CYBERGAMMON IS A CHEATER
Omar S Mamoun
Those who frequent Netgammon should beware that Cybergammon, the neural net computer
opponent, is a cheater. I played an 11 point match with the beast. He committed
such absurd technical errors as leaving blots (in several instances) on his goal
board while I still had men situated there. He must have left about 10-12 blots
during that match and I only hit two or three of them. In a couple instances he
left two blots and I didn't hit any! We're talking about blots 6 spaces away or
less. He got several "miracle throws per game, including entering three blots in
with a double, an instance where he entered his man and hit one of my blots (7%
chance at the time), and rolling multiple doubles in bear offs, among other
instances. He plays with
terrible technical skills, but makes it up by effectively destroying my positions
(he calls the dice) and giving me dice rolls which make it almost impossible to
close off positions. There was an instance when I was bearing off, and only a
6-1 would force me to leave a blot and (as they say in netgammon.fr) "voila!" he
got the roll. Of course, human bias may have a role when accusing a computer of
cheating, but when a computer keeps getting rolls "that accomplish two great
things at a time" (so to speak), surely something crazy must be going on,
especially if the computer leaves blots on its goal board not because it has to,
but rather as part of its strategy.
Peter Fankhauser
osma...@mail1.sas.upenn.edu (Omar S Mamoun) wrote:
>
<snip - see subject>
It is some time ago since I've played cybergammon. Yes it is
a weak player, strategically and technically. But I have some
doubts that it cheats for several reasons:
(a) what is its rating currently? - at the time
I last played on netgammon it was somewhere in the 1300. Now
with cheating it should do better.
(b) the implementor of netgammon and cybergammon would be pretty
foolish to implement a cheating strategy, which could be easily
uncovered (e.g. by the test Stephen Turner once suggested on r.g.b -
oops r.g.bg :). Netgammon would have lost its credibility forever -
not a good start.
(c) implementing such a cheat involves quite some overhead in the
protocol the server has to handle (not to talk of the rules which
determine lucky rolls for the bot).
(d) the "evidence" you mentioned just showed cybergammon is weak
in bearoff (I know of a couple of programs making similar mistakes).
With the small sample, you can not conclude it cheats.
funk
Goto Informatique
Peter Fankhauser wrote:
>
> osma...@mail1.sas.upenn.edu (Omar S Mamoun) wrote:
> >
> <snip - see subject>
>
> It is some time ago since I've played cybergammon. Yes it is
> a weak player, strategically and technically. But I have some
> doubts that it cheats for several reasons:
>
> (a) what is its rating currently? - at the time
> I last played on netgammon it was somewhere in the 1300. Now
> with cheating it should do better.
Thank you for trusting me !! Cybergammon doen't cheat and the reasons
you give are the best proof !
But just a point : Cybergammon is today rated 1520, his elo moves
between 1470 and 1630. his overall performance is 1540.. I know
that Cybergammon is not so strong, but please not so bad :)
Xavier Dufaure de Citres (XKlibur and Sysop on Netgammon)
Goto Informatique
This the kind of words I can't let without a reply ! I fell particulary
insulted by this message. Do you react like that each time you loose a
game ? Don't you know that luck is a part of backgammon ?
Your message involve that you think that I am a cheater ! that i've
programmed this software with the idea to cheat !
I know that Cybergammon has weakness (especially during the end of the
game) but as Peter already answer: do you think that Cybergammon could
be rated only 1540 if he cheats ? and what can be the aim of making
something like that ?
I am really upset by people who are not good enough in backgammon to be
able to see when they are lucky and remember only the games they lose
because their opponent was lucky. I can understand that you are not glad
if you loose a game, but you know : some people are better than you, and
some has more luck than you in a game : do you think they all cheat ?
I close the argument and i don't want to ask you any apologizes. Anyway
i hope you have fun playing on our server !
Xavier Dufaure de Citres
Cybergammon & Netgammon develloper
System Operator on IBS
Omar S Mamoun
Peter Fankhauser (fank...@darmstadt.gmd.de) wrote:
: osma...@mail1.sas.upenn.edu (Omar S Mamoun) wrote:
: >
: <snip - see subject>
: (d) the "evidence" you mentioned just showed cybergammon is weak
: in bearoff (I know of a couple of programs making similar mistakes).
: With the small sample, you can not conclude it cheats.
: funk
NO! Not the bearoff! We're talking blots during the opening game and middle game. If it were
bearoffs, that would be mildly understandable. But CG never leaves blots in the bearoff. Its dice
are too beautiful!
--OM
michael rochman
Goto Informatique <con...@goto.fr> wrote:
>But just a point : Cybergammon is today rated 1520, his elo moves
>between 1470 and 1630. his overall performance is 1540.. I know
>that Cybergammon is not so strong, but please not so bad :)
Xavier...
It recently beat me pretty bad in a match to 9.
However, I can remember a couple of critical rolls
I had where I needed specific numbers and got them
to win one of my two wins in that match.
How about 2 men on his one pt behind his 5 pt prime
(bar through two pt). My next two rolls were 1,1 followed
by 6,6 to win a gammon from him.
The point is, it doesn't cheat. It's cube usage is strange,
but it sure doesn't cheat.
mike
Peter Fankhauser
osma...@mail2.sas.upenn.edu (Omar S Mamoun) wrote:
>
>
> NO! Not the bearoff! We're talking blots during the opening game and middle game.
Oh! so cybergammon left blots in the early stages of the game?
Did it sometimes even have the nerve to leave two blots - like
with an opening 23 or so?
Now that is kind of weird...
funk
Omar S Mamoun
Peter Fankhauser (fank...@darmstadt.gmd.de) wrote:
: funk
>>>Nope. I'm talking blots where there is no sound reason to leave them. Opening 23 with two goal
board blots is unwise, but CG's blots were even more unwise then this. CG has decent opening
strategy, I think. It is not difficult to program a computer with all the right openings.
Peter Fankhauser
> board blots is unwise, but CG's blots were even more unwise then this. CG has decent opening
> strategy, I think. It is not difficult to program a computer with all the right openings.
Omar, you're perfectly right. It is of course silly to voluntarily
leave goal board blots. But those neural net computers still DO it!.
For example, when master loner (mloner on fibs and GG - I suppose),
has fallen behind in the match, because he failed to control the
dice for a while, he opens 21 by bringing down one checker from the
midpoint with the 2 and VOLUNTARILY!!! leaving a blot in its goal
board with the ace. Now what is that??? Any man with senses sees
that this just risks a fundamental loss in the race by 20!!! pips
(that is a 55), and thus either brings down one checker
from the midpoint by 3 points, or tries to prepare a backchecker
for escape. Whenever I see master loner teasing me with such
silly plays a red light starts flashing in my head. I just know
what will be happening. Either I don't roll a 4, master loner covers
his blot, hits me when I try to bring one of my back checkers home,
and I never come into the game again. Or I do roll a 4, but master
loner just uses the hit checker to make an annoying point in my
goal board, maybe gives me the chance to send a few more checkers
back, and then waits patiently for a late shot, to hit me when it
has already closed many points in its goal board. NO MATTER what
I roll, I lose a gammon.
It is a wicked game. There are only very few mortals, like
the famed M.I.A Rookman, who sometimes can cope with these neural
nets by adopting their sick strategy. You and I are not one
of them - alas.
funk
Omar S Mamoun
Peter Fankhauser (fank...@darmstadt.gmd.de) wrote:
: osma...@mail1.sas.upenn.edu (Omar S Mamoun) wrote:
: >
: >
: > >>>Nope. I'm talking blots where there is no sound reason to leave them. Opening 23 with two goal
: > board blots is unwise, but CG's blots were even more unwise then this. CG has decent opening
: > strategy, I think. It is not difficult to program a computer with all the right openings.
: Omar, you're perfectly right. It is of course silly to voluntarily
: leave goal board blots. But those neural net computers still DO it!.
: the famed M.I.A Rookman, who sometimes can cope with these neural
: nets by adopting their sick strategy. You and I are not one
: of them - alas.
: funk
>>>Sure they do it: if the calculations make sense there is no reason for them not to do it. I
have seen transcripts of mloner and jellyfish games. They are both very strong players, and I can
tell if a player is weak or somewhat strong or very strong. Why don't you try playing a 10 point
match with Cybergammon. He is no where near the playing strength of either ML or JF. I would say
that he is slightly below intermediate level. Play some games with him and see how silly his
blot strategy is.
michael rochman
Peter Fankhauser <fank...@darmstadt.gmd.de> wrote:
>For example, when master loner (mloner on fibs and GG - I suppose),
>has fallen behind in the match, because he failed to control the
>dice for a while, he opens 21 by bringing down one checker from the
>midpoint with the 2 and VOLUNTARILY!!! leaving a blot in its goal
>board with the ace. Now what is that??? Any man with senses sees
>that this just risks a fundamental loss in the race by 20!!! pips
>(that is a 55), and thus either brings down one checker
>from the midpoint by 3 points, or tries to prepare a backchecker
>for escape.
Funk,
Would you please post a diagram of an exact situation. From
your description it appears that MLoner, JellyFish, et all
are trying to quickly establish some offense at very little
risk as they are already behind in the race. This would
probably be correct strategy for that situation.
Regards, Mike
Peter Fankhauser
michael rochman wrote:
>
> Funk,
>
> Would you please post a diagram of an exact situation. From
> your description it appears that MLoner, JellyFish, et all
> are trying to quickly establish some offense at very little
> risk as they are already behind in the race. This would
> probably be correct strategy for that situation.
>
> Regards, Mike
argh, my posting was supposed to be a joke. Of course it
is appropriate to play an opening 21 with 13/11, 6/5, especially
when behind in the match (match that is, not game, race
should be even for a starting position, should it not?).
Probably I should refrain from trying to be subtle in a foreign
language...
funk
James Eibisch
>Probably I should refrain from trying to be subtle in a foreign
>language...
How long before you get flamed for elitist subtlety I wonder?
--
_
James Eibisch ('v') N : E : T : A : D : E : L : I : C : A
At home in (,_,) http://www.i-way.co.uk/~jeibisch/
Reading, U.K. ======= http://www.revolver.demon.co.uk/
Kit Woolsey
com> <32085F34...@darmstadt.gmd.de>:
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
Peter Fankhauser (fank...@darmstadt.gmd.de) wrote:
: michael rochman wrote:
: >
: > Funk,
: >
: > Would you please post a diagram of an exact situation. From
: > your description it appears that MLoner, JellyFish, et all
: > are trying to quickly establish some offense at very little
: > risk as they are already behind in the race. This would
: > probably be correct strategy for that situation.
: >
: > Regards, Mike
: argh, my posting was supposed to be a joke. Of course it
: is appropriate to play an opening 21 with 13/11, 6/5, especially
: when behind in the match (match that is, not game, race
: should be even for a starting position, should it not?).
: Probably I should refrain from trying to be subtle in a foreign
: language...
: funk
Your English is fine, Peter, as is your sense of humor. Everybody should
have no trouble seeing that your posting was a joke (well, almost everybody).
Kit