Terrible algorithms

Forum rules
Please DO NOT POST BUGS on this forum. Please report (and vote) bugs on : https://boardgamearena.com/#!bugs
User avatar
BackGa
Posts: 32
Joined: 14 October 2021, 00:57

Re: Terrible algorithms

Post by BackGa »

Romain672 wrote: 03 June 2023, 17:27
BackGa wrote: 03 June 2023, 17:09 In this game my opponent wins the game in 21 rolls. From the 7th roll, he does 7 doubles which is 1/2! I wouldn't mention if it happened only once....

https://boardgamearena.com/archive/repl ... s=91199929;
Probability to roll at least 7 doubles in 21 rolls: 4.8% (1 chance out of 20,9).

So since you got 12k games, it should have happened in average 600 times for you, and 600 times for the opponent.
And 30 of those times it was for both players at the same time :)
Wanted to share this one too, I was defeated in 16 rolls of which 8 doubles !!
https://boardgamearena.com/table?table=384307312
User avatar
Romain672
Posts: 1012
Joined: 05 April 2016, 13:53

Re: Terrible algorithms

Post by Romain672 »

BackGa wrote: 05 June 2023, 21:26Wanted to share this one too, I was defeated in 16 rolls of which 8 doubles !!
https://boardgamearena.com/table?table=384307312
8 doubles / 16 is 0.2% (1/465).
So again with 12k games, that should have happenned in average 26 times to you and 26 times against you.
And there is 5.4% chance to have at least one game where it happened to both players at the same time.

You can take the conclusions you want, it just depend of how far you goes to find a game / how much you played.
But it's nothing out of the ordinary.


edit: You played around 110 games between your two posts. That give you 21% chance to have this event occuring to you after 110 games. (note that this number is slightly biaised because you start your serie at a specific point, but can stop it whenever you want. So maybe you will find something unlikely after 1 games (which would get a very high probability), maybe after 5 games (which would get an high probability), or after 10, 20, 42, 50, 100 games).
And note we are outside an usual range I would advise to look at (ie most unlikely occurence in 100 games).
User avatar
Jellby
Posts: 1343
Joined: 31 December 2013, 12:22

Re: Terrible algorithms

Post by Jellby »

No amount of "look what happened in this game" will prove anything.

Unless someone rolls a 15 or something.
User avatar
BackGa
Posts: 32
Joined: 14 October 2021, 00:57

Re: Terrible algorithms

Post by BackGa »

Jellby wrote: 06 June 2023, 08:26 No amount of "look what happened in this game" will prove anything.

Unless someone rolls a 15 or something.
not trying to.....
User avatar
smattathias
Posts: 9
Joined: 06 December 2021, 04:28

Re: Terrible algorithms

Post by smattathias »

I had a professor in college who said if a few people get a question wrong, I know I wrote a good, challenging question. If a lot of people get my question wrong, I know that the fault is not in the students. It's in my question, and that's on me.

Feedback is meant to be constructive. It seems like a lot of backgammon regulars have some intuition that things are off. Don't get defensive. Just investigate it, swap out the random number generator with another, actually look into it.

Strings of nothing rolls should be happening all the time to both players constantly. They don't. We have double heavy games. It's weird. Sucks to both win AND lose with a lot of doubles, and it'd be great if some of the programmers would just take a beat and check under the hood. You can check the stats all you want. I think it's time to check the code.
User avatar
euklid314
Posts: 287
Joined: 06 April 2020, 22:56

Re: Terrible algorithms

Post by euklid314 »

Since the human perception is so bad at evaluating random number sequences it proves nothing that some people are complaining. At every real life tournament people are complaining (although precision dice are used there) - most of the time it is the same that complain.

The RNG (random number generator) of BGA is well known (read the FAQ) and is used in many high level applications. The code is not written by BGA but BGA uses a well-tested PHP function. Since all the statistics show that the RNG is working fine there is nothing further to look into from BGA's side.

You are mentioning number of doubles but as many users before you have not done any statistical analysis. You only talk about your feelings of many doubles - but you surely already know that the number of doubles on BGA is exactly as it should be.
User avatar
RicardoRix
Posts: 2100
Joined: 29 April 2012, 23:43

Re: Terrible algorithms

Post by RicardoRix »

smattathias wrote: 16 June 2023, 00:48 and it'd be great if some of the programmers would just take a beat and check under the hood. You can check the stats all you want. I think it's time to check the code.
Yes this has happened. Do you think that changes anything?
You can even check for yourself, everybody has payer and global statistics for each game they play. Backgammon highlights the level of doubles.
https://boardgamearena.com/playerstat?i ... 12&game=53

The professor-question analogy is a bit off. The players complaining is very much a subjective thing - it's how they feel. Everyone has the sense of aggrievance when the randomness doesn't quite fall their way. Getting a question wrong in an exam isn't in any way relatable.

And who's to say that the percentage of the number of people complaining is 'high' ? You'd expect some kind of level. Quite often it's the same people complaining about it multiple times.
User avatar
dschingis27
Posts: 548
Joined: 27 June 2015, 18:30

Re: Terrible algorithms

Post by dschingis27 »

smattathias wrote: 16 June 2023, 00:48
What I really hate about posts like this is the strong conviction that it must be the code. Apparently you know very little about how RNG is done on a site like BGA. You could have raised an open question, what could be the reason for results seeming off, how is this technically implemented? No, you immediately jump to "Please check the code."
veggivet
Posts: 48
Joined: 21 March 2022, 21:16

Re: Terrible algorithms

Post by veggivet »

smattathias wrote: 16 June 2023, 00:48 I had a professor in college who said if a few people get a question wrong, I know I wrote a good, challenging question. If a lot of people get my question wrong, I know that the fault is not in the students. It's in my question, and that's on me.

Feedback is meant to be constructive. It seems like a lot of backgammon regulars have some intuition that things are off. Don't get defensive. Just investigate it, swap out the random number generator with another, actually look into it.

Strings of nothing rolls should be happening all the time to both players constantly. They don't. We have double heavy games. It's weird. Sucks to both win AND lose with a lot of doubles, and it'd be great if some of the programmers would just take a beat and check under the hood. You can check the stats all you want. I think it's time to check the code.
AMEN!!
User avatar
NancyArce
Posts: 2
Joined: 22 May 2022, 02:10

Re: Terrible algorithms

Post by NancyArce »

We all know it. It is literally one of the worst FAIR games on the web. For people who don't buy it...gimmie a break. Your opponent gets ALL PERFECT moves and no matter what you move, they get the number they need. Lets not even mention the amount of doubles they get to boot. It is quite obvious and just plain sad.
Post Reply

Return to “Backgammon”