Is there a free backgammon program

[Aaron]

Aaron, 1/22/2006,8:38:11 AM, wrote:

[about GNU backgammon]

I sense you are trying to start something with me, but I'm
probably wrong.

I'm just trying to see if you really think GNUBG wins unfairly.

So you think it's programmed to cheat?

That's exactly what would be considered cheating in a program.

Your sample size is too small and your methods are too subjective to
reach a conlusion.

Exactly. Head over to rec.games.backgammon and you will see people
accusing backgammon programs of all kinds (including Jellyfish, GNUBG,
Snowie etc), or online servers of cheating.

They even have a humourous standard form for writing such complain!
http://tinyurl.com/domwg

But correct me if I'm wrong, no one has ever brought anything remotely
suspicious.

The fact is when you play a very good backgammon player and GNUBG is
world class, he just knows how to play the odds, so that if they pay off,
they look like miracles. The winner is always lucky.

If you don't believe, me, use the manual dice option, you will find that
the program kicks your ass just as handily.

My guess is that either gnubg was really lucky (possible for short
durations) or it plays at a higher level than JF, so it seems luckier.

Certainly in the match i cited above, *I* got far luckier throws of the
dice which allowed me to pull ahead 6-5 in the match before GNUBG won 7-
6.

It seems very unlikely that the programmer would make the dice-rolling
routine aware in any way of the current state of the board. The source
is available, and he'd be bombarded with flames.

I've being reading some threads about accusations of cheating at
rec.games.backgammon, and I'm amused at the lengths some people go to try
to convince the accusers (who have no evidence beyond some subjective
feeling) that the programs are not cheating.

For example jellyfish offers an option to read dice rolls from a text
file. GNUBG offers a choice of different psuedo random generators etc.

And of course, even these measures are not 100% proof it doesnt cheat,
not if you assume a devious programmer. For example even if all rolls are
predetermined in the txt file which prevents the bot from creating lucky
rolls on the fly when it needs it, it can still cheat by 'looking ahead'
all the rolls, and factoring that information into it's play.

Some even claimed that even with open source programs it's exceedingly
hard to prove it isn't cheating, how much worse for none- open source
programs??

In fact, because JF predates gnubg, complains of JF cheating far
outnumber that of other programs.

 

[George]

Jellyfish will not permit you to make normal backgammon moves in every case.

I ran into two instances where I was forced by the program to make unnatural
moves.

I haven't experienced this problem with Jellyfish. There is something
that took a while for me to figure out. After the dice rolls, the game
defaults to playing the highest dice roll first. You can right click
on the dice and play the lower dice roll first if you want. This may
resolve the unnatural dice moves that you talk about. It can be a
little confusing but if you click on a checker and the move for the
higher dice roll is blocked, it will play the lower dice roll first in
that case.

George

 

[George]

Opps! I am sorry everyone. I thought I tried all the freeware
backgammon games but I didn't know about GNU backgammon. I did try all
of the following:

WoodyGammon - http://www.caiman.us/scripts/fw/f1389.html
Quick Backgammon - http://www.caiman.us/scripts/fw/f636.html
JellyFish - http://www.caiman.us/scripts/fw/f914.html

I found the checkers on Quick Backgammon a little hard to move.
WoodyGammon has a nice visual appearance but the game play didn't seem
as good as Jellyfish. I am going to try GNU backgammon and will let
you know what I think. I am encouraged that GNU backgammon is supposed
to use the same neural net ability as Jellyfish.

 

[badgolferman]

This statement is probably rather accurate.

The fact is when you play a very good backgammon player and GNUBG is
world class, he just knows how to play the odds, so that if they pay
off, they look like miracles. The winner is always lucky.

I have played GNU some more the past few days and tested out one more
tool of it. The one where it gives you hints about your play and
assigns percentages to your play.

The first few times I played the game the dice roll of the computer was
ridiculous in my opinion. I have already made comments about that.
Since my last post I have played it five more times and have defeated
it each time. The last time I won it pointed out nearly ten moves of
mine it thought bad. The result of that game was a backgammon against
it.

I am not sitting here blowing my own horn and declaring I am so much
better than GNU Backgammon. However an experienced player can learn
the tendencies of his opponent and change his style to neutralize their
strengths. No doubt if I play it more the luck will shift over or it
will learn my style. Playing people is more challenging and fun
anyday, but I realize this thread is about computer games.

GNU is an excellent program, Jellyfish is an excellent program and
suits me better. You can't go wrong with either one if you want to
sharpen your skills. Another poster declared Jellyfish didn't allow
him to make normal moves. I have yet to see that in the four years I
have been playing it.

 

[George]

I have tried GNU backgammon for Windows. I installed it from this
site: http://www.gnubg.org/index.php?itemid=21

Initial impressions are good. It looks like a professional pay version
and has a great many options. It will take me a while to figure
everything out. It is much more complex than Jellyfish. GNU is still
in beta form. I'm betting it won't stay freeware.

 

[Craig]

GNU is an excellent program, Jellyfish is an excellent program and
suits me better. You can't go wrong with either one if you want to
sharpen your skills. Another poster declared Jellyfish didn't allow
him to make normal moves. I have yet to see that in the four years I
have been playing it.

Wrt Jellyfish "normal moves," I notice that it'll only allow me to move
the piece w/the greater number rolled. That made for some awkward
rethinking on my part. Other than that, but Jelly & Gnu seem like nifty
proggies for the backgammon fanatic.

-Craig

 

[Alastair Smeaton]

Wrt Jellyfish "normal moves," I notice that it'll only allow me to move
the piece w/the greater number rolled. That made for some awkward
rethinking on my part. Other than that, but Jelly & Gnu seem like nifty
proggies for the backgammon fanatic.

-Craig

There are options there to set it up so you move the larger dice
number with a left mouse click, and the smaller one with a right mouse
click - so you choose which one to move first.

I prefer jellyfish to gnu purely as I have not yet had the chance to
explore all the options in gnu - very powerful but quite complicated
to a hobbyist like me

 

[Aaron]

This statement is probably rather accurate.


I have played GNU some more the past few days and tested out one more
tool of it. The one where it gives you hints about your play and
assigns percentages to your play.

The first few times I played the game the dice roll of the computer was
ridiculous in my opinion. I have already made comments about that.
Since my last post I have played it five more times and have defeated
it each time.

Since you managed to beat it, does that mean you think it's not cheating
now? Or do you maintain it cheats only the first time to look better

Okay sorry about that, I presume you are now maintaining that it was just
freak chance the first time? No foul play involved?

In any case, I'm impressed but not that surprised if you can beat the
default 'expert' level which is actually midway between it's difficulty
levels.

If a newbie like me can give it a game (with good luck), an experienced
player like yourself should be able to beat it if the dice go your way.

I would be really suprised if you could beat the highest "grandmaster"
level consistently though (say over a dozen matches), if you can, you are
at least a top pro. I bit of googling shows that most people think that
at that level it's probably stronger than Jellyfish level 7 by how much I
don't know.

The nice thing about backgammon is that even at such levels, you can
still win a few, if you are lucky. In the chess field, It's almost
impossible to beat say Rybka or Fruit fairly, though employing anti comp
tactics might give you a slightly larger chance.

The last time I won it pointed out nearly ten moves of
mine it thought bad. The result of that game was a backgammon against
it.

Well it could be wrong and there's a hole in it's understanding of such a
position, or it could be objectively right, but you had a lot of luck.
What's your 'luck rating' for that game? That's the fun thing about
backgammon, luck plays a part.

Crank up the analysis level to World class, or Grandmaster level,
sometimes when it looks ahead 1-2 ply it gives quite different advise for
racing games.

Still one way to tell though who's right is to do a 'rollout' and run the
position through say a thousand times and see on average who's right to
eliminate the factor of chance.

Rollouts are yet another feature available in gnubackgammon but not in
jellyfish.

I am not sitting here blowing my own horn and declaring I am so much
better than GNU Backgammon.

Sure sounds like it.

However an experienced player can learn
the tendencies of his opponent and change his style to neutralize their
strengths. No doubt if I play it more the luck will shift over or it
will learn my style.

I don't think it learns about specific styles. Anyway from what i have
read these computers tend to be relatively weaker in cube play, and
handling back games, and some human experts have tried to exploit that.

GNU is an excellent program, Jellyfish is an excellent program and
suits me better. You can't go wrong with either one if you want to
sharpen your skills.

Personally I would think that if you just want to play JF is good enough.
It just plays, that's all.

Less strong players who want to learn could and will benefit from mentor,
and analysis features.

And for the more advanced players there are some cool advanced features,
like rollouts which can help them settle the objective truth of the
position.

Another poster declared Jellyfish didn't allow
him to make normal moves. I have yet to see that in the four years I
have been playing it.

Perhaps some hidden bug that triggers only in rare circumstances. I admit
That's just as likely as a bug that gives gnubg better rolls in the first
game only.

 

[badgolferman]

Alastair Smeaton, 1/24/2006, 3:08:22 AM,

There are options there to set it up so you move the larger dice
number with a left mouse click, and the smaller one with a right mouse
click - so you choose which one to move first.

Maybe that's why I haven't noticed any problems. I have it set to Drag
& Drop.

 

[George]

When I played GNU last night, I decided to try the "Hints". On several
occations it did suggest a better move. On one occation though, I sent
my opponent bacj and was on the bar. I made a made my move and before I
let it continue I hit "Hints" and found my move was ranked a 4. I
looked at the 1, 2 and 3 suggestions and noticed that each of them left
one of my men unprotected in my home court. If my opponent had gotten
the right dice roll, he could have sent me back when he came out. I
decided to stay with my #4 ranked move. The move I selected left all
my men in my home court protected (and I won). I suppose GNU may be
using a different stratergy than I use or there may be a few flaws in
the logic (this is a beta game). Anyway, I think the hints are worth
reviewing to see if there is a better move.

 

[Klaatu]

I have tried GNU backgammon for Windows. I installed it from this
site: http://www.gnubg.org/index.php?itemid=21

Has anyone put GNU backgammon on a USB drive so as to make it portable?
Does it work?

 

[Aaron]

When I played GNU last night, I decided to try the "Hints". On several
occations it did suggest a better move. On one occation though, I sent
my opponent bacj and was on the bar. I made a made my move and before I
let it continue I hit "Hints" and found my move was ranked a 4. I
looked at the 1, 2 and 3 suggestions and noticed that each of them left
one of my men unprotected in my home court. If my opponent had gotten
the right dice roll, he could have sent me back when he came out.

Well yeah but if he didn't get the right roll, you might be way better
off. You can't tell just by looking if the risks is worth the rewards.
You can't even say you are right because you won this time because you
might have made a lucky roll.

You need to do a roll out, run the position through say 1000 times using
both moves, and see if on the balance you did better or worse.

Most beginner players tend to play cautiously to avoid leaving possible
blots if possible. But many of the time, it's better to take risks since
the rewards outweight the risk.

 

[Aaron]

When I played GNU last night, I decided to try the "Hints". On several
occations it did suggest a better move. On one occation though, I sent
my opponent bacj and was on the bar. I made a made my move and before I
let it continue I hit "Hints" and found my move was ranked a 4. I
looked at the 1, 2 and 3 suggestions and noticed that each of them left
one of my men unprotected in my home court. If my opponent had gotten
the right dice roll, he could have sent me back when he came out.

Well yeah but if he didn't get the right roll, you might be way better
off. You can't tell just by looking if the risks is worth the rewards.
You can't even say you are right because you won this time because you
might have made a lucky roll.

You need to do a roll out, run the position through say 1000 times using
both moves, and see if on the balance you did better or worse.

Most beginner players tend to play cautiously to avoid leaving possible
blots if possible. But many of the time, it's better to take risks since
the rewards outweight the risk.

 

[badgolferman]

I would be really suprised if you could beat the highest
"grandmaster" level consistently though (say over a dozen matches),
if you can, you are at least a top pro. I bit of googling shows that
most people think that at that level it's probably stronger than
Jellyfish level 7 by how much I don't know.

I played Grandmaster level five times. It was a bit annoying the
amount of time it took to decide upon moves but I can live with that.
The first three times it creamed me. The fourth time I was ahead 6 to
5 when it came back to win. The fifth match I reversed the same
conclusion upon it. I'll continue playing it for a while to compare to
Jellyfish level 7. At this point I would have to give GNU the slight
edge, but I haven't played it enough to make a real judgement.

I will recant my objections to its "lucky" rolls. Either it heard me
complaining about it or the odds have shifted. Either way I will give
the game its due and include it in recommendations to others. GNU is
pretty and has many additional features. Jellyfish is simple and plays
a strong game.

 

[George]

badgolferman, Don't feel bad. I installed Jellyfish on a friend's
computer and he accused it of cheating on getting the right dice rolls.
It seems clear to me that GNU and Jellyfish play the game a bit
differently. I think most backgammon players would probably want both
games installed on their computer, beause it is like playing against 2
different players.

No one answered the question about whether GNU is portable. I haven't
tried it. I'm betting Jellyfish would probably be a better portable
game, because it is smaller in size and requires no installation.