Many thanks in advance.
PS: I once had Backgammon software for my previous PC, (Win 95), but could
have sworn the strength of the game for the PC, was dictated by the dice
rolls. Maybe I'm just paranoid, but the double 6's always seemed to go the
PC's way when bearing off.
Ray H
Taree
Since you are already complaining about the dice of a program you don't
even know what it's called anymore, by reasonably strong, I guess you
want a weak program that barely understands the rules? Try the above:
Try not to laugh.
Incorrect assumption. By reasonably strong I mean reasonably strong. Period.
The only laugh I have had today is your reply. As per usual, most newsgroups
are full of idiots. You are one of them. Thanks for nothing.
Ray H
Taree
Many thanks for a helpful reply. I did go to some pain to stress that I
didn't actually believe the dice might have been biased, on the previous
program, and stressed it was probably undue suspicion on my part. I forget
the program now, but a very short list of viable names will jog the memory.
Ray H
Taree
sooth wrote:
> www.gnubg.org
> The program is free and is world-class strength. It's default setting
> on installation, however, is only expert strength. It can also be set
> to weaker levels if you wish. I will caution you, though, if you are
> typical of backgammon players without a great deal of experience, you
> may convince yourself the program cheats. It does not. If you don't
> believe this, set it to use manual dice and roll them yourself.
Yes indeed sooth
www.gnubg.org is a great program that can help even the most seasoned
players,but still has tutor mode and other features for beginner to
intermeditate players.
I've got to give kudos to GNU, however. GNU has to be the best there
is. World-class play, a full range of features (analysis, tutor mode,
etc.), and eye-popping graphics. All that at a competitive price --
free!
Jellyfish version 3 rings a bell now. Not bad, and the version where I
suspected "loaded" dice.
But the GNU program is just awesome. Many thanks to all concerned for this
program. Loaded like a dream and installed in a jiff. The analysis is super
great too, and even gives me an analysis of the 'luck' side of it wrt the
dice. Needless to say, I won't report on its analysis of my moves for one
match<g> And I had super dice rolls too, compared to GNU.
Embarassing.
Ray H
Taree
> <mwse...@yahoo.com> wrote in message
> news:1118163584....@f14g2000cwb.googlegroups.com...
> > Jellyfish isn't bad, either, and plays at a world-class level. You can
> > download the most basic version (it just plays, no analysis, or
> > anything) for free. http://www.jellyfish-backgammon.com/
> > I've got to give kudos to GNU, however. GNU has to be the best there
> > is. World-class play, a full range of features (analysis, tutor mode,
> > etc.), and eye-popping graphics. All that at a competitive price --
> > free!
> Jellyfish version 3 rings a bell now. Not bad, and the version where I
> suspected "loaded" dice.
JF3 was world class or close (though GNU is somewhat stronger). You
suspected loaded dice because it was just that much better than you.
Don't worry, it's a lot better than almost everybody else too.
Michael
Nobody would work on a free program for years and then manipulate the
dice. But there's nothing wrong with rolling the dice yourself. But
wait, you sure your dice are correct? Judging from your observations,
your dice are pretty bad. I'd recommend you use the fair dice of gnubg
instead.
I do that quite a lot.
> You'll find that you get
> your butt kicked, losing at least 60% of the points, but probably a lot
> higher percentage. Yet if you roll it out manually (assuming you are a
> decent player) you can hold your own, winning about 50% (on the top
> setting) of the points.
You compare your performance against two different settings of Gnu. Do
you have the slightest idea, how many games you would have to play (for
each setting), to make your observation statistically convincing?
Backgammon is a casino game. You will win LESS than 50% against Gnu
(any setting) in the long run even when relying on the tutor, because
the tutor doesn't detect ALL mistakes, unless you change it's settings.
> maual rolls. I've created a quick roller. For those interested, I got
> a clear glass jar and put dice in it and put the lid on. Then you just
> turn it upside down and shake a few times.
I thought there may have been an independant little bit of software, that
gave truly random dice throws. Tests for randomness would be required and
statistically, (forcing me to get my stats books out), there are definite
tests for degrees of randomness.
But surely an independant bit of software is all that is needed, n'est-ce
pas, in order to clear the software one is playing?
Btw, the clear jar is a good idea, but the throws would need to examined and
tested statistically in order to prove true mathematical randomness.
Ray H
Taree
Also try Googling in Groups for "gnubg cheat dice"; you'll
find this conversation repeated ad nauseam.
> For anyone here, now and in the future, who tends to believe the cheat
> gospel of Saint Murat.....there is a relatively simple and efficient
> way to prove to yourself that a certain bot does not cheat:
>
> -get two computers,
> -place them side by side,
> -install the program under investigation on both computers,
> -run a game or match on computer #1 via the internal dice engine,
> -copy the same rolls via manual dice on computer #2.
>
> Lo and behold ! Surprise, surprise...
> ...the moves and cube offerings of the bot on computer #2 are
> identical to the ones on computer #1.
>
> And indeed, if this is not the case, there is "cheating"going on.
> I thought there may have been an independant little bit of software, that
> gave truly random dice throws. Tests for randomness would be required and
> statistically, (forcing me to get my stats books out), there are definite
> tests for degrees of randomness.
>
> But surely an independant bit of software is all that is needed, n'est-ce
> pas, in order to clear the software one is playing?
gnubg->Settings->Options->Dice->random.org
Rather than wasting time rolling real dice and then inputting the roll into
Gnu, use that time to improve your game by asking Gnu to point out your
errors. As your game improves you may notice that Gnu "cheats" less and
less.
Raymond Hall wrote in message
That's interesting, not that it matters to __me_.
But when I looked that up I noticed another option
"File". Didn't check it out, but if "File" works
anything like in Jellyfish this means you can set
gnubg to take its dice rolls from a text file
(maybe you put the filename in the Seed box,
maybe it has to be a certain filename?)
If that's the way it works then this gives a person
a way to make it use whatever random number generator
you want, with a little programming maybe: Make a
program that uses your favorite random number generator
to write a file of random numbers and then starts
gnubg.
************************
David C. Ullrich
I never suggested gnubg cheated. In fact playing it, so far, I am very
impressed, not only with its play, but the piece of software itself is a
gem. Its analysis of my games so far suggest that my cube strategy is
"awful". Will have to bone up on some reading about efficient use of the
cube.
Ray H
Taree
At any rate, this isn't a conclusive test. All the recommended moves
are is a suggestion based on what is most likely to happen given the
current state of the board/match/cube/etc. Unfortunately, of all valid
rolls, none is impossible, and you can't eliminate the highly
improbable (I once rolled 6-6, 5-5, 6-6 during a particularly close
bearoff -- impossible? No. Improbable? You bet. I got accused of
cheating). Just because GNU suggests a particular move as best, doesn't
mean that the next roll will make the suggested move backfire.
Personally, I've found that the better I get at backgammon, the worse
GNU and Jellyfish seem to play.
Create a text file with one roll per line:
==========
61
33
42
54
61
11
13
15
46
32
16
42
55
44
34
13
...
==========
The first time gnubg wants a roll, it will ask for the name of the
file. When the file is exhausted, gnubg starts over at the beginning.
--
Jim Segrave (j...@jes-2.demon.nl)
On Tue, 14 Jun 2005 22:02:42 -0000, Jim Segrave <j...@nl.demon.net>
wrote:
************************
David C. Ullrich
Talk about naive!!
DK
"Simon Woodhead" <si...@uniq.com.au> wrote in message
news:42ae29ba$1...@duster.adelaide.on.net...
> Isn't it obvious that the program can identify the sequence
> of numbers the rolls came from?
With manual dice ? I'd like to know how...
> Not to mention program-to-program communication ...
In the test below, the computers are not connected ...
> Talk about naive!!
Indeed!!
> DK
SW
It is trivial with any bot to prove that it doesn't choose different dice
based on the current position. Nor can you demonstrate that knowing what
will come next (in the upcoming sequence of dice that will be coming from
the RNG) that it will ever choose a different play based on a different
sequence of rolls that will come next.
You aren't naive, just ill-informed and wrong.
--
GPC