Biased Shuffling about deck-of-cards

[pakastin]

I know, I use Fisher-Yates a lot. The point here is to have it realistic: the more you shuffle the deck, the more random it gets - just like in the real world.

You could make another module for card games, when you need better randomness. These modules will become external and you can choose which ones to load in the near future.

Thanks anyway, I will keep this open until it haves Fisher-Yate as an option.

[agrif]

Ah ok. I was mostly worried that somebody would make a card game out of this expecting the shuffle to be unbiased. I get the feeling most programmers would look at your shuffle code and not see any problems.

[pakastin]

I was thinking that I should add a parameter to the shuffle -function:

deck.shuffle(true) // realistic shuffle
deck.shuffle() // Fisher-Yates

[Marak]

Reference:

https://www.npmjs.com/package/knuth-shuffle
https://github.com/coolaj86/knuth-shuffle

[pakastin]

When shuffling two times (like in the Deck of Cards example), I'm getting equal results as with Fisher-Yates:

function shuffle(array) {
  array.sort(function() {
    return Math.random() - .5;
  });
  array.sort(function() {
    return Math.random() - .5;
  });
}

http://bost.ocks.org/mike/shuffle/compare.html

So I might as well use Fisher-Yates since it's much faster.. :)

Thanks for sharing the link!

[agrif]

Oooh interesting. On Chrome, the old algorithm has a nice solid green streak down the diagonal, and in Safari it does something weird with the top row, but on both of them running the old shuffle twice looks pretty unbiased. I wonder why.

I love this problem, I find it really interesting for some reason.

[pakastin]

Yeah, the problem itself is very random but the results not so much :D