Think i know why RNG is inconsistent in PoE
" A good point to start is the Wikipedia page on the geometric distribution: https://en.wikipedia.org/wiki/Geometric_distribution There you can find the formula for the cumulative distribution function (CDF) as CDF(k) = 1 - (1-p)^k Thus, the probability that if you start now, you have to wait more than k tries until you successfully corrupt a gem into a level 21 gem is: P(k) = 1-CDF(k) = (1-p)^k So if the chance of a level 21 corruption is 1/10, then the probability that after 40 tries, you still haven't gotten one is: P(40) = (1-1/10)^40 = 1.48%. Or, the other way round, about one in... 1/P(40) = 67.65 ... approximately 68 corruptions will require you to corrupt at least 41 gems to get it. Remove Horticrafting station storage limit.
|
|
|
Read the whole thread xD.What a waste that was.
Now to my actual mathematical findings regarding the inconsistency of rng in poe: If it wasnt inconsistent it wouldnt be pseudo-random generation of numbers(please understand that with the same seed the pseudo random number generation would infact be consistent(just minimizing angles for attack by hyper aggressive forum posters)). Op trying to skew rng in his favour by corrupting 2 bad gems after a successful corruption is pure superstition. The 2 bad gems are as likely as any other you corrupt to have a good outcome. I didnt do good in math in school. You have to remember to chase and catch your dreams, because if you don't, your imagination will live in empty spaces, and that's nowhere land.
|
|
" Seems like you got the gist, though, be it because of math in school or for some other reason. Other than the OP. So I guess it doesn't matter what your math grades were ;). Remove Horticrafting station storage limit.
|
|
|
A seuquence of randomly generated numbers between 1 and 7. Each individual number has a chance of 1:7 to be 1. Chance for 1 to appear 2 times in a row is 1:49, 3 times 1:343 etc. Dont have the data anymore, but from top of my head in a sample of 10000 numbers there were:
1050 of 1s standing alone; 150 groups of 11; 15 groups of 111; and 4 groups of 1111. If we were to apply certain patter, how that would affect the end result? For example: if we get 1, skip the next number. If skipped number was 1, skip again. Skip until last skipped number is not 1. Continue picking until next 1. Here is how it looks visually:
Spoiler
 It never was a statement, but a theory. Some people missed that. Yo can look at original post, it wasnt edited. Put that theory to a test and made sure that there is no difference. It was an interesting excercise. "The only thing necessary for the triumph of evil is for good men to do nothing" - Edmund Burke Last edited by Toshis8#1464 on Feb 5, 2021, 11:27:36 PM
|
|
|
so youre saying i have more chances to get tails if i got heads prior?
thats just not how things work |
|
" You should ask your math teacher something about random numbers distribution and probability instead of stare on screen. It's not easy to get it, but it's very interesting. Give it a try. Last edited by Kminek666#4837 on Feb 6, 2021, 9:57:09 AM
|
|
|
Any time one of these threads comes along I like to post a link to wolfram alpha:
https://www.wolframalpha.com/input/?i=100+values+from+the+geometric+distribution+p%3D1%2F1200 " This is 100 values from the geometric distribution with p=1/1200 (let's say this is the 6 link probability). In other words, how many rolls/attempts until you hit that 6 link if you kept on trying. The reason RNG is "inconsistent" is that although the average number of rolls until success will average the probability, you can expect a massive variance, and the kinds of numbers you will see are quite scary (either very large or very small). This is why you may get two 6 links within 100 fusings or spend an entire league trying to 6 link. Using 2 vaal orbs on trash gems after every success is not doing anything. It's as if you were rolling a die for sixes, and after every 6 you take out another die and flip it twice, trying to "use up" the bad rolls. Not going to work is it? The only way this could work is if the rolls were not independent. I don't see why GGG would need to make this the case because the natural behaviour you get from the geometric distribution is already enough to get people addicted. edit: accidentally said "flip a coin" when I meant roll a die Last edited by Userbla#2018 on Feb 6, 2021, 9:36:22 PM
|
|
|
Told you... Heads.
|
|
" With that specific way of choosing random numbers, it would not affect the result at all, because you decide whether to use the next random number before you look at it, and all the random numbers are independent. With those two conditions satisfied, it doesn't matter how you choose your random numbers (because they are all identically distributed, i.e. uniform distribution between 1 and 7, and independent, just like one throw of a coin is independent of the one before). It would be a different story if you rolled until a 1 happened, and then skipped the next number if it was a 1. Because then you would be deciding whether you use a random number only after looking at it, and that affects your results. Mathematically speaking, the numbers in your string of random numbers would not be identically distributed any longer (the ones appearing after a 1 would be less likely to be a 1). Remove Horticrafting station storage limit.
|
|
