Is the shuffling algorithm truly random?

[lgaamorim]

I've been playing a lot of Forest Shuffle on BGA, and I was wondering if anyone else has had the same impression.

In many games, I've noticed what seems to be a recurring pattern: one player appears to receive significantly more favorable card draws than the others, more often than I would expect from pure randomness.

I understand that unlikely sequences can happen in any game with random shuffling, and I'm not claiming that there's necessarily a problem. However, after playing many matches, this feeling has become increasingly difficult to ignore.

It also made me wonder whether there's any difference in the shuffling algorithm between Premium and non-Premium accounts. I assume there isn't, but it would be reassuring if the BGA team could clarify this or confirm that exactly the same randomization system is used for all players.

Has anyone else had the same impression, or am I just seeing patterns where none exist?

[Jellby]

more often than I would expect from pure randomness.

Really? Do you have data or just your gut feeling? How often does it happen? How often would you expect it from pure randomness?

Just a reminder: randomness is not "even distribution".

Has anyone else had the same impression, or am I just seeing patterns where none exist?

Probably both. I'm sure many people have the same impression, but it's only because they (and you) are most likely seeing patterns, that's what our brains do.

[Silene]

[...] but it would be reassuring if the BGA team could clarify this or confirm that exactly the same randomization system is used for all players.

They have put that clarification into the FAQ (it's linked on the bottom the page).
Link to the specific section: https://boardgamearena.com/faq?anchor=f ... ing_random

[Jellby]

I wonder why they use different algorithms for dice and shuffling. Shuffling an N-card deck can be trivially done by "rolling an N-sided die", then an (N-1)-sided die, then an (N-2)-sided die, etc. I guess it's a less efficient way of doing it. Given that shuffling is not that frequent (compared with all the other things that happen during a game), is the performance hit that significant?

[BarnardsStar]

I wonder why they use different algorithms for dice and shuffling. Shuffling an N-card deck can be trivially done by "rolling an N-sided die", then an (N-1)-sided die, then an (N-2)-sided die, etc. I guess it's a less efficient way of doing it. Given that shuffling is not that frequent (compared with all the other things that happen during a game), is the performance hit that significant?

Some games enable players to "peek" at the deck, or to put cards on the top or the bottom of the draw pile, etc. A predetermined order is necessary for such operations.

And in general, I think that if the card order is predetermined (and players know and trust that it is) it makes disbelief in the algorithm harder. "How could the game know which player was going to draw the 17th card and what they would have in their hand at that point?"

[Jellby]

I didn't mean "choose a random card every time one needs to be drawn", but "shuffle the deck by picking cards randomly one by one". That should give a predetermined order all the same.

[BarnardsStar]

I didn't mean "choose a random card every time one needs to be drawn", but "shuffle the deck by picking cards randomly one by one". That should give a predetermined order all the same.

Oh! Sorry, misunderstood.

And agreed - as I understand it, shuffling algorithms do essentially boil it down to what you suggest (with a few optimizations), so it is curious they would specify that dice-rolling is based on a better randomization algorithm (PHP's cryptographic-level rand_int) than deck shuffling (Mersenne Twister - not as good but still plenty good enough for the purposes at hand)...