Who programmed the dice?

[Bechster]

Another solution might be to implement the "event cards". It is an expansion from "Traders and Barbarians", where 36 cards replace the dice in the game, and usually leads to a much fairer distribution of resources. I think this addition would be much welcome.

[Jellby]

And here we go again.

At least they didn't create a new thread

But most my games have a number that is rolled a wild number of times. Each time it's often a different number.

You're describing quite accurately a random distribution.

[Rufasu]

Historically speaking, the random handling of PHP, a development language, has been problematic and debated for a long time.
So, there is always a need to examine random number programs.
Despite the fact that it is the domain of computer science, we continue to take only the convenient parts of statistics and run away with them as an excuse.
The BGA is not willing to do it.

[Stroom]

Historically speaking, the random handling of PHP, a development language, has been problematic and debated for a long time.
So, there is always a need to examine random number programs.
Despite the fact that it is the domain of computer science, we continue to take only the convenient parts of statistics and run away with them as an excuse.
The BGA is not willing to do it.

Dude, just stop. You do not know what you are talking about. The problems with RNG happen in a very miniscule scale when you are handling very specific stuff. For a game that reduces the RNG to only 6 numbers, it is negligible. You won't even know what RNG is used if it would be changed to anything else.

[Rufasu]

There seems to be some misunderstanding, but my point is that the BGA is running away with the word statistics and has no desire to do any programming.
If they had time to make excuses, they could improve.

[Nekojin]

Yes, there's misunderstanding going on here. The problem is that the misunderstanding is yours.

None of the people posting here are BGA's programmers. None of the people correcting your "ideas" are people who can get down under the hood and make changes to the way the site's engine applies an RNG.

Similarly, I highly doubt that the people who handle the Catan implementation on BGA are sitting here dueling with you over this. But even if they were, the idea that you're representing is a huge amount of programming work with questionable value. For someone who is NOT a BGA-employed programmer, trying to switch to a different algorithm means discarding these two lines of code:

// Roll dices
$dice1_value = bga_rand(1, 6);
$dice2_value = bga_rand(1, 6);

... and instead adding new code to call on another algorithm that isn't already coded. I don't know exactly what it would take to kitbash in a different algorithm onto the BGA engine, but I'm pretty sure that it's at least an order of magnitude more complicated than two lines of code. Possibly several more orders of magnitude.

And, again, nobody, nobody in this thread or any of the other threads complaining about the site's RNG setup has demonstrated any actual knowledge to show that the Mersenne Twister algorithm is somehow too flawed to generate two d6es or a deck of cards. We're talking about a generator that is capable of generating numbers higher than the number of atoms in a human body*... and you think it's somehow not capable of generating 2d6?

Change your shorts, your ignorance is showing. Again.

* Edit: For people who think I'm throwing big numbers without a care... the number of atoms in an average human body is around 7 octillion (a 7, followed by 27 zeroes). The Mersenne Twister's range is 2^19937... which is a much larger number.

[euklid314]

Also, Rufasu is always coming back to the old PHP function rand() having had "problems". This is ridiculous because even this old and simpler algorithm was perfectly fine for every aspect necessary on BGA. This function could determine perfectly well a random number between 1 and 6. But it is not cryprographically safe, i.e., you should not encrypt your life savings based on the RNG function rand().

But BGA is NOT using rand() but is (and always has been) using the cryptographically safe PHP function random_int(). The function is available for everybody to check and has no flaws - you can use it for every high security application you wish - and this is done worldwide. If Rufasu is suggesting that somebody at BGA could write a better RNG than random_int() is simply ridiculous and shows lack of mathematical knowledge.

[Rufasu]

メルセンヌ・ツイスターアルゴリズムに問題があることは、PHPのライブラリ開発者たちも認めている。
また、乱数アルゴリズムの専門家たちも認めている。
さらに、新しい乱数アルゴリズムがいくつも開発されている。
それを知らないというのは、知識の不足を意味している。
PHPを扱う開発者ならば常識レベルの知識だ。
意図的に統計学の一部を引用するというのは、コンピューターサイエンスを知らない人を欺く行為だ。

[Rufasu]

The problems with the Mersenne Twister algorithm are acknowledged by PHP library developers.
Experts in the field of random number algorithms have also admitted this.
Furthermore, there are a number of new random number algorithms being developed.
Not knowing this implies a lack of knowledge.
It is common knowledge for developers who work with PHP.
To intentionally cite a piece of statistics is an act of deception to those who do not know computer science.

[Jellby]

The problems with the Mersenne Twister algorithm...

What are the problems? How do they affect dice roll simulations?

I'd bet big money to the fact that whatever non-randomness can be found in any of the common "flawed" RNG algorithms (including the one used in BGA) is much less significant than the non-randomness in real physical dice (which are, to the extent they follow Newtonian physics, not random).

PS. Even an "algorithm" that simply uses a stream 1 2 3 4 5 6 1 2 3 4 5 6 1 2 3 ..., but shared among *all* BGA games in progress, would probably be perfectly good.