Rigged Dice Distribution - An hypothesis.

[Alex]

Hello everyone. I am a lover of Catan, played the game since I was 14 years old, in the Netherlands with my family. I know that on Catan Universe I am just one of the many who have gotten incredibly frustrated by the dice distribution that appears to be so much against reality.

I have played/observed many games, chatted with many players and analyzed many distribution charts. In the end, I formulated a theory as of why is the dice distribution so seemingly unrealistic.

Let me start by quoting a developer:

Catan Universe CM eredeti hozzászólása:

The dice RNG does what it is supposed to do. It produces random numbers. We get feedback about changing RNG results (it’s getting worse or better) after every Update but the RNG hasn´t been touched in a veeeery long time.
This shows us, most Catanians do perceive the RNG to be changing when it fact it doesn´t. It has been the same and it works perfectly fine.
For a regular game with -let’s say 50 rolls - chances are quiet high to not even get close to a standard distribution. Or let’s put it in other words… the results might be all over the place. Seemingly favouring certain numbers over others despite them being less probable to get rolled. Now why is that the case? It´s because we do not use loaded dice.
We encourage everyone to do a small experiment at home. Just write down every roll for every player when playing with your friends or family at home sitting around the table. It’s very likely that someone gets lucky during the game and someone will have bad luck. Now would you think the person who is lucky used loaded dice? Or rigged the dice?
Also feel free to use https://www.random.org/dice/?num=2 and write down the 50 dice results. You might be get quite random results ;)

----------

This folk is a half-lie or a half-baked truth. Or rather, it's misleading information.
Anyone who knows a little about software development and computers in general, should know as a fact that perfect randomness is simply not achievable.
(here an interesting article about it
Stack Overflow [softwareengineering.stackexchange.com] ).
Mr. Developer's comment refers to one of those algorithms that emulate randomness, but it does not address how Catan Universe probably does not use one of those, or uses an adjusted algorithm.

So that was point 1: No true realistic random number generator exists.

And here's point 2: How did they adjust the algorithm?

Please understand that this is speculation, however, I tried to find a logical explanation as to why Catan Universe doesn't make an extra effort to address the dice problem.

The idea is: Catan Universe is a business. Much like any major social network, their business model expects to create a big community of users, who will then purchase items in-app over time, and provide a cash flow, which will then generate revenue.
Hear me out here, it's a business with the sole purpose of making money: they have not developed Catan because they are good samaritans and you like to play Catan with your friends so much.

Once you understand this, you will see how, from a business perspective, it would very natural and pertinent to the stable business plan, the goal of maximizing the total user count: a small community would simply be not profitable.

So how do you maximize the community? By making sure that everyone wins!
Here it's where it becomes delicate.

I have the dreadful suspect that there is some kind of hidden ranking list that records losses vs wins on any given player. For example. If the player loses 4 games and plays a firth game, the algorithm will make sure to favor him in that game, to not make him tired, to not make him wanna give up on the platform, to keep the adrenaline going.

That is how Catan Universe makes sure that nobody leaves (deletes their account)
That is how Catan Universe makes sure that even the worst players win now and then. That is how Catan Universe maintains the platform addictive, just like the slot machines in a casino: they always make you win a little to keep you going.
That is how Catan Universe meet its business requirement to keep a huge community of players.

But if all of this is true, how is a player advantaged vs others in a game? Many variables come into play. Please note that these variables only affect the optimal scenario in which other players don't interfere with the preset handicaps. NOTE: These observations apply only to the base game, although some of them might be present also in the expansions.

1 - Starting order: you will notice how sometimes the board is arranged horribly in favor of the first player, who in this case will be the advantaged player (AP).
2 - Dice distribution: what everyone is complaining about. The dice will favor the AP only.
2.1 - The AP cannot be kept blocked by the robber. AP will often roll 7 to unblock themselves.
2.2 - Game 90% stars with a 7 in the first round. Usually by the AP, easy way to block others and get a head start.
2.3 - 7 Often strikes players with more than 7 cards that are not the AP.
2.4 - 7 Often gets rolled in succession until the robber gets parked on a spot that will effectively block the players who are not the AP.
2.5 - 7 Often gets rolled after a knight card is played wasting the efforts of buying and using the dev card. This happens to everybody but the AP. (BTW the AP doesn't need a knight card: see point 2.1)
2.4 - Dices often roll numbers on which AP is settled. Often many times in a row, creating an irrecoverable gap.
3 - AI: bots act completely irrationally.
3.1 - on a 7 bots almost never rob the AP. It can happen that the bot robs the player who is winning, but that's just because the current leader happens to play better than the AP.
3.2 - bots tend to avoid trading with players who are not supposed to win.
3.3 - bots trade until the is at 2 points from the win (3 if I am not wrong in C&K) when it's already too late.
3.4 - bots trade easily more than 1-1.
4 - Development cards: development cards tend to advantage the players. You might have noticed more often than not, how you will pull a card that you don't need at all for your tactics.
4.1 - You will invest in dev cards but the very first card you pull from the deck is a victory point card (pulling a victory point card as the very first card is a complete disaster, some of you might understand). The probability of pulling a VP card as the very first card is 1/5. But the probability of that happening often is way lower than that... yet you are left with what you experience.)
4.2 - You will pull a road builder in a game on which it's completely useless.
5 - New player account: if the player is recently signed-up there's a good chance he will win in the first 5 games.


----------

Conclusion

These were my observations and speculations as on how Catan Universe keeps it's community and sustains its business model by predetermining an advantaged player for any one game. I have observed several methods, one of which is by messing with the RNG.

If you find this helpful please contribute by replying with what you have experienced. Also, if you think that some of my analyses were wrong please let me know.

[Atomic Priest]

Catan Universe CM eredeti hozzászólása:

As part of the Catan Universe team, I do feel it is my duty to answer the here posted hypothesis despite the fact that we already answered a lot to the mentioned topics above.
1. Of course, most of the feasible RNG solutions for games are pseudo RNGs. We used to describe it as such in many many many posts in the past and tried to explain how they work and that it is perfectly normal to use such a pseudo RNG.
Well… that always backfired! Because our users thought that this was the smoking gun! That pseudo RNGs aren´t real RNGs and therefore the RNG is flawed.
Also please keep in mind not every user does have either programming experience nor a deeper understanding of statistics. By describing the inner works with more common terms we do have a bigger chance of more Catanians understanding what is going on.

2. There is a lot of unpack here but I´ll give it a try.
In your hypothesis you are trying to prove, you should be able to find one major flaw. If one player wins up to 3 players are losing and having a bad time. If there was such a thing as an advantaged player, 3 players would just have a bad time and they would be much more likely just to delete the game.
But without speculations, here is what we do:
The board:
The setup of the map is the so-called spiral method you´ll also find in the physical board game as well. Catan Universe had a more random approach in the past but Catanians asked for the spiral method because it produces more similar but more balanced board overall. So we implemented that.
Please look up the board game rules here: https://www.catan.com/service/game-rules. You´ll be able to find the random map method and you´ll be able to find that every base game board will be according to those setups. Please note: The ore for Wool scenario is excluded in this case because due to the “special task” a certain setup is needed.
The 7 /dice in general:
We just recently added a new feature to the game. It´s the “live dice statistics”. Now every player can track every player dice throws throughout the game. Why did we do that? Because Catanians wanted to know who rolled what. Is there any value in knowing that your opponent already rolled the 8 three times. Absolutely not ^^ But player wanted it and we gave it to them. If there was anything fishy with the RNG, why would we lay this info open like that?
Catan Universe doesn´t give any player an advantage when rolling the dice. The pseudo RNG isn´t influenced by anything happening on the board. It is a thing on its own. It works independently from any influence. The only thing that influences the pseudo RNG is the so-called seed. The seed is pre-determined before the game even starts. It’s a combination of different hard values we are in control off and that can´t be changed by any player for example server time and date. But the seed doesn’t involve any value connected to any Catanians profile.

Dev Cards:

The dev card stack is shuffled one time while the game gets initialized. It only includes a certain amount of cards. You can read up on the dev cards here as well: https://www.catan.com/service/game-rules.
After the first shuffle, the stack just sits there. There is no line of code in Catan Universe reorganizing the stack based on any event in the game.

AI:

The AI isn’t perfect but it doesn’t care about any Catanians overall stats nor does it have any “knowledge” about it. The AI that comes in after one player drops, isn´t the smartest one we offer in Catan Universe. So it will make mistakes by design. Those “mistakes” are solely based on not being super aggressive towards players in general and being more inclined to trade. If we had put in the strongest AI as a substitute, it might give players a challenge that is too big to overcome.

General statement:

Catan is a luck-based game where the portion of luck you need to win can be reduced by understanding how the game works … your skill. But skill alone won´t shield you from having bad luck from time to time.
But we can see that over the last years the same top players not only do well in international championships but also do well in-game events here in Catan Universe.
Our goal was to create a platform where you can play the game as true as possible to the board game. CATAN is such a great game, we do not need dirty tricks to get people hooked. We do not need voodoo magic to let people have a great time. The design of the board game is what makes this game special

Because you give the roll stats after each game i decided to keep track of them on excel because i noticed weird dice rolls (the most extreme example was not a single 8 in 40 rolls which happens 1.27/10000 or 0.0127%) and i concluded that the algorithm has one major "flaw".

The RNG tries to balance itself out in 4 to 5 five games depth by rolling more of the numbers that are missing to complete the gaussian bell. In more simple words, if you get, let's say, more 5s than expected in one game, the RNG will try to balance on the gaussian bell and roll less 5s in the next 4-5 games so the combined outcome of a certain number of games added up will fit as much as possible.

My hypothesis has some missing parts though, how many games the RNG keeps track on each account and it tries to do the fitting and after how many games it resets (or maybe it's a timer that decides that). Sadly i don't have any mathematical proof of that only my own observations with my own eyes and understanding. I can't tell you what numbers are going to be rolled in the next game but after 2 games i can tell which numbers are not going to be rolled so much on the 3rd just by remembering which ones were excessively rolled on the previous two.

[N o i r]

MICatan eredeti hozzászólása:

semantics. again. i wont argue with you about constructions like "clearly false" and "proof/evidence". I wont argue with you about the difference of the terms feeling and experience. there is no point in discussing lack of access to some required fundamental data to back up my claims, since i am not a member of the devs, who own this specific data.
no need for me to play intellectual ping pong with you, i would not stand a chance against your self-projection. i instead bow to your intellectual superiority.
have a nice day

Semantics? No, it's just a result of understanding how statistics work.

You and other posters wrote a bunch of nonsense, and instead of retracting when facing actually informed discussion you just leave and treat the discussion as a jest.

[Ice-Nine]

Atomic Priest eredeti hozzászólása:

Because you give the roll stats after each game i decided to keep track of them on excel because i noticed weird dice rolls (the most extreme example was not a single 8 in 40 rolls which happens 1.27/10000 or 0.0127%) and i concluded that the algorithm has one major "flaw".

The RNG tries to balance itself out in 4 to 5 five games depth by rolling more of the numbers that are missing to complete the gaussian bell. In more simple words, if you get, let's say, more 5s than expected in one game, the RNG will try to balance on the gaussian bell and roll less 5s in the next 4-5 games so the combined outcome of a certain number of games added up will fit as much as possible.

My hypothesis has some missing parts though, how many games the RNG keeps track on each account and it tries to do the fitting and after how many games it resets (or maybe it's a timer that decides that). Sadly i don't have any mathematical proof of that only my own observations with my own eyes and understanding. I can't tell you what numbers are going to be rolled in the next game but after 2 games i can tell which numbers are not going to be rolled so much on the 3rd just by remembering which ones were excessively rolled on the previous two.

That's literally crazy talk. There's no way in hell that some RNG algorithm is trying to balance out die rolls for each player. There are 2-5 other players in each game, and it would be sheer programming insanity to try and find multiple players who all needed the same die rolls to balance out their individual distributions. Nobody would ever be able to find a match.

Y'all just have very poor understanding of statistics.

[PikeStance]

N o i r eredeti hozzászólása:

Semantics? No, it's just a result of understanding how statistics work.

You and other posters wrote a bunch of nonsense, and instead of retracting when facing actually informed discussion you just leave and treat the discussion as a jest.

You have n such thing. In fact the responses have been similar to your first point.

[Atomic Priest]

Ice-Nine eredeti hozzászólása:

Atomic Priest eredeti hozzászólása:

Because you give the roll stats after each game i decided to keep track of them on excel because i noticed weird dice rolls (the most extreme example was not a single 8 in 40 rolls which happens 1.27/10000 or 0.0127%) and i concluded that the algorithm has one major "flaw".

The RNG tries to balance itself out in 4 to 5 five games depth by rolling more of the numbers that are missing to complete the gaussian bell. In more simple words, if you get, let's say, more 5s than expected in one game, the RNG will try to balance on the gaussian bell and roll less 5s in the next 4-5 games so the combined outcome of a certain number of games added up will fit as much as possible.

My hypothesis has some missing parts though, how many games the RNG keeps track on each account and it tries to do the fitting and after how many games it resets (or maybe it's a timer that decides that). Sadly i don't have any mathematical proof of that only my own observations with my own eyes and understanding. I can't tell you what numbers are going to be rolled in the next game but after 2 games i can tell which numbers are not going to be rolled so much on the 3rd just by remembering which ones were excessively rolled on the previous two.

That's literally crazy talk. There's no way in hell that some RNG algorithm is trying to balance out die rolls for each player. There are 2-5 other players in each game, and it would be sheer programming insanity to try and find multiple players who all needed the same die rolls to balance out their individual distributions. Nobody would ever be able to find a match.

Y'all just have very poor understanding of statistics.

I know it's crazy but you can test it yourself and notice some crazy dice rolls happening. I played over 1000 real games in the past 3 years with nothing close to what i saw happening in 50 games on steam.

Also by saying "y'all just have a very poor understanding of statistics" when i clearly used examples using statistics makes me think you are arrogant and dumb af.

[Ice-Nine]

Atomic Priest eredeti hozzászólása:

Ice-Nine eredeti hozzászólása:

That's literally crazy talk. There's no way in hell that some RNG algorithm is trying to balance out die rolls for each player. There are 2-5 other players in each game, and it would be sheer programming insanity to try and find multiple players who all needed the same die rolls to balance out their individual distributions. Nobody would ever be able to find a match.

Y'all just have very poor understanding of statistics.

I know it's crazy but you can test it yourself and notice some crazy dice rolls happening. I played over 1000 real games in the past 3 years with nothing close to what i saw happening in 50 games on steam.

Also by saying "y'all just have a very poor understanding of statistics" when i clearly used examples using statistics makes me think you are arrogant and dumb af.

You can "notice" whatever you want, but it's nothing but hokum confirmation bias. And no, you didn't provide any examples using statistics. You provided ONE example of a probability, but that's not "statistics" nor is it anywhere close to being useful or reliable information to support what you're trying to argue.

[Atomic Priest]

Ice-Nine eredeti hozzászólása:

Atomic Priest eredeti hozzászólása:

I know it's crazy but you can test it yourself and notice some crazy dice rolls happening. I played over 1000 real games in the past 3 years with nothing close to what i saw happening in 50 games on steam.

Also by saying "y'all just have a very poor understanding of statistics" when i clearly used examples using statistics makes me think you are arrogant and dumb af.

You can "notice" whatever you want, but it's nothing but hokum confirmation bias. And no, you didn't provide any examples using statistics. You provided ONE example of a probability, but that's not "statistics" nor is it anywhere close to being useful or reliable information to support what you're trying to argue.

Tell me what i did wrong.

-I kept all the numbers from dice rolls i got after each game

-Added them for each die and all together to get the averages

-Based on the total average dice rolls on each game and the real propabilities of each die i found the canonical for in the given # of dice rolls

-Ploted them together and NOTICED (cause i don't have a PhD in statistics but i know my s**t) that the numbers that were NOT rolled so much came MORE on the next one or two games af if the propabilities where conditional.

[afrothunder]

I'm trying to figure out how to buy this game outright. The only option I can find is to buy suns for pay for use options. Pppplease help!

[fearenough]

N o i r eredeti hozzászólása:

MICatan eredeti hozzászólása:

semantics. again. i wont argue with you about constructions like "clearly false" and "proof/evidence". I wont argue with you about the difference of the terms feeling and experience. there is no point in discussing lack of access to some required fundamental data to back up my claims, since i am not a member of the devs, who own this specific data.
no need for me to play intellectual ping pong with you, i would not stand a chance against your self-projection. i instead bow to your intellectual superiority.
have a nice day

Semantics? No, it's just a result of understanding how statistics work.

You and other posters wrote a bunch of nonsense, and instead of retracting when facing actually informed discussion you just leave and treat the discussion as a jest.

no most of us said there is a problem with rng that it loves to favour 2 numbers and miss out 2 numbers. i don't care about semantics or statistics just facts. oh and otherwise i think now other that dice devs have made a great game. sometimes dice roll well and i enjoy win or lose sometimes dice roll bad and i hate so much i have quit even when winning

[kubli3]

I think what should be questioned more than the programming of the game is the hacking of the game. The leader boards for cities & knights makes no sense. There is either developer abuse as users or there has been hacking involved.

[Deornwulf]

A few notes to throw some gasoline on the flame...

One - An RNG could generate the number "9" 1000 times in a row and one could argue that there is no way to know that it was not random while someone else could take serious issue with how well the RNG is working.

Two - Just about every single major computer gaming company has issues with RNG code. Some are more honest about how they attempt to tweak the code to be as close to a true RNG as possible.

If players are noticing extreme oddness in the small samples of 50 to 100 throws, declaring that at some point the numbers "smooth out" to an even distribution is not much comfort. Just as there would be some games with skewed number rolls, there should be games with almost perfect distribution.

[fellow_daoist]

I care not about RNG, from my experience it is busted, that is how I feel it like a player. I don't care if it is a good or bad algorithm. what I care is my user experience, and that my friends is ♥♥♥♥ most of the times. What I want to is an experience closer to the real game, which it has never been, is frustrating and disappointing.

Also, this is not the tread to talk about it but since I'm venting, what the hell, when is the pairing going to be fixed!?

The last 6 months I kid you not, most of my pairing is with bronze, silver maybe platinum.. between 950-1050 ELO. I'm around ♥♥♥♥♥♥♥ 1200. The amount of games dropped but this bad player is crazy. I end up playing another person (supposly) and 2 NPC. that is totally bulls%%**£. As you all know playing with the NPC is like dying to paper cuts. So not only I'm not enjoying any game, my ELO drops at about 12-15 points if I'm unlucky enough to finish 3rd.

Please fix the pairing. I can make my piece to the crap of Dice, put the former issue really kills it for me.

[MSISR]

My experiences so far

1. The AI ​​prefers one player per game. There are already many posts in the forum by players who had to make this frustrating experience over and over again. I can confirm this accusation from many games myself.

2. A ranking system like in chess is used. However, this is a dice thwowing game by chance in which I cannot control the result of the dice or the game. Since the game is predetermined by the AI, the ranking system does not reflect the skills of a player, but is also controlled by the AI.

3. The higher players go in ranking, the less points they get if they win, but the more points lose if they don't win. As a grandmaster, I have to win three games to balance my score if I lose one. The only strategy to succeed: create multiple accounts and cheat.

4. Like this game has frustrated many others, it only frustrates me because it is controlled by the AI ​​and every progress in the game is destroyed over time.

5. Unmoderated insults to other players are never punished, the support never reacts. You can insult any other player in the worst possible way, the support never takes action. A great playground for bullies and cheaters.

6. I deleted my account. Shame on KLAUS TEUBER for putting his name on that scam.

[PikeStance]

I have noticed that "7" is rolled a lot more times than before. I think there was one game where the "7" hit on 7 out of 8 rolls. The game also seems to get caught in a rut with numbers.

-----
On a side note:
1. "Rank" is still misspelled, it is currently "Rang."
2. There is a "shadow effect" where the game would indicate receiving a resource that will come in the next to rolls. It is done in a flash.

[fearenough]

the thing i find funny is take time in beta then 3 years at lest to get game play to run so now players can mostly finish games but they can make a perfect pseudo RNG on first attempt