson down, oOOo (_) oOOo---------------------
Be of stout heart and true. --Tom Lehrer: Fight Fiercely, Harvard
Michael Manolios
stev...@my-deja.com wrote:
>
> > Excellent news! Now the "JF cheats clan" can get whopped with dice
> they
> > can't whinge about lol. Well done amni for finding that! I'm
perfectly
> happy
> > with JF's dice however.
> >
> > Webby's Backgammon Site
> > http://marina.fortunecity.com/frog/303/BGHome.htm
> >
> >
>
> with 312 matches played,(since my last reset), using jellyfishes dice.
>
> In a 5 point closed situation with JF on the bar, it definately gets
out
> on the first roll over 50% of the time, and never dances more than 3
> times. How many times have any of you danced more than 3 times on a 5
> point board? Why doesn't JF? And getting out on the first roll over
50%
> of the time is well over the 33% that it should.
>
> A statistical fluke? Maybe, as I suspect it would take thousands of
> matches to produce meaningful data, but I have read a large enough
> quantity of comments from others saying the same thing to make me
> suspect that my experience is not a fluke.
>
> And, since you are so certain that those of us who feel JF has a
problem
> are so easily "whopped", how bout an 11 point match, me and you, on
GG?
>
> Steve
>
> Sent via Deja.com http://www.deja.com/
> Before you buy.
>
JF cheats.
Stop complaining sirs! Prove it!
How ? Scientifically:
1) Play LOTS (say 1000) of money games against JF, with JF's dice.
2) Save all the games and write down every single result.
3) Play LOTS (say 1000) of money games against JF, with YOUR dice.
4) Save all the games and write down every single result.
5) Do a statistical test to find out if the difference between the two
overall results is statistically significant or not.
6) Take into consideration the improvement in your game during a 2000
money games session against JF... :)
6) If you find out that the dice are biased, announce proudly in r.g.b.
your thriumph.
7) If not, admit in r.g.b. your fault and let the case rest at last...
8) In any case, give us the saved games and the results so we can check
the test.
9) In any case, thank JF for the free lesson...
Until you are done with such an experiment, let's talk about
something else in this newsgroup, please.
Michael Manolios
(mann on FIBS)
--
We play one and only money game through our whole life...
stev...@my-deja.com
I have a better idea. How about you do all of the above in order to
prove that it *does not* cheat. :-)
(when it comes to JF being on the bar with 5 points closed, which is my
only contention).
Rodrigo Andrade
>prove that it *does not* cheat. :-)
Pardon my ignorance, but how do you prove a negative?
--
RODRIGO
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Descending from heaven
The angel sworn to bring him down
The hunter the thunder
The wrath of heavens coming down
-Iced Earth
Murat Kalinyaprak
> amni schrieb in <3811e73d...@news.infomall.co.il>...
>> After an additional long search in the NET I found a site which
>> supplies non biased rolls for JellyFish 3.5.
> Excellent news! Now the "JF cheats clan" can get whopped with
> dice they can't whinge about lol. Well done amni for finding
> that! I'm perfectly happy with JF's dice however.
And the "JF doesn't cheat clan" just doesn't get it
that such schemes are completely irrelevant to JF's
cheating or not cheating with its own dice rolls...
I had argued before that JF's and SW's offering
external dice file or dll capabilities are nothing
more than useless bullshit...! Why can't apparently
intelligent people who can post reasonable articles
on other issues here (including you) can't bring
themselves to agree with or initially make logical
arguments on this particular subject...?
For example, if one way JF could cheat were by
looking ahead to future dice rolls, it could do
that with both external dice files and external
dice dlls. So, additional arguments need to be
made such as JF's always making the same moves
given any pairs of board positions and dice rolls.
But, by the time you would make such an argument,
you also render any schemes like external dice
files or external dice generators totally useless
to the overall argument of whether JF cheats...
Is this so hard for you folks to see/comprehend?
If not, why can't you folks have the decency to
acknowledge that such features offered by JF and
SW are just useless bullshit? In fact, why can't
authors of those software prove that they have
the intelligence to comprehend this and then have
the decency to not insult our intelligence by not
offering such bogus bullshit as proof in the first
place or take them out of future relases of their
software...?
Here is my open appeal to the authors of JF and SW:
"Please take those features that may possibly[?] be
insulting to the intelligence of some of your users
out of your future software". Now let's wait and
see if they'll ever hear it, get it and do it...?
It should be obvious to the dimmest of minds here
that anything external to the JF's dice generator
is irrelevant to it as far as being any proof for
JF's cheating or not cheating with its dice. This
is just as simple as this... Anybody who finally
may gets this, would you please raise your hand?
MK
Murat Kalinyaprak
> Really sorry, but I 'm fed up with all those who are
> so "sure" that JF cheats.
> Stop complaining sirs! Prove it!
> How ? Scientifically:
Really really sorry, but I'm fed up with people like
you who seem to be purposefully trying to give science,
logic, etc. bad names...
> 1) Play LOTS (say 1000) of money games against JF, with JF's dice.
> 2) Save all the games and write down every single result.
> 3) Play LOTS (say 1000) of money games against JF, with YOUR dice.
> 4) Save all the games and write down every single result.
> 5) Do a statistical test to find out if the difference between the
> two overall results is statistically significant or not.
> ...............
I can only feel sorry for you and/or anybody else who
may see any "science" in this...
JF may very well play a strong game with manual dice,
mop the floors and do laundry also. But what does that
have to do with whether it cheats or not with its own
dice (besides your and others' false assumption that
rich don't steal)...?
And what if I told you that I did what you suggested
above and found out that I beat JF 62% of the time
with JF's dice vs. 54% of the time with manual dice?
What would you say then...? Let's hear it but make it
sound really "scientific", ok...?
MK
Donald Kahn
Ah, so good to have back in the fold our own "Chariots of the Gods",
"Worlds in Collision", and "Mysteries of Crop Circles" man!
dk
Michael Manolios
> > 1) Play LOTS (say 1000) of money games against JF, with JF's dice.
> > 2) Save all the games and write down every single result.
> > 3) Play LOTS (say 1000) of money games against JF, with YOUR dice.
> > 4) Save all the games and write down every single result.
> > 5) Do a statistical test to find out if the difference between the
two
> > overall results is statistically significant or not.
2000
> > money games session against JF... :)
> > 6) If you find out that the dice are biased, announce proudly in
> r.g.b.
> > your thriumph.
> > 7) If not, admit in r.g.b. your fault and let the case rest at
last...
> > 8) In any case, give us the saved games and the results so we can
> check
> > the test.
> > 9) In any case, thank JF for the free lesson...
> > Michael Manolios
> >
> > (mann on FIBS)
> I have a better idea. How about you do all of the above in order to
> prove that it *does not* cheat. :-)
> (when it comes to JF being on the bar with 5 points closed, which is
my
> only contention).
>
> Steve
How about treating JF as innocent until PROVEN guilty ? Would
anyone of us dare to accuse publicly another human player without any
proof ? Just because that player "seems" to enter quickly from the bar,
hit many shots, or whatever ? JF cannot defend itself. This doesn't
mean that we humans have the right to accuse it unless we have in our
hand something that can be verified and treated as substantial proof.
Finally I must point out that I personally don't know if JF
cheats. Because I have no proof that it doesn't cheat either. And
instead of wasting all this time in an experiment such as I described,
I prefer to use JF as an excellent tool and to concetrate to the game
itself.
Michael Manolios
"Murat Kalinyaprak" <mu...@compuplus.net> wrote:
> Michael Manolios wrote news:7vbuil$gp7$1...@nnrp1.deja.com...
> > 1) Play LOTS (say 1000) of money games against JF, with JF's dice.
> > 2) Save all the games and write down every single result.
> > 3) Play LOTS (say 1000) of money games against JF, with YOUR dice.
> > 4) Save all the games and write down every single result.
> > 5) Do a statistical test to find out if the difference between the
> > two overall results is statistically significant or not.
> > ...............
>
> I can only feel sorry for you and/or anybody else who
> may see any "science" in this...
>
> JF may very well play a strong game with manual dice,
> mop the floors and do laundry also. But what does that
> have to do with whether it cheats or not with its own
> dice (besides your and others' false assumption that
> rich don't steal)...?
>
First of all I must make clear that I don't know if JF cheats. And
that is so because I have no proof that it does, and I have no proof
that it doesn't.
But this doesn't mean that someone with enough time and patience
cannot accumulate data sufficient to give an answer. Now, it is
irrelevant whether you see any science in a statistical test or not.
And if we don't agree that such a test could give us an answer, then
this discussion is over as far as I am concerned.
> And what if I told you that I did what you suggested
> above and found out that I beat JF 62% of the time
> with JF's dice vs. 54% of the time with manual dice?
>
> What would you say then...? Let's hear it but make it
> sound really "scientific", ok...?
>
> MK
I would say CONGRATULATIONS !! Because if you BEAT JF (and I 'm
talking about JF 3.0, Level 7, Time Factor: 1000) then I can only
assume that your backgammon skills and knowledge have made you a
millionaire!
Concerning your results now, I would like to ask:
Have you saved all the games, so anybody can check them ? You can
imagine that only in this case can an experiment be approved.
Second question: what is your sample ? How many games did you play
with JF's dice ? How many with yours ? (You know, every now and then I
manage to get a positive score against JF in 100 consecutive money
games too, but I always lose in 200 games.)
Moreover a 62% or 54% percentage doesn't mean anything by itself.
In order to come to any conclusions one needs the overall points per
game, and of course the number of games. It would be even better if we
had every single result. Again if we don't agree in this one, then we
cannot go any further.
Finally, let's suppose that you HAVE saved the games and you HAVE
kept all the needed results. This 8% difference may or may not be
statistically significant, depending on the number of the games played.
So as I see it, we can only have three possible results.
a) This 8% difference is NOT significant. This means that JF's dice
generator is OK. In other words, JF does not cheat.
b) This 8% difference IS significant. This means that JF's dice
generator is not OK. In other words, JF cheats. BUT IT CHEATS ITSELF!
c) The number of games is not big enough to allow us to conclude any of
the above (which, allow me to believe, that propably is the case).
Ian Shaw
Murat Kalinyaprak wrote in message <7vfof6$gq...@taisp3.in-tch.com>...
[snip]
>It should be obvious to the dimmest of minds here
>that anything external to the JF's dice generator
>is irrelevant to it as far as being any proof for
>JF's cheating or not cheating with its dice. This
>is just as simple as this... Anybody who finally
>may gets this, would you please raise your hand?
>
>MK
I have always fully understood this, so consider my hand raised. You are
correct. But oh! Murat, Murat!
I thought we'd moved on from this. I was under the impression that you were
now convinced that JF did not cheat by knowing its own rolls. I think that's
why we have not discussed this particular method of cheating for a while.
To recap,
a) JF always produces the same rolls for a given seed and counter.
b) JF always plays the same move for a given position, roll and match score.
In combination, the above two statements demonstrate that JF does not cheat
by knowing what the upcoming rolls are. If JF cheats in this manner, one of
the above statements must be false. Do you believe either or both statements
are false?
The fact that:
c) JF always plays the same move regardless of how the dice are generated
also supports the argument that JF does not cheat by knowing the upcoming
rolls. However, (a) & (b) do the job by themsleves, so as you point out, (c)
is irrelevant.
--
Regards
Ian Shaw (ian on FIBS)
stev...@my-deja.com
Michael Manolios <mm...@tee.gr> wrote:
>
> > > 1) Play LOTS (say 1000) of money games against JF, with JF's dice.
> > > 2) Save all the games and write down every single result.
> > > 3) Play LOTS (say 1000) of money games against JF, with YOUR dice.
> > > 4) Save all the games and write down every single result.
> > > 5) Do a statistical test to find out if the difference between the
> two
> > > overall results is statistically significant or not.
> 2000
> > > money games session against JF... :)
> > > 6) If you find out that the dice are biased, announce proudly in
> > r.g.b.
> > > your thriumph.
> > > 7) If not, admit in r.g.b. your fault and let the case rest at
> last...
> > > 8) In any case, give us the saved games and the results so we can
> > check
> > > the test.
> > > 9) In any case, thank JF for the free lesson...
>
> > >
> > > (mann on FIBS)
>
> > prove that it *does not* cheat. :-)
> > (when it comes to JF being on the bar with 5 points closed, which is
> my
> > only contention).
> >
> > Steve
>
> How about treating JF as innocent until PROVEN guilty ?Would
> anyone of us dare to accuse publicly another human player without any
> proof ?
Never. But JF isn't human. It's a program, and in my opinion all
programs, when they don't execute as they should, should be treated as
guilty until proven innocent!
>Just because that player "seems" to enter quickly from the
>bar,
With all due respect, saying that JF "seems" to enter quickly from the
bar is like saying that the ocean "seems" to be a bit moist.
> hit many shots, or whatever ? JF cannot defend itself. This doesn't
> mean that we humans have the right to accuse it unless we have in our
> hand something that can be verified and treated as substantial proof.
> Finally I must point out that I personally don't know if JF
> cheats. Because I have no proof that it doesn't cheat either.
And
> instead of wasting all this time in an experiment such as I described,
> I prefer to use JF as an excellent tool and to concetrate to the game
> itself.
>
> Michael Manolios
>
> (mann on FIBS)
I agree with you here Michael. And the fact that I think JF doesn't sit
on the bar as much as it should in a 5 point closed situation does not
diminish its usefulness as a backgammon tool for me. I am certain my
game as improved due to many, many interesting matches against the Fish.
Steve
amni
<we...@hannover.sgh-net.de> wrote:
>
>amni schrieb in Nachricht <3811e73d...@news.infomall.co.il>...
>>
>>Hi,
>>
>>After an additional long search in the NET I found a site which
>>supplies non biased rolls for JellyFish 3.5.
>>
>>The setting for JF3.5 can take rolls from a text file, in which the
>>rolls are written as pairs of numbers in each line. For example, the
>>file which includes the following three lines
>>
>> 3 4
>> 6 2
>> 4 4
>>
>>will be interpreted as the rolls "3 and 4", "6 and 2" , "4 and 4"
>>(in that order).
>>
>>Now the site WWW.RANDOM.ORG supplies (free) files with such lines.
>>You have to fill a simple "form" in which you are asked to tell, how
>>many numbers you need (maximum 10,000 numbers each download)
>>the lowest number (fill 1) and the highest number (fill 6),
>>the nubmer of colums (fill 2).
>>
>>The rolls are claimed to be very un biased because they are created
>>fro a "natural data" (atmospheric noise or something like that).
>>
>>
>>amni
>>
>
>
>Excellent news! Now the "JF cheats clan" can get whopped with dice they
>can't whinge about lol. Well done amni for finding that! I'm perfectly happy
>with JF's dice however.
>
amni
Hi,
I know for sure that JF internal DICE GENARATOR is biased, i checked
this during 10,000 games or so (once in 50 games or so there are at
least 5 doubles in consecutive 20 rolls). That's why i suggested
using external rolls from files like those in RANDOM.ORG.
Beeing biased doesn't neccessarily means that JF cheats deliberately.
Using those external files doesn't help to avoid cheating (because
JF can read ahead that file).
I have no idea if JF cheats or not but i must say it doesn't
give a fair chance to test this claim. To get meaningfull
results (like "average lost/win points") one should play samples of
1000 single games with manuall rolls and this is practically not
possible (because the manual rolling is ackward). Moreover, the
manual rolls are so ackward that one who use it cannot concetrate in
the game: his thinking is interrupted by opening again and again
the menu and clicking on the skew table of roll buttons. Since
concentration is lost --- his results will be inferior (comparing his
results in real life situation).
amni
On Mon, 01 Nov 1999 07:57:14 GMT, Michael Manolios <mm...@tee.gr>
wrote:
>In article <7vfpd4$gq...@taisp3.in-tch.com>,
> "Murat Kalinyaprak" <mu...@compuplus.net> wrote:
>> Michael Manolios wrote news:7vbuil$gp7$1...@nnrp1.deja.com...
>
>> > 1) Play LOTS (say 1000) of money games against JF, with JF's dice.
>> > 2) Save all the games and write down every single result.
>> > 3) Play LOTS (say 1000) of money games against JF, with YOUR dice.
>> > 4) Save all the games and write down every single result.
>> > 5) Do a statistical test to find out if the difference between the
>> > two overall results is statistically significant or not.
>Michael Manolios
>
>(mann on FIBS)
amni
amni
TEST POSTING (IGNORE)
Steve Harris
>
>
>
>
>
>Hi,
>
>I know for sure that JF internal DICE GENARATOR is biased, i checked
>this during 10,000 games or so (once in 50 games or so there are at
>least 5 doubles in consecutive 20 rolls). That's why i suggested
>using external rolls from files like those in RANDOM.ORG.
>
>Beeing biased doesn't neccessarily means that JF cheats deliberately.
>
>Using those external files doesn't help to avoid cheating (because
>JF can read ahead that file).
>
>I have no idea if JF cheats or not but i must say it doesn't
>give a fair chance to test this claim. To get meaningfull
>results (like "average lost/win points") one should play samples of
>1000 single games with manuall rolls and this is practically not
>possible (because the manual rolling is ackward). Moreover, the
>manual rolls are so ackward that one who use it cannot concetrate in
>the game: his thinking is interrupted by opening again and again
>the menu and clicking on the skew table of roll buttons. Since
>concentration is lost --- his results will be inferior (comparing his
> results in real life situation).
>
>amni
I do this all the time, using a separate dice generator and entering them into
Jellyfish manually. It doesn't affect my concentration at all. It's as automatic
as picking up dice and rolling them.
--Steve
Daniel Murphy
>I know for sure that JF internal DICE GENARATOR is biased, i checked
>this during 10,000 games or so (once in 50 games or so there are at
>least 5 doubles in consecutive 20 rolls). That's why i suggested
>using external rolls from files like those in RANDOM.ORG.
1. You say you know the JF dice generator is biased, but did you tell
us HOW you know and WHAT the bias is?
2. Would you consider it unusual, statistically speaking, if once in
50 games or so there are at least 5 doubles in consecutive 20 rolls?
Eulfe
rac...@cityraccoon.com wrote:
> On Thu, 11 Nov 1999 22:33:00 GMT, no_...@mail.com (amni) wrote:
>
> >I know for sure that JF internal DICE GENARATOR is biased, i checked
> >this during 10,000 games or so (once in 50 games or so there are at
> >least 5 doubles in consecutive 20 rolls). That's why i suggested
> >using external rolls from files like those in RANDOM.ORG.
> 2. Would you consider it unusual, statistically speaking, if once in
> 50 games or so there are at least 5 doubles in consecutive 20 rolls?
The probability of getting at least 5 doubles in 20 rolls is about
.231250781. If this happens only once in 50 games (which probably are
longer than 20 rolls), this _is_ unusual.
Eulfe
amni
The probabilty for 5 doubles in sequence of 20 rolls in a game
is *much much* lower than 0.25 or so; maybe it's 0.0025 (namely,
about ONCE IN 400 GAMES, i couldn't make correct computation now
because my relevant mathematics books are not handy but a rough
computation shows this may occure once in 400 games, definitely
not once in 50 games).
amni
JP White
> Hi,
>
> The probabilty for 5 doubles in sequence of 20 rolls in a game
> is *much much* lower than 0.25 or so; maybe it's 0.0025 (namely,
> about ONCE IN 400 GAMES, i couldn't make correct computation now
> because my relevant mathematics books are not handy but a rough
> computation shows this may occure once in 400 games, definitely
> not once in 50 games).
>
> amni
>
Sorry Amni, I agree with Eulfe.
The assumption I have made is that the probability can be calculated using
the binomial probability distribution. I may be wrong, but the Encyclopedia
Britannica defines the conditions that must be met for the binomial
distribution to be appropriate as follows :-
> A binomial experiment has four properties: (1) it consists of a
> sequence of n identical trials; (2) two outcomes, success or failure,
> are possible on each trial; (3) the probability of success on any trial,
> denoted p, does not change from trial to trial; and (4) the trials are
> independent.
>
This would appear to fit the bill for the purposes of estimating the
probability of 5 doubles in 20 rolls.
The formula to calculate the possibility is defined as
f(x) = (n/x)p^x(1-p)^(n-x)
Without HTML this formula looks weird, I use ^ to denote raising to the
power of.
To aid calculation I used excels built in Binomial function to calculate the
probability of 0 through 20 doubles in a trial of 20 rolls.
The probability of 5 or more doubles is 0.231250781 (I.E. the same as 1
minus the probability of 4 or less). This is the number Eulfe came up with.
The probability of *exactly* 5 doubles is 0.129410292. I believe for the
purposes of this discussion that we talk about the probability of 5 or more
I.E. 0.231250781. I say this because of your statement as follows.
> On Thu, 11 Nov 1999 22:33:00 GMT, no_...@mail.com (amni) wrote:
> >
> > >I know for sure that JF internal DICE GENARATOR is biased, i checked
> > >this during 10,000 games or so (once in 50 games or so there are at
> > >least 5 doubles in consecutive 20 rolls). That's why i suggested
> > >using external rolls from files like those in RANDOM.ORG.
>
The Excel table is below. (I hope it comes out OK over usenet)
(You'll notice that the probability you quote of .0025 is fairly close to
the probability of getting 9 doubles in a sample of 20 rolls!)
Number Trials Probability Probability Running 1-Cumulative
Of of of Total of Running
Doubles double X doubles f(x) total of
(x) (n) (p) f(x) f(x)
0 20 0.166666667 0.026084053 0.026084053 0.973915947
1 20 0.166666667 0.104336213 0.130420267 0.869579733
2 20 0.166666667 0.198238805 0.328659072 0.671340928
3 20 0.166666667 0.237886566 0.566545638 0.433454362
4 20 0.166666667 0.202203581 0.768749219 0.231250781
5 20 0.166666667 0.129410292 0.898159511 0.101840489
6 20 0.166666667 0.064705146 0.962864657 0.037135343
7 20 0.166666667 0.025882058 0.988746715 0.011253285
8 20 0.166666667 0.008411669 0.997158384 0.002841616
9 20 0.166666667 0.002243112 0.999401496 0.000598504
10 20 0.166666667 0.000493485 0.999894981 0.000105019
11 20 0.166666667 8.97245E-05 0.999984705 1.52949E-05
12 20 0.166666667 1.34587E-05 0.999998164 1.83622E-06
13 20 0.166666667 1.65645E-06 0.99999982 1.79765E-07
14 20 0.166666667 1.65645E-07 0.999999986 1.41201E-08
15 20 0.166666667 1.32516E-08 0.999999999 8.68527E-10
16 20 0.166666667 8.28226E-10 1.0 4.03015E-11
17 20 0.166666667 3.89753E-11 1.0 1.32616E-12
18 20 0.166666667 1.29918E-12 1.0 2.69784E-14
19 20 0.166666667 2.73511E-14 1.0 0.0
20 20 0.166666667 2.73511E-16 1.0 0.0
Total probability 1.0 0.0
If I have used the wrong method of calculating these probabilities (It's a
*lllloooooonnnngggg* time since I was at school) I welcome corrective and
constructive criticism.
Basically I believe you to be wrong Amni.
JP White
<snip>
>
> Basically I believe you to be wrong Amni.
Maybe 'wrong' is the wrong word, I'll take that back. I should have said I
believe you to be inaccurate.
If the probability of 5 doubles in 20 rolls is close to 23%, then the data you
collected indicating this happened once in every 50 games or so (2%) would
appear to prove that there is in fact bias in the dice, in so much as there are
too few occurrences of 5 or more doubles in a sequence of 20 rolls. This is the
point Eulfe was making I believe.
So your advice to use RANDOM.ORG would appear to be correct if this fact really
troubles you. Of course the dice are the same for both contestants, so this is
only of significance in calculating the correct probability of being hit etc.
However, there are some doubts I have about your initial post.
> i checked this during 10,000 games or so (once in 50 games or so there are at
> least 5 doubles in consecutive 20 rolls).
>
10,000 games or so. 50 games or so. These are quite ambiguous statements which
cast doubt upon the accuracy of your results.
I am surprised that anyone has enough time to play 10,000 games, or even collect
data for 10,000 games other people play against Jellyfish, or even Jelly vs
Jelly. Your track record in making inaccurate statements about probabilities
would lend me to question the validity of your numbers.
Your posts appear to have been done from 'memory' of your earlier work on this.
This raises an interesting point. As humans we tend to remember what we want to
(I.E. Jellyfish gets too many doubles, I never get them etc. etc.) and since you
are doubtful about Jellyfish's RNG you may selectively remember certain facts or
figures, which become distorted over time. All your statements prove is that you
don't trust Jellyfish's RNG, not the fact that it is untrustworthy (which may or
may not be true, but you have failed to demonstrate this).
I have no numbers, but I would not be surprised to hear that the RNG is biased,
merely because a computer calculates it, which at best is pseudo random. What
would be interesting to hear is by how much is it biased and is it enough to
worry about. Personally I am not troubled by it.
Igor Schein
> On Thu, 11 Nov 1999 22:33:00 GMT, no_...@mail.com (amni) wrote:
>>I know for sure that JF internal DICE GENARATOR is biased, i checked
>>this during 10,000 games or so (once in 50 games or so there are at
>>least 5 doubles in consecutive 20 rolls). That's why i suggested
>>using external rolls from files like those in RANDOM.ORG.
> 1. You say you know the JF dice generator is biased, but did you tell
> us HOW you know and WHAT the bias is?
> 2. Would you consider it unusual, statistically speaking, if once in
> 50 games or so there are at least 5 doubles in consecutive 20 rolls?
I believe Mersenne twister is the best algorithm for random
number generator available. I'd like to see it employed in
backgammon programs as well as on FIBS. It's period is
~10^6002, which is pretty good for all practical purposes.
Also, I've always been a believer for using 2 independent seeds
for both sides. I think FIBS uses the same seed for both players.
Igor
Daniel Hollis
>
>I believe Mersenne twister is the best algorithm for random
>number generator available. I'd like to see it employed in
>backgammon programs as well as on FIBS. It's period is
>~10^6002, which is pretty good for all practical purposes.
>Also, I've always been a believer for using 2 independent seeds
>for both sides. I think FIBS uses the same seed for both players.
>
>Igor
AFAIK, FIBS uses the same seed for _all_ players. Each game queries the
same random number generator for a dice roll. So, even if you did know
the seed that you're on, you don't know how many rolls were requested
before the next roll.
This is a pretty good indication against bias based on the current roll
(for instance, when I roll a racing number like 65, fibs always seems to
give my opponent doubles in return).
Instead of your next roll being a deterministic function of the current
seed, it's also now a function of the current seed and the number of other
players rolling - which depends on the time you're taking to roll, the
number of people online, and so on.
There probably are not enough people online rolling for this effect to
increase the seed between rolls significantly, so there is still
predictability based on the current seed. However, it does make the
predictability less; specifically, the next roll you get no longer depends
on the current seed 100%. I'm assuming that we can't measure how many
rolls other players get in between yours, so we can model that with some
sort of probability.
Dan
alanback
significant sample, all other things would be equal and a preponderance
of doubles in favor of JF would be evidence of what we all know
intuitively.
In article <7v8vuf$cou$1...@news01.btx.dtag.de>,
"VSG" <vsg...@t-online.de> wrote:
>
> Rodger Poppleton <ro...@rpoppleton.freeserve.co.uk> schrieb in im
Newsbeitrag:
> 7v83sc$hv8$1...@news5.svr.pol.co.uk...
> > Of interest to those jellyfish cheats fans, I recently checked
doubles over
> > 40 games using j's dice, and the stat's were - jellyfish 170, mine
114.
> > I'm used to the vagaries of dice, and in my opinion, jf cheats.
> >
>
> Doubles can be good and bad. There are very many situations where
doubles can be
> the last thing you wan't. Coming in against a back game; high doubles
whilst on
> the bar; blocked doubles due to duplication etc. etc. You shouldn't
judge a bot
> on whether it cheats by how many doubles it/you roll imo.
>
> Alan Webb
>
> Webby's Backgammon Site
> www.isg-vsg.de/backgammon/BGHome.htm
alanback
In article <7vfg8s$h11$1...@bgtnsc03.worldnet.att.net>,
"Rodrigo Andrade" <gamm...@REMOVEatt.net> wrote:
> >I have a better idea. How about you do all of the above in order to
> >prove that it *does not* cheat. :-)
>
> Pardon my ignorance, but how do you prove a negative?
>
> --
> RODRIGO
>
> =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
>
> Descending from heaven
> The angel sworn to bring him down
> The hunter the thunder
> The wrath of heavens coming down
>
> -Iced Earth
>
>
Michael Manolios
> > Rodger Poppleton <ro...@rpoppleton.freeserve.co.uk> schrieb in im
> Newsbeitrag:
> > 7v83sc$hv8$1...@news5.svr.pol.co.uk...
> > > Of interest to those jellyfish cheats fans, I recently checked
> doubles over
> > > 40 games using j's dice, and the stat's were - jellyfish 170,
mine
> 114.
> > > I'm used to the vagaries of dice, and in my opinion, jf cheats.
Let me point out again (and again in this newsgroup) that we cannot
reach to any conclusions when we say that "JF did this x times and I
did it only y times so, IN MY OPINION, JF cheats".
It seems that it is really hard to emphasize enough the following
statements:
1) When we refer to number or events we must ALWAYS announce the trials
needed to get the specific results. When we say "170 doubles in 40
games" we don't mean anything at all, because we don't have any exact
expectation of how many doubles must result on average per game. But we
know how many doubles we should expect in one roll: 1/6. Thus, we
should say "I noticed 170 in (e.g.) 800 rolls".
2) When we try to compare two results, the simple fact that one of them
APPEARS to happen more frequently doesn't also mean that it happens
more frequently indeed. And, most important, OUR OPINION is not worth a
penny. What it counts is STATISTICS' OPINION only. Which means that we
should perform again a statistical test to find out if this difference
between the two results is statistically significant.
3) When we want to accuse ANYONE, human or bot, of cheating we should
ALWAYS have the proof to do it. Which means, we should have saved all
the facts (results, rolls, games, matches etc.) that we refer to, so
anybody can check and reexamine our allegations.
(I 'm afraid I 'm repeating myself and some of my previous posts,
but it seems I have to)
I believe that this discussion with JF cheating or not has gone too
far, so I think I 'll perform such a test myself and post it here. I
know of course that I won't convince many people that have
already "convince" themselves, but it will surely be a much more
accurate way to treat accusations, and something that can be provided
as substantial evidence to both parties.
Michael Manolios
(mann on FIBS)
--
We play one and only money game through our whole life...