3DFibs fool

[Thentar]

Strange things happen.

Does the computer in offline play seem to get lucky rolls?  Hell yea.
I win about 1/3rd of the time agianst the computer but some strage things I notice is

1) I often get a 2:1 roll for the first roll very often but the computer I've only seen get it about 2 times

2) The computer seems to be able to get its single stacks to double normally on the next roll.  I'm lucky to do that 1/3rd of the time.

3) The computer hardly ever gets stuck with a piece in the middle more then 1 roll even if I have all but 1 spt open, ie it gets the number it needs far more then the 1/3rd chance it should.

4) The computer gets great streaks at the end when the match is close and my rolls seem to goto the toilet.  I just played a game and rolled 2:1 five times in a row....5!!!!

5) Last and most desturbing is I have played quite a few games where my roll didn't let me move when I get like a 4:2 on my first roll.  When I click on the dice it rerolls to something like 5:1, 2:1 or 6:2 most of the time.  I'd pass this off as a graphics glitch but it also happens when I'm second on my first roll....kind of like the program rolled the dice, displayed the dice then realise it was an ok first roll and randomly makes me reroll my 1st roll.

I do good agianst human apponents.  I realise the computer is "World Class" but its streaks seem a little dodgy.   Not that it cheats every roll but more like if it got a crappy roll it rolls agian.



#5 just happened to me agian....here is the text
Starting a new game with GnuBG.
You rolled 5 , GnuBG rolled 3
It's your turn to move.
GnuBG moves 13-7 8-7


At this point I see my roll at 5:3  Why does it say GNUGB moves?  It has not moved...I'm first....I've got a good roll!  So I click my dice and it rerolls

Starting a new game with GnuBG.
You rolled 5 , GnuBG rolled 3
It's your turn to move.
GnuBG moves 13-7 8-7
You roll 2 and 1.


WOOOO what a surprise!!!!  2:1

Now tell me there isn't something dodgy going on there!

[diane]

Are you talking about using GNU to play - against it - or the feature in 3D?

If you use GNU - you can roll your own dice and manually enter the rolls.  If you do that - you might get a better feel for just how often unpleasant rolls come up with real precision dice, rolled by your own hand - and no cheating mind...

Course it is a bit long winded....

[Thentar]

I'm using 3dFiBs, Not to farmiliar with the program.  Press F3 to start ... pick the number of point and go.  I haven't found any option where you can roll the dice manually

[diane]

Ahh, no, in 3d fibs you have to have the computer rolls.  I still suggest trying GNU - google it and get it for free, the  try the manual rolls, you will be surprised - I am sure    

[Thentar]

Thanks I'll give that a go.

[Tomawaky]

Just analyse your matches and you will see that Dices are not the problem    

[bowe310s]

3dfibs is definitly broken - I used to beat it 80% time - now I'm lucky to win 1 in 10.
Just used Jellyfish for the first time and beat it 5 out of 5 so its not my playing.

Like a previous poster I also get the bug where your first roll won't work, then you roll some new dice.

Another one is interesting bug that works in your favour - try messing with lots of doubling - you get an extra turn!
eg, the computer doubles and you double and double again and then throw 5-2 - you get to move 5 then 2 then 2 again! (always the lower dice value!).

And as another poster said - if you need a 2 to get out, you will probably get 6-6 three times in a row, then 4-5
6-1 and every conbination apart from a 2. As soon as the computer locks the 2, you get 2s!

If its a simple run off, the computer will get consistantly high dice while you get 2-1 three times in a row.


I suspect that the programme works out what dice would be favourable for the next move and then weights the probability of it getting those numbers depending on your skill level - so the best roll you could get might need to be 4-3, it lowers the chances of you getting it by adding extra useless numbers like 6-6 to the 'pot' it
draws from.

[ah_clem]

While there may be some obscure bugs in the 3DFibs interface when using it for human vs computer play, your dice whining is off base.

It doesn't cheat with the dice.  That's a common perception but there's zero evidence that it does, and quite a lot of evidence (e.g. the gnubg source code) that it doesn't.

This is a common topic, as every month or so a newbie shows up to claim that "gnu/ snowie/jellyfish cheats with the dice".  Try Googling the topic before furthering this nonsense.

[rebcalale]

Hi,

It is quite normal to suspect the bots since they win so many games. If you have saved your matches, why not convert them to jf.mat format and evauate them for errors using gnubg.  You may be surprised at how many obvious moves you have missed. Gnubg also has a pretty good luck factor analysis so you can see if you really were unlucky but played good, or if you played bad and were just lucky. It is a good way to improve your game. It is used by almost all high level players and it is an accepted software in the backgammon community.

maareyes (tucsonAZ on fib

Do the same u might be surprised what a jackass u r.  I have witnessed bots play and they win for one reason and one reason only, THEY GET ROLLS.  PLEASE ENOUGH with the bots r so good BULLSHIT.  FIBS dice r so obviously biased it is sad and if u can't see this u r more of a moron than I already suspect. 

[ah_clem]

...every month or so a newbie shows up to claim that "gnu/ snowie/jellyfish cheats with the dice". 

As I was saying...

[rebcalale]

While there may be some obscure bugs in the 3DFibs interface when using it for human vs computer play, your dice whining is off base.

It doesn't cheat with the dice.  That's a common perception but there's zero evidence that it does, and quite a lot of evidence (e.g. the gnubg source code) that it doesn't.

This is a common topic, as every month or so a newbie shows up to claim that "gnu/ snowie/jellyfish cheats with the dice".  Try Googling the topic before furthering this nonsense.

Cheat is a strong work but u r full of yourself.   There is NO , let me reapeat that, NO evidence that bots do not get favorable dice.  U can see they "anticipate" (expect certain rolls) rolls now maybe this is "good" software and it might not be classified as cheating but it is close.  If u do not complain about fibs dice u r an idiot.  These r so obviously biased it is sad.  A simple way to settle this is to keep stats.  Surprise, surprise, no one seems to get this done. Guess what, get the facts and u might realize the nosense is accepting that fibs dice r legit without proof.

[diane]

A simple way to settle this is to keep stats.  Surprise, surprise, no one seems to get this done.

Oh goody - you have obviously done this - I can't wait for you to show us and explain how the bias works 

[Yvon]

The truth is, it is extremely difficult to make statistics on backgammon. The reason is because there are at least 5 different parameters each of which has to be examined separately. Åven if you do statistics on all of them you have to manually decide which was the one that decided the game. Was it the too many doubles, was it the sequences of best rolls for one worst for the other, was it the out of bar luck, was it the never missing a hit? Needless to say that there are cases all your statistics won't worth a sh**  simply because the game was a backgame, or a closed tactical game.

I have personally programmed a macro on Excel that calculates the out of bar luck as a percentage of available possibilities and although it is mathematically 100% correct it cannot be used to evaluate more than say 40%-50% of the games. Results are really amazing, it can even prove you that you lost a game while you were actually very lucky!

If you would like to have one of your games evaluated according to this parameter attach it zipped here in 3Dfibs or mat format and I will let you know the results.