Loads of doubles

[dschingis27]

I took the effort and analysed your last 30 games. And it shows that your view on your results is totally biased. In 17/30 games, your opponent got 6-6. And only in 3 of the 30 games did your opponent start with a double.

Also, having a 6-6 in one roll has a probability of 1/36 and the average game of Backgammon on BGA has 25 rolls per player. So it is totally normal that your opponents kind of "usually" get a 6-6 during a game.

[Ze Monstah]

[Ze Monstah]

[Ze Monstah]

[dschingis27]

I think you should have looked at the last 30 games (or 10, for a smaller detailed view) when I wrote that comment.

I exactly looked at the last 30 of your games from the time you made this comment:

[...] (and in general, any of my opponents get at least one 6-6 in a game). [...]

You were claiming that strange things happen to you "in general" and "usually" and now you are telling me that I should look at a specific sequence of your games??

If you want that people do not dismiss your arguments, you shuold be more precise and consistent about what you argue..

Also you don't read other posts thoroughly and dismiss well-intentioned suggestions as insults.

At least we agree on that we should stop "discussing" with each other. Bye.

[Ze Monstah]

[Benoit314]

We hear so many different complaints about dices.
Some say it gives better dices to new players, better dices to paying players, and now about doubles.
To me better dices can be so much different from a game to another, these don't make sense.

When the code calls bga_rand, this function doesn't know it is called how many times and what are the relationship between these calls. It doesn't even know it will be called again. It doesn't have a player id as an input so it doesn't know who requested a random value.
The simplest explanation is that it just returns a random value... A cryptographic level random value, which is strong.
If your claims would be true, it has much more implications than just BGA. It would means that random_int isn't safe for security purposes.

I'm sure there are researchs on this also, maybe go read what researchers think about that random_int PHP function.

[Romain672]

Ok I get where I mess up. I mix the sum of number of doubles for both player, with the number of turns in average for one player. Which just multiply the number of doubles by 2, which is enormous.

163852413: 2y 4r 27
163842398: 3y 4r 24
163760484: 5y 2r 25.5
163750180: 5y 5r 31.5
163749910: 2y 3r 22
163733497: 6y 4r 16.5
163731807: 1y 2r 11
163737793: 3y 3r 18.5
163730921: 3y 2r 17
163729488: 5y 6r 19

So we got those numbers: 6>54, 7>48, 7>51, 10>63, 5>44, 10>33, 3>22, 6>37, 5>34, 11>38.
Which gave these probabilities to do this number of doubles or less: 18%, 44%, 37%, 52%, 24%, 98.5%, 58%, 49%, 58%, 49%, 98.3%

So the probabilities I got should be normally (nearly) completely random.
And yeah it's the case here. We got two number of 98.x, which is pretty high, but which still happen one game out of 60. This isn't weird at all.

@Ze Monstah:
When I said you (or humans) didn't got a good feeling about probabilities... (and me too btw, was too focused on my calculations which I assumed were right)
I'm trying to be in your camp, but if it's the most weird thing you can find out of 170 games, everything is normal (would said the same even for 30 games here). (I will personnally start to put less value on your posts from now on, since I got this one, which is completely into the camp of you who didn't got a good feeling of probabilities)

I would be interested to know how these probablities were calculated. I feel pretty confident that these are incorrect.

Yup they were wrong. My table was right but where I looked at were not. Just a bernouilli table which I have done more than 10 times in excel (since it's easy and fast to do).

Ty all.

[Ze Monstah]

[Jest Phulin]

When i made the "check my 10 games to see what he is talking about"... i meant those last 10 games for instance.

Please specify the series of 10 games you are referring to by table numbers, so we don't have to cross-reference play times and forum post times.