silentProtest wrote: ↑23 July 2022, 23:30
Stroom wrote: ↑23 July 2022, 07:44
What Meeplelowda said. People who complain about randomness do not understand randomness.
I actually thought about finding papers I read, about how to evaluate RNGs/pseudo randomness ( at what point it could be considered random enough that it doesnt matter, in relation to predictability, that it's only pseudorandom ), or just googling it, because that might be faster, but it's not worth the time (for me, right now), and it wouldnt help without the large data set. My point: there are valid reasons to complain about algorithms, that are generating these "random" numbers; not that all here are, but neither are none of them.
You don't even need to go that far into it. The RNG issues that people have, deal with highly precise measurements. Not when reduced to 6 results (x2).
One RNG roll outputs a value between 0 and 1, with 14 digits. Then that value is multiplied by 6 and rounded to the first integer. So you will get a roll of 1-6. All this precision is lost in this calculation so it does not really matter how good the RNG actually is.
What is a problem for games like Catan is the
low amount of dice rolls. Usually there are about 70 dice rolls in a game of Catan. For two 6-sided dice there are 36 combinations of dice roll results. You can rarely get a normal distribution in 70 dice rolls. Pretty much every game will feel like the dice are not favoring at least one of the players.
It's almost like flipping a coin 2x and hoping that it will always be either heads-tails or tails-heads - because that would be balanced. But it is as probable that it will be heads-heads or tails-tails. Basically you could compare it to Catan - about 50% of the time you will get a total dice statistics that seem skewed to a human observer. Not following the perceived distribution.
Also, when flipping dice, you will often get the same result many times in a row. You could have HHHHHHHHTTTHTTTHTTTT - Most of the heads appear in a cluster but if you flip long enough it will even out. Humans would perceive it as unfair but the reality is that this can still happen. If not exactly like this then some other way that seems "unfair". And there are enough of these "unfair" distribution combinations that it would exceed the amount what a human would think is "reasonable". But this is the reality and you have to get over it. For a game like Catan, the analogue is that you roll all 6s at the start of the game and later roll all the 8s... evening it out in the end. This is not good for the players who built on 8. The player who built on 6, can spread to 8 by using the previous outcomes... And he benefits from the 8 as well.
https://www.geogebra.org/m/UsoH4eNl - go and try it out - roll 100 times, reset, roll 100 times, reset. The graph will be skewed quite many times.
You have to realize some things:
Previous dice rolls do
NOT affect the future rolls. Therefore, rolling the same result 4-5 times in a row is perfectly fine to happen
To achieve a perceived normal distribution with 2 dice, you'd most likely need 100x as many rolls as there are different combinations. And even then it is not guaranteed.
Saying anything like "I have 8+ cards so it is more likely that a 7 is rolled" or whatever you seem to perceive is most likely confirmation bias.
