Jellyfish a way of cheating?
Ludovico
perfect programme and whether it ca be used to input moves and therefore
cheat on other sites. I don't wish to do that myself but it is something
that I was discussing with a partner the other day.
Kind regards,
Ludo
Patti Beadles
better than almost all of them in practice, since it will never
overlook a play.
Yes, it could in theory be used to cheat online. It's a relatively
slow process to do it manually, though.
-Patti
--
Patti Beadles |
pat...@gammon.com | Thinking about you as innocent
http://www.gammon.com/ | is sort of like thinking about
or just yell, "Hey, Patti!" | a viking playing badminton.
Ludovico
limit be enough for cheating with Jelly? How do I know if i'm being cheated?
Ludo
"Patti Beadles" <pat...@mauve-new.rahul.net> wrote in message
news:b140j9$c64$1...@blue.rahul.net...
Peter Schneider
1. Two acquaintants of mine who I know from my Hamburg time tried the following
simple setup: One of them played on FIBS while the other one was operating
Snowie on a second computer, entering the dice and opponent's moves and telling
how to move. Even at 3-ply, it reportedly is as fast as if you think yourself.
If you play for significant amounts of money, you'll want to think a little then
and when, won't you? Nobody can tell whether your buddy feeds Snowie instead.
2. Since gnubg is available in source code, it's no problem to program a FIBS
client which moves gnubg's moves without any manual intervention except the
kibitzes. Examples are the gnubg bots playing on FIBS and otaku's bots. gnubg is
probably stronger than JellyFish.
3. Some sites which live from money games claim to check games with Snowie to
detect whether one side actually moved the bot's moves. Well, gnubg allows you
to set a "noise" factor. That means, it plays generally very strong but
introduces a second or third best move from time to time, as any human would do.
This is afaik not distinguishable from (still very strong) human play.
For all these reasons, I can only strongly discourage anybody from playing
online for significant amounts of money, except against good friends. And who
plays good friends for significant amounts of money?
Regards,
Peter aka the juggler
Ludovico
think this is why there aren't that many backgammon website for money?
ludo
"Peter Schneider" <schneiderp...@gmx.net> wrote in message
news:b144df$vhg54$1...@news.hansenet.net...
Patti Beadles
Ludovico <nicc...@hotmail.com> wrote:
>Most websites I play on have a time limit to make the move, would that time
>limit be enough for cheating with Jelly? How do I know if i'm being cheated?
I've never played on a site that had moves on any sort of short timer,
nor would I want to... while I tend to play quickly, I like being able
to think about moves from time to time.
How long would it take? I'd recommend setting an experiment and
trying it yourself. You'll probably find it's somewhat slower than
normal play, but not impossibly so.
How do you know if you're being cheated? The best way is to do
offline analysis of previous matches. If you find that someone always
makes the "best" move, they're probably using the same software to
play that you are for analysis.
-Patti
--
Patti Beadles |
pat...@gammon.com |
http://www.gammon.com/ | The deep end isn't a place
or just yell, "Hey, Patti!" | for dipping a toe.
Douglas Zare
> Hi Ludovico [is this from Clockwork Orange?],
>
> > I would like to know whether anyone believes that jellyfish is a near
> > perfect programme and whether it ca be used to input moves and therefore
> > cheat on other sites. I don't wish to do that myself but it is something
> > that I was discussing with a partner the other day.
>
> 1. Two acquaintants of mine who I know from my Hamburg time tried the following
> simple setup: One of them played on FIBS while the other one was operating
> Snowie on a second computer, entering the dice and opponent's moves and telling
> how to move. Even at 3-ply, it reportedly is as fast as if you think yourself.
> If you play for significant amounts of money, you'll want to think a little then
> and when, won't you? Nobody can tell whether your buddy feeds Snowie instead.
There is a certain rhythm to normal human play. Some moves are played almost
instantly. The method of manually entering every move will cause a delay on
"obvious" moves, and moves which are not obvious but where humans tend to play
quickly anyway. On the other hand, there won't necessarily be a delay when humans
would ordinarily think for a while.
> 2. Since gnubg is available in source code, it's no problem to program a FIBS
> client which moves gnubg's moves without any manual intervention except the
> kibitzes. Examples are the gnubg bots playing on FIBS and otaku's bots. gnubg is
> probably stronger than JellyFish.
I'm not sure about "no problem," but it is certainly doable. I think most
programmers would not find it worth doing for rating points.
> 3. Some sites which live from money games claim to check games with Snowie to
> detect whether one side actually moved the bot's moves.
One site does, and I am rather skeptical of the effectiveness of their methods. Are
there more?
> Well, gnubg allows you
> to set a "noise" factor. That means, it plays generally very strong but
> introduces a second or third best move from time to time, as any human would do.
> This is afaik not distinguishable from (still very strong) human play.
No, it would not resemble strong human play. Try it, and compare with matches
against top human opponents. The errors look completely different. This is a flaw
with current backgammon programs. Their errors when weakened are very strange, and
practicing against them does not prepare you for playing against human opponents of
the same strength.
A more common concern is that someone might be using a bot on a few decisions,
perhaps take/pass decisions. It is normal to think for a while then anyway, and
there is no way to detect is someone cheats once or twice. However, it is not
possible to go from an intermediate level of play to an expert level by this method.
An intermediate player would blunder too much on plays he/she thinks are clear.
Again, there are timing issues.
> For all these reasons, I can only strongly discourage anybody from playing
> online for significant amounts of money, except against good friends. And who
> plays good friends for significant amounts of money?
I've seen no evidence of cheating for small stakes (say, $25/point or $100/match),
but I have been told by a pro that he has found several online cheats for high
stakes, and that is quite plausible.
Far more common than cheating is for people to select their opponents to be weak.
There is nothing wrong with this if it is done openly. I think artificially lowering
one's rating should be more strongly discouraged, though. I think TMG does itself a
disservice by not keeping ratings at all. They cater to the sharks. Most players
should worry more about accidentally playing a strong human player (some of whom
still have ratings of 1500-1600 on GamesGrid) than they should worry about cheating.
Douglas Zare
Ludovico
Ludo
"Patti Beadles" <pat...@mauve-new.rahul.net> wrote in message
news:b14ckg$fs7$1...@blue.rahul.net...
bob
Douglas Zare wrote:
>
>
>>2. Since gnubg is available in source code, it's no problem to program a FIBS
>>client which moves gnubg's moves without any manual intervention except the
>>kibitzes. Examples are the gnubg bots playing on FIBS and otaku's bots. gnubg is
>>probably stronger than JellyFish.
>
>
> I'm not sure about "no problem," but it is certainly doable. I think most
> programmers would not find it worth doing for rating points.
It was trivial. You're probably right about "most" and "think".
>
>>Well, gnubg allows you
>>to set a "noise" factor. That means, it plays generally very strong but
>>introduces a second or third best move from time to time, as any human would do.
>>This is afaik not distinguishable from (still very strong) human play.
>
>
> No, it would not resemble strong human play. Try it, and compare with matches
> against top human opponents. The errors look completely different. This is a flaw
> with current backgammon programs. Their errors when weakened are very strange, and
> practicing against them does not prepare you for playing against human opponents of
> the same strength.
It is far better to introduce the errors yourself, often using the most
"human" of the choices displayed, but at other times having to make up
moves on your own to appear credible.
>
>
>>For all these reasons, I can only strongly discourage anybody from playing
>>online for significant amounts of money, except against good friends.
STFU you bastard.
>
>
> I've seen no evidence of cheating for small stakes (say, $25/point or $100/match),
> but I have been told by a pro that he has found several online cheats for high
> stakes, and that is quite plausible.
Flying under the radar is far more lucrative in the long run. $10/hour
working from the comfort of your own home, doing what you love, is
awesome. Thank you all.
Peter Schneider
> Thanks, I suppose it all depends on how rich they are I suppose. Do you
> think this is why there aren't that many backgammon website for money?
Well, the bg market is certainly somewhat limited, compared to general gambling
in the net. I also think that it needs a lot of professionality to run a site
which deals with real money, which is another hurdle.
Regards,
Peter/the juggler
Peter Schneider
Ooops, bob changed his nick and is again wasting my and my computer's valuable
resources? Time to define another kill filter...
Peter Schneider
Thanks for your comments... profound as always even if I'm not totally
convinced, see below.
I wrote
> > 1. Two acquaintants of mine who I know
> > from my Hamburg time tried the following
> > simple setup: One of them played on FIBS while the other one was operating
> > Snowie on a second computer, entering
> > the dice and opponent's moves andtelling
> > how to move. Even at 3-ply, it reportedly
> > is as fast as if you think yourself.
> > If you play for significant amounts of
> > money, you'll want to think a little then
> > and when, won't you? Nobody can tell
> > whether your buddy feeds Snowie instead.
And you answered ([] from me):
> There is a certain rhythm to normal human play.
> [1]Some moves are played almost
> instantly. The method of manually entering every move will cause a delay on
> "obvious" moves, and moves which are not obvious but where humans tend to play
> quickly anyway. [2] On the other hand,
> there won't necessarily be a delay when humans
> would ordinarily think for a while.
[1]: Not necessarily... Even without money, some of my opponents think about
every move twice, and they are probably right to do so. There can be also lag as
a constant add-up to the delay which cannot be easily falsified (that you can
ping your opponents in 100ms does not mean that they can ping the server in
100ms). I think in a $25/pt game, I'd certainly take my time in every move. And
having a second person enter the moves is, reportedly, *very* quick.
[2] Well, to fake a delay shouldn't be a big problem ;-), even at the right
places.
> > Well, gnubg allows you
> > to set a "noise" factor. That means, it plays generally very strong but
> > introduces a second or third best move from
> > time to time, as any human would do.
> > This is afaik not distinguishable from (still very strong) human play.
>
> No, it would not resemble strong human play. Try it, and compare with matches
> against top human opponents. The errors look
> completely different. This is a flaw
> with current backgammon programs. Their errors
> when weakened are very strange, and
> practicing against them does not prepare
> you for playing against human opponents of the same strength.
Interesting. I have never done that. But do you think a rather weak player could
sense it? Even if so, could they *prove* it or would it stay a mere feeling? And
also, aren't the player's "error profiles" as different as their personalities?
I tend to make these "oh, I could hit there??" 0.5 errors then and when... on
any skill level, there may be players who tend to make mistakes that
everybody else shakes their head about.
> I've seen no evidence of cheating for small stakes
> (say, $25/point or $100/match),
Not small for me and enough to make a living from it.
Ludovico
backgammon4money.com but it then closed down on me and couldn't retreive the
money I had on deposit (I presume that was due to the lack of a decent
payment provider) i've read trhough gammonvillage.com and they claim to be
the leading website for backgammon for money. They state 20,000 actvie
members which sounds to good to be true, in backgammon4money.com you
certainly didn't have as many. Would you regard their website as
professional? I never played on it and would like to join, can you give me
some extra info on them before i join?
Ludo
"Ludovico" <nicc...@hotmail.com> wrote in message
news:b144tn$q7o$1...@sparta.btinternet.com...
Frank Mazza
I don't know how you got from Zare's post that: "Jelly Grandell" logs
on and Zare tells him that he must be cheating because he plays too
much like "Jerryfish"...
You are truly the master of the non sequitur.
Frank Mazza
On 2 Feb 2003 02:02:15 -0800, mu...@compuplus.net (Murat Kalinyaprak)
wrote:
>Douglas Zare wrote 3E35C39B...@math.columbia.edu
>
>> Peter Schneider wrote:
>
>>> 3. Some sites which live from money games claim to
>>> check games with Snowie to detect whether one side
>>> actually moved the bot's moves.
>
>> One site does, and I am rather skeptical of the
>> effectiveness of their methods. Are there more?
>
>Bla bla... So what if all sites did it and so what if
>it was effective...??
>
>Doesn't anybody remember what was written about "Jelly
>Grandell" playing too much like "Jerryfish" many years
>ago and then everthing else along the same lines ever
>since...?
>
>Some sumb ass pays $450 for a bot to learn how to play
>like it and after wasting years of his life on top of
>the $450, he finally gets to a point where he can play
>like that bot...
>
>And then, Zare and his ilk come along to tell him: "Oh,
>no! You can't ever possibly play like a certain bot"...
>
>"Jelly Grandell" logs on and Zare tells him that he must
>be cheating because he plays too much like "Jerryfish"...
>
>Hah hah haaaa... :)
>
>Dumb fucking, brainsless, lagicless morons... :))
>
>MK
Frank Mazza
Zare in his post was discussing the effectiveness of Internet playing
sites using Snowie to detect cheaters. If it can be shown that a
player often plays with an ET error rate (in matches, not
one-pointers) that hovers near zero, that is evidence that the player
may be using the assistance of a bot.
Jerry Grandell is nicknamed "Jelly Grandell" because he has an low
error rate when his matches are analyzed by the bots. But Grandell
doesn't have an error rate as low as some cheaters do. I've recorded
and analyzed him and many other top players on Gamesgrid, and their
error rates are low, but more erratic and not nearly as low as someone
using Snowie all the time would show. One match Grandell was even
rated "novice" by Snowie because he didn't cube when Snowie thought he
should have.
Zare and most players understand this about top players vs. cheaters.
In addition, even though Grandell plays more slowly than most, he also
by most accounts plays slowy when playing at live tournaments. His
success at live play, where there is no way he could use Snowie, gives
players confidence that he is not cheating when he is playing for
money on GamesGrid.
Hence I just found it absurd that you would think Zare would accuse
Grandell of cheating because of his excellent play.
I know from your previous posts that you have an exceedingly low
opinion of the morals, imagined sexual habits, etc. of backgammon
players who play for money.
For myself, I wouldn't worry I was being cheated against if I played
any of a number of famous players for money. In my opinion they have
enough pride in their game and morals not be cheating. I would just
worry about playing well enough to not embarass myself and also lose
my shirt.
But there are a number of players on GamesGrid who do cheat in various
ways. But since backgammon is a dice game, even using Snowie isn't
guaranteeing yourself a win in any particular match, just success over
time.
As for the scatological part of your post, it is what it is and you
are what you are.
Frank Mazza
On 6 Feb 2003 02:40:31 -0800, mu...@compuplus.net (Murat Kalinyaprak)
wrote:
>Frank Mazza wrote 1muv3v4uvlrcpcmr2...@4ax.com
>
>> I don't know how you got from Zare's post that: "Jelly
>> Grandell" logs on and Zare tells him that he must be
>> cheating because he plays too much like "Jerryfish"...
>> You are truly the master of the non sequitur.
>
>Well, let me explain more for the thick sculled then...
>
>If Jerry Grandell can play much like jellyshit, someone
>else could play much like gnudung and somebody else may
>play much like snowie, etc...
>
>Since all those "world-class" bots play similarly and
>agree on most moves, Jerry Grandell may be said to play
>much like gnudung or snowie also...
>
>Are you calling me "master of the non sequitur" because
>Zare was talking about "Snowie" and I was talking about
>"Jellyshit"...?
>
>Well, then, let's talk about "Bob Slower" playing like
>Snowie and because of that being called "Suck Blowie"...
>
>Of course, everybody including Zare would know him and
>if he logged on as "Bob Slower" or "Suck Blowie", they
>wouldn't say anything...
>
>But if some day "Bob Slower" felt playful and logged on
>as "Mrank Fazza" or "James Deer", Zare would pounce on
>him saying that he must be cheating because he plays too
>much like Snowie...
>
>Zare's ass is probably still hurting... So, keep licking
>his ass and you may alleviate his pain a little... :))
>
>MK
Kim
congratulations murat! once again you have proven without a shadow
of a doubt what a huge flaming asshole you are.
rebbeler
Bob
mu...@compuplus.net (Murat Kalinyaprak) wrote in message news:<2831c30c.03020...@posting.google.com>...
> Frank Mazza wrote 1muv3v4uvlrcpcmr2...@4ax.com
>
> > I don't know how you got from Zare's post that: "Jelly
> > Grandell" logs on and Zare tells him that he must be
> > cheating because he plays too much like "Jerryfish"...
> > You are truly the master of the non sequitur.
>
Frank Mazza
While nothing theoretically precludes a human being from having a near
zero error rate; no player to my knowledge has done it.
Players with far larger match databases then mine have studied the
error rates of those players who are by general consensus and
tournament success considered to be the best in the world, and no one
sustains an error rate near zero consistently in match play.
I don't have the book in front of me, but Kit Woolsey in New Ideas in
Backgammon discusses the difference in error rate between the world's
top players and the bots and notes that the bots play better to a
significant degree.
As for Grandell, there are any number of his matches recorded and
analyzed from both live play and online play. I don't think you would
see an appreciable difference in error rate at either.
But Grandell, for the sake of this discussion, is just a stand in for
any number of top players. Some of whom may well have error rates
lower than his.
But I don't think you will find anyone sustaining an error rate of
.41, say, over time in match play.
And yes Jerry Grandell could sign in as John Deere and KitWoolsey
could log on as Billy Thornbuckle and play for money against
unsuspecting players.
I certainly couldn't prove that they don't. But as I said, I don't
think they do. Call me naive and gullible if you want.
You seem to assume if it is possible to cheat, then most are cheating,
I don't.
As for the only winner in gambling being the "house," some Backgammon
players have built nice houses off of their winnings without needing
to cheat at all.
And yes ET means Extra-Terrestrial in Snowie parlance and thanks for
restraining the scatology.
Frank Mazza
PS: I was a Philosophy/Psychology/English triple major at a quite good
private college, so I'll skip the remedial night school classes at my
local community college. You may have problems with what you perceive
as my lack of thinking abilities, but most don't.
On 12 Feb 2003 01:31:46 -0800, mu...@compuplus.net (Murat Kalinyaprak)
wrote:
>Frank Mazza wrote 26a54vo9q9cs6oqsf...@4ax.com
>
>In response to your last comment, I will try to write without
>using any "scatological" expressions and see how you respond
>to it...
>
>> Zare in his post was discussing the effectiveness of Internet
>> playing sites using Snowie to detect cheaters.
>
>This is what I am discussing also... You didn't have to do it
>but thanks for trying to explain to me what the subject was...
>
>> If it can be shown that a player often plays with an ET error
>> rate (in matches, not one-pointers) that hovers near zero,
>> that is evidence that the player may be using the assistance
>> of a bot.
>
>What does "ET" error rate mean...? (Extra-Terrestrial?)
>
>Reagardless, that wouldn't be "evidence" of anything at all...
>
>Nothing precludes a human from being able to make "best moves"
>that coincide with the moves deemed as "best" by a robot. Even
>at a "near zero" rate...
>
>Your best bet may be to approach it from a different "finger
>print" of a robot. For example, unusual mistakes that robots
>would make but humans wouldn't...
>
>But even then, a human can learn to play so much like robots or
>a certain robot that he may also learn to make those same errors.
>
>Like I suggested many times before, try to "think". Taking a few
>classes at your local night school may help... While at it, take
>Zare with you...
>
>> Jerry Grandell is nicknamed "Jelly Grandell" because he has an
>> low error rate when his matches are analyzed by the bots.
>
>Yes, that was one of the points in my writing...
>
>> But Grandell doesn't have an error rate as low as some cheaters
>> do.
>
>Heh... :)) Anybody can be a "Grandell" but nobody better...??
>
>Up to "Grandell"s error rate, you are OK; you are "world class".
>
>But if you dare to be a notch better than "Grandell", then you
>become a "cheater"...??
>
>> I've recorded and analyzed him and many other top players on
>> Gamesgrid, and their error rates are low, but more erratic and
>> not nearly as low as someone using Snowie all the time would
>> show.
>
>So...?? You better have compiled and compared his error rate in
>"live" matches vs. his error rate in "online" matches per Snowie.
>
>Have you...???
>
>Since we know that he is not using Snowie to help him in "live"
>matches, his "online" error rate better not be less or else he
>will be suspect of cheating... :)
>
>And, he doesn't have to cheat by "using Snowie all the time"...!
>
>We know that some online *world-class gamblers* do "sand bag",
>"hustle", etc. by keeping their ratings artificially low, etc...
>
>So, if he just used it to help him "half of the time", wouldn't
>that be cheating and how would you have detected that...???
>
>> Zare and most players understand this about top players vs.
>> cheaters.
>
>They don't...! They only try to guess and assume. Not more...!
>
>If "Grandell" logged on as "Grandell", there would be no problem
>with him playing much like "JerryFish"...
>
>But if he logged on as "Jack Deer", then Zare and his ilks would
>accuse "Jack Deer" of using "JerryFish" to cheat...
>
>My point is very simple no...? I bet Zare will understand it for
>you, even if you won't understand it for him... :))
>
>> In addition, even though Grandell plays more slowly than most,
>> he also by most accounts plays slowy when playing at live
>> tournaments.
>
>"More slowly than most" and "slowly" don't quite sound the same
>to me. Watch out, I have a bad habit of picking on details... :)
>
>Unless "Grandell" plays "more slowly than most" in live matches
>(as slow as online), you are in a slight trouble here... :((
>
>> His success at live play, where there is no way he could use
>> Snowie, gives players confidence that he is not cheating when
>> he is playing for money on GamesGrid.
>
>"Assuming" that he always plays as "Grandell"...
>
>But what if he plays as "Jack Deer"...??
>
>And what if somebody logs on as "Jack Deer" and has an error rate
>similar to "JerryFish"...? Would you rather "assume" that he is a
>cheater using "JerryFish" or would you assume that he is "Grandell"
>logged on as "Jack Deer"...??
>
>Har har... :)) Normally, I would have said something like "my ass
>is busting from laughing too hard" but I won't say it now since I
>am trying too hard to be polite today... :((
>
>> Hence I just found it absurd that you would think Zare would
>> accuse Grandell of cheating because of his excellent play.
>
>But he can accuse somebody else or me of cheating because of my
>excellent play...? And he can do that just because my "log on"
>name isn't "Grandell"...? And how would he know who is who...??
>
>> I know from your previous posts that you have an exceedingly
>> low opinion of the morals, imagined sexual habits, etc. of
>> backgammon players who play for money.
>
>Their "morals" yes but not so much for their "sexual habits"...
>
>Otherwise sex-starved people like me would be out there "playing
>backgammon for sex"... :)) Har har har... :))) Sorry, my ass just
>busted from laughing too hard... :( Just couldn't help it despite
>my best efforts... :((
>
>The only winner in gambling is the "house"... Anybody too dumb
>to know this may end up as a "gambler"... And anybody too dumb
>to be a gambler can't be a good backgammon player...!!
>
>Please, etch this down in an unused corner of your brain as the
>least bit of wisdom "Murat" has contributed to backgammon...
>
>MK