FIBS Board backgammon forum

HI folks,

Im new to this board but have been on FIBs for some time.  I have a dice question.  

I have noticed, to a very stangly high rate, that FIBs dice returns doubles immeditially after your opponent accepts the doubling cube from you.  Has anyone noticed this trend?  Is it possible that the dice can be effected by actions in the game?  I would think a random number generator should not depend upon any game factors in order to obtain a result.  Does anyone know of a good article on how FIBs dice work?

Thanks for any information you can provide

http://www.gammon.com/fibs_dice/

QuoteHI folks,

Im new to this board but have been on FIBs for some time.  I have a dice question.  

I have noticed, to a very stangly high rate, that FIBs dice returns doubles immeditially after your opponent accepts the doubling cube from you.  Has anyone noticed this trend?  Is it possible that the dice can be effected by actions in the game?  I would think a random number generator should not depend upon any game factors in order to obtain a result.  Does anyone know of a good article on how FIBs dice work?

Thanks for any information you can provide

I have heard some people mention this phenomena. If I accept a cube and my opponent immediately slams me with a double roll .. yea I will remember it. However, very seldom do I remember rolling doubles after I cube though. Just like I dont remember ever rolling 66 of the bar to save a gammon or hitting an indirect shot.
Why should I remember them   that is what is SUPPOSED TO HAPPEN!.

Seriously though.. I think it is just the nature of the game that we remember or perceive some things more than others. Also the fact the dice are generated behind the scenes with the online servers and programs adds to the illusion.

I dont know of any articles that describe fibs random number generation but I would guess that if you diligently documented what happens after each time the cube was turned (100 times), you might discover that nothing is out of the ordinary.



maareyes (tucsonAZ on fibs)

Quotehttp://www.gammon.com/fibs_dice/

Thanks for the link.  i am going to use the percentages listed to see if what I feel I am experiencing occurs at a greater percentage then the norm.  The problem is that I can play enough games for my test to be statistically valid but I will keep track for a while just for laughs.

What I dont see referenced in your link is an answer to my question of whether or not the program that generates the random numbers uses factors that are dependent upon conditions in the game or conditions based upon factors related to the players ie number of letters in name or combined rating or something like that.  I'm sure it all too scientific for me to understand but I am curious.    I would assume that using factors such as those I mentioned would not be the case but I would love to read something on how it works.

Finally, after my post it further dawned on me that this strange occurance frequently returns either double 5s or double 6s.  I must be nuts!

matt

I was in on a shout conversation (some time ago) involving the subject of random dice or not, with Patti (who ought to know).  The ending consensus was that Fibs dice are randomly fixed, meaning that they are fixed only in that they repeat eventually, but not that any other factors play in how the dice roll.

There are those, who will swear the cube correctness influences the dice outcome.  The most prominent of those is vegasvic.  Ask him what he thinks about the dice sometime.  

Food for more thought, of course, can be found in the psychological study article at:  http://www.alef.co.uk/fibs/archive/dysfunction.html  If you haven't read this yet, it is a MUST.   It will bend your mind!  

socksey




"If it weren't for my lawyer, I'd still be in prison. It went a lot faster with two people digging." - Joe Martin

Quote[...] an answer to my question of whether or not the program that generates the random numbers uses factors that are dependent upon conditions in the game or conditions based upon factors related to the players ie number of letters in name or combined rating or something like that.  I'm sure it all too scientific for me to understand but I am curious.    I would assume that using factors such as those I mentioned would not be the case but I would love to read something on how it works.

matt

There are numerous Pseudorandom number generator algorithms, of which the Mersenne Twister is the best known to mankind (maybe only outside the NSA, who knows). It is used by e.g. GnuBG to generate its dice by default.

Fibs is closed source and written before the Twister was perceived in 1997, thus I presume it uses another algorithm. The most widespread ones are described in Donald Knuth's fundamental The Art of Computer Programming, Volume 2: Seminumerical Algorithms. For a good overview see e.g. here.. My bet is fibs uses one of the Knuth algorithms, most likely the one described on page 106 (Borosh, Niederreiter random number generator). It's amazingly simple yet effective: x_{n+1} = (a x_n) mod m with a = 1812433253 and m = 2^32 and x_{1} being the seed (see below). Note that 1812433253 is a prime number.

The Pseudo-Random generators I'm aware of take exactly one parameter, and that only exactly once. This is when they start generating, thus that parameter is called a seed. Given a seed, the generator begins producing numbers without any further input. The same seed will always yield the same number sequence. A sequence will repeat itself after about 10^6000 numbers for e.g. the Mersenne Twister, breaking randomness then. However, the universe is only about 4.4*10^17 seconds old (14 billion years, but biggles_two may object this). Thus I don't expect to see such problems on fibs in my lifetime.

Good seeds are derived from the current time, the free bytes on disk, the number of network packets received and other random data available to your computer. Note that seeds and the sequences they produce are uncorrelated, i.e. the smallest change in the seed yields a totally different pseudo random sequence. And to make that clear: even a bad seed will not spoil the randomness of the generators output.

There are also random devices in hardware, which use some physical source to produce real random. Most popular is atomic decay of radioactive isotopes and, of course, dice. Such boxes cost several hundred bucks (for the atomic decay) and are normally only used by extremely heavy duty applications, such as military strength encryption. Dice are more widely spread and less expensive, albeit  much slower.

Now that I blinded you with science, here comes the bottom line. A (pseudo) random generator does not take any parameter from any fibs related data (e.g. names, the cube, previous moves).  Those are far from random and a statisticans nightmare.

As the same generator serves all dice to all players, even more random is added. This is because while you think of your next move, a lot of numbers from the already random sequence are generated and used for games run by other players. The exact time interval between two of your consecutive moves (in milliseconds) is pretty random, too.

Hope that is good^h^h^h^hrandom enough  

Seeds I understand!  Beyond that, I am lost.      

socksey



"No matter how rich you become, how famous or powerful, when you die, the size of your funeral will still pretty much depend on the weather."  - Michael Pritchard