There is a bug or intentional code in POE's random number generator that is causing streakiness.

"

ampdecay wrote:

"

Inexium wrote:

I have a question, i will be very surprised to hear you ( all of you ) saw the same thing happening, and i hope os.


Did you ever experience when you have been hoarding an orb for a good moment then one day you spend all of them of some of them.
You go killing monsters.. and : The hoarded orb that stoped droping for a moment just start to drop again :o
?

I am very suspicious yet it's not that important to me.

Could that thing affect rng clouds ?

I experience this all the time with chaos, alc and fusing.

I remember when it happened to me for Chaos's

"

SL4Y3R wrote:

agreed. It started out alright, but saying what you have in your stash matters is beyond pushing the envelope.

See.. ?
You really never saw something like this happening ? Maybe you will start to notice it now..

Maybe it's not true, who knows!

"

Idioticus wrote:

Spoiler

"

mark1030 wrote:

I just flipped a quarter 20 times (really, I did). It's a US 2013 with "Great Basin, Nevada" on the back minted in the "P" mint. I can post a picture if you don't believe me.

The results, in order, were:

THHTT
HTTTT
HTTTH
HHTTT

There must be something wrong with the universe's RNG, because look how streaky that was. I had 4 Tails in a row once, and then after that I had 3 Tails, then 3 Heads, then 3 Tails again. In fact, I had 13 Tails and only 7 Heads. Out of 20 runs! No way can such streakiness be attributed to probability. There must be something behind the scenes to explain it. Maybe the universe wants me to trade my quarter with somebody who has one that favors Heads so everything will even out.

Come on, people. Drop rates are not set for each player and each run individually. They are across all the servers. So if something has a 1 in 10 chance of dropping, and 10 people do 10 runs each at the same time, what are the odds that each person got that drop once? If one person didn't get that drop and another got it twice, is that no longer random?

Ever notice how in groups of people smaller than 365, there are people that share the same birthday?

You know... GGG could run some bots on a farming spree (or just simulation, whatever, I'm not a programmer) to see if there's any truth to what people are saying.
Just scoffing this issue aside because "lol, I did something completely unrelated to the game" is no way to handle this.

If there is a problem, GGG can find it out. This seems to be a serious enough an accusation.
That's on one hand... on the other, could be this is what GGG has planned but wants to keep "secret"... ugh.

What would be the point of running a bot? No amount of proof will ever be enough because people think random means predictable for one player in a vacuum. People kind of ignore my example of an item that has a 1 in 10 drop rate with 10 players doing 10 runs each, but that's what this all boils down to. You're practically never going to end up with each player finding exactly one of the item. Now scale it up to however many thousand players doing however many thousand runs. People still want to base their "proof" on just one player at a time instead of the whole realm at one time.

A couple days ago I used 2 of my 8 fusings on a 6 socket armor and got a 5L. This is not proof that each fusing has a 50% chance to give me 5 links. If you took all the people who also attempted to 5L a chest that day, I imagine my good luck wouldl offset somebody else's bad luck, and realm-wide the odds of 5 linking would be true to whatever the value is set to.

"

RogueMage wrote:

To the OP - your thread is nothing but troll bait - you've presented zero evidence to back up your rash claim that "There is an error in PoE's random number generator".

The true answer is pretty much that I'm too lazy to do science. So I can't be sure, for sure. But anecdotal evidence of all community members + my intuition says it's so.

Take the Dilbert comic as an example and think for yourself instead of regurgitating what you've heard elsewhere.

The odds of winning the mega millions jackpot, according to Wikipedia, is "1 in 175,711,536"

That means, if you see "9999999999" in a "random number generator," then there are two possibilities:

1. It was randomly generated. There is roughly a 1 (or 9, depending on if you count 3333 etc.) in 1,000,000,000 chance of this happening,
2. There is a problem with the random number generator causing a sequence of numbers to repeat more than they should. There is roughly a 999,999,999 in 1,000,000,000 chance of this happening.

Tack on just 5 more nines and you can be 10,000 times more sure that you'll win the Mega Millions lottery than that the "random number generator" is random.

Would you bet on that lottery?

The point of the Dilbert comic is that you can never be sure.

The point of statistical science is that you can be AS sure that something is happening AS some other phenomenon.

However, it is up to you to determine whether the overwhelming number of patterns of streaks players have noted in POE is a 999, or a 99999999999999999.

The number of streaks I've personally experienced is pretty statistically significant from a very rough armchair calculation. For me, it's reading somewhere around a 99999 occurring in a randomly generated sequence of decimal digits.

For fun, let's compare the "heads or tails" poster's analysis with an outcome of 99,999 in a 1 out of 100,000 roll.

"

THHTT
HTTTT
HTTTH
HHTTT

The suspicious result in question was having four "3+" streaks in a row, or 1 4-streak and 3 3-streaks immediately afterwards. A crude estimation of that probability would be to find the chance of 4-streak, then multiply by 3 3-streaks. That's 1/8 * 1/4 * 1/4 * 1/4, or 1 in 512. Congratulations, you had roughly a 1 in 512 occurrence. Go bet on a horse race.

It is 200 times more likely that you would receive 1 4-streak and 3 3-streaks in a row one after another than it is that a true random number generator would produce 99,999, or that a true random number generator would produce six of the same digit in a row. Roughly speaking.

My work:

(actually used a decision tree to find 1/8 and 1/4, but they can be derived analytically too as follows:
3 flips = 2^3 possibilities - two possibilities (TTT and HHH) are "suspicious" 2/8 = 1/4
4 flips = 2^4 possibilities - teo possibilities are suspicious. 2/16 = 1/8

Now it's up to you to decide if the phenomena noted in the forums and in global chat by multiple people is more like 333,333 appearing in a sequence of digits (10^6 possibilities), or more like a coin being flipped a mere 20 times (2^20 possibilities)

"

Come on, people. Drop rates are not set for each player and each run individually. They are across all the servers. So if something has a 1 in 10 chance of dropping, and 10 people do 10 runs each at the same time, what are the odds that each person got that drop once? If one person didn't get that drop and another got it twice, is that no longer random?

If one person gets 4/2/1/0/0/0/1/0/1/0 ("cool coincidence!"), another person gets 0/0/3/1/1/0/2/1/2/0 ("intuitive."), another person gets 0/0/3/0/2/1/4/3/1/0 (cool... co..incidence?), and yet another person gets 0/2/0/5/4/0/0/0/3/0 (ummmm..........), then there is a statistically significant chance it's no longer random. I think. My examples might not have been strong enough. Basically, you calculate the chance that a person gets 4 in one run within 10, multiply by itself once (because it happened twice) and calculate the chance that a person gets 4 in two runs within 10, and multiply by the previous result. Then you add that up with any other "similarly suspicious phenomena," and compare with a familiar probability, like the chance of winning the mega millions, or the chance of two tellurium-128 isotopes decaying within a second of each other.

The people who are denying any form of RNG tweaking or streakiness based on factors that we don't currently know are just as bad as people who are claiming that it is 100% true.

Blind faith in a system you have absolutely zero information on is not being rational.
The answer is we don't know. Saying it's entirely random without knowing is just as ignorant as saying that GGG is pulling strings to make certain items drop at certain times for certain players.

For some more anecdotal stuff, i had three damnings drop in 3 back to back docks runs, followed by two last resorts back to back after the damnings.

"

Xendran wrote:

Blind faith in a system you have absolutely zero information on is not being rational.
The answer is we don't know. Saying it's entirely random without knowing is just as ignorant as saying that GGG is pulling strings to make certain items drop at certain times for certain players.

THIS!

See, it's a bit of my bad for stating my thread topic as if I were 100% sure of it. I'm not. I haven't done the science!

But neither have the people who are claiming that RNG is being RNG.

Personally, I don't think GGG is "pulling any strings," I think there's just a problem in their code generation that's unintentionally generating streakiness.

Check out the post of one of the people who (I think?) was trying to disprove my statements:

http://boallen.com/random-numbers.html

Would you rather have your drops determined by Random.org, or PHP rand() on Windows?

Which one do you think is closer to what GGG uses?

EDIT: The bottom line is that after GGG is done tackling more important issues like completely revamping the synchronization code, it's definitely worth it to take a look at their random number generator and make sure that it is functioning acceptably. The game has grown a lot, and maybe what was "acceptably pseudorandom" for 10,000 players is not acceptable for 1,000,000 players (or however many POE has).

"

adghar wrote:

The odds of winning the mega millions jackpot, according to Wikipedia, is "1 in 175,711,536"

That means, if you see "9999999999" in a "random number generator," then there are two possibilities:

1. It was randomly generated. There is roughly a 1 (or 9, depending on if you count 3333 etc.) in 1,000,000,000 chance of this happening,
2. There is a problem with the random number generator causing a sequence of numbers to repeat more than they should. There is roughly a 999,999,999 in 1,000,000,000 chance of this happening.

Tack on just 5 more nines and you can be 10,000 times more sure that you'll win the Mega Millions lottery than that the "random number generator" is random.

If that's an example of what makes you suspicious of PoE's RNG, it's no wonder your accusations are so incoherent and scattershot. You don't realize, do you, that there's nothing special about the number "9999999999", that's it's no more or less likely to come up than any other number in a lottery?

So basically, your argument is that when you happen to notice certain types of "special" events more or less often than you expect, it must be more than a coincidence, there must be some intricate, systematic underlying mechanism that explains why these things always seem to happen to you and so many people you know? Guess what, it's much bigger than PoE, it's universal, lots of people have caught on, and here's how it works:

http://www.horoscope.com/

"

Xendran wrote:

Blind faith in a system you have absolutely zero information on is not being rational.

It's impossible to be a functional human being without any faith.

Consider how many systems you interact with on a daily basis. Do you really know that much about every single one of them? From the device drivers on your PC to the produce at your grocery store to the maintenance of the building where you work, there are too many things in the world that we all interact with for each person to learn the process behind each of the end results. At a certain point you just have to take some things on faith.

That faith is based on trust. It's good to be passionate about things, but if you trust GGG then you should have some faith in them too. And if you don't trust GGG, then why did you give them your email address, your time, and your musings on any subject you have typed about on these forums?

"

RogueMage wrote:

Tack on just 5 more nines and you can be 10,000 times more sure that you'll win the Mega Millions lottery than that the "random number generator" is random.

If that's an example of what makes you suspicious of PoE's RNG, it's no wonder your accusations are so incoherent and scattershot. You don't realize, do you, that there's nothing special about the number "9999999999", that's it's no more or less likely to come up than any other number in a lottery?[/quote]

Have you studied any statistics, as in pre-calculus or calculus class, or beyond? (Can't remember if algebra covers it). Really hoping mods won't scrub that question, as it's a legitimate question. Oh, and biology is a good class for learning about statistics as well.

9,999,999,999 is no more or less likely to come up than 1,835,242,593. The problem is that it has 9,999,999,999 other competitors for an outcome, each of which is "unremarkable" in some way. You don't seem to have recognized the analogy I've posed. In this analogy, we aren't even looking at 9,999,999,998, we recognize it as a natural phenomenon. Naturally, it's up to us to select what is "suspicious" or not. But the term "Occam's Razor" persists in science for a reason.

If one poses the question, "what is the probability that a true random number generates 11 of the same decimal digits in a row?" you have a clear answer for that - 1 in 10,000,000,000 (may have carried one digit wrong, not overly important). Naturally, if you include other significant sequences like 1234567890, then the probability of a "suspicious" number increases, and the probability of a phenomenon distorting the probability decreases. But even if you had come up with 100 different "suspicious" numbers that could result from the production of 10,000,000,000 digits, that's still about a 1 in 100,000,000 chance, less likely than winning $250,000 from the mega millions, but more than winning the jackpot. This makes the analogy a bit more similar to the POE streakiness question, but the scale is still not congruous with the expected number of outcomes.

PolarisOrbit replied on page 1, "What are your expectations?" This is the important part. When you observe streakiness in isolated cases, as in my three anecdotes in my original post, you can make a case that it's just a coincidence chalked up to randomness. However, when you have potentially millions of players to look at, and pretty much literally countless examples of streakiness, you ask - Is that number of streaks expected for the amount of players we have? How many standard deviations away from the mean is it? Is this number of streaks in the 95th percentile (outer edge of the normal curve)? The 99th? 99.99999999th? Basically, the further away from the center of the normal distribution you go, the more suspicious you SHOULD get. Think about the probability of the sun rising each day. Is it randomized, or is there a natural phenomena behind it that is affecting the probability? Since the rising of the sun was so dependable, scientists got "suspicious" and went and found the phenomena. I think GGG might have a phenomenon worth looking for inside their code.

Also, why are you saying I'm making "accusations?" Isn't finding irregular results in programming, then attempting to find a solution, usually called "troubleshooting"?

Naturally, I must apologize for sounding "incoherent" and "scattershot." My mind prefers to think in analogies, and since POE isn't a science journal, I feel no obligation to reign things in and provide scientific proof because that's work, not play. (for me, at least. I prefer to nerd out without my ability to be heard being limited by the judginess of others.) I'm just a dude who wanted GGG to look at their random number generation code after other higher priorities are sorted out. That's the bottom line.

"

PolarisOrbit wrote:

And if you don't trust GGG, then why did you give them your email address, your time, and your musings on any subject you have typed about on these forums?

Because they made a fun game.
I don't trust GGG with a large number of things, including balancing of difficulty, item progression, and unique item design. I do, however, trust them with my not-so-personal information. You do not have to trust everything somebody or some group of people does in order to trust them.

Also, if i trusted GGG's game design decisions fully, i WOULDN'T be making feedback posts, because i wouldn't feel the need to point out flaws or necessary changes since i'd trust them to fix it.

Also, has GGG come out and stated matter-of-fact that there are absolutely no factors that can modify the RNG?

There's already one that is public knowledge in the game, quality fusing and jewelling. And we know how much GGG likes to hide things from players. It is not tinfoil hat to think that there is a possibility, and potentially a high one, of there being RNG influencing factors based on GGG's previous history.

It's not like they're afraid to hide things that change systems in a large way. See: 20% Gem recipe.