[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Bug-gnubg] Dice Rolls

From: Ian Shaw
Subject: Re: [Bug-gnubg] Dice Rolls
Date: Mon, 19 Aug 2013 10:28:32 +0000

> Hi Max,
> I was not playing against the automated bot.
> I was playing human against human, and I was playing both players. 

I would think that playing equally well for both sides is probably quite 
difficult, especially when you have already formed an opinion about what you 
expect to see. "Confirmation bias" is a well-known problem afflicting many 
areas of research far more important than backgammon.

Have you run a gnubg analysis on your 10 matches, and noted the error rates and 
luck rates for both sides?

Welcome to gnubg backgammon. I hope it gives you much enjoyment and helps you 
improve your game.

The developers would welcome genuine evidence of any bugs in gnubg, so that 
they can correct them. However, they will expect high standards of proof. 
Almost every backgammon bot and online server regularly gets accused of 
manipulating the dice in favour of the opponent, or other forms of cheating, so 
this is déjà vu for them. Many of the best players in the world use the bots to 
help them study, and none of them have noticed anything amiss, so I'm sure you 
can appreciate that the developers will treat the first impressions of an 
unknown person with a degree of scepticism.

To my knowledge, only one bug has ever been found, in a minor online server, 
which was found to be generating too few doubles. When they corrected the 
error, there were even more complaints that the dice were unrealistic! I can't 
remember where I read this, so I can't give you a link to it, I'm afraid.

In the case of gnubg, the code is open source, so anyone can read it, compile 
it, and even change it to suit themselves. There have been a large number of 
developers over the years (the original ones are no longer involved), and they 
are confident that the random number generator is correct. I'm not one of them, 
but I have read much of the code. I can find no evidence that the rng looks at 
the board position or who is on roll. I'm certain that the move generator only 
looks at the board position, not at the upcoming dice sequence.

Ian Shaw

reply via email to

[Prev in Thread] Current Thread [Next in Thread]