Try to look in this way (and then burn every thoughts in this way, because they will leave you in muddy waters):

Cheating is a human qualitiy. Cheating has always a goal. But the only goal bots have, is to maximize the equity for the given roll to play. A build in function for cheating must produce a significant pattern, easy to see in an analysis.
When i remember right, but i don't know names, i just heard the story one time, years ago, perhaps someone can tell more, there was a backgammonserver, and someone cracked the dice generator, so he knew the upcoming dicies. He was caught, because someone noticed, that he made from time to time not the best move, but the best one according to the upcoming dices.

No one serious has seen anything in this direction for the bots. The bots are trustable for research. Any cheating would give them a bias, and no one serious would work with them. Can we end now and foerever the cheating feature?

Well, re my op, i am not alleging that there is a cheat factor built in to GNU - i mean, i just can't imagine it.

The purpose for my post was to see if anyone else was experiencing a similarly huge gnu +ex over a large sample like mine.

I was not aware of the "luck rating" feature in the analysis and i am presently running another 300 game sample where i will track the results on a spreadsheet - vs the estimate/steno pad tracking i was doing with my earlier sample.

Again, i really hated my sky id falling op, i am well aware of variance, odds etc, i have been playing card games and board games for decades, proficient and comfy with math;

the thing about my 300 game run was just so bizarre, it seemingly defied likely probability and i was just wondering if there was something to it.

Thanks for input all. Cheers.




Having said that tho - online poker is definitely rigged; especially pokerstars.

LOL!

GNUBg is likely to have an edge against all humans. GNUBg is will have a considerale edge against anyone who think it cheats. What is your average error rate ? How much of an edge did GNUBg have against you over 300 games according to the numbers ? Numbers, CIs ?

Its all very simple really. Just go to excel and create a file dice1.txt or dice1.csv by basically creating a column that has the form on cell A1;

+INT(RAND()*6)+1

then copy this cell and paste on the entire column of A from A2 to A10000 say .

Now you have a column of 10000 rnd dice rolls. Save the file as say dice1.csv (select csv format) and answer yes to all questions.

By the way if you go to cell B4 and enter a spacebar entry from keyboard and then press enter the entire column A is recalculated and you have a new sequence.

Create as many such files you like.

Then go to GNU and in settings , options , dice , select read from file. Then it will prompt you to select the file so go where you saved the dice1.csv and select it.

From now on all the dice is taken from this file and you can actually observe that this is true for verification. In fact you can take a look at the file after 100 rolls and still see the sequence you saved so that it doesnt turn back on creating its own rolls later . Of course you shouldnt be looking at the sequence for other than test purposes because you then are the one cheating by knowing what will come. In any case now you have an external source of dice and no more a possible claim of cheating. Play this way and see if you have the same feelings of frustration that we all get when graded by this "sob" lol program at 1800 or 1750 or 1950 that then has the "audacity" to always claim we are always luckier (if you see the luck estimation) when we win a 17 point match every once in a while (say 40% of the time for me).


It sure feels very tilting at times but i guess its the nature of the streaks and how severe they prove when your pieces are not optimally placed to begin with . Basically what happens i think is that we make errors in placing pieces that improve slightly the probability to get hit and also if we then get unlucky to not enter in ridiculous spots of 3 openings or even 4 openings adn for the computer to start forming blots fast as well for the entire game to collapse. What i am saying is that the computer program plays in such a way as to maximally capitalize good luck for itself and bad for us. This in effect produces the impression that it is luckier because it gets what it wants more often and it feels highly convenient. Of course to be truly objective one has to study what it needs to get and see that it places itself in such a manner for typically the higher probability of good outcomes to emerge after every event of critical importance. It has managed to play in ways that allow it higher probability for good things in the future to happen. Just pay attention to detail and see what i mean.

I recall hearing that, back in the day when it was more difficult to generate long sequences of random numbers, certain publishers of books of sequences of random numbers would "adjust" their sequences as they seemed to follow patterns and, therefore, seemed not to be random when, in fact, they were. The human mind is good at finding patterns, even when they do not exist. Many players of roulette, lotteries and other games of chance will insist that they have observed a pattern that they can exploit.

Keep in mind that if tell GNU to read from a given file, every time you start a new GNU session it is going to start from the top of the file, so you will get the same sequence of rolls unless you tell it to use another file.

You can also have gnu get dice rolls from random.org, which bypass the dice PRNG and downloads a set of numbers form random.org. I am not sure that you can see what it fetched from random.org however, so if you are in the gnubg cheats crowd you might still claim it manipulated the numbers. You could also use random.org yourself to created a set of truly random number in a file like the excel method above, but the PRNG in excel I am sure is good enough for generating dice rolls.

The feeling that electronically generated dice, whether from bots or online servers, are somehow cheating or not correct is extremely common, and virtually always unfounded. (I know of one proven exception). There is something about the human mind that seems to attribute the jokers and antijokers in an electronic format to cheating, when the same rolls or sequences of rolls with real dice are readily accepted as just "bad luck" or "good luck". And if you play live much at all you will see the same sort of things happen with real dice.

Even for those of us that understand something about the behavior of random numbers it is hard to shake the feeling that something is amiss when you get some of these sequences. I know I find myself, particularly playing on line, mumbling things like "unbelievable, "happens every %^&* time" etc. when my opponent does something like rolls the one double that brings them off bar against a 5 point board and hits my blot all at once, or seems to roll the 1 in 18 shot multiple times in a match. But those are emotional responses to the situation coupled with selective memory, and I realize that in the long run it is just the nature of the game, and it happens in live play all the time as well. Somehow, we just don't seem to jump to the "cheating" conclusion in live play like we do with generated dice.

Full tilt is currently allowing a cheater to prosper on the 6 handed .25/.50 PLO game... I've reported it twice... in the last 3 hours, player SEEKPLAYER has managed to CONTINOUSLY hit 4 outer gut shots against set after set, top two pair and much more. I have NEVER seen ridiculous playing like I am seeing right now on Full Tilt! I have demanded my money back or I am NEVER playing on this site again.

Cheating for sure! Nobody ever makes top two or a set in PLO.

Of course GNU cheats! The stats don't stand up to scrutiny for random dice. GNU absolutely does run "hot". When this occurs, there is no outplaying GNU since the dice are rigged. I don't know why there are doubters or "holier than thou" types who question this. They must not be playing GNU. In order to back up "Am I nuts...", and with my own experience, I recruited a statistician, ...a real one, and asked him to watch the play. It was noted that GNU absolutely cheated the odds over and over with remarkable rolls and hits that didn't meet the random test. He took over the site, did quite a bit of study and after a while just simply said that GNU has a built in hot streak that must play out before it "allows" anyone else to win. It then goes into statistically correct mode and back again to "hot".
Being satisfied that neither "Am I nuts..." nor I am paranoid, I will say that GNU is the best tutorial around. If one can get past the cheating, GNU will make you a much better player. Oh, and it also teaches one to control the temper when rolls are way off the random mark.

Gnu is open source its simply not possible that it cheats.
Hmmm, btw it would be quite a difficult thing to programm a "good" cheating Bgsoftware anyway - and why should this bgoftware cheat if its stronger than any human anyway?!

Even if you are a good player, against GNUBG, you are a fish!

In backgammon and poker, the sharks love the fish, because the fish cannot see that they are fish. They think that they are sharks who are just unlucky. And against a computer, they think the computer cheats!

For your consideration:

How to Prove Your Bot Doesn't Cheat
by Bill Robertie

"The root of all superstition is that men observe when a thing hits, but not when it misses."

- Francis Bacon.

I joined this forum for the sole purpose to confirm those who thought GNU cheats. You guys are 100% correct. I've been playing with GNU for a year and now am going to quit to instead play on playok with my 2100+ account. Those who don't think GNU cheats probably haven't played enough against grandmaster level. It's obvious. I've played better players and never get the disadvantage I get against GNU. Even the help system of GNU is aimed at winning the cheating dice... you'll see what I mean if you play GNU on grandmaster.

Also... GNU doesn't always cheat, but it's too obvious once it starts to cheat.