Why so many doubles, though?

[vannik]

I've been meaning to post this since I don't know when, but dang, this is super freaking frustrating. Whoever wrote this iniquitous dice system for this game, I would like to inform you that it is exceptionally bad. I'm really starting to hate how this game works here. I've had opponents that "roll" 13 to 15 doubles in a single game. Now, I've been playing backgammon for over 20 years and the probability of rolling that many doubles is way way way lesser that in this online implementation. It's not because I lost half the games I played here that I'm writing this, it's because of how I lose them. It's like 4 out of 5 games the dice rolling is noticeably favoring one of the two players. It's only a game, yes, I know, but it gets annoying at some point.

[Romain672]

Yay another one of those

13 doubles /20 rolls has a probability of happening of 1/603k.
So if that really happen, it's unlucky, and that should not happen it again.

These posts are recurrent: humans are really bad of perceiving randomness.

Feel free to give me any of your past or future games, it's likely the number of doubles of yourself, the opponent, or both players would have a probability of happening of more than 1/10k.
Or at most, you will find one game which is below that shreshold (regularly people share some on the forum. People share too games which have a probability of ~1/30 of occuring because yeah, their perception of randomness is even lower).


Here is my bernouilli table: https://docs.google.com/spreadsheets/d/ ... 1846324187 . If you check the number of rolls and number of doubles you get, you can find the probability of that happening really easily, but I can do it else if you don't arrive to read that table.


I checked your two games before you did that post, I find for both players 8doubles/53 & 6doubles/32, really far from your 13&15.

[RicardoRix]

I've been playing backgammon for over 20 years and the probability of rolling that many doubles is way way way lesser that in this online implementation. It's not because I lost half the games I played here that I'm writing this, it's because of how I lose them. It's like 4 out of 5 games the dice rolling is noticeably favoring one of the two players.

You can see the number of doubles rolled for any given player in their statistics for that game.

Please can you provide some links / players that have an abnormal number of doubles...

[vannik]

I checked your two games before you did that post, I find for both players 8doubles/53 & 6doubles/32, really far from your 13&15.

Please can you provide some links / players that have an abnormal number of doubles...

I was not talking specifically about my last games, even though there were lot's of doubles in them too, I was talking about my experience with the game on here in general. I've played thousands of games IRL and I don't recall anyone I've played with rolling so many doubles as with the e-dice on here.

[RicardoRix]

I was not talking specifically about my last games, even though there were lot's of doubles in them too, I was talking about my experience with the game on here in general. I've played thousands of games IRL and I don't recall anyone I've played with rolling so many doubles as with the e-dice on here.

The statistics are over all games for a specific player and will prove whether there are any anomalies. These will be able prove you right if you can provide the username of the players.
https://boardgamearena.com/playerstat?i ... 02&game=53

[SDH24]

I agree with vannik. BGA dice games do not reflect true randomness. It's not just doubles. It's extremely unlikely rolls happening commonly. And I've been on both sides of it - benefitting from unrealistically lucky dice, and being punished by unrealistically unlucky dice. When I play in real life, occasionally the dice heavily favor one player, which you would expect with true randomness. In BGA, it happens regularly.

If you want an example, take a look at the absurdity of how this recent game ended: https://boardgamearena.com/table?table=307664403

When we both had all our checkers in our home board, this sequence took place: he rolled double 5; I rolled double 4; he rolled double 4; I rolled double 3. I'm pretty sure the odds of doubles being rolled on four straight turns is 1/1,296. But stuff like that happens regularly on BGA.

[Romain672]

I checked and yes, you got 4 doubles in a row which have a probability of 1/1296.
But it's really at a specific moment.

There was 54 rolls on this game.
So, you could have taken 51 start of streaks of 4 to prove your point.
So we are already at a probability of 1/25.9.
I really insist on that: in one game out of 26 of that length (which look like a game above average in number of rolls), you will get 4 doubles in a row.


But we can go even lower,
You could have instead taken your doubles or opponent doubles. Which give 51+24+24 start of streaks possibles, so 99 starting streak. That now give 1/13.6.

I will ignore the fact you could have taken streaks of like 3doubles then one non double then 3 doubles again since your streak was pure, but if you give me others game with that part, I could make the probability even more likely.


I will repeat what I said during my first post, give me probabilities of events under 1/10k. If you can't, that mean your claim can't just be the 'rng is broken' but must be something weirder like 'the rng sometimes give series of doubles and between those series give less doubles' which start to be really weird and nearly non sensical.

Another possibility is for your next 100 games, you note all unlikely games you got, you should not find something sensical to share.


And I suppose you took a really recent game, and not a really weird one, that's why you got those small probabilities, but still, you gave me a 1/13.6 which should have happen about ~56 times since you are on this site on backgammon.

[SDH24]

I just played another game: https://boardgamearena.com/gamereview?table=307664898

This sequence happened at the beginning of the game: I rolled double 2; he rolled double 1; I rolled double 5; he rolled double 5; I rolled double 5; he rolled double 5.

That's 6 straight doubles, including 4 straight double 5s. I believe the odds of six straight doubles is 1/46,656. The believe the odds of 4 straight double 5s is 1/104,976. Those two things happened in the same game.

So in the past two games I played, I had a game with 4 straight doubles (1/1,296) and a game with 6 straight doubles including 4 straight double 5s (1/46,656 & 1/104,976).

Is that sufficient proof to you that BGA dice are not random?

[SDH24]

Sorry, this might be a better link to that game: https://boardgamearena.com/table?table=307664898

[Romain672]

Is that sufficient proof to you that BGA dice are not random?

With all those others posts on the forum of course not. But it's a nice first step for it.


There was 30 rolls that game, so 45 streaks possibles for yourself, the other player or both players.
That give a probability of 1/1037, which is really not bad, especially for a single game after your last post.


I would like to calculate the probability of both events occuring but that will be hard.
To simplify let's count 144 streating sreaks, and lets count the second unlikely occurence.
Here is my doc, I don't suggest you to open it: https://docs.google.com/spreadsheets/d/ ... edit#gid=0
I did 2000 simulations for fun. 11 find something less likely than that.
Which give about 0.5% of occuring: 1/200.
(really though that would be lower, pretty surprised by that 1/200. Guess 4 doubles isn't that unlikely)


And those probabilities are independant: 1/1037 & 1/200, you can't multiply them.
But like I initially said, that's one step, but not much more.

(btw, I often challenge people on this subject and they rarely came back with something new, so you are the one which got your lucky game after it )