Random Number Generator (RNG) - Not so random

You seem to think that randomness is dichotomic: either you can predict or it’s fair game.

That’s not exactly the case: a RNG (be it true or pseudo) can be unpredictable and still biased, i.e. you don’t get to know when a queen is next in a [constantly reshuffled] deck even if the deck has five queens.

We can concede that unpredictability is sufficient. Or not.


It would also be a sign of non-randomness. Randomness is typified by not being consistent.

Edit: HAH I just noticed the original post was from 2018. That player probably stopped playing long ago due to the inconsistent (non)randomness.

1 Like

Bad analogy, the dice are an independent roll (and not as random as you think). A summons is part of a (pseudo-)random string generated server side.

Your point being, exactly?
PRNG outputs being non-independent?
Dicerolls being theoretically modelizable under the complete information assumption?

My point was that a randomly “chosen” subset (of adequate cardinality) of [in wide-sense] random events should exhibit the same stochastic properties of the full set of random events.
I.e. the fact I don’t see all the PRNG outputs is not relevant as long as I see enough outputs and we can assume I’m not shown outputs based on any bias.
Because we are assuming that, aren’t we? :innocent:


No, i did not. Random is random. It cannot be on two polar opposites as you have worded: dichotomic. If it predictable, then its not random anymore, right? And the game would suck since you dont need to solve the puzzle of the boards as you can win all the time in raids and in wars. Some people here think that the board is rigged and that everything that is supposed to be RNG-based is not allegedly random. Well, if they can just predict the tiles on every opening board, i will tend to agree with them. So far, none has attempted to present a clear and convincing evidence that the board is not random and they mostly state it based on their personal bias.

By the use of the word “can”, are you insinuating that the board is also predictable? If so, by all means, enlighten me, convince me? Have you already cracked the code? Did you already discovered a third-party app able to determine the opening boards, the numbers and positions of the tiles at the onset as well as to the resulting matches or cascades? I would love to learn that and perhaps join the hoard of tin-foil hatters claiming that the board is not truly random.

My wife plays a game advertised here in Mystic Vision, something to do with a vacationing butler about home and/or garden. And it has boards and missions that is so darn hard that even if i used her special tiles, i cant seem to be able to complete the board missions. And she is angry with me for depleting those special tiles.

As far as I know, RNG means random numbers generator. If it is predictable, then it cannot be random. Once a solid proof is proffered, that would be the time i may start my journey in weaning myself on this game.

Assuming the numbers are randomly generated on the server side and then randomly distributed amongst the player base, if there is any non-random skewing you need to consider all the results. A randomly “chosen” subset of adequate cardinality would require many thousands of results; for an individual that is months and months of data points.

Or individual accounts are being targeted in which case it has absolutely nothing to do with the RNG, and again requires thousands of data points.

As for your questions,

They are until the seed is changed. Hence “pseudo”.

They are. However, any complete information model must include the observer and thus is likely unachievable. The very act of observing changes the initial conditions. Nevertheless, dice rolls and coin flips (like the weather) are a poor example of random.

1 Like

The ultimate problem with RNG is that it’s virtually impossible for any individual to be able to prove whether it is or isn’t working “properly”.

And even if thousands of people keep spreadsheet results on it and the overall results show that “yes, X percent of the time you get Y result,” that doesn’t mean that everyone is getting the same results. Some people might be beating the odds, while others are getting totally screwed.

Which brings us back to the original problem. Is SG intentionally screwing those people? Or is that just RNG?

Because in the real world, it’s not at all unusual for random chance to screw over random people for no reason while also ridiculously rewarding random people for no reason.

Take two unlikely opposite scenarios: winning the lottery vs. getting struck by lightning. The odds of either happening to any individual even one time are generally considered to be incredibly low. Yet, there have been multiple examples of people both (1) winning the lottery more than once, and also others of people (2) getting struck by lightning more than once.

As much as I would love to blame the game when I get a bad board or a bad pull, it’s probably just as likely that the universe itself has aligned against me.

EDIT / ADDITIONAL / assorted nonsense:

I wonder if I’m the only one who ever tried this. Have you ever played a coin flip game with yourself? Calling out what you think it will land on each time, compared to what it actually lands on, and tracking the results.

I have. And when I did, I made the wrong call over 70% of the time.

I was flipping the coin. I was making the calls. Obviously my calls are most likely to end badly.

So then I tried it again, only this time, I did the opposite of what I would normally do. If my instinct was to call heads, I would call tails instead.

Did it using that method, in the same number of flips. Again, I was wrong 70% of the time.

Which meant that my normal instincts would have actually been right 70% of the time in the second sample. But the coin decided to reverse itself just to spite me.

1 Like

Absolutely not.
I’m saying that the fact you can’t predict results doesn’t mean a random phenomenon is working as it should.

That is a distortion.
A PRNG is predictable even if none of us can predict it.
Hence this point of discussion needs a deeper definition structure, otherwise it’s nonsense.


I know a thing or two about random, models and identification :slightly_smiling_face:

My questions were more or less rhetorical :wink:

That’s great! When it comes to the forums here my default position has become that people don’t know anything about random.

1 Like

Just another anecdotal fact to add to the “you hardly lose because of random” pool.
I just barely lost to this team

and I just can’t think of no other reason than the fact my opponent’s Kelile and Scarlett dodged 6 consecutive times in a row combined.
(Li Xiu might have evaded too as I was using CTib, but I was more keen on the fast reds doing high damage and opening a lane for ghost tiles)
Chances of every single event happening: 20%.
Yet again we are in the single digit chance in 10’000 territory.

We can assume that’s bad luck.
But we can also assume that conceiving a skill that out of a diceroll - i.e. nothing the player can really do about it - can take away thousands of damage is poor design.

N.B. as per usual this was the first day I was playing in tournament because… because, really, enough of the very unlikely events already.
Especially when fighting dynamics take away the chances to stall and counter.

Random, again.

Fast tournament, red forbidden.
In around 30 rounds I haven’t been able to charge a purple.
That is basically TWO useless colours.
(N.B. I switched to stacking purple because in the first four days I struggled with green)

Then you face a Proteus that blocks another colour managain and you are doomed.

Now, let’s face it.
For whatever reason E&P exposes you to colour shortages - there’s really no other way to describe impossibility to connect 9 tiles of one colour in at the very least 125.
In this type of tournaments if it hits one of 4 colours you can bring* you are basically doomed.

Also… It would be nice if the most overwhelming skill for such tournaments (no mana gain) could be found in an ordinary hero. Like really ordinary. One you could get by stumbling on RNG enough times without postponing your renovation.

  • truth be told the forbidden colour easily translates in almost two.
    Let’s say red is forbidden.
    Tank (if not others) will almost certainly be green so by bringing blue heroes you compromise your damage and are exposed to high damage from the defending team.

In the very same tournament I had only one defense on the first day. It was a loss.
The second day 1 win and 1 loss.
Third day I was on par at 3-3.
Ever since my defense has only won, i.e. it has theoretically win all the matches as they were getting harder.
I know this has already been discussed but… again, it’s influenced by “random”.
How about we cut a little bit with the imponderable nature of the game instead of enhancing it, no matter the fairness of RNG?

Because, you know, a universally accepted way of reducing RNG-related problems/discussions/innuendoes is getting rid of it, not embracing it.

1 Like

I’ve decided to revive this old thread, because more and more I’m feeling that RNG is not so random.
Is it a coincidence? Maybe. But it has happened more than once and I’m getting suspicious.

I would like to share my experience in summoning heroes. Usually I have enough heroes of each colour to work on, so I always work on 5 heroes at the same time. A few months ago I run out of blue heroes, I finished all rare, epic and legendary heroes. Maybe just a week passed and I got my first Magni form TC20, Junior Chick from BF portal and Vodnik from HA5. Yesterday I finished Magni and again I had nothing to work on. Today I did two pulls and got Captain of Diamonds. Why him, not Hansel or Peters, which I don’t have and I have too many green heroes than I can ascend? Similar thing happened to me with red heroes, when I finished all heroes, their costumes, I LB some of them and then I got Khagan and Bauchan. Maybe it’s silly (and probably is), but it’s funny.

It feels that SG is spying on me and to keep me interested they raise chances of pulling a unique hero of a specific colour. :upside_down_face:


Actually adaptive PRNG is an issue that many governments who are scrutinizing gacha consider critical: as a matter of fact there is no proof they are not reality.

Whatever the forum maintains as true, pure PRNG is a bad business choice and small adjustments could be possible without an actual chance for any single or small group of players to ever have a chance at aggregating enough data on customized experiences.

As for myself I believe I’ve noticed off things about the PRNG and while I get that this is not proof, I have a condition that makes my feel for numbers very above average.

But this is not a discussion for a forum anymore: we should start considering “mandatory third party certification of PRNG in any environment that makes money out of it” something to be brought to the attention of regulators and app store certificators.


I couldn’t agree more and thank you for posting this.

1 Like

I like that argument about third party certification. Good luck pulling it off and regulating it in an international market.


Well… You only need to convince a certain company with a certain fruity logo :wink:

Fruity company only hosts the app. They do not own or manage the app.

Fruity company will only cancel hosting if App Owner pulls out or it is a violation in particular jurisdictions.

Yup, but when China decided that quasi-slot machines without explicit percentages were a no-no fruit of the doom forced every app that wanted to stay on its store to comply.

As a matter of fact in this instance stores policies can be more powerful than country regulations.

Exactly what I meant. Jurisdiction dictates. Fruit company will push buck to App owner.

If app owner refuses to comply, then no hosting.

1 Like