Dice Algorithm Sucks


  • administrators

    @MonkeyZ We do keep track of the dice rolls and I can assure you that the RNG is working fine.

    If you look at all the theories out there on how the RNG might be broken, you will see based on the different testimonials.... it really is random.

    It is simple and straightforward. We didn´t tamper with it nor are we keeping track about single games and where people might have positioned their settlements or cities.

    We keep track of the overall stats.

    Random rolls are an important part of playing Catan properly. If you take away that element of "luck" you will lose a big part of the fun gameplay elements and the spirit of Catan. That is why we do not offer card stack (dice mode) for our more competitive game modes.

    All digital games using some kind of RNG get at some point accused of tampering with it to their advantage. But for as long as I am working on Catan projects no one could ever tell me what advantage we would gain from using a broken or tampered dice RNG :)

    Being "yelled" at because of it all the time ... isn´t really fun nor an advantage ;)

    In my personal opinion, it´s the "stats" that will drive the people to the conclusion that there is something wrong with it. Playing with your family and friends you will most likely not keep track of each dice roll. And if you lose despite the best positions on the board someone will probably console you. And tell you have been out of luck.

    Playing online you don´t have that. And the only thing you can blame for the loss is the: "broken RNG" because it has cost you the win.



  • @Administrator said:

    If you look at all the theories out there on how the RNG might be broken, you will see based on the different testimonials.... it really is random.

    Sorry Admin, this is not a serious evidence and I think you can give us better Information.

    Any RNG on regular programming languages using their standard libs are NOT random.

    If you say, that the RNG, that catan universe is using for:

    Dice,
    Mixing Cards,
    Board Layout,
    Number Generator (except the spiral mode)
    Start of Number at spiral mode
    and any other randomness

    is really random,

    it means, that these numbers are NOT generated by any computer based algorithm!

    So you are using already a system or service like random.org which provides a random number series
    created by some kind of physical properties?

    Quote from random.org(e.g.):

    "What's this fuss about true randomness?

    Perhaps you have wondered how predictable machines like computers can generate randomness. In reality, most random numbers used in computer programs are pseudo-random, which means they are generated in a predictable fashion using a mathematical formula. This is fine for many purposes, but it may not be random in the way you expect if you're used to dice rolls and lottery drawings.

    RANDOM.ORG offers true random numbers to anyone on the Internet. The randomness comes from atmospheric noise, which for many purposes is better than the pseudo-random number algorithms typically used in computer programs. People use RANDOM.ORG for holding drawings, lotteries and sweepstakes, to drive online games, for scientific applications and for art and music. The service has existed since 1998 and was built by Dr Mads Haahr of the School of Computer Science and Statistics at Trinity College, Dublin in Ireland. Today, RANDOM.ORG is operated by Randomness and Integrity Services Ltd.
    "

    So, Admin please, can you confirm, that catan universe is using a service like this for RNG an NOT an algorithmic based RNG?

    I think a lot of people would be happy if you can confirm this.

    (In case you are NOT using a service like this, please accept, that the dissatisfication of the crowd is entitled and you should change it in this direction)

    CETERVM CENSEO LVDVMCATANEM CONSERVANDAM ESSE

    1


  • Just wanted to confirm that I have seen this strange numbers lately very often in "the first island" game (especially after the last upgrade). It was suggested in another thread, publish all the statistics from all games for dice numbers to see how the curve looks like.

    I have a picture from a recent game and 9 was rolled 20 times while 5 was rolled 3 times. I play catan in real life often and most of the time the curve is in normal parameters. It happens sometimes to have a few more 3s or 11s but as an exception.

    3 and 11 are favourites lateley which is odd. Amd as somebody else said you can see from the begginign if it is a "5" game or "8" and so. I have also noticed that 8 is much better than 6 and I would be curious to see overall how the totals look like.



  • How most computer generated RNGs work is that they need a "seed" and they will always generate the same numbers based on the same seed. We have seen bugs where the seed has always been the same so the numbers rolled are the same throughout the game. The fact that this has already happened makes me personally suspicious about the nature of the RNG you have selected.

    Is the "RNG of the seed" random enough? How is the seed generated? Maybe there are like 10 or 100 different seeds to choose from?



  • @Stroom said:

    How most computer generated RNGs work is that they need a "seed" and they will always generate the same numbers based on the same seed. We have seen bugs where the seed has always been the same so the numbers rolled are the same throughout the game. The fact that this has already happened makes me personally suspicious about the nature of the RNG you have selected.

    Is the "RNG of the seed" random enough? How is the seed generated? Maybe there are like 10 or 100 different seeds to choose from?

    A random Seed is an initial value for a RNG which will offers the same number sequence every time.

    https://en.wikipedia.org/wiki/Random_seed

    in productive areas the aeed should NEVER be set manually. It is just set at start (e.g. by the actual time) and after this it is never touch - otherwise you could produce same number sequences.

    Only for test purposes or using random values for crypt/decrypt you can set it to a Special value.
    But if you want to use the rng NEVER set it.

    And as I wrote above: NEVER use RNG implemented by a computer algorithm.
    Use physical properties which guaranties real random numbers.

    CETERVM CENSEO LVDVMCATANEM CONSERVANDAM ESSE

    0


  • @Administrator

    Könnt Ihr hier einmal bitte antworten?
    Welcher Zufallsgenerator wird in eurer Software benutzt?
    Diese Frage dürfte doch einfach und schnell zu beantworten sein.
    Einfach mal den Programmierer fragen.

    So ein wichtiger Bestandteil der Spielmechanik sollte dokumentiert sein!

    @Honie said:

    @Administrator said:

    If you look at all the theories out there on how the RNG might be broken, you will see based on the different testimonials.... it really is random.

    Sorry Admin, this is not a serious evidence and I think you can give us better Information.

    Any RNG on regular programming languages using their standard libs are NOT random.

    If you say, that the RNG, that catan universe is using for:

    Dice,
    Mixing Cards,
    Board Layout,
    Number Generator (except the spiral mode)
    Start of Number at spiral mode
    and any other randomness

    is really random,

    it means, that these numbers are NOT generated by any computer based algorithm!

    So you are using already a system or service like random.org which provides a random number series
    created by some kind of physical properties?

    Quote from random.org(e.g.):

    "What's this fuss about true randomness?

    Perhaps you have wondered how predictable machines like computers can generate randomness. In reality, most random numbers used in computer programs are pseudo-random, which means they are generated in a predictable fashion using a mathematical formula. This is fine for many purposes, but it may not be random in the way you expect if you're used to dice rolls and lottery drawings.

    RANDOM.ORG offers true random numbers to anyone on the Internet. The randomness comes from atmospheric noise, which for many purposes is better than the pseudo-random number algorithms typically used in computer programs. People use RANDOM.ORG for holding drawings, lotteries and sweepstakes, to drive online games, for scientific applications and for art and music. The service has existed since 1998 and was built by Dr Mads Haahr of the School of Computer Science and Statistics at Trinity College, Dublin in Ireland. Today, RANDOM.ORG is operated by Randomness and Integrity Services Ltd.
    "

    So, Admin please, can you confirm, that catan universe is using a service like this for RNG an NOT an algorithmic based RNG?

    I think a lot of people would be happy if you can confirm this.

    (In case you are NOT using a service like this, please accept, that the dissatisfication of the crowd is entitled and you should change it in this direction)

    CETERVM CENSEO LVDVMCATANEM CONSERVANDAM ESSE

    1


  • @Administrator I don't care that this comment is old, I love the game but RNG is still Broken!!!! Consistently getting games where a a number like 5, 6, 8 and 9 will not come up, games where the frequency is only 1 or 2 in a whole game! have been common in my experience. I have more chance than winning the lotto with these statistics. RNG is BROKEN!



  • @Ned-Kelly I'm uninstalling



  • @Administrator I completely agree that RNG is working properly & most complaints are people just not understanding the sample size but I don't understand your POV on dice roll RNG being essential to gameplay at all. Catan has RNG in the variable board layouts, your card draws & the order of dice/stack results already, choosing which hexes to develop based on your needs & their combined probabilities IS THE SKILL GAP part of Catan. USM's Catan Classic app included 2 stack options that worked really well IMO especially for Cities & Knights which has deeper strategy while the stack algorithm in Catan Universe doesn’t even solve the issue it’s supposed to unless the Almanac description is inaccurate about it reshuffling the same 36 tiles so this is a major step back in functionality that forces me to stick w/ the old app & prevents me from recruiting more gaming friends to invest in Catan Universe for now. I've explained it all in the review I left in the App Store copied below which I look forward to updating to 5 stars when a future update has better "Stack" functionality.

    ——— 3 star App Store review ———
    I tried this again after all the improvements since its release when I found its gameplay far too slow compared to the Catan Classic app. I really wanted to like it given the community potential w/ guilds, chat & ELO and the gameplay was much improved but its "Stack" mode is too limited for competitive play IMO. You can't activate "Stack" mode unless you invite a friend so it’s not even an option for your usual quick auto-match games & even if you do the implementation of "Stack" as explained by the Almanac is far worse to the simple but effective "Full Stack" & "Stack minus 5 random tiles" options from Catan Classic. What's even the point of tracking ELO in a game driven completely by RNG dice roles??? Probability is a central aspect of Catan and Cities & Knights has ample strategic depth but all of that is wasted when you can't normalize the dice rolls to some approximation of expected probabilities via decent "Stack" options. Me, my potential revenue & all the friends I recruit to play will be sticking w/ the Catan Classic app until Catan Universe supports "Stack" at least as well as that very old app :-/



  • Agreed. Some people really like the stack modes.



  • There are hundreds of other online board games out there that do not have such a huge luck factor. If you don't like Catan, go play something else. Boardgamearena has a lot of fun games. Tzolk'In is quite challenging for instance.

    Even if you get a normal dice distribution, it might not matter because the order in which the results appear could still make or break the game. It's completely fine.



  • @Stroom I don't understand how having an option you can simply choose to never use is a problem? If you like more RNG in your game no one is forcing you to play with a stack algorithm, keep enjoying your games w/ the standard RNG dice while those of us who enjoy stack modes create our own games in that format. I doubt matchmaking will be that much slower just by offering 1-2 stack options in addition to standard RNG but if it is you can always speed it up by choosing to play regardless of the dice mode...

    You are correct that every game has a mix of strategy, skill & chance but your pt about the "roll" order proves there will always be chance even in a "full stack" game mode. Catan and especially C&K have too much strategy & too much chance (card draws & roll order) that it will never be just a matter of who gets the best tiles even in "full stack" games — stack modes just reduce the number of games where you never even had a chance to win bc you were completely starved of resources relative to other players (& relative to what probability expected for your tiles) by RNG. The issue @Developers don't seem to grasp here is that while RNG balances over the long run in any 1 game it’s going to be significantly out of whack w/ probability pretty frequently, an annoyance which people will remember even more frequently bc of cognitive biases which then discourages them from playing the game, which leads to longer queues, less revenue and on & on :-/

    I do realize that RNG dice is exactly like playing w/ real dice in a table top game but that doesn’t mean you shouldn’t offer other options. If I was playing for ELO on table top Catan I’d use a physical stack anyway! Just to reemphasize, I’m only proposing that "full stack" & some "stack w/ some RNG" OPTIONS be available, again you would never be forced to play them if you don’t want to! USM clearly understood this at some pt bc their prev Catan Classic app included options for "Stack" & "Stack minus 5 random tiles" in their game settings which both work better than the stack algorithm that is only available in custom games w/ friends in Catan Universe :-/



  • To make different options work, they should throw away the matchmaker. If you add so much customization to the game, you divide the player base among a lot of decisions so finding players who want to play the options you choose will become pretty much impossible. If there would be a lobby, yes, choose whatever you want. But the admin has said that they will not even consider it.

    Let's say we have 10000 people online. How do we divide the player base?
    Currently 4 modes: base game, Seafarers, C&K, Seafarers + C&K (And Incas but let's be fair, no-one plays that). So, 2500 players per mode. (Yes, they are not evenly distributed. But that even further means that some mode will be harder to find a game for).
    Now, you want people to be distributed to games based on their elo (although that is not a good metric overall). Maybe 5 groups. Within 300 elo or something. That means 500 players per that combo.
    Now, there is already friendly robber that divides everything to 2 groups. 150.
    And then there's card stack or normal dice. 75.
    Now, you have 75 players to choose from who might want to play with you. Some of them want to play with 3, some with 4 players. Finding a game usually takes 2 minutes but the game lasts 50 minutes. That roughly means that there are 2-3 players online who would want to play with you at the same time.
    And now, people really want 5-6 player modes and other expansions introduced. That would completely ruin all the calculations and divide the player base even more.

    The card stack is not "fair" because you can track what numbers have rolled less so you try to build towards these resources. It completely changes how you play the game. "Oh, 6 has been coming up so much. No point placing the robber there because it's not coming up anymore." Similarly, if the robber has blocked a 6 a lot of times, the player is completely ruined and he knows that the number is not coming up anymore. People who want to control the randomness do not understand how it does not work.



  • @Stroom Have you actually played many games with either Stack or "Stack-5" on any of the apps/sites that support? I've yet to find a regular player who thinks it worsens the experience especially for C&K so maybe try it before criticizing it even being an option.

    Seeing as I regularly play on these settings I can say w/ certainty that it does not completely change how you play the game. Full stack is only 36 "rolls" so there will be shuffle(s) based on the game length & if you include any RNG like w/ "Stack-5" then you can't know exactly what's coming next. I do prefer some RNG in base Catan since the robber & Monopoly cards can really swing the game so "Stack-5" works better than a full stack there IMO but being able to make better Robber & Inventor decisions in C&K is part of the skill gap between players so any stack option is better than RNG dice there. Why even play Catan if you just want dice rolls to determine the outcome, go play craps or roulette xD

    A limited number of options shouldn’t hamstring matchmaker too much but since RNG dice makes ELO pretty meaningless until you’ve played 100s of matches I’d be pretty OK ignoring that if queue time exceeds a personal preference. Faster queues vs better matched opponents is something that can be tuned & can also be dealt with via personal preferences so I don’t really see an issue there. If the problem is trying to get closer ranked opponents to stabilize ELO I'd argue you just fooling yourself as ELO is kind of a joke for RNG dice games until you've completed 100s of matches when skill eventually overtakes chance, just label people Beginner/Intermediate/Expert & call it a day. I mean how much RNG is there in chess?



  • @acallan18 said:

    A limited number of options shouldn’t hamstring matchmaker too much but since RNG dice makes ELO pretty meaningless until you’ve played 100s of matches I’d be pretty OK ignoring that if queue time exceeds a personal preference.

    Limited number? I just showed how 10000 online players can be reduced to 75. And even that is probably an optimistic number. Now, if you add more options, each halves that number even more.
    Yeah, I guess if people want these options then it would be fine. Implement that stuff. I'm just saying that it will not really solve the problem, it just makes it less obvious. Yes, all numbers came up more "evenly" but the order in which they come can be gambled. I already showed how it can be done.
    With so many options we really would need to get rid of matchmaker and use a lobby. So you can actually coordinate the game and choose who you play with. Maybe you know that one of the players hates you so you do not want to play with him... Ignore and block list actually has a limited number it can hold. 200 or something. If you play 100s of games, that will fill up quite fast with this community.

    I've always said that ELO is a meaningless metric for Catan. It would be better for the community if we didn't have that. It just causes players with lower ELO to target players with high ELO because then they will win more points. It has been reported to be a thing.

    Faster queues vs better matched opponents is something that can be tuned & can also be dealt with via personal preferences so I don’t really see an issue there.

    Have you seen any "tuning" of the matchmaker? There's constantly problems with being in a game with players who just are not as good as you.

    If the problem is trying to get closer ranked opponents to stabilize ELO I'd argue you just fooling yourself as ELO is kind of a joke for RNG dice games until you've completed 100s of matches when skill eventually overtakes chance, just label people Beginner/Intermediate/Expert & call it a day.

    100s of games of Chess is much faster than 100s of games of Catan.


Log in to reply