In Settlers of Catan, there's a variant where instead of rolling the dice, you use a stack of 36 cards with each combination. This reduces the luck component, but has as disadvantage that you can predict future rolls to some extent.

The iPhone Catan game also allows this rule variant.

Have you tried this and do you think it's an improvement?

EDIT: Have you tried other methods of reducing the importance of luck?

  • 15
    My method of reducing luck is to play Go. If I want a game with no luck, I play an abstract strategy game. Luck is an essential feature in Catan for me; it means that weaker players can get a little boost sometimes, and that you need to be flexible and adapt to anything when playing. Oct 22, 2010 at 0:18
  • Part of the point of the game is reducing dependency on luck, by adding more settlements, converting to cities, colonizing seaports...
    – Kyralessa
    Aug 23, 2018 at 17:13
  • Alternatively the OP could play any one of the civilization building games out there that don't rely on dice.
    – Stephen
    Apr 11, 2019 at 6:21

13 Answers 13


The card system that you describe is not just a variant, it is an official expansion, Catan: Event Cards, which includes not only the rolls (in the form of the totals with the appropriate distribution) but also red dice for Cities and Knights and a selection of minor game events.

Sample Cards

  • Note: everyone I play with hates this expansion since everyone just starts card-counting. Jun 19, 2018 at 3:52
  • 2
    @YetAnotherUser are you shuffling before reaching the bottom of the deck, as instructed by the rules that come with the deck? if people are still counting, you can move the New Year card up higher in the stack, incrementally increasing the randomness.
    – Sparr
    Jun 24, 2018 at 17:38

I played a diceless variant where every village has a worker, we use glass beads to represent them.

  • When you create the village, place the worker in an empty tile next to it
  • At the beginning of your turn
    • you get one resource for each worker based on the tile it is on
    • Or you may skip acquisition to remove another player's worker (he'll be able to re-assign it at the end of his next turn)
  • At the end of your turn, you may re-assign your workers (still into empty tiles next to their village)
  • Cities have two workers

The game is very different from the original version, and very interesting.


I haven't tried this variation, and to be honest, I don't think I would want to either.

Randomness in games can bother me, but Settlers of Catan is one of those games where I don't mind the randomness. With the deck-of-cards variation, you know at some point in the next 36 cards that, for example, a 2 and a 12 will come up at some point, and if you can count cards, you can most certainly predict what the next results will be.

Me, I rather be amused when we spend six turns in a row rolling a 3. That's just part of the game.

  • 10
    Some games are simply spoiled by the dice, that can be pretty frustrating. There's also a variant where you remove 5 random cards from the 36 possibilities, so that you can't be quite sure what the last cards in the deck will be. Oct 21, 2010 at 23:38
  • 1
    the 2 and 12 are not certain to appear. the expansion comes with a card called 'new year' which is shuffled into the last few cards (sorry cant remember how many). when this card appears the deck is reshuffled, so you very rarely go through the deck in entirety. Sep 25, 2016 at 10:05

The large majority of my games of Settlers, or more recently Cities and Knights, were done using a computer program to simulate the deck of 36 dice. It would randomly reshuffle at some point between 34 and 36 cards through, so you can't count rolls precisely, but it was a blissful solution to the frustration of the "Settlers probability distortion field". I much prefer Settlers played with "enforced probabilities" like this: there's still plenty of randomness in what order the rolls come up, but you at least know that building on a 6 and an 9 you will get more resources than the person who built on 4 and 11, where in the dice version the opposite seems to happen frustratingly often.

The program we were using also provides for the Cities and Knights dice by colouring one number red, and giving a background colour of black (3 in 6) / yellow (1 in 6) / green (1 in 6) / blue (1 in 6).

You can download it from http://members.lycos.co.uk/qqzm/downloads/Projects/Other/SettlersDice.zip.


I bought the deck of cards, but everyone I play with just wants to roll dice.

So I bought precision dice.


I use an iPod/iPhone Settlers Dice App. The benefits I see are: 1.) They're quiet, 2.) They're fast, 3.) They don't make a mess and get hung up on piles of cards or the edge of the board, and 4.) The probabilities are assured by the programming.
The probabilities exactly match the theoretical probability of each dice roll, which is not the case with physical dice.

I have used two iPod Touch/iPhone apps: "Settlers' Stats" and "Dice of Catan". "Stats" has a so called feature that allows you to track point changes of the various players. This turns out to be more trouble than counting again when you wonder. This app stinks. "Dice of Catan" is a different story.

"Dice of Catan" (On the App Store) costs $0.99. It is a very pleasant and simple app to use, and it takes the physical dice out of the game, which I think is a big plus. You tap the dice, and they display a pseudo-random (probability driven) Settlers of Catan roll.

Features: Cities and Knights (C&K) support (has a die for barbarian movement/progress card.) Safe mode for early rounds, which allows no sevens (and no barbarian movement in C&K). Multicolored dice (one red, one green, see below.) Tracks barbarian movement and has a pop-up reminder for the Robber/Bandit and when the Barbarian strikes.

Example Roll Example Bandits Always starts in Safe Mode, tap to change


I wrote a Mathematica script to handle the dice rolling. It keeps track of all previous dice rolls, and weights the next dice roll probabilities according to the difference

(expected # times rolled) - (actual # times rolled),

so that the actual distribution converges faster to the expected. (And dice rolls on different turns are not independent.)

Though with this method, people could predict things like "a 7 just came up, so it probably won't come up for several more turns", so there's also an option to effectively ``increase randomness''. With probability p, the dice roll will actually just be random according to correct distribution for 2 dice, and with probability 1-p, the weighted history will be used to roll the dice. This makes it much harder to predict things while still making sure the distribution converges pretty quickly. You can of course still predict things like "a 5 hasn't come up for a long time, so now it's due to come up".

link to Mathematica cdf

Note: Eventually I may end up writing it in python or javascript (the code to just roll the dice is quite simple), but for now I only have it in mathematica. You can use the cdf file with the free Wolfram CDF player though.


  • Can you relink this? Link is broken would appreciate it Aug 27, 2020 at 20:34

I've got (but don't use) and Android App called steady roller. It auto-adjusts the probabilities to even out the randomness, but in a less predictable way than the event deck / dice deck. I don't know its exact algorithm.

You can also set it to reduce the chances of 7s, if you like that sort of thing.

It's a freebie on the app market.


We tried a variant where the amount of resources are produced according to the probability of the dice roll that would produce this resource.

Say you've got an "8" and "3" on ore and a "12" on wood. Each players turn, after the dice roll (for bandit, barbarians and C&K) resources are produced in this way: to roll an "8" with two dice the probability is: 5/36, for a "3" it's 2/36 and for a "12" it is 1/36. So in each players turn you would receive 5 + 2 ore and 1 wood marker. If you reach 36 you immediately discard 36 markers and gain one card. This needs a lot of administration, but makes the resource management more planable and cuts off a litte from the importance of settlement/city placement.

If a bandit enters a resource field, all players with a city or settlement adjacent to that field loose all their markers for that resource. Other bandit rules still apply.

You can do a similar version with development cards, where you have to gather development points according to your progress in this area of development (two for the first, one for each one after the first), but this leads to very fast or very slow developments, depending on the number of players. You would have to set the necessary development points to 36 to have the same average in developments as with dice.

With barbarians and the bandit this works too.

However we found that especially with development, barbarians and bandit, the randomness makes the game more exciting than annoying. So we roll the dice as usual, ignore resource production and carry out just the events development, barbarians and bandit.

If you cut out all randomness (exept for the player influence) some strong players did manage to think ahead a number of turns and act accordingly, which left weaker or inexperienced players no chance.


As @panavia pointed out, a weighted dice which re-weights the probabilities every turn based on the historical outcome is the best choice.

The problem as I see it is that since one game is not long enough, the true distribution is never reached and events such as 3 5's in a row can have fatal consequences for the game.

I wrote a post (click here) on the topic in cross validated with R-code for simulation.


The problem with such variants is that randomness is one core part of the game.
Yet, I'd agree that Catan depends too much on randomness.
...but isn't finding the "right degree of randomness" something extremely subjective ?

I would suggest the 'dice results cards' system, but with variations to keep uncertainty.
The official expansion already features a "new year" card, just move it earlier in the deck (like after 18/36 cards, instead of 31/36) This way, you'll just "soften" randomness, but keep the game away from an ugly a boring fully equitable distribution.

Reducing/preventing "absurdity" like getting 4 ou 5 of a same result in a row, but keeping uncertainty and giving no guarantees looks like a decent compromise to me.

I suspect the "Cities and knights" expansion might be another way of reducing randomness, since the games seems to require more dice rolls.
Well, I personally HATE C&K, which looks slow & burdened to me, so this is not what I would recommend... but this is a question of taste :)


Similar to the Event Cards and Deck of Dice, there's also Dice Cards: https://www.thegamecrafter.com/games/dice-cards


just use a 12 sided die.its so easy

  • 3
    All this does is shift the odds more toward low and high numbers. It doesn't prevent it from having a big luck component where, say, you might get unlucky and get very few resources from an 8 while an opponent gets more from an 11 - in fact it makes that particular thing more likely.
    – Cascabel
    Oct 28, 2014 at 5:53
  • 2
    Also, in the future it's probably worth writing a longer answer, including why you think this is a good solution to the problem. Oct 28, 2014 at 20:55

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .