The drop rates weren’t changed. You’re just getting unlucky
Without a doubt in my mind im sure they have been changed
Its not like im making this post after opening a single batch and got unlucky these crystals were reliable when and months after i became paragon and were change a couple bg's season ago 50 crystals could get you like 5 refills now you get 0
I’ve got over 100, when I need to open them for content, I will let you know how many I get! Hope I can open them soon, too many potions and revs expiring
What OP is saying and theory of probability is all correct and justified in ideal and theoretical maths.
But my take from whatever little I understand algorithm implemented to generate random number which mimics probability is not actually a true random number but it have its own periodic and cyclical feature some with higher frequency and some with lower.
So, a particular account might face higher probability than usual for a period of time and also lower probability for another period of time.
It all depends on how they have implemented their randomizer for an account.
LMAO, they haven’t changed because you people say so? Nah.
The burden of proof isn’t on anyone. You are stating your experience and the OP (and I) are stating ours. Unless you’ve seen the source code, these responses is nothing more than anecdotes the OP is proclaiming.
What OP is saying and theory of probability is all correct and justified in ideal and theoretical maths.
But my take from whatever little I understand algorithm implemented to generate random number which mimics probability is not actually a true random number but it have its own periodic and cyclical feature some with higher frequency and some with lower.
So, a particular account might face higher probability than usual for a period of time and also lower probability for another period of time.
It all depends on how they have implemented their randomizer for an account.
Also, because there are many overlapping crystal you pull and if I think is correct kabam must be using same randomizer for all type of crystal and basically multiplying the probability as one of it's coefficient, so your luck aka possibility of getting expected results may be getting pulled toward opening other crystal as well.
Because of having these many interdependence, it is very hard to actually predict and create an ideal situation for every type of crystal.
LMAO, they haven’t changed because you people say so? Nah.
The burden of proof isn’t on anyone. You are stating your experience and the OP (and I) are stating ours. Unless you’ve seen the source code, these responses is nothing more than anecdotes the OP is proclaiming.
Regarding Randomizer I am completely sure, it always have a pattern and mostly follow a cyclic behaviour.
Noone till now able to provide an algo to create a perfect random number.
This can be checked if you transform the numbers which are generated by algorithm using Fourier transform and check in frequency domain. It is very easy to visualise.
Kabam is just a gaming company, they might be even using generic inbuilt randomizer and not even some multi order function (because of load to server).
And even in reserch white super computers are used till date cyclic features are very prevelent in both nature/environment and artificial RNG algo.
But my take from whatever little I understand algorithm implemented to generate random number which mimics probability is not actually a true random number but it have its own periodic and cyclical feature some with higher frequency and some with lower.
Also yes and no, but also mostly no.
We don't know exactly which pRNG MCOC uses for crystals, but we can make educated guesses. If the developers rolled their own, it would most likely be horribly broken to the point where this would be easily detectable, so we can rule that out right now (I've seen such implementations, and I've discovered them in other games. Its always obviously horrible, because most programmers understand how to invent their own pRNG about as well as most players understand randomness in general).
One of the more common pRNGs from the old days that is still commonly found in computer software (it is actually one of the most common algorithms found in slot machines, for example) is something called Mersenne Twister. This is considered a mediocre generator with a few spectral flaws (difficult to explain, but relate to how the random numbers correlate with each other over certain sophisticated tests) and it does repeat itself (like all pRNGs must) but the period for MT is 2^19937 - 1. Or about 10^1995. If everyone on Earth started opening crystals at one per second per person, over the lifetime of the universe, you would not make a dent in this number. Mersenne Twister repeats in theory, but there is no computer software we can or ever will build that will ever actually see MT repeat, ever.
MT does have some flaws, but they aren't the kinds of flaws a human would see when opening crystals in a game. And these days the pRNG included with most software programming languages is much better than MT (which, as I said, casinos use in slot machines that process way more money than MCOC does). But this brings up the question of what is a "true random number" and what difference does it make? "True random number" is not something with a solid commonly agreed upon definition. In practical terms, we (meaning software designers that deal with random numbers) talk about "random enough."
Suppose I hand you something and say "this is a *true* random number generator." How would you know? What test could you perform to determine if I was telling the truth or not? There isn't one. There's no way to prove a generator is "truly random." That's why "true randomness" is often considered a meaningless statement. However, we can define what we mean by "random enough." For something like a video game lootbox we really just want a few specific properties of the pRNG:
1. There's no measurable bias. When we test the pRNG across trillions of drops, we don't see the generator drop any one particular result more often than any other. Of course, in any test *something* will be more common than other things, because no random result generates exactly equal amounts of everything. However, over longer and longer tests, no one result is consistently preferred. Sometimes one result is more common, sometimes another result just happens to be more common, and overall there's no preference.
2. There are no apparent correlations. We don't want, say, a particular result to happen more often than average after other results. We want every result to appear independent of the previous ones. Whether you open ten crystals at once or one a day over a week and a half should make no difference.
3. There should be no way to predict future results from any reasonable amount of measurements of previous results. In other words, players should not be able to watch the drops from a crystal and be able to determine when it is more or less likely to generate the drops they want, even if they were to somehow have a record of every crystal drop in the history of the game.
To put it simply, we want the pRNG to be fair, independent, and unpredictable, on time scales of the lifetime of the universe. More than that is overkill. And there are pRNGs that satisfy all three conditions. Any pRNG with these properties would generate results that no human could ever find a statistical flaw with. You would never see such a flaw, because you aren't going to live long enough for it to happen even in theory.
But my take from whatever little I understand algorithm implemented to generate random number which mimics probability is not actually a true random number but it have its own periodic and cyclical feature some with higher frequency and some with lower.
Also yes and no, but also mostly no.
We don't know exactly which pRNG MCOC uses for crystals, but we can make educated guesses. If the developers rolled their own, it would most likely be horribly broken to the point where this would be easily detectable, so we can rule that out right now (I've seen such implementations, and I've discovered them in other games. Its always obviously horrible, because most programmers understand how to invent their own pRNG about as well as most players understand randomness in general).
One of the more common pRNGs from the old days that is still commonly found in computer software (it is actually one of the most common algorithms found in slot machines, for example) is something called Mersenne Twister. This is considered a mediocre generator with a few spectral flaws (difficult to explain, but relate to how the random numbers correlate with each other over certain sophisticated tests) and it does repeat itself (like all pRNGs must) but the period for MT is 2^19937 - 1. Or about 10^1995. If everyone on Earth started opening crystals at one per second per person, over the lifetime of the universe, you would not make a dent in this number. Mersenne Twister repeats in theory, but there is no computer software we can or ever will build that will ever actually see MT repeat, ever.
MT does have some flaws, but they aren't the kinds of flaws a human would see when opening crystals in a game. And these days the pRNG included with most software programming languages is much better than MT (which, as I said, casinos use in slot machines that process way more money than MCOC does). But this brings up the question of what is a "true random number" and what difference does it make? "True random number" is not something with a solid commonly agreed upon definition. In practical terms, we (meaning software designers that deal with random numbers) talk about "random enough."
Suppose I hand you something and say "this is a *true* random number generator." How would you know? What test could you perform to determine if I was telling the truth or not? There isn't one. There's no way to prove a generator is "truly random." That's why "true randomness" is often considered a meaningless statement. However, we can define what we mean by "random enough." For something like a video game lootbox we really just want a few specific properties of the pRNG:
1. There's no measurable bias. When we test the pRNG across trillions of drops, we don't see the generator drop any one particular result more often than any other. Of course, in any test *something* will be more common than other things, because no random result generates exactly equal amounts of everything. However, over longer and longer tests, no one result is consistently preferred. Sometimes one result is more common, sometimes another result just happens to be more common, and overall there's no preference.
2. There are no apparent correlations. We don't want, say, a particular result to happen more often than average after other results. We want every result to appear independent of the previous ones. Whether you open ten crystals at once or one a day over a week and a half should make no difference.
3. There should be no way to predict future results from any reasonable amount of measurements of previous results. In other words, players should not be able to watch the drops from a crystal and be able to determine when it is more or less likely to generate the drops they want, even if they were to somehow have a record of every crystal drop in the history of the game.
To put it simply, we want the pRNG to be fair, independent, and unpredictable, on time scales of the lifetime of the universe. More than that is overkill. And there are pRNGs that satisfy all three conditions. Any pRNG with these properties would generate results that no human could ever find a statistical flaw with. You would never see such a flaw, because you aren't going to live long enough for it to happen even in theory.
Here, you are talking about repeatability but issue is not repeatability but where that number is falling there is a cyclic behaviour on which area that number is going to land and not the number itself. More like you divide something between 0 to 100 and make 50 as boundary. So how much number is going to land below 50 and how many above 50 have very small cyclic period that repeatability itself.
But my take from whatever little I understand algorithm implemented to generate random number which mimics probability is not actually a true random number but it have its own periodic and cyclical feature some with higher frequency and some with lower.
Also yes and no, but also mostly no.
We don't know exactly which pRNG MCOC uses for crystals, but we can make educated guesses. If the developers rolled their own, it would most likely be horribly broken to the point where this would be easily detectable, so we can rule that out right now (I've seen such implementations, and I've discovered them in other games. Its always obviously horrible, because most programmers understand how to invent their own pRNG about as well as most players understand randomness in general).
One of the more common pRNGs from the old days that is still commonly found in computer software (it is actually one of the most common algorithms found in slot machines, for example) is something called Mersenne Twister. This is considered a mediocre generator with a few spectral flaws (difficult to explain, but relate to how the random numbers correlate with each other over certain sophisticated tests) and it does repeat itself (like all pRNGs must) but the period for MT is 2^19937 - 1. Or about 10^1995. If everyone on Earth started opening crystals at one per second per person, over the lifetime of the universe, you would not make a dent in this number. Mersenne Twister repeats in theory, but there is no computer software we can or ever will build that will ever actually see MT repeat, ever.
MT does have some flaws, but they aren't the kinds of flaws a human would see when opening crystals in a game. And these days the pRNG included with most software programming languages is much better than MT (which, as I said, casinos use in slot machines that process way more money than MCOC does). But this brings up the question of what is a "true random number" and what difference does it make? "True random number" is not something with a solid commonly agreed upon definition. In practical terms, we (meaning software designers that deal with random numbers) talk about "random enough."
Suppose I hand you something and say "this is a *true* random number generator." How would you know? What test could you perform to determine if I was telling the truth or not? There isn't one. There's no way to prove a generator is "truly random." That's why "true randomness" is often considered a meaningless statement. However, we can define what we mean by "random enough." For something like a video game lootbox we really just want a few specific properties of the pRNG:
1. There's no measurable bias. When we test the pRNG across trillions of drops, we don't see the generator drop any one particular result more often than any other. Of course, in any test *something* will be more common than other things, because no random result generates exactly equal amounts of everything. However, over longer and longer tests, no one result is consistently preferred. Sometimes one result is more common, sometimes another result just happens to be more common, and overall there's no preference.
2. There are no apparent correlations. We don't want, say, a particular result to happen more often than average after other results. We want every result to appear independent of the previous ones. Whether you open ten crystals at once or one a day over a week and a half should make no difference.
3. There should be no way to predict future results from any reasonable amount of measurements of previous results. In other words, players should not be able to watch the drops from a crystal and be able to determine when it is more or less likely to generate the drops they want, even if they were to somehow have a record of every crystal drop in the history of the game.
To put it simply, we want the pRNG to be fair, independent, and unpredictable, on time scales of the lifetime of the universe. More than that is overkill. And there are pRNGs that satisfy all three conditions. Any pRNG with these properties would generate results that no human could ever find a statistical flaw with. You would never see such a flaw, because you aren't going to live long enough for it to happen even in theory.
Here, you are talking about repeatability but issue is not repeatability but where that number is falling there is a cyclic behaviour on which area that number is going to land and not the number itself. More like you divide something between 0 to 100 and make 50 as boundary. So how much number is going to land below 50 and how many above 50 have very small cyclic period that repeatability itself.
Check, this is also a RNG plot, you can clearly see there is cyclic features even though there is no repeatability.
LMAO, they haven’t changed because you people say so? Nah.
The burden of proof isn’t on anyone. You are stating your experience and the OP (and I) are stating ours. Unless you’ve seen the source code, these responses is nothing more than anecdotes the OP is proclaiming.
So because no one has presented evidence that people aren't microchipped and mind controlled by the CIA, then it's anecdotal to say that's not a thing. Interesting approach.
LMAO, they haven’t changed because you people say so? Nah.
The burden of proof isn’t on anyone. You are stating your experience and the OP (and I) are stating ours. Unless you’ve seen the source code, these responses is nothing more than anecdotes the OP is proclaiming.
Regarding Randomizer I am completely sure, it always have a pattern and mostly follow a cyclic behaviour.
Noone till now able to provide an algo to create a perfect random number.
This can be checked if you transform the numbers which are generated by algorithm using Fourier transform and check in frequency domain. It is very easy to visualise.
Kabam is just a gaming company, they might be even using generic inbuilt randomizer and not even some multi order function (because of load to server).
And even in reserch white super computers are used till date cyclic features are very prevelent in both nature/environment and artificial RNG algo.
Definitely agree on all front. I don’t KNOW if it’s changed, it FEELS like it’s changed. Of course I may very well be wrong and I appreciate well thought out responses like yours.
It’s the gaslighting that I take exception to. Responses like “go do a 10k sample” is incredibly asinine.
But my take from whatever little I understand algorithm implemented to generate random number which mimics probability is not actually a true random number but it have its own periodic and cyclical feature some with higher frequency and some with lower.
Also yes and no, but also mostly no.
We don't know exactly which pRNG MCOC uses for crystals, but we can make educated guesses. If the developers rolled their own, it would most likely be horribly broken to the point where this would be easily detectable, so we can rule that out right now (I've seen such implementations, and I've discovered them in other games. Its always obviously horrible, because most programmers understand how to invent their own pRNG about as well as most players understand randomness in general).
One of the more common pRNGs from the old days that is still commonly found in computer software (it is actually one of the most common algorithms found in slot machines, for example) is something called Mersenne Twister. This is considered a mediocre generator with a few spectral flaws (difficult to explain, but relate to how the random numbers correlate with each other over certain sophisticated tests) and it does repeat itself (like all pRNGs must) but the period for MT is 2^19937 - 1. Or about 10^1995. If everyone on Earth started opening crystals at one per second per person, over the lifetime of the universe, you would not make a dent in this number. Mersenne Twister repeats in theory, but there is no computer software we can or ever will build that will ever actually see MT repeat, ever.
MT does have some flaws, but they aren't the kinds of flaws a human would see when opening crystals in a game. And these days the pRNG included with most software programming languages is much better than MT (which, as I said, casinos use in slot machines that process way more money than MCOC does). But this brings up the question of what is a "true random number" and what difference does it make? "True random number" is not something with a solid commonly agreed upon definition. In practical terms, we (meaning software designers that deal with random numbers) talk about "random enough."
Suppose I hand you something and say "this is a *true* random number generator." How would you know? What test could you perform to determine if I was telling the truth or not? There isn't one. There's no way to prove a generator is "truly random." That's why "true randomness" is often considered a meaningless statement. However, we can define what we mean by "random enough." For something like a video game lootbox we really just want a few specific properties of the pRNG:
1. There's no measurable bias. When we test the pRNG across trillions of drops, we don't see the generator drop any one particular result more often than any other. Of course, in any test *something* will be more common than other things, because no random result generates exactly equal amounts of everything. However, over longer and longer tests, no one result is consistently preferred. Sometimes one result is more common, sometimes another result just happens to be more common, and overall there's no preference.
2. There are no apparent correlations. We don't want, say, a particular result to happen more often than average after other results. We want every result to appear independent of the previous ones. Whether you open ten crystals at once or one a day over a week and a half should make no difference.
3. There should be no way to predict future results from any reasonable amount of measurements of previous results. In other words, players should not be able to watch the drops from a crystal and be able to determine when it is more or less likely to generate the drops they want, even if they were to somehow have a record of every crystal drop in the history of the game.
To put it simply, we want the pRNG to be fair, independent, and unpredictable, on time scales of the lifetime of the universe. More than that is overkill. And there are pRNGs that satisfy all three conditions. Any pRNG with these properties would generate results that no human could ever find a statistical flaw with. You would never see such a flaw, because you aren't going to live long enough for it to happen even in theory.
Here, you are talking about repeatability but issue is not repeatability but where that number is falling there is a cyclic behaviour on which area that number is going to land and not the number itself. More like you divide something between 0 to 100 and make 50 as boundary. So how much number is going to land below 50 and how many above 50 have very small cyclic period that repeatability itself.
Without getting too much into the weeds, pRNGs are tested for this sort of thing. It can be shown mathematically that if a pRNG contained such a cyclic period for a 0-100 partition, or in fact *any* partition you can imagine, this would measurably impact the entropy generated by the generator. This would show up in compressibility tests among other things (I'm assuming hyperplane correlations would also be impacted).
Incidentally, there's a fencepost error in your example. Yes, in the range of integers from zero to 100 inclusive 50 is the mid point, such that there are fifty numbers above and fifty numbers below that mid point. However, no one would test a pRNG like that, because such a test fails to include the mid point itself. 50 should come up about 0.99% of the time, which means the sum of the other two possibilities will not sum to one. You'd more likely do a test from zero to 99 with the partition being 49/50, testing for the odds of landing in the 0-49 range and the 50-99 range, which should be 50% in the long run.
Except, of course, you wouldn't test 0-99 either, because 100 is not a power of two. This complicates the analysis, because you would have to factor out partition bias.
But my take from whatever little I understand algorithm implemented to generate random number which mimics probability is not actually a true random number but it have its own periodic and cyclical feature some with higher frequency and some with lower.
Also yes and no, but also mostly no.
We don't know exactly which pRNG MCOC uses for crystals, but we can make educated guesses. If the developers rolled their own, it would most likely be horribly broken to the point where this would be easily detectable, so we can rule that out right now (I've seen such implementations, and I've discovered them in other games. Its always obviously horrible, because most programmers understand how to invent their own pRNG about as well as most players understand randomness in general).
One of the more common pRNGs from the old days that is still commonly found in computer software (it is actually one of the most common algorithms found in slot machines, for example) is something called Mersenne Twister. This is considered a mediocre generator with a few spectral flaws (difficult to explain, but relate to how the random numbers correlate with each other over certain sophisticated tests) and it does repeat itself (like all pRNGs must) but the period for MT is 2^19937 - 1. Or about 10^1995. If everyone on Earth started opening crystals at one per second per person, over the lifetime of the universe, you would not make a dent in this number. Mersenne Twister repeats in theory, but there is no computer software we can or ever will build that will ever actually see MT repeat, ever.
MT does have some flaws, but they aren't the kinds of flaws a human would see when opening crystals in a game. And these days the pRNG included with most software programming languages is much better than MT (which, as I said, casinos use in slot machines that process way more money than MCOC does). But this brings up the question of what is a "true random number" and what difference does it make? "True random number" is not something with a solid commonly agreed upon definition. In practical terms, we (meaning software designers that deal with random numbers) talk about "random enough."
Suppose I hand you something and say "this is a *true* random number generator." How would you know? What test could you perform to determine if I was telling the truth or not? There isn't one. There's no way to prove a generator is "truly random." That's why "true randomness" is often considered a meaningless statement. However, we can define what we mean by "random enough." For something like a video game lootbox we really just want a few specific properties of the pRNG:
1. There's no measurable bias. When we test the pRNG across trillions of drops, we don't see the generator drop any one particular result more often than any other. Of course, in any test *something* will be more common than other things, because no random result generates exactly equal amounts of everything. However, over longer and longer tests, no one result is consistently preferred. Sometimes one result is more common, sometimes another result just happens to be more common, and overall there's no preference.
2. There are no apparent correlations. We don't want, say, a particular result to happen more often than average after other results. We want every result to appear independent of the previous ones. Whether you open ten crystals at once or one a day over a week and a half should make no difference.
3. There should be no way to predict future results from any reasonable amount of measurements of previous results. In other words, players should not be able to watch the drops from a crystal and be able to determine when it is more or less likely to generate the drops they want, even if they were to somehow have a record of every crystal drop in the history of the game.
To put it simply, we want the pRNG to be fair, independent, and unpredictable, on time scales of the lifetime of the universe. More than that is overkill. And there are pRNGs that satisfy all three conditions. Any pRNG with these properties would generate results that no human could ever find a statistical flaw with. You would never see such a flaw, because you aren't going to live long enough for it to happen even in theory.
Here, you are talking about repeatability but issue is not repeatability but where that number is falling there is a cyclic behaviour on which area that number is going to land and not the number itself. More like you divide something between 0 to 100 and make 50 as boundary. So how much number is going to land below 50 and how many above 50 have very small cyclic period that repeatability itself.
Check, this is also a RNG plot, you can clearly see there is cyclic features even though there is no repeatability.
Out of contest I can't be 100% certain what that plot even represents, but it seems highly unlikely this represents any sort of random plot. No random plot would be marked with those axes, and having two obviously correlated data sets seems to imply the data contains structure, and not random.
There are many mathematical structures that have some cyclic features but are not repeatable, but here you're using those words in a completely different mathematical context from the study of pRNG algorithms.
At some point this becomes me arguing with wikipedia tossed into a blender, and unless you demonstrate some actual field knowledge about this topic and not just randomly throwing words and graphics onto the wall to see what sticks, I'm going to assume this is just another someone who thinks they've found a hole in Relativity or has invented an entirely new form of arithmetic and just came along too late to evangelize it on USENET.
Simply put, they're not required to. We don't purchase Free Crystals.
As far as I'm aware, basically that. Kabam is not obligated to list drop rates for all crystals. Apple app store guidelines specify that games that offer lootbox like items "for purchase" must list the odds for that item. Most game publishers have interpreted "for purchase" to mean "for purchase with cash through the in-app purchase mechanism" as this guideline is listed under the section related to the use of the in-app purchase features.
Why not just list them for all crystals anyway? Well, I can think of two reasons. First, it takes extra work. Second, if they don't list odds they can't be accused of lying about the odds. If they ever make a mistake and list the odds incorrectly on a crystal, they would be accused of deception.
Beyond the obvious reasons, there is another possibility that I've discussed with other developers of other games but not with Kabam specifically. Off the record, many game developers do not like to list too much numerical information about the games they design, because they don't want to encourage calculation based min/maxing. They would prefer players exercise intuitive judgment based on experience. Things like the four hour crystals are meant to be opened and used. Of course, they know players will hoard them, or save them up for some future use, and that's fine, but they might not want to actually explicitly lay out precisely what the numerical pros and cons are for doing that and encourage people to calculate the "optimal strategy." They might rather have players all guess individually.
I don't know that last one for a fact: that is speculation on my part based on conversations having nothing to do with MCOC. The first two reasons are more logical objective observations.
But my take from whatever little I understand algorithm implemented to generate random number which mimics probability is not actually a true random number but it have its own periodic and cyclical feature some with higher frequency and some with lower.
Also yes and no, but also mostly no.
We don't know exactly which pRNG MCOC uses for crystals, but we can make educated guesses. If the developers rolled their own, it would most likely be horribly broken to the point where this would be easily detectable, so we can rule that out right now (I've seen such implementations, and I've discovered them in other games. Its always obviously horrible, because most programmers understand how to invent their own pRNG about as well as most players understand randomness in general).
One of the more common pRNGs from the old days that is still commonly found in computer software (it is actually one of the most common algorithms found in slot machines, for example) is something called Mersenne Twister. This is considered a mediocre generator with a few spectral flaws (difficult to explain, but relate to how the random numbers correlate with each other over certain sophisticated tests) and it does repeat itself (like all pRNGs must) but the period for MT is 2^19937 - 1. Or about 10^1995. If everyone on Earth started opening crystals at one per second per person, over the lifetime of the universe, you would not make a dent in this number. Mersenne Twister repeats in theory, but there is no computer software we can or ever will build that will ever actually see MT repeat, ever.
MT does have some flaws, but they aren't the kinds of flaws a human would see when opening crystals in a game. And these days the pRNG included with most software programming languages is much better than MT (which, as I said, casinos use in slot machines that process way more money than MCOC does). But this brings up the question of what is a "true random number" and what difference does it make? "True random number" is not something with a solid commonly agreed upon definition. In practical terms, we (meaning software designers that deal with random numbers) talk about "random enough."
Suppose I hand you something and say "this is a *true* random number generator." How would you know? What test could you perform to determine if I was telling the truth or not? There isn't one. There's no way to prove a generator is "truly random." That's why "true randomness" is often considered a meaningless statement. However, we can define what we mean by "random enough." For something like a video game lootbox we really just want a few specific properties of the pRNG:
1. There's no measurable bias. When we test the pRNG across trillions of drops, we don't see the generator drop any one particular result more often than any other. Of course, in any test *something* will be more common than other things, because no random result generates exactly equal amounts of everything. However, over longer and longer tests, no one result is consistently preferred. Sometimes one result is more common, sometimes another result just happens to be more common, and overall there's no preference.
2. There are no apparent correlations. We don't want, say, a particular result to happen more often than average after other results. We want every result to appear independent of the previous ones. Whether you open ten crystals at once or one a day over a week and a half should make no difference.
3. There should be no way to predict future results from any reasonable amount of measurements of previous results. In other words, players should not be able to watch the drops from a crystal and be able to determine when it is more or less likely to generate the drops they want, even if they were to somehow have a record of every crystal drop in the history of the game.
To put it simply, we want the pRNG to be fair, independent, and unpredictable, on time scales of the lifetime of the universe. More than that is overkill. And there are pRNGs that satisfy all three conditions. Any pRNG with these properties would generate results that no human could ever find a statistical flaw with. You would never see such a flaw, because you aren't going to live long enough for it to happen even in theory.
Here, you are talking about repeatability but issue is not repeatability but where that number is falling there is a cyclic behaviour on which area that number is going to land and not the number itself. More like you divide something between 0 to 100 and make 50 as boundary. So how much number is going to land below 50 and how many above 50 have very small cyclic period that repeatability itself.
Check, this is also a RNG plot, you can clearly see there is cyclic features even though there is no repeatability.
Just from looking at the axes, I suspect that this random data may be generated by a physical source. The vertical axis indicates the data may be a measurement from an ADC (Analog to Digital Converter), which leads me to believe this is a Hardware Random Number Generator (HRNG). There is precedent for this kind of RNG, you can find plenty of good journal articles detailing several implementations. Ironically, HRNG applications are often thought of as 'better' than the PRNG algorithms used in more mainstream applications. In fact, some refer to these generators as 'True' Random Number Generators (though as DNA pointed out, the concept of 'true' randomness is nebulous at best). If this graph is indeed displaying a natural source of randomness, it may not have any periodicity at all. A more in depth study of the specific source would need to be conducted to be sure, but it doesn't look too great from this graph alone. It is likely that a standard PRNG algorithm would perform far better.
Just from looking at the axes, I suspect that this random data may be generated by a physical source. The vertical axis indicates the data may be a measurement from an ADC (Analog to Digital Converter), which leads me to believe this is a Hardware Random Number Generator (HRNG).
My first thought was a jitter source, but that doesn't seem right. I also wondered if the graph was not an actual RNG stream but more of a random walk, but that also doesn't work and no hardware source outputs that sort of incremental data. It could be a derived random walk from a source, that would line up with the ADC/ADU axis and I guess the trace point axis, sort of, but if that's a random walk from a hardware generator it is completely broken.
I suppose it is possible this is the raw output from a thermal jitter source, before it has been properly sampled. On (relatively) long time scales HRNGs can appear random, but on sufficiently short time scales successive data points are strongly correlated. Perhaps this graph came from some research paper looking for the maximum bit rate possible from this source. Which would mean that ironically, these are non-random values from a hardware random number generator.
Comments
But my take from whatever little I understand algorithm implemented to generate random number which mimics probability is not actually a true random number but it have its own periodic and cyclical feature some with higher frequency and some with lower.
So, a particular account might face higher probability than usual for a period of time and also lower probability for another period of time.
It all depends on how they have implemented their randomizer for an account.
The burden of proof isn’t on anyone. You are stating your experience and the OP (and I) are stating ours. Unless you’ve seen the source code, these responses is nothing more than anecdotes the OP is proclaiming.
Because of having these many interdependence, it is very hard to actually predict and create an ideal situation for every type of crystal.
Noone till now able to provide an algo to create a perfect random number.
This can be checked if you transform the numbers which are generated by algorithm using Fourier transform and check in frequency domain. It is very easy to visualise.
Kabam is just a gaming company, they might be even using generic inbuilt randomizer and not even some multi order function (because of load to server).
And even in reserch white super computers are used till date cyclic features are very prevelent in both nature/environment and artificial RNG algo.
We don't know exactly which pRNG MCOC uses for crystals, but we can make educated guesses. If the developers rolled their own, it would most likely be horribly broken to the point where this would be easily detectable, so we can rule that out right now (I've seen such implementations, and I've discovered them in other games. Its always obviously horrible, because most programmers understand how to invent their own pRNG about as well as most players understand randomness in general).
One of the more common pRNGs from the old days that is still commonly found in computer software (it is actually one of the most common algorithms found in slot machines, for example) is something called Mersenne Twister. This is considered a mediocre generator with a few spectral flaws (difficult to explain, but relate to how the random numbers correlate with each other over certain sophisticated tests) and it does repeat itself (like all pRNGs must) but the period for MT is 2^19937 - 1. Or about 10^1995. If everyone on Earth started opening crystals at one per second per person, over the lifetime of the universe, you would not make a dent in this number. Mersenne Twister repeats in theory, but there is no computer software we can or ever will build that will ever actually see MT repeat, ever.
MT does have some flaws, but they aren't the kinds of flaws a human would see when opening crystals in a game. And these days the pRNG included with most software programming languages is much better than MT (which, as I said, casinos use in slot machines that process way more money than MCOC does). But this brings up the question of what is a "true random number" and what difference does it make? "True random number" is not something with a solid commonly agreed upon definition. In practical terms, we (meaning software designers that deal with random numbers) talk about "random enough."
Suppose I hand you something and say "this is a *true* random number generator." How would you know? What test could you perform to determine if I was telling the truth or not? There isn't one. There's no way to prove a generator is "truly random." That's why "true randomness" is often considered a meaningless statement. However, we can define what we mean by "random enough." For something like a video game lootbox we really just want a few specific properties of the pRNG:
1. There's no measurable bias. When we test the pRNG across trillions of drops, we don't see the generator drop any one particular result more often than any other. Of course, in any test *something* will be more common than other things, because no random result generates exactly equal amounts of everything. However, over longer and longer tests, no one result is consistently preferred. Sometimes one result is more common, sometimes another result just happens to be more common, and overall there's no preference.
2. There are no apparent correlations. We don't want, say, a particular result to happen more often than average after other results. We want every result to appear independent of the previous ones. Whether you open ten crystals at once or one a day over a week and a half should make no difference.
3. There should be no way to predict future results from any reasonable amount of measurements of previous results. In other words, players should not be able to watch the drops from a crystal and be able to determine when it is more or less likely to generate the drops they want, even if they were to somehow have a record of every crystal drop in the history of the game.
To put it simply, we want the pRNG to be fair, independent, and unpredictable, on time scales of the lifetime of the universe. More than that is overkill. And there are pRNGs that satisfy all three conditions. Any pRNG with these properties would generate results that no human could ever find a statistical flaw with. You would never see such a flaw, because you aren't going to live long enough for it to happen even in theory.
More like you divide something between 0 to 100 and make 50 as boundary.
So how much number is going to land below 50 and how many above 50 have very small cyclic period that repeatability itself.
Interesting approach.
It’s the gaslighting that I take exception to. Responses like “go do a 10k sample” is incredibly asinine.
Incidentally, there's a fencepost error in your example. Yes, in the range of integers from zero to 100 inclusive 50 is the mid point, such that there are fifty numbers above and fifty numbers below that mid point. However, no one would test a pRNG like that, because such a test fails to include the mid point itself. 50 should come up about 0.99% of the time, which means the sum of the other two possibilities will not sum to one. You'd more likely do a test from zero to 99 with the partition being 49/50, testing for the odds of landing in the 0-49 range and the 50-99 range, which should be 50% in the long run.
Except, of course, you wouldn't test 0-99 either, because 100 is not a power of two. This complicates the analysis, because you would have to factor out partition bias.
I wasted 10 crystals just to check this nerf for myself. I cannot believe how nerfed this is
There are many mathematical structures that have some cyclic features but are not repeatable, but here you're using those words in a completely different mathematical context from the study of pRNG algorithms.
At some point this becomes me arguing with wikipedia tossed into a blender, and unless you demonstrate some actual field knowledge about this topic and not just randomly throwing words and graphics onto the wall to see what sticks, I'm going to assume this is just another someone who thinks they've found a hole in Relativity or has invented an entirely new form of arithmetic and just came along too late to evangelize it on USENET.
We didn't change anything.
Thanks.
Why not just list them for all crystals anyway? Well, I can think of two reasons. First, it takes extra work. Second, if they don't list odds they can't be accused of lying about the odds. If they ever make a mistake and list the odds incorrectly on a crystal, they would be accused of deception.
Beyond the obvious reasons, there is another possibility that I've discussed with other developers of other games but not with Kabam specifically. Off the record, many game developers do not like to list too much numerical information about the games they design, because they don't want to encourage calculation based min/maxing. They would prefer players exercise intuitive judgment based on experience. Things like the four hour crystals are meant to be opened and used. Of course, they know players will hoard them, or save them up for some future use, and that's fine, but they might not want to actually explicitly lay out precisely what the numerical pros and cons are for doing that and encourage people to calculate the "optimal strategy." They might rather have players all guess individually.
I don't know that last one for a fact: that is speculation on my part based on conversations having nothing to do with MCOC. The first two reasons are more logical objective observations.
I suppose it is possible this is the raw output from a thermal jitter source, before it has been properly sampled. On (relatively) long time scales HRNGs can appear random, but on sufficiently short time scales successive data points are strongly correlated. Perhaps this graph came from some research paper looking for the maximum bit rate possible from this source. Which would mean that ironically, these are non-random values from a hardware random number generator.
Someone else that gold is very Short in the crystals. Now the Energy refills.
When they remove the pots there will only be 7* Drax in These?
Jax said back then the Rates a coded. And when Displayed the promt Take the rates Out of the code,Not from a dev that wrote it in the promt.
Thats why people can Tell when featured Droprates are wrong!
I think Jax rather Performs Frozens "into the unknown" in the a Livestream then a dev shares their sorcecode (and explain it for dummies).
I am looking forward for Halloween Livestream and Elsa-Jax 😁