Rapport • Board Game Arena

Tous les rapports

Rapports de bug sur Backgammon

#16741: "Implement a correct randomizer for the die rolls"

rejected: Les développeurs estiment que ce n'est pas une bonne idée ou que le ratio coût/bénéfice est trop élevé

Quel est le sujet de ce rapport ?

Backgammon #83256267

Que s'est-il passé ? Merci de choisir ci-dessous

Suggestion : à mon avis, la modification suivante améliorerait grandement l'implémentation du jeu

Que s'est-il passé ? Merci de choisir ci-dessous

Suggestion : à mon avis, la modification suivante améliorerait grandement l'implémentation du jeu

Veuillez vérifier s'il existe déjà un rapport sur le même sujet

Si oui, veuillez VOTER pour ce rapport. Les rapports ayant le plus de votes auront la PRIORITÉ !

#	Status	Votes	Game	Type	Title	Last update

Description détaillée

Historique du rapport de bug

4 May 2020 12:49 • Anaglyph • Cette suggestion n'a pas encore été analysée par les développeurs:

I really like this site and the games, but this one breaks it. Games where dices are involved should be as fair as possible.

4 May 2020 18:52 • Red Nebula • Cette suggestion n'a pas encore été analysée par les développeurs:

The 1 in 6 rolls being a double is true only if you roll the die an infinite number of times. In all other cases you can get any number of doubles. When you play your game you can't expect to get 1 in 6 rolls being a double in any way. You can even get all doubles or no doubles at all, but the random number generator is still a correct random number generator.

The fairness of the random number generator has been questioned many times so far on the forum.
You can find better explanations by searching on the forum. Here some threads I have found:
boardgamearena.com/forum/viewtopic.php?f=3&t=13433
boardgamearena.com/forum/viewtopic.php?f=71&t=14562
boardgamearena.com/forum/viewtopic.php?f=3&t=15221

9 May 2020 11:44 • enki60 • Cette suggestion n'a pas encore été analysée par les développeurs:

I, and many players share the opinion that something is wrong about the dices and their doubles, and i implement in my job rng, as well as I studied math. And it feels as if in the formula they don't fix the dices, making 2-4 and 4-2 are the same, what is wrong, and with that making 2-4 and 4-2 together are the same with 6-6, what is wrong. What you comment about being just true for an infinite series, that's right, but even though that, you are able to check if that is right or wrong with a certain level of confidence, 99%, for example. It is possible that you throw 1000 the dices and it is always 6-6, but statistically you would say that there is wrong. In the dices case, the probability of getting doubles should be 1/6 and not 2/7. I guess you already checked that, the formula. But it is already to many people who share my feeling.

Sorry for the bother, and thanks.

Lluís Piñero Casòliva.

13 May 2020 06:41 • tim_is_awesome • Cette suggestion n'a pas encore été analysée par les développeurs:

i do not feel the dice are properly random. have played several games and seen too many instances of god rolls on both sides. have discussed same with opponents and several feel the same way. i have removed the game from my favorites as a result of this issue.

also a partial suggestion, when rolling doubles, would be nice to select more than one checker to move. perhaps that already exists and i just unaware.

13 May 2020 07:05 • tim_is_awesome • Cette suggestion n'a pas encore été analysée par les développeurs:

just found the random dice generator threads. i have read those now and it is interesting. the code for the generator is pretty simple and straightforward. while my specific view is a small sample size, it is certainly enough for a perception. and if that is being felt by many others at the same time, perhaps this code is not acting as folks think it should be acting.

if you keep getting reports like this perhaps something is in fact going on that is not readily apparent or just with this small section of code.

i do agree that it would be odd that someone would program a game to make it less random (as it currently feels). there is no point to that clearly. but still the perception is out there and prevalent from my limited number of opponents in such a quantity that i took the effort to put in a vote for the dice generator to be investigated. note i am not complaining of losing or winning as a result of this seeming abnormal dice. it is just that it detracts from my personal enjoyment when i do not think the dice are as random and unbiased for either side as they should be given the issues with computers and randomness.

22 Jun 2020 05:53 • Shorty747 • Cette suggestion n'a pas encore été analysée par les développeurs:

I have played a few games now. And It is always in favour of the other player rolling 5 doubles to my 1 or 2 doubles.

25 Jul 2020 21:28 • _Redlum_ • Cette suggestion n'a pas encore été analysée par les développeurs:

@admins: Please look at all the games played and provide the ratio is number of doubles over all rolls. This will definitely tell us if the function is truly random. I've played many games where the ratio is close to 50% and no games ever to balance that. In my last game, my opponent got 9 doubles put of 19 rolls. This happens often on BGA, and NEVER in real life. Please share the raw data for all games so that we can draw our own conclusions. Thanks.

15 Sep 2020 10:29 • swiss_nick • Cette suggestion n'a pas encore été analysée par les développeurs:

Just played a game where there were 7 doubles between the two players. All of them were double 6!!!

16 Sep 2020 17:20 • CaractacusPots • Cette suggestion n'a pas encore été analysée par les développeurs:

Dice are simply not realistic to real life throws simple as. It's not sufficient to just say that over many 1000s of rolls it all evens out to where it should be. The generator is essentially broken if, within a single game, it puts way too many of those doubles into that one game. It simply breaks the game.

I've had games with 18 doubles in them. Others have had games with 20+ doubles in them.

Played an opponent yesterday these were her first 11 throws

#112398793

11 throws - 5 doubles

36,16,14,34,55,66,22,33,64,22,61

A string of 5 doubles in 6 consecutive throws.

This was her next game

17 throws - 7 doubles

35,33,33,46,46,35,25,13,22,36,66,13,25,66,46,22,66

There is a need to present dice rolls in a way that suits the game. Computer dice generations which work from seeds just don't seem to do that. They are ok for huge quantities of rolls as it evens out eventually, but along the way they produce many strings of numbers which are simply unacceptable to the game.

You could theoretically say that with infinite rolls you could easily get 20 consecutive doubles thrown.

But Backgammon is not a game of infinite rolls. On average there are 30-50 rolls in a game. It is therefore simply unacceptable to have 20+ doubles in a game, random or not. It breaks the game makes it pointless.

Listen to what people are saying. Please. The dice are BS. It's not just the doubles either. The number of times I see 3 exact same dice rolls occur one after the other is for me too often. Should only happen 1 in 324 rolls. However unless you provide a practical simple way to download ALL history games to say a CSV file then no-one can realistically check and prove it. Scraping dice rolls from your histories one game at a time and cut/pasting is not an acceptable solution. No-one will ever acquire a sufficient sample size that way.

The dice are just laughable end of. If you provide a way for us to download all the histories to a file I'll happily analyse it.

16 Sep 2020 18:17 • diamant • Cette suggestion n'a pas encore été analysée par les développeurs:

When you use a pseudo-random function, you simply perform a calculation based on non-random data.

Whatever the complexity of the calculation, the result will always be rigorously identical if the non-random data is always the same, a series of calculations using the previous result as non-random data will always provide the same series of numbers from the same initial data.

Personally, I will not trust a pseudo-random function not validated by a mathematician and associated with comments specifying the limitations of use of the function from a mathematical point of view.

I find it quite plausible that a bias in the pseudo-random function could significantly increase the probability of obtaining two identical results in successive jets.
If this is the case, one way to overcome this defect would be to have several completely independent pseudo-random functions (that is, not sharing any data), and to assign a separate function to each of the dice used simultaneously in any game.

28 Oct 2020 16:40 • King of Games • Cette suggestion n'a pas encore été analysée par les développeurs:

Without looking at the code, I can assure you it would be more difficult to write a bad RNG than a good one. You are all likely just experiencing survivorship bias.

Survivorship bias happens when people change their behavior based on an event. In this case, the 99% of players that saw an "average" number of doubles, continued playing and didn't notice any problems. The 1% of players who got a large number of doubles became convinced that the system is broken.

But in reality, they are just the 1% of players who got lucky (or unlucky) and rolled more doubles than everyone else.

25 Dec 2020 00:01 • TheAncient001 • Cette suggestion n'a pas encore été analysée par les développeurs:

I just rolled double-sixes three times in a row- table ID #133739143 . Doubles come up FAR more frequently than they should statistically. It seems very likely that the random number generator being used in this game is flawed.

18 Mar 2021 23:25 • crispy116 • Cette suggestion n'a pas encore été analysée par les développeurs:

Dice still a joke sometimes in this game: 83256267 opponent rolled 7 doubles in 11 dice - really......the odds of this happening are miniscule; but in here happens all the time.

19 Mar 2021 08:05 • bidulpik • Les développeurs estiment que ce n'est pas une bonne idée ou que le ratio coût/bénéfice est trop élevé:

Statistic for the amount of thousands games played are corrects.
The php random function used is the best function available in this case, and we have no advantage to modify the randomizer, only trouble 😢

If you calculate the probability to have 4 continuous double it is very low. But the probabilty for you to roll another double after a double is still 1 in 6 chance.

It is rare, but 4 continuous double six may happen.

Ajouter à ce rapport

Veuillez ajouter ici toute information qui semble pertinente pour reproduire ce bug ou comprendre votre suggestion :

Un autre n° de table / n° de coup
Appuyer sur F5 a-t-il résolu le problème ?
Le problème est-il apparu plusieurs fois ? À chaque fois ? Aléatoirement ?
Si vous disposez d'une capture d'écran du bug (bonne pratique), vous pouvez utiliser le service d'hébergement de votre choix (snipboard.io par exemple) pour la télécharger et en copier/coller le lien ici.

Annuler Mettre à jour le rapport

#16741: "Implement a correct randomizer for the die rolls"

Quel est le sujet de ce rapport ?

Que s'est-il passé ? Merci de choisir ci-dessous

Que s'est-il passé ? Merci de choisir ci-dessous

Veuillez vérifier s'il existe déjà un rapport sur le même sujet

Description détaillée

Historique du rapport de bug

Ajouter à ce rapport

Changer le statut du rapport en :

Aide

Contribuer

Naviguer

Nous suivre