the dice are not random



  • I just played a game, where the 6s were rolled 34 times, 5s were rolled 29 times, and 8s were rolled twice.
    I dont even have to run a chi-square test to know this is not a random dice system.
    To be honest, it happened to me a few times.
    So the question is, why it is so hard to implement a better dice system?



  • I have had similar things happen with real dice. If you want more even distribution in custom matches you can turn on "card stack".

    You have to remember the dice have no memory. They don't know what they rolled previously. Each roll is independent. Probability doesn't predict outcomes, it is only a mathematical representation of what the likelihood of something happening is. Winning the lottery is a one in billions chance, yet it still happens.

    Also, that math is only a somewhat accurate representation when the sample size is large enough. You need several hundreds or thousands of rolls before probability calculations can come close to representing reality.



  • So I have played over 700 Catan rounds online. I have also played about 300 Catan board games. Furthermore, given that I am 55 years old and have played dice games my entire life, I know what ordinary random dice rolls look like throughout a game. I completely understand probability and game theory. On flip side, I also understand that in order for a game to appear to be fare, developers can tweak algos to lessen the pain of adverse events or increase the chances of positive events. These tweaks are advantageous to the developers when attracting younger players or new players. Think of this Dev intervention is a way to passify the angry(think of being robbed). Or to heighten a new player's addiction to playing more(think about less probable numbers coming in repeatedly for less advanced players). Finally, think about the real concept of imperfect algo that get hung up on 5 -9 loops or 4 -11 loops. These loops NEVER occur ever when using real dice. My observations of suspected tweaks and algo flaws are as follows: 1) With approximately a 75% probability, a player who was just robbed the prior turn, will get a replacement of the stolen resource back on the next turn. So you lost a sheep on the previous turn after being robbed? No problem, on the next turn a 12 will be rolled and your settlement on a sheep territory will replenish a sheep. 2) In approximately 30% of the games that I play, there is the annoying as hell glitchy algo that makes me want to quit the game online forever. That is, only two numbers (not 6-7-8) get rolled repeatedly. For example, I have been in over 6 games in the last two weeks where 11 & 4 or 5 & 3 are the only numbers rolled in succession for over 7 rolls. While the odds of that event occurring is possible, the probability of that sequencing occurring so frequently is virtually impossible. I am ready to stop playing this game due to this one glitch alone.

    My solution to this problem you ask? I suggest this to the Devs... Give players the option of playing the game in two alteranative dice modes. 1) traditional which uses the current dice algo. and 2) a new truly random algo that removes the tweaks. For appearance purposes, you would need to call the play options something different to distract from the fact that the dice rolls are juiced in the original game. So maybe simply offer two modes of play? lets say Beginner and Advanced?



  • You are right. I also have observed that the probability of dices is flawn by developers, when playing against the machine, and I think it's intentionally. It's why I have lost the taste to play against the machine.
    I think there is not an algorithm to increase difficulty when you improve your level. So, this is a trick to assure that this is harder for you to win. Sometimes I was surprised also for a quick and easy victory. The dices did everything again, to prevent you get discouraged and quit the game forever. It's disgusting as you have the feeling that the game is always controlled and fake.



  • Yes, the game is rigged: it biased against the human, especially when he’s ahead. High prob rolls are consistent only when blocked by robbers: uncovered, the machine weights against.

    (24 resource cards? What human could go that many rounds without getting sevened?)

    Catan is supposed to be strategic and tactical decision making, with dice the random element. The algorithm fiddles the randomness, and that destroys the game.

    Using a truly random dice table, cutting algorithm designed bias out of the rolls at least, is all we ask.



  • I agree 200% with all of the posts here! I was beginning to think that it was my imagination ! Hope that they got the message right and do it quick!



  • The dice are neither rigged nor are there any malfunctions with the RNG.
    Sequences of rolls, as well as certain numbers not getting rolled, can and will happen. This happens in real life and in the App.

    If you want to learn more about probabilities, randomness, and why only 50-90 rolls ( a CATAN game) won´t produce a standard distribution, please read the following articles:

    http://stungeye.com/archive/by_date/2013/12/17
    https://cs.stanford.edu/people/nick/settlers/DiceOddsSettlers.html

    Keep in mind. Thousands of games are being played daily on our platform. “Odd” results will happen and users will post them, creating the impression that every game looks like that. But (1) no one will ever post just a normal game without anything that will get them mad and or attention.



  • @Administrator Why are you pretending that the dice rolls are not rigged?
    It's so obvious that nearly every single game there are 2-3 numbers are rolled nearly all the time.
    Yes, sometimes it can happen, but every single game? At least don't insult us by lies. Grow a spine and admit that dice rolls are tweaked, so we can live with it or don't play.


  • administrators

    @MidgetJones did you read the articles? These are mathematical facts. Do you think we want to waste our time trying to teach people about probabilities, randomness, and how their brain is tricking them into seeing sequences and patterns?

    https://en.wikipedia.org/wiki/Apophenia
    https://en.wikipedia.org/wiki/Gambler's_fallacy

    We developed many board games over the years and many CATAN adaptations. It would have been easier to actually create a rigged dice algorithm to address the pain points of users that feel unfairly treated by the dice. But ... we didn't! Because luck and randomness are part of the game. That's what makes it fun and maybe also frustrating at times.



  • @Administrator Well, you may not have rigged the dice on purpose, however your algorithm is heavily flawed causing to favour a single player in an extremely high portion of the game. No, it's not natural to have nearly every single game 2 or 3 numbers rolled a lot more than others especially when the numbers have low probability. Try to replicate in real life that kind of dice run. Mathematically it's possible to draw 1,2,3,4,5 on the lottery, however it has never happened in real life. Your dice rolls pattern barely can happen in real life, however they happen nearly all the time in the game. Maybe request screenshots from players of dice rolls of finished games if you don't have that data, and face the facts. I will start taking screenshot of every single game dice rolls from now on and will post it. That way you might be willing to face the facts and do something about it.



  • In the lottery the chance of drawing 1,2,3,4,5 is equal to the change of drawing 4,11,12,16,20,22. As a mathematician I am pleased to see that 2 or 3 numbers are rolled a lot more than expected and of course in favour of a single person (not myself 3 out of 4 times). The aim of playing is to be prepared for bad luck and beat the odds.



  • I have played nearly 20,000 games of Settlers, 4500 of which on Catan Universe, mainly of Seafarers + Cities. I can honestly say that Catan Universe dice are not a true representation of a natural distribution curve. Unfortunate facts. Sorry Catan Universe. I actually strongly believe that it has been skewed even further by the most recent update where we were given new colors and new map layouts. I have 38 games of data so far and the results have been startling to say the least. 38 games is still not necessarily enough data but it is not looking good. Shame on Catan Universe for lying to their userbase.

    P.S. if you're reading this and you know my username, sorry for being a dick in game, just know that it is out of love ;)


  • administrators

    @LeBron-James-Srsly You should also look at this article: http://stungeye.com/archive/by_date/2013/12/17 There are almost no better explanations out there for what you are experiencing. And please note the person who wrote that article had nothing to do with Catan Universe.

    In general:

    All the "accusations" are all over the place from "hot" numbers to, "hot" pairs to certain accounts being favored to certain accounts being mistreated, to Elo favoring top player or paid accounts or new account and low Elo accounts. The only thing they have in common is that they are very inconsistent....

    That should make you think. Why are so many different patterns apparently visible? For every (!) update we do, we get feedback that the dice got "worse".

    Please ask yourself from a rational standpoint:

    How often do you think are we making changes to the dice? Every update?!

    Why would we integrate a flawed dice algorithm and spend the time and effort to talk for hours and hours for years and years on our forums, social media, and so on with you guys about it not being flawed? Why would we waste all that time and energy?

    What do we gain from a flawed dice algorithm?

    On Catan Universe, hundreds of thousands of games are being played every day! This means "odd" results will happen and people will post them! No one has ever posted a screenshot of a game and said: well that was just a normal game. This creates the myth of the dice being somehow not working as intended.



  • It bothers me that no administrator will admit that the algorithm takes into account the point differential of the players. I noticed this a year ago when I started playing and searched the forum and found a response from an administrator that began “the occurrence sevens depend upon several things...” I can’t find the post now as I think they deleted it. I only play card stack mode and the distribution is fairly close to what you’d expect if the game lasts long enough BUT... if a player gets a large lead there is a period of time over the next 5-7 rounds ((15-21 rolls) where the player not with the big lead will get a virtual improbable number of sevens and the player in the large lead will be decimated. They are trying to “make it seem more fair” or “even it up” but it is maddening! I mean like threesevens in five rolls or 7 in 10. It is so obvious what is happening and so dependable I told my wife and daughter to watch and they started laughing! Yet they won’t admit they do this.



  • I have this now 8 games in a row.
    Grandmasters get the dice so often and kill the others.
    games are not even anymore.

    https://ibb.co/Chc3nsL



  • @Administrator said:

    https://cs.stanford.edu/people/nick/settlers/DiceOddsSettlers.html

    The whole "Card Stack" model is an admission that the random number generator is flawed (as are all software RNG's). In this case the results are MECHANICALLY adjusted to converge on the normal distribution as the game goes on. But the seriously NON-RANDOM streaks early in the game often provide such an insurmountable advantage to the lucky player or players that the game is already lost by the time the adjustments are applied - the final distributions do look more normal, but by then the point is moot. The developers could solve this problem easily enough by ponying up to use a truly random (i.e., physical-noise based) number generation service like random.org. Instead, they make it much, much worse by biasing the rolls in various ways (like increasing the chance of rolling the number the Robber is currently sitting on, or increasing the number of "7"s in general) in hopes of increasing the entertainment value. Such a fantastic game but it could be so much better by getting the hell out of the way and letting true probabilities govern play.


Log in to reply