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

"

adghar wrote:

"

mark1030 wrote:

Are you comparing winning the lottery 6 times in a row to getting the same unique to drop 3 times in an hour? Because that's the extent of this streakiness that people are saying proves that RNG is broken. I think the drop rates are a little higher than lottery odds.

Streaks are just part of randomness. Random doesn't mean you will always have different outcomes, which seems to be what people think the RNG should deliver. Rolling a 6-sided die 6 times will more than likely not roll each number once. Rolling the number 3 twice or three times in a row does not mean it isn't random. That's the extent of this streakiness people are claiming. I haven't seen any that say "I found 6 of the same uniques in a row". Three is the most I've seen.

One point - I'm looking at a bigger picture. Rolling a 6-sided die 6 times with 3 3's in a row is pretty normal. Several people rolling 6-sided dice and getting 3- and 4-streaks (like, say, 85 out of 100 people) is grounds for suspicion.

I think this is the point of yours that I'm stuck on. You just said 85 out of 100 is where there'd be ground for suspicion. I totally agree. But when you bring that into this game, you're way way way over exaggerating the number of people who are reporting this behavior. How many thousands of players have played this game? And how many people have reported seeing this "problem"? 85%? 75%? 50%? 25%? 10%? 5%? I'm probably overestimating when I say that 1% of the players have complained about it. Just because they are vocal and repetitive about it doesn't increase their numbers. How many actual individual people have you counted? Small streaks are common in random events. Long streaks are less common but not impossible. But I'm not hearing about long streaks. I'm hearing about 2 or 3 of an outcome.

I don't watch twitch but I often see the name of Krip mentioned who apparently played a ton. Is there a thread in here somewhere where he talks about this "problem"? If anybody would have noticed, it'd be him.

There's a fascinating high amount of people who dont know statistics.

It seems the same with all games. Like in poker. 2 hands, 2 Times AA? Game is rigged! 8 times AA in half an hour? Dealer cheats! 15 times red in roulete? Can't be, this is against the law of physics!

Yet, all this happens. Why? Because there are a lot of people playing.
Some dozen occurences of some mildly streaks of a few people do NOT prove that the RNG is bugged - it's THE PROVE it works correctly.

FYI: I check IT security for a living. Real randomness is a basic prerequisite of every crypto system. The RNG implementation for PoE has even not to be cryptografically secure - means it can use the same high-speed implementations of RNG's also used in statistics, like for monte-carlo simulations of future bond development in the finance sector.

"

siliconlife wrote:

There's a fascinating high amount of people who dont know statistics.

It seems the same with all games. Like in poker. 2 hands, 2 Times AA? Game is rigged! 8 times AA in half an hour? Dealer cheats! 15 times red in roulete? Can't be, this is against the law of physics!

Yet, all this happens. Why? Because there are a lot of people playing.
Some dozen occurences of some mildly streaks of a few people do NOT prove that the RNG is bugged - it's THE PROVE it works correctly.

FYI: I check IT security for a living. Real randomness is a basic prerequisite of every crypto system. The RNG implementation for PoE has even not to be cryptografically secure - means it can use the same high-speed implementations of RNG's also used in statistics, like for monte-carlo simulations of future bond development in the finance sector.

Amen to that.
People often do not realize that a statistical experiment (and playing PoE can be called one) has to be repeated a lot of times to give any significant data. So either you do thousands (literally, and since the word is often used exactly the wrong way round, I mean literally literally) of Piety runs, collect the data and let mathematician analyze it (if you're not one yourself) or you accept that with RNG anything is possible.

I don't know which types of lotteries are popular in other countries, but here we have a game called Lotto. 6 numbers out of 49 are drawn randomly and if you have 3 or more correct on your ticket, then you win an amount of money depending largely on how many of those six you had correct.
The drawing machines are controlled by an official before every game. Yet I know many people (my mother, other relatives) who think that something must be wrong when the numbers are too regular, like 2,4,6,8,10,18. I guess news and papers would be abuzz with speculation if the machine was somehow corrupted if ever 6 consecutive numbers were drawn or 1,2,3,4,5,6.
Yet at some time even 1,2,3,4,5,6 must be drawn if it is really a random event.
Like siliconlife said: Streaks are prove the RNG is working correctly and not (arbitrarily) filtering out "unlikely" results.
I'm not that rich a player, only have 3 exalteds (all in all I've got more than 3 exalteds worth of stuff, though). These 3 dropped in a relatively short time frame. I'm playing since February, Found the exalts in June I think. None before, none after.

just popping in to report that I found FOUR 17% Groundslam Gems in normal fellshrine yesterday in a span of 45 mins...2 in the same run. Strange things are afoot at the Circle-K.

"

kolp wrote:

just popping in to report that I found FOUR 17% Groundslam Gems in normal fellshrine yesterday in a span of 45 mins...2 in the same run. Strange things are afoot at the Cirle-K.

WOW!


COuld you drop that comment here please, if it's true : http://www.pathofexile.com/forum/view-thread/517989

"

SL4Y3R wrote:

While I will agree the streakiness does seem unusually high, ive actually thought the rng is tied to something. Not gonna say what though.

I also wouldnt count rgb links in that though. Personally, those always seem to drop properly random.

all random number generators are tied to something a lot of times programmers will tie it to the time or the system clock because it is very hard produce the exact same number. I doubt this is how GGG has done it though

It's actually quite surprising to see this thread still going and people comparing AMAZING stories of that one time something memorable happened and thus it disproves the tens of thousands of times it didn't that you don't remember.

The mind loves a good pattern and it will even create them for you. It doesn't mean it really is there.