fix the dice



  • dice algorithm can not be good, do something to fix it.
    1 picture 1000 words :
    https://i.ibb.co/SryLGxH/bad-dice.jpg[



  • @BloodForPeace said:

    https://i.ibb.co/SryLGxH/bad-dice.jpg

    At least there were 6s and 8s. I keep playing games where 5s and 10s and 11s are way more frequent.



  • I agree. Here's mine: https://imgur.com/a/kXo2KUI



  • Okay, these complaints are just anecdotal, they have legitimate issues to fix, this is not one of them. I've played real games where one number will roll 20 times in a row. This happens. Probability is based on iterations approaching infinite. I had a game where we rolled snake eyes 15 times in a row. The odds of that are impossible, but it happened.



  • Feels like the dice chooses to randomly favor 2 numbers each game and boosts their chance to roll. Really unbalanced when you see the stats at the end of the game and an unlikely number rolls twice as much as a "higher probability" number. How could a 3 roll 15 times when 6 rolled only 7 times? Seems that a lot of players are noticing these patterns, which makes it unlikely that it's just bad luck. Oh well guess it's time to jump ship to colonist.io.



  • Please fix the dice (or unfix them, as it were). It's not fun as it is, and I wish I'd read all the complaints before I made purchases.



  • I find that most recently whatever number I have my settlement are the least rolled. I understand that in a real game sometimes the most popular numbers don’t always come up most often but it seems unlikely that in almost all games the AI players get the most rolled numbers ? Makes it not fun - probably won’t keep playing if they don’t improve the dice algorithm.



  • Yes, it does often seem that when playing against bots the dice tend to favor the bots. Thankfully the AI doesn't play intelligently (which is why I usually call them bots rather than AI). :wink:



  • There's nothing wrong with the dice IMO. 1 tiny sample size does not mean anything. Show me a graph with a 1000 rolls and I guarantee it will be a nice bell shape.

    If you want closer to a bell shape over a small sample size, select the "Card Stack" option



  • 1 tiny sample size does not mean anything. Show me a graph with a 1000 rolls and I guarantee it will be a nice bell shape.

    Exactly this. These posts are always from people that don't understand probability. A game of Catan only has ~60 dice rolls total. You simply cannot expect a uniform distribution over the dice rolls with so few samples.

    There is simply zero evidence that the dice are biased in any way, other than "the dice didn't favor ME, so they're broken".

    To be extra sure, I wrote a program that rolls two six-sided dice 60 times with 100% RNG (no bias), and low and behold, you end up with very non-uniform results about 50% of the time (just like in CU).

    At the risk of adding more anecdotal evidence, games we've played on the physical board also have hot streaks of the same roll. That's just the way she goes. Asking for "completely random" results, and then getting made when a 9 is rolled 4 times is insanity. You don't want "random", you want enforced uniformity, which is the event deck (and CU even has an option to play like this, although admittedly not in auto-match).

    Catan has an large element of randomness baked into it. If you don't like that, this maybe isn't your game.


Log in to reply