RNG conspiracy thread #718

Grim Dawn Gameplay Discussion
1

Play a different character. Que the RNG cult 3,2,1. I am convinced the seed is somehow generated from your class, maybe some other things, but it is character specific. I believe if you keep running circles with the same toon the items that drop rarely will rarely/never drop. It will get denied, ill be called a conspiracy theorist, but in every case where a piece of a set would not drop on one character, and where certain items rolled over and over, changing the character changed the items that rolled over and over and I got the item that would never drop. Ran for hours and hours through level 35 content looking for miasma hood…started a new toon and got it on first run. Could be coincidence, but it is not a matter of sample size. I have had approaching hundreds of identical level 35ish epics drop at the exclusion of others on some characters, only to get it frequently on another character. RNG is not random, it is Pseudorandom, it is a deterministic approximation of randomness. It is very possible to get repeated, deterministic outcomes, excluding some possibilities, if not used with care. Its not the sample size or in your head, its the small event pool size and something to do with the seed , that causes repetitive drops.

And this is why i play 5 characters. Mid level epics, I have most of them, only after a few hundred hours of getting identical ones. Epic wands still seem to be a unicorn though, while Nightblade gear in general drops copiously on most characters, but that might be because half of the gear is for nightblade.

2

And I have, uh, 23 I think :stuck_out_tongue:

Your whole RNG story is, sorry to say, nonsense.

3

you can be convinced all you want, it simply is plain wrong. The seed is the time you start playing, nothing else.

It will get denied, ill be called a conspiracy theorist, but in every case where a piece of a set would not drop on one character, and where certain items rolled over and over, changing the character changed the items that rolled over and over and I got the item that would never drop. Ran for hours and hours through level 35 content looking for miasma hood…started a new toon and got it on first run. Could be coincidence, but it is not a matter of sample size.

it is coincidence and not only is your sample size way too small, your memory is playing tricks on you too as you only remember the cases where this happend, not the ones where you found the item on the first char or not on the second char either.

By focusing on items that did not drop for a long time, you are already creating a self fulfilling prophecy by ignoring all the items you found and therefore do not fit your narrative

RNG is not random, it is Pseudorandom, it is a deterministic approximation of randomness. It is very possible to get repeated, deterministic outcomes, excluding some possibilities, if not used with care.

yes, the above is correct in theory, there are a few problems though

  1. there are some easily implemented RNGs readily available, this takes care of the ‘not used with care’ part
  2. the RNG result is a 32bit number while you have maybe 100 items to choose from, so whether you get 33,458,961,013 or 33,458,961,014 as the result makes next to no difference when mapping the full range down to 100 values. So excluding possibilities, even if it were happening, would not actually make a noticeable difference

Yes, it actually is all in your head.

4

Hi Mamba, its been a while, Ive missed you.

Seed is time you start playing, interesting. So someone typically sits down at 6:30 to play a game they are going to get the same seed every time? Time to what precision? Because people are remarkably precise machines of habit.

You win, not gonna argue. Everyone else play some alts and you will find your stuff.

5

not sure how precise your clock actually is or what routine they use to get the time they then use as the seed, certainly down to at least milliseconds, if your PC clock were that precise

You win, not gonna argue. Everyone else play some alts and you will find your stuff.

yeah, that is kinda arguing since you double down on what you initially said :wink:

So just to reiterate as well, the seed is not at all based on the char you play. So any theory you derive from that assumption is utter nonsense.

6

That is why I say “and other things perhaps”. It is something that persists, or has a high chance of persisting from play session to play session. Influenced by class at the very least. Like the number that would have made Bloodhound Grieves for Class A makes Brimstone Pauldrons fall for Class B.

In the Necropolis around level 50, for Druid it was Eastern Legs, Eastern Legs, Eastern Legs. Now for Witch Hunter it is Shadescale Legs, Shadescale Legs, Shadescale Legs. I’m keeping a log.

7

There are no class-based parameters for the Loot System.

The only persistent rule between sessions is Blueprints, where the game will try to reroll a blueprint drop several times if you already have the blueprint it generates.

Within a session, the game attempts to reroll unique items that you got in That session, but duplicates can occur, especially in level ranges where there are few options.

8

If it is not based on your char, it is also not based on your class, which is part of your char :rolleyes:

It really is very simple, you just refuse to accept it: nothing whatsoever about the RNG is dependent on your char (its masteries, items equipped, name, time of creation, what have you…)

9

Pixies and gremlins it is then.:stuck_out_tongue:

10

And now for the comment that’ll trigger Rhis…

Something-something “Good Seed Bad Seed”

In TQ the drop system always felt a little wonky and extensive research into the system was done by the community with mixed results. Sadly, most of that research has disappeared alongside most of the TQ forums during the latest hoopla. Though the TQ devs continually denied that there was really anything wrong with the RNG behind the loot system, patch 1.32 of Titan Quest: Anniversary Edition notes the following:

Update #4 (1.32) Random Number Generator Fix, Editor Fix, Better cheat protection, Gameplay Item Tweaks

Update #4
19th October 2016
GENERAL

  • Fixed random number generator bug that caused duplicate item drops.

Which rather makes me wonder what was really going on, whether that ‘bug’ was made by THQ Nordic or was in the base TQ game, and whether that ‘bug’ exists here in GD.

I don’t really know how we got here, but a discussion of GD’s loot system, while certainly a valid topic, is not relevant to this thread as per the first poster’s topics.

11

GD did not inherit TQ’s…RNG oddness…

12

It is not. I apologize, I made a side comment to someones comment directed at something else and the thing went super nova. Conversations evolve. Thank for your inflammatory contribution while reminding us that we have strayed from topic. I so want to run with it and gloat but Ill hush.

13

Ok lets discuss why Rng issue threads had an increase lately.

So player amount is about to reach 1.1m and unlucky amount is also increasing as you suggest?

14

more players = more superstition, as you suggest ?

I am not even sure I agree that we had an increase in RNG related threads lately.

If you want the code for the GD RNG to dissect it and settle this once and for all, I am sure it can be provided :wink:

15

or more players = more meta data that looks perfectly random and reflective of the distribution in total as it should but /= perfectly random at the user level. Or is that not collected when we opt in?

16

I would be surprised if this were collected - and unless there were a huge mistake in it, I doubt collecting it would show anything.

Also, if something looks perfectly random and of the right distribution across users, then it is, regardless of how a single user during one session feels about it :wink:

Users get just a small sample of values of one specific seed (the full range is longer than your game session, most of the numbers rolled are probably for fight calculations, not item drops) and somehow think that this minuscule data set is enough to extrapolate to the entire RNG.
What is more, they remember the few cases they found odd, not the many where they did not.

I do not think looking at any sample will ever prove anything here, for that we have far too small a sample set.
Up to this point the side that maintains that there is an issue has shown nothing but a lack of understanding of RNGs in general and a lack of understanding of this RNG in particular, frequently claiming things things that have been debunked long ago already (like what makes up the seed…).

Unless you group together and collect a huge dataset that at least shows that there is reasonable doubt (which so far has not been done), you are better off looking into the algorithm itself if you want to prove your point.

17

No it isn’t. No one cares how overall random it is, they only care is it random enough on my end that im actually someday going to get such and such an item.

Question was, do they get the same small sample each time they play, does it start back at the top when you start anew, and does it persist in any way from session to session.

Hundreds of hours, hundreds of drops, on the same character at level 50+ is not too small a sample, at least not for the legendaries that drop globally after 50. In fact, that distribution is exactly what you need to know in order to settle the matter.

After hitting 50 and rolling your first legendary, with the number of legendary items there is no reason to expect the same item again very soon. It is not impossible but should only happen for a small minority of people, yet right in line with my previous perceptions, i just sold off 3 Venom Grieves. Confimation Bias or Bias Confirmed? I have only been 50+ for a very short time. I must be the most unlucky guy in random loot history, along with everyone else who is observing it.

If it were incidental, happened once or twice, id agree, stroke of luck. But my luck with this RNG has been fairly constant and predictable.

18

actually, yes it is, even if your one session is a big exception, such exceptions do not mean the RNG is not working properly

Question was, do they get the same small sample each time they play, does it start back at the top when you start anew, and does it persist in any way from session to session.

when you start anew, you have a different seed. If you magically happened to get the same seed again, you would get the exact same numbers in the exact same sequence as last time, but unless you could completely replicate your session you will still get different drops as RNG is not just used for drops but also eg for damage dealt, both by you and mobs. Nothing persists.

Hundreds of hours, hundreds of drops, on the same character at level 50+ is not too small a sample, at least not for the legendaries that drop globally after 50.

hundreds of uniques at level 50 ? I agree, your RNG is broken, I have a lot less by then :wink:

And 100s is way too small a sample size, the first size that is anywhere close to relevant is maybe 10,000 - personally I would want one order of magnitude more still

In fact, that distribution is exactly what you need to know in order to settle the matter.

yes, but for a way bigger sample…

After hitting 50 and rolling your first legendary, with the number of legendary items there is no reason to expect the same item again very soon.

most are higher level, so excluded from the drop pool for a level 5x char

19

Points in order.

As a matter of fact, all of the legendaries i have gotten after 50 over 3 characters running the end game…Plains to Bastion, have been in duplicate, sometimes triplicate, the epics as well. The eastern Legs definitely on the Druid, the Shadescale Legs on Witchhunter, but that is the very, very short list. I have sold many and kept many.

I am not sure about the seed. Maybe it is supposed to get a new seed, but everything on my end looks like the seed is staying put producing a high probability of getting the exact same rolls. And its not like I am farming this content, this is just to and fro, complete a quest, end session, come back later, advance more.

hundreds of uniques at 50? that is not even remotely close to what i said or implied. [Aisde: But a close to a few dozen (of the same 4 or 5, over, and over, and over again) for me at just a few hours of 50+ content already, they aren’t as rare as I expected.]

And no, when i can go back to that same content the next day and start predicting the drops, it has nothing to do with sample size, it has to do with your code. I may just look at it sometime, and when i find your problem it will be a very, very expensive solution.

20

are you expecting to not get duplicates until you have the better part of all legendaries ? That seems to be a deeply flawed assumption…

I am not sure about the seed. Maybe it is supposed to get a new seed, but everything on my end looks like the seed is staying put producing a high probability of getting the exact same rolls.

not even the seed staying the same would result in this, I already explained why

And its not like I am farming this content, this is just to and fro, complete a quest, end session, come back later, advance more.

that might actually be part of you problem as the game already tries to reduce the number of duplicates you get within one session… not that this by itself would explain why you get a lot of duplicates, but certainly why you would get more than average

hundreds of uniques at 50? that is not even remotely close to what i said or implied.

it certainly sounded like that to me. You said

so unless you were referring to legendaries (not even epics), I am not sure how the 100s of (then presumably yellow and green) drops would be relevant for the legendary drops you got

But a close to a few dozen (of the same 4 or 5, over, and over, and over again) for me at just a few hours of 50+ content already, they aren’t as rare as I expected.

And no, when i can go back to that same content the next day and start predicting the drops, it has nothing to do with sample size

a) you still have a tiny sample size
b) bias confirmation at work
b) running the same areas over and over instead of going through the game might compound the problem. For one I am not sure which mob uses what loot table, but that might contribute to you getting the same drops, for another short runs definitely mess with the logic that makes duplicate drops less likely

it has to do with your code. I may just look at it sometime, and when i find your problem it will be a very, very expensive solution.

I am sure we will hear from you if you find something :wink: