# Random Number Generator (RNG) - Not so random

I’m sure people will mock me, i have already been dubbed the Oracle, the colour whisperer.
I spotted a series of patterns on the RNG… One of them being ’ clumpiness ’ as it was described by the OP.

I won’t post my system again on this thread , the first time i posted my system on these forums which is based purely on observation, i was called a cheat…i won’t even go into that ridiculous comment and the following 1000 posts that ensued, but if you find my original post on my system for predicting colors its on these forums.

The short of it is, the PRNG tries to mimic the law of averages, and in doing so it tries to self correct by knowing what tiles have passed.
Now link this to my theory that boards are not individual, it’s one long conveyor, that’s how the PRNG knows what has been in the past and can auto correct… Thats also how if you do 100 raids the colors almost codes out perfect at 20% each… People on these forums thoughts that was proof of random… Actually it’s not, it’s proof of pseudo random.

Anyone who knows simple maths knows what this means and can work out the rest.
But finally, if i can predict boards with high accuracy, how can it be true random.? it’s not… It’s pseudo random and the algorithm which tries to replicate random is fairly predictable.
You can still of course get the odd freak anomolies, there is still some random to the none randomness.

3 Likes

It’s an interesting theory, and one tbh I’m not entirely sure why the previous threads on the topic of pRNG got closed… But oh well, that’s done now.

The fact that sgg use pRNG is widely documented… I mean it’s impossible for programmed algorithms to create perfect randomness.

The concept of it being “self correcting” is different, I’ve heard people make that claim but I’ve never really (A) paid enough attention or (b) been able to make use of the theory in the past.

Anyways, just a mod popping by to give a star and say thanks for prompting me to have a bit of a think.

• Do you think that the pRNG algorithm is still self correcting? I know sgg used to use a common seed for up to 20 minutes for device side RNG; so that could definitely have been reduced.

• Can you still make use of your “method”? If so how successful would you say it is? 9/10? 19/20? 99/100?

2 Likes

Hey Guvnor
I will repost my system , i don’t think it ever got removed, it got buried under a 1000 replies by a certain individual.
My pre attack system for wars is based on how barren the boards are of a certain color, the more barren, the more it increases the chance of it being in abundance very soon. I.e that’s of course with my theory that the algorithm tries to self correct and it’s a conveyor.
I would say 19/20 it helps me avoid those really really bad situations.

On the 1/20 where a barren wilderness is followed by more barren wilderness, i go on the theory that the clump of my tiles is very near and i really look for a big combo.
Everyone in my alliance will swear i get 6/6 every war for as long as they can remember. Although the only that can seem to replicate my system with the same success is my wife.

We noticed a change on a previous version…i think it was v42…? Where the system stopped working as it did previous, But later we realized it was only on the starting board. The system still brings the tiles that have been lacking in a clump closer to where you start.

My theory started when watching videos on my raids when going mono… When i watched the videos the colors would come in a clump just as i was dying, and i would get frustrated and rematch… sure enough the same pattern emerged… Colors came in abundance just as i died , and i realised, i am simply out of sync. I was starting in the barren area of the conveyor.

One thing i will say is the system requires time and patience… You need to video each attack and watch it back careful, it’s not how the board ends as you might have forgotten a few moves before the end that you exploded a crystal of that color.
Sometimes we watch a video and agree it’s inconclusive and we don’t risk it, other times when one color is missing so much its crazy ridiculous, we go straight to war with that color with full confidence.

On the rare occasion yes, we get freak anomolies, but they are very rare, and the system only reduces the chance of experiencing a missing color, It doesn’t eliminate it.
One thing i see is often is people losing and claiming the boards are not random…i’m the opposite , i claim that the reason i always win is because the boards are not random. That is, they are to a certain extent predictable.

The most compelling evidence used to be friendly battles, it seemed to be working on a different algorithm, the colors were so evenly distributed it was impossible to lose… But again we noticed this changed on about V42, but don’t hold me to that, i can remember specific details on a certain day , but i couldn’t tell you if that day was 2 months ago or 5 months ago

3 Likes

Oops, sorry, got lost in transactions

I’m sorry, I’ve read quite a few papers on math and statistics but your use of “baron” is novel to me…

Typo – Barren is the correct word; albiet not used per the intended definition.

2 Likes

I believe he means “barren”.

As for clumpiness, that is very much a common outcome from random events. In fact, it is more common in truly random than in humans trying to simulate random. There is a video on Youtube somewhere of a mathematician guessing numbers chosen by an pRNG and by a human trying to be random. He was twice as successful guessing the human’s choices. The human avoided clumpiness as being “non-random”. The pRNG didn’t.

The most familiar natural demonstration of random is the digits of the number PI. There are sections in the decimal component of PI that have (so far observed) up to 9 of the same number in a row. Yet PI is an irrational (hence random) number.

Since PI is an infinitely long random string of digits it likely contains every possible sequence of numbers you can think of, like 123456789, 111111111, 121212121212, etc. Any random sequence will have such “non-random” looking sequences, even pseudo-random ones.

If the algorithm is changed to “self-correct” than that makes SG liars as that is no longer random.

I don’t see any possible reason SG would ever do this. It would require much more effort in order to what - obtain the same result that would be obtained by letting the native pseudo random algorithm run its course and even out over time. Makes no sense whatsoever

LOL I have been saying that for a few years now.

I also said I would quit if anyone could prove that the pRNG algorithm was not random. I am still playing.

Thanks for this…

… but not for this , which is not true.

Irrationality does not imply randomness of sorts, most common example is probably a number constructed as follows:
0.121122111222111122221111122222…

As a matter of fact your argument that a number we know over 30 trillion digits of only has consecutive 9 digits patterns (I don’t know if it’s true nor am I calling you a liar, I’m just assuming it is true) weakens its likeliness to be random.

Pi is supposed to be normal but, as far as I know, no irrational constant has ever been proved to be normal.

Oh, just to add to the fray some “it might be a chance or not thought” a few days ago I popped to help Alliance on a titan.
I didn’t remember, but I had a HA10 training going (and god knows when completed).
It came down to this:

Incidentally, and I say this because another forum user observed it, I was totally dry on red.

Now a real question: has anyone ever noticed if new tiles replacing popped ones actually come uniformly?

I think this topic excites some people because they can hulk smash thread after thread where no direct revelations will ever be revealed.
I imagine that makes them feel good.

And some other’s are mildly interested, but the overtones usually seethe with smoke from a dumpster fire so they steer clear.

I don’t think you are either @nevarmaor

PI is irrational. That doesn’t mean random.
All of its digits are fixed.
You will get the same …patterns each time you calculate the number.
I understand your preface, but think the example is skewed.

Not that it’s overly important one way or the other but… since this topic is where I was born into the forum and my time is running out, it’s fitting that I perish where I began.

Why is it such a terminally important topic for you?

1 Like

Hello, i wondered how long it would take for you to turn up.

I respect you as a player i watch your videos , your attack teams combinations are well thought out and you know how to move tiles to maximum efficiency, but i don’t think we will ever agree on this.

But to be honest i am happy you reject my system, otherwise you would be as good as me

Just to clarify, Guvnor was interested in my theory, so i expanded and gave an explanation.

Apologies to everyone on my repeated use of the word baron, yes i did mean Barren

I think I may regret this, but I’m going to ask you a question.

Q. Do you believe that if you have a bad board without your colors, then there is more chance that your colors will come soon?

1 Like

To be honest i don’t think it makes them liars, you can change the percentage chance and it’s still random.

And self correct is my wording to try and explain how i predict.

Over a 100 raids the colors evened out… fact…
everyone said that’s proof of random… my system is based in the fact it evens out over time… So really the argument goes full circle.

I.e wu kong has a 34% chance to miss doesn’t mean it’s still not random whether he actually misses or not.
They could change it to 64% chance to miss… It’s still random to whether he actually misses or not.

SG need to define random… Does random mean 20% chance of each color , or does it mean 20% of each color randomly placed?
Both are random but yet very different.

1 Like

It would prove that SG lied and are not trustworthy. They have explicitly stated that the boards are random. I tend to not support companies that lie to me.

The numbers in PI are unpredictable, that is what makes them random, or as close to random as we can truly demonstrate. Yes, we can get the same sequence every time up to what we have calculated, but the next number cannot be determined by what has come before. You can think of it as pseudo random. A pRNG sequence will be the same given the same seed starting value. Unless you know the sequence you still cannot know or calculate the next number (without knowing the exact algorithm).

Are you saying that, for instance, if the boards have not had many blue that the percentage for blues is increased? That skews the randomness of the boards and yes, makes them non-random. But over 100 boards there will be a natural tendency to even out without having to skew the percentages. And over 1000 boards it will be even more so. You are saying they do a lot of excess programming to accomplish the same thing.

As for your question – not directed to me — but I expect a 20% chance of getting each colour in replacement tiles regardless of the initial spread.

I don’t know if this will add anything new to the discussion, however while playing the tutorial fights n times (and I played them a ton due to multiple alts creation) I’ve noticed that the board of the 1st province fights (and before that) have not only the same starting tile distribution - they also have absolutely the same identical tiles coming from below after any particular move is made.

However I can only confirm that about the first fights of the tutorial (up to the end of province 1) since those are actually the ones which have pre-set identical starting boards.

1 Like

Ooooh, c’mon, no worries.
I was just wondering if I was missing something, either in semantics, notation or sheer theory

1 Like

You must try to avoid the baron at all costs.

If 3 is the first digit of pi
Then 5 is the millionth.
Moreover… 5 will always be the millionth digit.
If it is not 5 then indeed, a random number will take it’s place. In which case, I would argue that you have been lied to.
But…
I just looked.
It’s still 5.

As of April last year, pi has been calculated out to 62.8 trillion digits. None of them will ever change.

I prefer the argument:
I personally can not calculate the next number,
therefore it must be random.
Vs. "because they said so"
The latter seems a bit childish.
But again… from everything I’ve read…
The latter is also the strongest argument to hang your hat on.

But being that all pseudo RN generators…
ALL have a margin for error where “patterns” emerge. I wouldn’t say that you have been lied to @nevarmaor
But I would question the PR perhaps:
"That they have developed the algorithm to simulate “random” as much as humanly possible." would have been a more accurate statement. One that no doubt would have poured gasoline on all these fires…

Even the lynchpin itself, of seeing the code displayed will not happen for obvious reasons…
Here is one:

But viewing the lynchpin still…
The debate would rage on.

For those who KNOW…
That hitting a Titan without pants on, will produce spectacular results. I applaud you!
And recommend your path forward be to seek out Discrepancy

Over 1,000 games each,
Try counting how many times a board reshuffles when using mono and rainbow

2 Likes

My system is no different to watching roulette tables in a casino , red or black.

13 in a row of one color is extremely high odds.

Imagine you start to play and double down on black, the chances of it coming red 5 times in a row is 1:32

So imagine you wait until it gets to 8 red in a row and then you start to bet on black and double down.
Its now a 1:8192 chance that is will continue to come red for 5 more turns in a row.
Because it’s not 5 in a row , it’s 13 in a row.

See how starting to bet on 8 changes the percentage chance from 1:32 to 1:8192

What I’m saying is start placing your bets when you get closer to 13.

Maybe that’s a better explanation?

2 Likes