Lorem ipsum dolor sit amet, elit eget consectetuer adipiscing aenean dolor

Gems of Wars RNG is currently broken

… But because it’s RNG. It’s impossible to illustrate let a lone prove in any justifiable way. So the devs can seriously look into it and make the game better and fairer for those who bother to still play/support them.
Or dismiss it as:

Don't click on me if you're sensitive to flashing lights/colors

giphy - 2020-11-03T152936.461

(I picked that gif that’s hard to see because the current broken code is hard to see game play wise.)

Perhaps it’s an issue with Unity though? :man_shrugging:
All I know is too many RNG streaks are happening in multiple sectors. And not the good kind…
tenor - 2020-11-03T153154.927

I could list them all… But essentially it wouldn’t serve any good.
Those who believe me will take my word on face value. Those who don’t would argue that the sky isn’t blue…or the world isn’t round. And if a Dev reads this… They can look at the code or not. But my complaints not just about the RNG streaks happening but the frequency of streaks occuring would be dismissed as “RNG streaks”. Because due to the nature of RNG, even if it’s broken, it’s impossible to tell without seeing the code. :man_shrugging:


Just curious… what were you trying to go for?

1 Like

I would argue that it is very much possible and even not that hard to determine if a pRNG is not producing random output. There are multiple way to do that with pretty high reliability. But seems like you already made up your mind, so think as you wish then. :wink:


The RNG has been “broken” since game launch. The devs (read: Sirrian) either do not understand the math involved, or it’s skewed on purpose to give the AI a better chance. I’ve never been quite sure which it is. I don’t think Unity has anything to do with it, because I don’t think they’re using the standard Unity RNG calls, they’re using custom code that was written before the switch to the Unity engine, at least on the PC platform. Based solely on gut instinct, I think the Switch platform has better RNG, because in that case they’re forced to use the Unity calls.

1 Like

Just sayin’, the current iteration of the game is written in C# or another .NET language, and the .NET library of course comes with built-in pRNG. The move to Unity was definitely a complete rewrite (or actually a port of the console rewrite).

Whether they use the built-in pRNG or not is another matter, of course. But complaints about the biased RNG in the game (or any game, really) are as old as the game itself. Humans are bad at evaluating randomness; without a statistical analysis, this is just anecdotes. The plural of “anecdote” ain’t “data.”


Lyya, You’re going to dismiss personal observations from someone who knows the game better than most of the devs lol?

lol, but as you stated very well humans are bad at evaluating randomness which has actually evaluated to be a real thing unlike bias in RNG. To illustrate this point a little better, think about all times RNG was in your favor? In most cases you don’t think about RNG favoring you because you won’t have a visceral reaction when it does. Most players play fast and only focus on bad RNG when it hurts them. A player will lose two games in a row to skull cascades and think the game is cheating, but forget the same streak they won due to the same RNG.

There are also a couple of variables players forget about RNG. First off they over evaluate how good they are at the game and the gameplay choices they make. Humans are imperfect and it’s most likely one of the ways the devs catch cheaters that use overlays. They can look at hundreds or thousands of player matches and see abnormal patterns. Players aren’t perfect, even though some players believe they are.

Secondly, the player has all the advantages against the AI. Players always cry foul in GW, delves, pvp based on RNG, but seemingly forget they always get first turn, have access to the same quick start troops(empowered converters) and can take advantage of predictable AI behavior.

Which brings me to the third point, the AI is actually predictable to a degree. Players use this to their own advantage all the time. For example skull baiting. taking mana on the board knowing the next turn the AI will almost 100% of the time take the skulls. The order of operations for the AI’s skills can be predicted in some scenarios and used to your own advantage. You can even use their own empowered converters against them as they sometimes fire off with no 4/5 matches on the board, setting you up perfectly.

So no, RNG is very likely not off. OP’s misperception, understanding, ego about game mechanics, and inability to accept the fact that you the player gets many advantages the AI doesn’t offsetting any perceived bias, is likely the culprit to “RNG is wrong!” assessment. But than again, he does know more about gems than the devs so he must be right.

1 Like

I didn’t even say what I thought was broken about it. But yep. I must be wrong. Because it can’t be broken or bugged in anyway. Impossible when everything else about the game works so well.
But because you have to be right… For your own ego strokes. I must be wrong. Weird.

They can’t even flag items to be released when they are supposed to be, but we are to believe something as intricate as an rng algorithm is working fine?

Are you kidding me?

1 Like

If I may suggest a constructive and informative approach to settle at least some of the issues. You can state what you think may be broken, and I can try to suggest, if possible, a reasonable and acceptable way to test it, which should produce some data that can be analyzed with minimal bias and sufficient reliability and power. This should give a general idea of the amount of data that needs to be collected. If that amount and means to collect it are reasonable, you or other players can try to collect the data based on the requirements. I then will try to process it to determine whether the data are random or not sufficiently random and to estimate possible bias. It might be also possible to evaluate independence, skewness, and streakiness depending on specific dataset. The latter three parameters may be more complex to assess. However, randomness and bias are usually straightforward and require a limited dataset.


Then I refer you to the OP.
The abundant amount of streaks in the game. Anywhere RNG is. I am aware that’s a lot. Some streaks… Makes sense.
Seeing streaks with Legendary Tasks, with devours, with chaos portals, with dodge rates…ect… That’s second level. I realize you’re trying to help. But many hyenas are salivating at the mere mention of details so that’s all I’m going to say on the matter.
Those who have access to the code, can easily run tests to see if there is an issue. Players will have a tremendously more difficult time. You can’t see in a matter of a hour or two what I’ve noticed over the course of weeks.

So perhaps in the OP instead of saying it’s impossible for players to test all these factors. I should have said… It’s unreasonable.

If the code is sufficiently complex and has numerous interactions, it is actually easier to determine the overall randomness/streakiness of the result. Some interactions, while being pretty straightforward based on the code, may have a number of unintended consequences due to multiple interactions with non-random or simply heavily skewed factors. Moreover, keep in mind that even if a simple single pRNG output may be random, a combinations of conditional pRNG outputs does not automatically mean that it is random. Hence, I would argue that the end-product users are uniquely qualified to test the overall randomness of totalled output because the analysis of the code is not helpful from a statistical standpoint in this case.

What I mean to say is that developers are happily thinking that everything is random and declare so. However, complexity of interactions of multiple random conditional outputs does not necessarily yield anything even remotely resembling random. That is one of the reasons while multiple games that use random mechanics have specific tests and intermediate random processing to endure that the declared randomness of the output actually matches the description. It is more complex than actually seems even to a seemingly competent developer. There is rather complex math to substantiate these statements. I can provide some references if somebody is actually interested. It is heavily reliant on general multidimensional theory of measurements and probability density functions. Actual implementations are not necessarily exceedingly complex.

I typically don’t give up easily. So, IMHO, it is always worth a try. Baseless declarations will always be less convincing than some partial assessment if complete assessment is unreasonable. Also, making something that seems unreasonable at a first glance into a reasonable approach is a challenge, which makes it actually interesting. So, I am willing to put some effort into this as it is a creative problem. Data collection may be mundane though, but it should not normally take more than a few hours and may be possible to split between few participants once you know what is actually needed.


You can collect all the data in the world.
The devs will say yep “that’s rng being streaky”.
Then they’ll decide if they think it’s happening too much (which I believe it is = broken) or it’s not.
But essentially it wouldn’t be up to you, nor I. So you’re welcome to take a set of 300 legendary tasks…see how many in a row share the same exact reward and amount. Then how many times a troop like Mimic devours on its first cast in the arena.
Back to back dodges for the AI or the player in any mode. I’m sure that’s an excessive amount that I haven’t even noticed where RNG streaks are happening at above average frequency. Hell even an an abundance of Clan orbs from Chaos Orbs from the Vault a month or two a ago.

Again (because others besides Akots have difficulty reading) I am aware that RNG streaks… It’s the amount it’s streaking that’s problematic and raising flags. And since it’s not just isolated to chests or Legendary Tasks.

Plus this isn’t even a bug report or me asking support from other players. It’s essentially an opinion piece where I request the devs look into something.
Yet, even that is somehow “not allowed”.
If only some individuals put in as much effort to raise others up as they do stalking threads waiting to bring others down… Then perhaps the community and the game would be in a better place. :man_shrugging:

I won’t be closing this thread, but please remain civil in it.

We had evaluated our code many times and the Ai does not favour the computer. In fact, at lower levels, the AI purposefully favours the player.

There is nothing to be gained by having bad RNG that hurts the player. It would destroy our bottom line, cause a huge loss in players, and be overall bad for the long term health of the game. As such, we have evaluated it several times, as I mentioned earlier. There is nothing to be gained from us having rigged (in the favour of the PC) AI or RNG.


Not sure what you are looking for here AW.
If the devs run a check and confirm the RNG is not broken, how many of us will believe it more than before?

Personally I feel the RNG is less streaky than a couple months ago, but this is only from one person who never tracked data on this stuff.

1 Like

Well, there you have it, TC…

Thank you for the reply. You said the RNG isn’t favored towards the AI, but could you also confirm that the RNG is pure RNG and not designed to “streak”?

Not necessarily in favor of the AI, but I’ve noticed several times where things that have a ~10% chance of occuring have streaked 6+ times in a row. It’s definitely not impossible, but these streaks should be incredibly unlikely yet occur often enough to be noticeable. Whether it’s me dodging 7 skull matches in a row or watching my 96 attack Gut fail to devour 4 times, the RNG definitely feels like it’s designed to streak.

1 Like

It’s not designed to streak, but streaks aren’t that uncommon when looking at RNG as a whole.

Also, at lower player levels we actively STOP the RNG from streaking too much in the AIs favour when it comes to gem matches. (This pertains to waterfalls, and board control.)

I’ve asked the devs about percentage chances for spells, spawns, etc. EDIT TO ADD: The RNG is not DESIGNED to streak, but my guess is that there aren’t measures in place to completely stop streaking in these circumstances like there are for gem cascades at lower levels. In this case that would make it pure RNG. We haven’t purposefully made it streakier, as that wouldn’t benefit us or our players.


So there are methods in place to manipulate RNG.

I have mentioned this before, it isn’t something new. Quite a while ago we did this to make the game easier at lower levels.

I want to re-iterate that we have at no point made the RNG favour the AI, as doing so would negatively impact our game. Not only is there no reason for us too, it would actively harm the long term health of Gems of War.


Appreciate the time and effort you’ve taken to discuss this. Sadly its comments like [quote=“Tabu, post:18, topic:66918”]
So there are methods in place to manipulate RNG.
[/quote] that prove to me players don’t want to be told the truth, they just have cognitive dissonance when it comes to gems. They are looking for evidence to fit their narratives.

For example our guild has completed 1,000’s of LTs since 4.7 and have received 5-7 mythics and one week we got 3. So does that show a nerf of LTs? No, it shows its anecdotal evidence. Without looking specifically at the code people will look for evidence to fit their narrative to explain what they perceive as something being fixed against them.

Do I believe its all perfect RNG? No. But im not a programmer. I’ve been in this game as much as any of the longtime players here have and across multiple platforms. The only difference to RNG gameplay is my team builds and ability. Players complain GW is rigged RNG all the time. Except when we have players stream battles to help them and slow down their decision making, they get better and RNG plays less of a factor in wins or losses. More proof that players control more of their destiny in gems than RNG.

I know I’m seen as the enemy to the players here who believe they are fighting the good fight against devs and a game they perceive is cheating them or broken. Gems isn’t perfect and critism of the developers design and update choices are warranted. But its gotten to a point where certain players here just want to nit pick daily for the sake of nit picking. And from their own nit picking they perceive that as the devs doing it on purpose or that they are incompetent.

I will say It once more. I don’t side with the players or the devs. I side with facts. Fact is, there is no proof rng for the mid to late game is off. Devs have confirmed this is the case. If you then don’t believe the devs, stop asking for them to address it and just go on believing your own truth. You do you. But understand you are actively ignoring facts to fit your bias. The burden of proof is then on you to prove something is wrong with the code.

EDIT: I know this is off topic and I apologize, but I don’t want to make a separate post. Thank you @Saltypatra for combing through all the posts recently to report and comment on bugs which has always occurred regularly. I’ve had to watch these forums divulge into chaos and hysteria over insignificant things but just because I perceive them as ridiculous, doesn’t mean those players don’t deserve responses.

Maybe I’m just niave, but I believe if you communicate respectfully and present evidence to your points good discourse can occur. As for me personally, im just matching the level of the person im replying to lol. I’m not a dev and I don’t have to be filtered. If I get flagged so be it. But at some point the players here have to moderate the toxicity.