Dice Probability

[PhilipK_]

With 2 d6 die, the probability of rolling the same number on both die is 1:6. the probability of rolling doubles 24 times with 50 rolls is 0.0000224%. 2 in 10MM. Yet this occurs on 2 different days. 24 of 50 on one day, and again 24 of 50 on a different day (same week). "random" number generator my *. I should let BGA play the lottery for me

*Moderator edit: please refrain from swearing on BGA, this is a family-friendly platform.

[Silene]

Yes, unlikely things happen all the time. It's because something has to happen and there will always be some unlikely aspect about it.

Defining the relevant parameters only after it happened makes your observation worthless.

[Gulchen]

Which tables were these?

[gmrsmrt]

Report this as a possible bug (including table numbers). I am pretty sure that random number generator is ok, but it's weird enough to have a look into the code and see wether the random numbers are properly used. I am sure the developer (if he is still around) will have a careful look into it.

By the way the probability is much lower than stated: 2.25 10**(-7) to get exactly 24 double and 2.82 10**(-7) to get at least 24.
Already one occurrence is suspicious, but two close ones point definitely to a bug

[Romain672]

A better advice would be to wait for a third instance of it @PhilipKDick.
Now that you have defined what's the problem is, it will really be 1/10millions.

[Cile]

Already one occurrence is suspicious, but two close ones point definitely to a bug

I agree that 24 out of 50 is quite the outlier, but my guess is confirmation bias over bug. Every couple of months we get one of these threads from someone who doesn't understand randomness or the basics of code.

It's really hard to create a bug in "RND(1,6); RND(1,6);"
It's much easier to see doubles 16 times (twice what one would expect, happens once in every 200 games, and there's currently over 6000 games of Castles of Burgundy being played right now) and saying it must have been 24 times. The fact that it was exactly 24 in both games makes me think OP didn't actually count the number of doubles, but I'm willing to be proven wrong if they want to link the tables.

[gmrsmrt]

I am assuming good faith and accurate information.
Of course any claim should be supported with the reference to the tables where it happened.
I agree that the problem cannot be in the random number generators, but if you by chance generate a random and unwillingly overwrite its memory location (just an example) the random itself is fine but you still have a problem.

[euklid314]

With 2 d6 die, the probability of rolling the same number on both die is 1:6. the probability of rolling doubles 24 times with 50 rolls is 0.0000224%. 2 in 10MM. Yet this occurs on 2 different days. 24 of 50 on one day, and again 24 of 50 on a different day (same week). "random" number generator my *. I should let BGA play the lottery for me

*Moderator edit: please refrain from swearing on BGA, this is a family-friendly platform.

If a player reports a 8*10^(-14) event (and the maths is perfectly correct) one should not assume good faith and accurate information but some misunderstanding on the players side.

This player has 16.43% of doubles in his 1287 games of "Castles of Burgundy", i.e., the correct amount that is to be expected.

He claims 24 out of 50 rolls were doubles on a specific day but he does not provide the day. At the moment I think he misinterpreted some statistic that BGA provides and he had 24% of doubles out of 50 rolls - which would be a 11.7% event and not a 0.0000224% event. Having this event twice in a row is a 1.4% event and not a 8*10^(-12)% event.

Of course I am speculating, but having 24 doubles or 24% of doubles would explain everything without believing in miracles or conspiracies.

[Nekojin]

Looking back at a slice of events that happened and calculating the odds of that happening is always wrong. Probability is for determining the chances of future events. Looking backwards at an event that has already happened is practically guaranteed to be an anomaly, because it involves cherry-picking things that the speaker thinks to be unusual.

Look at it this way. If I win the California State Lottery, the Super Lotto Plus, I've just achieved a one in 42 million chance. But it's already happened - because it happened in the past, it's now a 1 in 1 chance, the probability has collapsed on a result. It happened. If I look at it again, it still happened. And talking about how unlikely it was inherently ignores the millions of other tickets that were purchased.

Stop cherry-picking past events. If you believe that the BGA random number generator is broken, the way to prove or disprove that is to start recording every single roll that is made for the next month of every game you play. Not one specific game. Not just the ones you think are unusual. ALL of them. Thousands (or millions) of die rolls across hundreds of games. And then post the results.

But that's hard work. Hardly anyone wants to do that.

[euklid314]

Looking back at a slice of events that happened and calculating the odds of that happening is always wrong. Probability is for determining the chances of future events. Looking backwards at an event that has already happened is practically guaranteed to be an anomaly, because it involves cherry-picking things that the speaker thinks to be unusual.

Look at it this way. If I win the California State Lottery, the Super Lotto Plus, I've just achieved a one in 42 million chance. But it's already happened - because it happened in the past, it's now a 1 in 1 chance, the probability has collapsed on a result. It happened. If I look at it again, it still happened. And talking about how unlikely it was inherently ignores the millions of other tickets that were purchased.

Stop cherry-picking past events. If you believe that the BGA random number generator is broken, the way to prove or disprove that is to start recording every single roll that is made for the next month of every game you play. Not one specific game. Not just the ones you think are unusual. ALL of them. Thousands (or millions) of die rolls across hundreds of games. And then post the results.

But that's hard work. Hardly anyone wants to do that.

Whereas everything you write is absolute correct it is not a sufficient reply to the OP who thought that he has detected a 1:100 000 000 000 000 event. Such an event should happen only in millions of years on BGA despite the many players and the many random events.

In such a case it is feasable also to look back into the past! If a nationwide lottery is won by the same person three times in a row the reply "everything is possible" is true but perhaps not sufficient.

But the OP should have asked in the forum what he missed/what he misunderstood - instead he attacked the RNG. This shows bad judgement on his part...

I gave one explanation where the OP could have gone wrong - but the OP gave not enough information to definitely solve the riddle...