RNG conspiracy thread #718

Grim Dawn Gameplay Discussion
81

ok, as far as I can tell one has not been proven in any way, shape or form yet however (admittedly it is not easy to prove either, even if something were wrong), so let’s focus on 2 for now

When I say that perhaps Crate should cater more to RNG complainers, I was referring to consideration of the second issue. Because when people complain about the RNG, often what they’re really complaining about is the lack of dependency on previous events.

as I mentioned earlier, to a small degree we have that with the game keeping track of which uniques dropped during your session and rerolling if the same one drops again (but if it then still is a duplicate it still drops) to reduce duplicates.

If you want to go beyond that, it might become rather involved, but it would be an option, at least in theory.

I don’t know what you’re saying here: first you said “given the number of units sold I’d say we have a negligible amount of RNG complainers” implying an inverse relationship between # of units sold and # of RNG complainers. Then you said, “the more users there are, the more users that will complain about something” implying a direct relationship.

I did not imply an inverse relationship, I was pointing out that despite the sales this is a very small number of players, so a very small percentage, something like 0.002% or so.

Whatever the case, like I said both of us are merely speculating at this point with zero actual evidence.

agreed

  • HTML scrape reviews on Steam for keywords like “RNG”, “dupes”, and “shitty drops”. The official Steam API doesn’t support retreiving review information (yet).

anecdotal evidence at best. If I roll a die 10,000 times, do you expect me to get four 6s in a row or not ? The likelihood of that happening more than once in that sequence is actually quite high. So if enough people roll a die 10 times, some are also bound to have four 6s in a row and then wonder, even if that is only to be expected if you just do it often enough.

82

I gotta say that it doesn’t really make me wonder that there are some kinda esoteric views regarding §RNGs.
Judging from my experience RNGs actually are a pretty esoteric matter as they are so complex and there are so many properties that can be attributed to them that even widely renowned experts get them wrong (see Marsaglia’s “KISS” for example); and that’s just on top of them being basically blackboxes.

I’d wager that it is entirely possible that for any given seed there would be items that would, de-facto, be rolled less often than they should be for the usual player - just because the sequence of numbers that would have to be generated to get this item would be generated less often (or even never) within the usual duration of usual gameplay.
And the only thing that Crate could hypothetically do would be to craft a mathematical model to statistically analyze the state of the RNG throughout gameplay and basically feed it every possible seed and do this for every possible state of the game, that is for every possible succession of actions within the game that involve the RNG, just to see if all items are really generated as often as they should be during that usual duration of usual gameplay.
Given the absurd amount of possible states of the game this would be a pretty futile attempt though and Crate might rather consider working on establishing a colony on Alpha Centauri.

I mean, every time the RNG returns a number this advances the state of the RNG, so every entity generated and every attack made and every skill used (and so on…) changes the state of the RNG.
And the sequence that these happen in obviously determines which result you get for e.g. determining the drops of a lootable container.
Maybe the number that the RNG spat out when you delivered the final blow to the boss would have been the number needed to drop that one sought-after item from it’s loot orb? If only you had not attacked that one lonely cultist standing at the entrance to the lair!

So while it might be possible that items, for a significant number of players, based on their usual style and duration of play, are not generated exactly as often as their drop-chance might imply, the RNG would really have to have a huge flaw for that flaw to be actually perceivable, in a game that generates that much entropy by itself just through gameplay.
And so all in all I’d say that Crate can’t do much more than seeing if players complain en masse about this or that item never dropping even though it should, or dropping far more often than it should.
If that would be the case and adjusting the drop-chance wouldn’t help then that would be an indicator for that huge flaw in the RNG.
Something potentially possible, something very unlikely and, given that I can’t remember having heard of such an outcry and the subsequent inability to fix the problem, something almost certainly not being the case.

And I guess that’s what mamba wanted to say - the more players that play the game the greater the chance should be that such a huge flaw in the RNG would come to light. As the game has sold pretty ok it’s pretty likely that we would have been able to reveal such a flaw by now if it existed.
I too think that what quite some people complain about is in fact randomness itself, not the lack thereof, and that they want the way that items are generated to be more deterministic as to accommodate for the players’ needs. I’ve had that feeling myself quite some times.
Sadly it’s not that easy to fix this without making it too easy to get your hands on items that are supposed to be rare.
Because in the end it’s those rare items that people will mostly be hoping to get dropped, right?
What would you care to have every existing epic item in your stash when that legendary you desperately need hasn’t dropped in twohundret hours of hardcore-grinding?

83

Pretty much this. I meant that if there were a huge flaw in the RNG, we would see more complaints about it. The absence of these complaints is therefore a strong indication that there is nothing wrong.

I too think that what quite some people complain about is in fact randomness itself, not the lack thereof, and that they want the way that items are generated to be more deterministic as to accommodate for the players’ needs. I’ve had that feeling myself quite some times.

would agree with this assessment too

84

creating/implementing a really good random number generator is an art. There
is a range from flawed to nearly perfect.
It is not just “it´s working or it does not”. So it could be possible, that there are some smaller flaws in it.
Considering Titan Quest, there were some really obvious anormalities in the
drop rate which cant be explained by RNG.

In Grim Dawn, hmmm… there are some hints of repetitions in item drops, which could indicate a discrepancy from RNG.

regads Arkon

85

By session, do you mean after the game starts and before the game exits? Or when you’re playing a certain character?

Yeah, going further might become more involved. Two options I can think of:

  • Keeping track of previous unique drops, for each character, as key-value pairs (item id => # of times dropped) stored in the character’s file. Anytime a unique is rolled to drop, the game would query how many times it’s dropped before for this character and make a decision accordingly (keep or re-roll). Going even further, the value could also contain a timestamp of when this item was last dropped and this data could also be factored in.

  • Scanning either the character’s or all characters’ stashes, including the shared stash, for rolled unique item existence and making a decision accordingly.

anecdotal evidence at best. If I roll a die 10,000 times, do you expect me to get four 6s in a row or not ? The likelihood of that happening more than once in that sequence is actually quite high. So if enough people roll a die 10 times, some are also bound to have four 6s in a row and then wonder, even if that is only to be expected if you just do it often enough.

The point of that scraping is to get a better idea of what % of players are RNG complainers. With this data we can better assess the psychological impact of the current RNG/loot drop implementation. Like it or not, perception is reality for most people, even if there’s nothing techically wrong with the RNG from a maths/CS point of view.

The point of that scraping is not to provide evidence of RNG broken-ness. I thought we weren’t talking about that now?

Pretty much this. I meant that if there were a huge flaw in the RNG, we would see more complaints about it. The absence of these complaints is therefore a strong indication that there is nothing wrong.

I think what you mean to say is that if the RNG were broken, we would see significantly more complaints than we see now. That is probably true.

But, we can agree that insofar as these forums is concerned, we still get a lot of RNG complaints. Ever heard of, “the customer is always right, even if they’re wrong?” Most people aren’t very good at maths and don’t understand probability well. The key question is, assuming the RNG implemenation is correct, should Crate do more to satisfy these complainers?

86

One session = select char, press ‘Start’, play, press ‘Exit’ to stop playing with that char

If you want to keep stats across that definition of session, it would be harder than if you only tried to do so within one imo.

Yeah, going further might become more involved. Two options I can think of:

  • Keeping track of previous unique drops, for each character, as key-value pairs (item id => # of times dropped) stored in the character’s file. Anytime a unique is rolled to drop, the game would query how many times it’s dropped before for this character and make a decision accordingly (keep or re-roll). Going even further, the value could also contain a timestamp of when this item was last dropped and this data could also be factored in.

yeah, that probably would be the way to go if Crate wanted to go down that route

  • Scanning either the character’s or all characters’ stashes, including the shared stash, for rolled unique item existence and making a decision accordingly.

the downside to that one is that you dismantle the items you do not want and keep the ones you want (either for this char of to hand them to a different one), so this could increase the likelihood of unwanted drops

Like it or not, perception is reality for most people, even if there’s nothing techically wrong with the RNG from a maths/CS point of view.

I agree and I do not like it :wink:

There are so many cases where people are convinced of the stupidest BS against factual evidence to the contrary, just because it is their conviction, that I see where this is hard to overcome.
At the same time it is hard to fix an imagined problem… which leads us to the above attempt at actually steering away from a proper RNG

The point of that scraping is not to provide evidence of RNG broken-ness. I thought we weren’t talking about that now?

fine, but then to me collecting that data also becomes less interesting as it no longer is about facts but about perception. You can still do so, but to me it is futile then.

I think what you mean to say is that if the RNG were broken, we would see significantly more complaints than we see now. That is probably true.

as far as I can tell that is exactly what I did say. I definitely agree with this.

But, we can agree that insofar as these forums is concerned, we still get a lot of RNG complaints. Ever heard of, “the customer is always right, even if they’re wrong?”

I agree we have a few complaints, I do not consider them a lot.

I also do not agree to ‘we have to accept every opinion as valid and valuable, no matter how delusional’ even when you repackage it as ‘the customer is always right’.

That being said I have no issue with making the RNG more ‘fair’ (which really means less perfectly random).

The key question is, assuming the RNG implemenation is correct, should Crate do more to satisfy these complainers?

depends on how easy that is, I definitely am not opposed in principle

87

One thing I’ve forgotten about is - what if I actually wanted more dupes of certain items?

Suppose that my first character is a DW lightning build and I wanted dual Crystallums? Or if I had a lot of level 50-ish characters and wanted Runic Bracers for many of them?

We hear of people getting dupes of stuff they don’t want, but what about dupes of stuff they do?

Going back to the stash-scanning proposal: Maybe the game could also keep track of uniques that were left on the ground, sold, traded, or dismantled as well? That way, the game can know that kinds of things players want and what they don’t. But then that brings up another issue: what if a patch made a certain item much more attractive? Beronath Reforged comes to mind. If the player sold/dismantled this item many times before, the game would decrease its appearance, despite how valuable it might be or would become.

Given this, other unexpected consequences, complexity, and time/labour of implementation, I would now advocate for the status quo, assuming the RNG implementation is technically correct. In fact I would go so far as to request that the “re-roll if unique dropped before in session” feature be repealed, just to maintain purity.

88

I think the easiest way would be to simply store e.g. the last 5 epics and legendaries (each separately) that dropped and make sure the next one is not one of them.

If you want to refine it, just keep track of the last 5 sold / dismantled of each. That way items you want keep dropping and you may get a better version / second one for items you want two of.

89

This could be a reasonable option if it were documented and if it were made an option in the game settings.

90

Take it as you want, as I’ve not really read all nine pages of the thread.
Since I started dropping legendaries, I kept all of them in stach and reroll toons. After a bit of time it started to be confusing as hell so I made an excel sheet with every legendary of the game listed (by lvl and “type” : weapon, armor, jewelry) and those I have.
Turns out, I have an awful time dropping shoulder armors, it’s one of the less available part of stuff for me, while I have fairly more than half of the legendaries in the game (approaching two thirds currently). There’s also a sort of decrease by lvl of the relative amount I drop, but it can be explained by the high number of stuff lvl 75 (and the crafts around, not only for helmets, so :blueprints:), and the rules of dropping linked to the level of your character/your ennemies.
I seriously can’t explain why I don’t drop shoulder armors, because for blue and green stuff it seems to drop normally (even if I had a hard time getting some).

Edit : if you want to look at my sheet, I can upload it later today after work.

91

Honestly I simply feel there is less Shoulder armor than a lot of other piece. If I’m not mistaken there is only 4 pieces outside of the Set (and far from all Set got Shoulder). Just like there is less Pants than Chest.

In both case I feel this is saved by the MI Pants (3 Gardian) and MI Shoulder (all Nemesis have 2 versions each). You should get more Chest/Pants and basically 0 Helm considering everything is a BP (except Outcast Secret)

92

Problem is that they are in set, so I can’t complet most of them because of that (while I have a fair number of blueprints for helmets ahah).
I don’t remember the exact number of legendaries shoulders, I think it amounts to 20-30, but on that I may have one or two (and one I didn’t even drop myself).

But, for example, while I own almost all lvl58 legendaries, I miss the two shoulder armors (and another thing if I recall right). And lvl58 stuff isn’t that hard to get. I don’t track the number of duplication I get, but of course there’s those too.

What I want to say is that, even without doing numbers and percentages, it clearly struck that there’s less shoulder dropped than any other piece of stuff, and I find it weird.

93

Legendary shoulder armor and epic jewelry always evades me as well.

94

instead of wasting time writing you guys could’ve played and gathered data instead /facepalm :rolleyes: