maybe improvement for randomizer necessary (better seed?)

[magnum108]

I played a lot of games over the past few days and I can tell you for sure, the dice roll on this website is freaking disgustingly broken. It's so, so, so, irritatingly moronic that it is definitely due to human interference. There is no way a player can get 6-6 doubles three times in a row if the algorithm is not f....ed in the a.. You definitely dont need a science degree to notice how amazingly idiotic dice rolls are in this game.

Sir, you are very funny

[veggivet]

Au contraire! Don't you believe all the admins who parrot the same line over and over again that the dice are totally random?

[Inthy]

I played a lot of games over the past few days and I can tell you for sure, the dice roll on this website is freaking disgustingly broken. It's so, so, so, irritatingly moronic that it is definitely due to human interference. There is no way a player can get 6-6 doubles three times in a row if the algorithm is not f....ed in the a.. You definitely dont need a science degree to notice how amazingly idiotic dice rolls are in this game.

Actually, there is a way. The probabiltiy of this event is slightly less than 0.5%. Thus, for sure you observe the row of 6-6 once per several games.

[veggivet]

Actually, the odds are .4629%. If the dice were truly random and fair, three consecutive doubles (of any number) should occur 46 times in 1000 games. We all know this sequence occurs much more frequently than that, especially during the endgame.

[dschingis27]

Actually, the odds are .4629%. If the dice were truly random and fair, three consecutive doubles (of any number) should occur 46 times in 1000 games. We all know this sequence occurs much more frequently than that, especially during the endgame.

The thing is that such a line of three consecutive doubles should occur in 0.4629% of all sets of 3 consecutive rolls, NOT only in 0.4629% of all games.

Let's say the average backgammon game has 24 rolls for one of the players, it means it consists of 22 sequences of 3 consecutive rolls but these are not independent sequences. You could say for simplicity that a game with 24 rolls consists of 8 independent sequences of 3 consecutive rolls.
So the true probability of 3 consecutive doubles in one game is somewhere between 0.4629%*8=3.7% and 0.4629%*22=10.2%. It seems complicated to me to derive the precise solution but it must be between these 2 values. So it happens roughly every 10th to 20th game for one player.

The probability for 3 consecutive 6-6 doubles is 0.00214%. So the true probability of 3 consecutive 6-6 doubles in one game is somewhere between 0.00214%*8=0.017% and 0.00214%*22=0.047%, so roughly every 2000th to 6000th game meaning it must happen here and there.

[veggivet]

Thanks for proving my point. You are correct in the games vs rolls. My mistake. I think anybody who plays backgammon (or backwardsgammon, as I call it) on this site knows that these roll sequences occur much more frequently than that!

[Jellby]

The average rate of double rolls seems to be about 16.56%. Are you saying the dice are rigged such that double rolls occur in streaks of the same number, even though the overall rate is "correct"?

[Silene]

I think anybody who plays backgammon [...] on this site knows [...]

Then you think wrong. The truth instead is "You and some others [...] on this site BELIEVE [...]".

Your perception is fooling you. Our brain makes us notice intensely, when our suspicion is met while quickly forgetting about so many cases, where it wasn't. This psycological effect is called confirmation bias (https://en.wikipedia.org/wiki/Confirmation_bias).

[veggivet]

You don't need to inform me of confirmation bias. I am well aware. I'm also well aware that the frequency of certain rolls is statistically significantly higher than expected by chance. For example, rolling double sixes when you have a man on the bar and the only point covered by your opponent is the 6 point. Happens WAY more frequently than dictated by fair dice.

[Silene]

You don't need to inform me of confirmation bias. I am well aware. I'm also well aware that the frequency of certain rolls is statistically significantly higher than expected by chance. For example, rolling double sixes when you have a man on the bar and the only point covered by your opponent is the 6 point. Happens WAY more frequently than dictated by fair dice.

First you claimed (in another thread) that the number of doubles is too high in general. That was disproven by pointing out that overall stats are tracked by the game stats which shows 1/6 of overall doubles. So obviously you have already fallen for perception bias before.
So now you say it is the number of doubles (double 6s) is too high in just one specific situation, which of course is harder to disprove because we can't just look at an automatically tracked stat...
However: what data do you have to back up your claim this time? (and with data I mean a large enough data-set. not some cherry-picked examples)

The implementation of BG-rolls was checked by a dev here:
https://boardgamearena.com/forum/viewto ... 0&start=14

So if the shown code is what is actually in place for BG, there would be no way for the random function to know what situation the game is in. I'm not a dev, so I cannot check myself - but I trust Ricardo to not lying there

So are you claiming that the code that can be looked up by the devs is actually not really running at this site, but instead a very complicated code, that's tracking the game situation, is secretly running instead? (And why would that even be?)
Or is it even more complicated than just tracking the game situation but instead is in addition tracking WHO the players are and whether they should be favored or disfavored?