Catan Is Broken. Unplayable

[Cile]

If I say I know how random numbers work, believe me.

Except you immediately contradict yourself:

70 trials is not very few. 4 is few. 70 is a lot and you should expect the numbers to approximate a bell curve.

Not with two dice.

I've found this tool for you: https://academo.org/demos/dice-roll-statistics/
Go ahead and set it to two dice then rapidly click roll 70 times, see if you get a "normal" distribution (the site will keep track for you).
Refresh the page and do this 10 times, will probably take less than 2 minutes. How many of those attempts were a "normal distribution"?

I find the games I play on BGA do not match the distribution of rolls of the games I play in real life.

I find when playing Catan in real life there is almost always an outlier: some number rolled more frequently than it "should" have been. The thing is we never know what number that will be before the game starts. After the game we'll look back and say "5 was rolled a ton that game" but next game it will be a different number. That is how randomness works.

And like I said that is not the first game where I got a strange distribution, that's just the one where I finally came to the forums.

I'm not saying it's not fair, I'm saying it's not seeming to behave as though it's random based on my experience in 150 games.

So in what way is it not random? Do you think 11 is always rolled too frequently? In what direction are the rolls biased?

And what I'm hearing from folks responding is you have faith in the RNG technology on BGA, and no faith in people's experiences so you blindly trust that the RNG will produce a normal distribution over time, but nobody has actually verified that.

Extraordinary claims require extraordinary evidence. You've come here with an extraordinary claim (the dice are not truly random) and your evidence is just a relatively small sequence of rolls. Then you're asking other people to do the math for you on how likely that specific result was (not an easy calculation) in order to help you 'prove' that this result was unlikely and therefore not random. Meanwhile you're insisting you understand how randomness works while you argue that this random pool of numbers is not "truly random".

That is why you feel people are being condescending: You are asking them to explain how randomness works, which would be a reasonable question if you weren't simultaneously insisting you know how randomness works and that this isn't random. You aren't the first person to claim the dice aren't random, you won't be the last, and you've presented nothing new to the argument.

[nik592]

...but if they think someone has gone through the effort of making the Catan dice not truly random for some reason, despite "RND(1,6)" being incredibly easy to code...

This is an important point. It's much harder to write code that DOESN'T use true random numbers, than code that does. If you think a dev has just been lazy, then you'll see the same pattern of numbers come up time after time, which I don't think is what you're claiming. If you think a dev has written something to (for example) benefit premium players, that's a whole pile of work. Let's assume there's just a single premium player at a table (because what if all players at a table are premium? Then we're going back to true randomness?), how is the "rigging" going to work? The game is going to keep track of what numbers the premium player has and randomly choose one of those every other turn (cause doing it every turn would be too obvious, and also not lead to the distribution you posted)? Even that could give you a game where there's minimal benefit to the beneficiary of the rigging - sure, it rolled 9 heaps and I got loads of brick, but at a certain point I need wood and sheep and whatever too. And maybe I have more resources to trade, but again, everyone else only needs a finite amount of brick and if some other resource is minimally available, that will make trades harder for everyone - people who naturally have it won't want to trade it, no matter how much brick I offer, and others may have other things to offer, better than my massive supply of brick.

So ok, let's say it's not just randomly choosing between numbers that benefit the target of the rigging and it's instead implemented some sophisticated AI that somehow figures out just what resources the target needs and giving them to them at just the right times. Stop and think for a moment about how much work THAT would be. And yes, AI systems exist out there, but they still need adaptation to the purposes you want, and the code framework being used. And the AI that determines the rigging would need to be developed for every game with random chance.

Thinking about all that, why would an unpaid volunteer dev choose to do that, rather than just code RND(1,6)?

Extraordinary claims require extraordinary evidence. You've come here with an extraordinary claim (the dice are not truly random) and your evidence is just a relatively small sequence of rolls. Then you're asking other people to do the math for you on how likely that specific result was (not an easy calculation) in order to help you 'prove' that this result was unlikely and therefore not random. Meanwhile you're insisting you understand how randomness works while you argue that this random pool of numbers is not "truly random".

The other point here is that an "unlikely" result still has to happen at some point. Let's take a really simple probability example of a ticketed raffle. There's 1 million tickets and I have just one. It's unlikely that I will win (literally 1 in a million chance). But given that someone has to win, it's just as likely that my ticket will be drawn as any other ticket. Let's say there were 1 million raffles, where each person had just one ticket. It's actually very likely that some people won't win any raffles and some people will win more than one. It's entirely possible that I could be the unlucky person who never wins.

[Jellby]

I'll tell you that the odds of getting exactly the rolls you got in 70 rolls is 1 in approx 10^-107 (= 1/6^140). Infinitely small for all practical purposes. But so is every other possible result. And yet, some result must happen, no matter how unlikely. As soon as you start grouping results to try to explain that some results are more likely than others (e.g. you say that 6+3 = 3+6, or that the order of the rolls does not matter, or that you only care about cases where there are more 10s than 7s) you're introducing a bias.

[Ehronde]

The developers of CATAN on BGA could just add statistics for each number (2 - 12). That should easily prove over time that the distribution of numbers is just as expected.

If someone thinks the dice are not random, it's their job to prove it. It's not the game developers job to refute any accusations without proof. Up to now, there hasn't been a proof for a lack of randomness. People only showed their lack of knowledge about randomness and statistics.

[Silene]

ok here's a dice distribution from my recent game

I have built a little something:
You can do a "Spot the Plot"-Test here: https://onecompiler.com/python/3zvmdjamm

On the right side a selection of Histograms that represent possible outcomes of 56 rolls with 2 dice are shown. (56 because your example had 56 rolls too)
To have a good view at it, best select the full screen button in the upper right corner and make the window wide. (Propably not possible to have a useful view on mobile)

The weird outcome you presented earlier is mixed into those in a random position. How fast can you spot it?
It is indeed one of the more extraordinary results among them (and it makes sense given that it was handpicked). But does it really stand out in it's "weirdness" as much as you thought it would? On all runs I did, I could find a couple of very unusual results.

Feel free to execute the code again with RUN to get a new set of 100 Histograms. Please don't edit the code - but you can check in the code on the left side, how the Results are created.

[Cile]

ok here's a dice distribution from my recent game

I have built a little something:
You can do a "Spot the Plot"-Test here: https://onecompiler.com/python/3zvhkmfbm

You spent more time on this than it deserved, and I appreciate it. The "All samples summed up" at the bottom is *chef's kiss*

[Kayvon]

70 trials is not very few. 4 is few. 70 is a lot and you should expect the numbers to approximate a bell curve.

Not with two dice.

I've found this tool for you: https://academo.org/demos/dice-roll-statistics/
Go ahead and set it to two dice then rapidly click roll 70 times, see if you get a "normal" distribution (the site will keep track for you).

That sounded like a fun exercise, so I did it. Unsurprisingly, 70 rolls doesn't yield a bell curve, but you can see one starting to form if you squint. Take a look!
https://imgur.com/a/SPBaQYv

I have no issues with the site's RNG. Humans excel at finding patterns, even when none exist.

[Fromper]

You want proof that the dice were randomizing correctly in your 150 games? Go back and look at the game results for all 150, track how many times each number rolled in each game, and add them up. (150 games) x (50-70 rolls per game) should be a large enough number of rolls to see something close to the "expected" bell curve.

If you don't see the expected results from going through ALL of them (no picking and choosing based on which games look right or wrong to you!), then come back here and let us know.

In the mean time, here's a tip for winning more games: Put settlements on ALL the numbers. That way, you get a resource no matter what rolls. Except the 7, of course.

[Cile]

Humans excel at finding patterns, even when none exist.

Or in this case: Humans excel and being offended when a pattern doesn't show up the way the expect it to.

[nik592]

Well, human brains are built to look for and make use of patterns. It's part of how we function in a world which generally contains too much info to take it all in at the same time. In fact, our brains will happily lie to us about what we are really seeing/hearing in order to fit the patterns it thinks it has found (that's how many optical illusions work). So the instinct is to seek out a pattern - the problem is that randomness, by definition, is not a pattern.