...

[ReelMcKoy]

I liked the first post in this thread and enjoyed reading the comments that followed.

One thing I've noticed in playing for a long time now is that the best players always do have the best luck. They really do understand the randomness of the game, but also the probabilities involved. So since any die can only have a possibility of 6 outcomes they use that knowledge to their advantage.

Long ago I used to play against an AI program called snowie. I almost always lost. Even in one game where I took all my checkers off (but one) while the opponent had his complete board left. My very last blot was hit and snowie proceeded to hit and block me in so many ways that in the end removed every one of its checkers while I was still on the board with my one. I couldn't believe that I lost and was sure that it had cheated. Now I get it, and in fact have had similar experiences right here at BGA.

I can't tell you how many times I have been put on the bar with only the 6 point blocked when I get 6,6. It happens. (I also can't tell you how many games I've played... probably a lot) And I tend to remember those seemingly impossible rolls very well.

Anyway, I will say, though it often feels tilted or unfair, I do believe the game designers made every attempt to be truly random. The challenge in this game is to see what best outcome I can make with the random (and unexpected) rolls I get.

[nmego]

The one in the Developers forum is especially interesting, as it is a developer stating that something looks wrong. They are then presented with independent ways of verifying the randomness, and do not come back and say those tests failed.

That was a fun read, thanks for sharing.

[Daavids]

I have noticed the same. But don't just look at 6-6.

Hey, great mathematicians! How big is probability to get 3-3 three times in a row? My opponent dit it today. He still lost.

[Jest Phulin]

Hey, great mathematicians! How big is probability to get 3-3 three times in a row? My opponent dit it today. He still lost.

Well, not a great mathematician, but with a good interest in probability

The probability of getting any particular number on one die is 1:6. That makes the probability of any throw (1:6)*(1:6), or 1:36.* Therefore, the probability of getting 3 in a row is 1:36^3, or 1:46,656

Sigh. I thought I had commented on how many games a day were being played with numbers available to everybody, and someone with better access said exactly how many games had been played, so I could give you an idea of how many times per day something like that happened. Unfortunately, I can't find it now.

*Note: This is without the possibility of 2 throws looking the same. IE, I have a 1:36 chance of getting a 3 on die 1 and a 5 on die 2. But I also have a 1:36 chance of getting a 5 on die 1 and a 3 on die 2. Either way will look like a 3-5 to me, so the probability of any non-pair roll is actually 2:36, or 1:18.

[Technogressist]

I noticed some strange pattern by myself to: I sometime deliberately leave a lone checker in my last three lines, directly after catching one of the opponent. The odds to made a counter-catch are slightly above 1/3 (precisely 12/36 for line 24, 13/36 for line 23 and 14/36 for line 22): Well, when my opponent has an ELO higher than mine by 50 or more he always rolls a right combination for a counter-catch, otherwise he doesn't.

I have to play tons of games in order to check a somewhat relatable pattern, but of course I don't want to waste my time in such an effort and quite frankly I was tired about this "randomness". Is the PRNG used by BGA open-sourced?

[nmego]

. Is the PRNG used by BGA open-sourced?

Quoted from the FAQ

How do you manage random events (dice throws, card shuffling, ...)?
For dice and equivalent 'pick a number' actions, we use the PHP function 'random_int'. This function has a cryptographically secure randomness, which means that it may be even better than real dice

For card shuffling, we use the PHP method 'shuffle', based on the 'Mersenne Twister' algorithm used by most online Poker websites.

Please note that there are thousands of games played each day on BGA, so a rare event whose probability is really low may still happen dozens of times per day on BGA.

Finally, for those asking themselves "But is that random number generator truly random?", please check this page from random.org which is a great resource to understand the complexity of this question.

[dsyring]

I will support the claim of excessive double sixes immediately after being taken out with a first blot in the game. It happens to me in many games. I don’t have stats for it, but really think it happens in many more games than seems realistically random.

[Jest Phulin]

I don’t have stats for it,

Any discussion about randomness based on "feels" rather than stats has been met with "please provide statistics to back up your claim." All games on BGA are replayable, so the stats can be obtained.

So, uh, "please provide statistics to back up your claim."

[Cullich]

I'm not on any side in this argument, but I do agree that if you want to deal with randomness, you have to deal with statistics, and you have to deal with proof. Statistics doesn't deal in "feelings." I just can't figure out how someone who is tossing statistics into this conversation is somehow the bad guy. (He doesn't deserve insults just because someone doesn't understand statistics.)

And yes, this conversation has happened before. It was probably before we had game-specific forums. I can remember when Backgammon first came online and there were tons of complaints about it (and no, they weren't all about this issue), and I'm pretty sure that was before this forum existed. Therefore, those threads are either in the announcement post, in general discussion or in suggestions, and some are probably in the bugs. If anyone wants to find them, try looking at one of those.

There is definitely a such thing as confirmation bias, i.e. you're seeing what you want to see and noticing what proves your point. I've seen many, many complaints about this type of issue, on various sites, for various games. If you want to prove your point, then do the work required to prove it. Don't expect someone else to take the time to do it for you! And getting a few people to agree with you isn't proving your point. It's all part of the confirmation bias - that someone else believes what you do whether it's true or not.

I used to believe that my opponents were getting more doubles than I was. Then I started tracking it, for real. And discovered that really wasn't true, but something I was just noticing when my opponent got more. (Some day I'll track what those doubles actually are, since double 1's is worth less than an average non-double roll, but I don't play it much anymore.)

[Cullich]

All threads I could find were in Discussions, in date order:

https://boardgamearena.com/forum/viewto ... mon#p35242
(love the last comment, based on ONE game - there's some good scientific proof!)

https://boardgamearena.com/forum/viewto ... mon#p28692
(not much worth it cause it just refers to another thread)

https://boardgamearena.com/forum/viewto ... mon#p56431
(statistics but those who don't like them in this thread won't like them in that thread, but it's relevant)

https://boardgamearena.com/forum/viewto ... mon#p58573
(about can't stop but same concept)

There's also a few posts in the bugs.