Make the RNG Less Sticky

[surreal barn]

While I noticed this a long, long time ago, I actually have the data to back it up now. I've kept a thorough record of my last ~125 runs and it's very clear that the RNG continues to make similar choices in consecutive runs that last from 3-5 runs.

For example, Jumbo will be in the first Companion choices , you won't get a shop at the end, the first Lumin Vase part will take you away from the shop after the third battle, and you won't get a second companion until after the fourth battle.

Clusters of very specific things like that—and often there are more than just four—will occur five times in a row. I play this game a lot, so at first I thought the RNG must factor in the current time because this would make sense as I rarely play just one run per session. So I tested this and played at very different times of day and it didn't make a difference.

I don't know what the answer is, but there's something weird about how the RNG works that is prohibiting variety. It feels like it comes up with the content for two runs and then over the course of five runs slowly makes changes (as if one's morphing into the other).

I hope this makes sense because it's hard to explain. I've made 1,226 runs since the game's release. While I wasn't winning much in the early days, ever since I improved at this game I have seen this "sticky" RNG situation. It's hard to suggest what to do about it more specifically without knowing the underlying algorithm but I'm guessing it's a mix of using a pseudo-RNG that cycles through parts of an enormous value and efforts to control the choices it makes to make the game more fair/enjoyable. I have some ideas on how to fix this but I don't want to make this suggestion even longer with strategies that may not be applicable. I'd be happy to help if there's any interest in addressing this issue, though.

[nimble relic]

I think you've just gotten particularly unusual luck with the rng seeds you've been getting {or selection bias with your memory of specific events happening?}, I personally rarely see the the lumin vase take me away from the shop, and skipping over hundreds and hundreds of runs while testing with console mods, I've seen tons of times where I'd get companions after fight 2.

In fact, a lot of the time, you have the option to take 2 companions after the first fight.

[cursive hare]

i think they meant to say IF jumbo is in the first companions choice then X happens

that's the only way what they are saying makes any sense

i don't understand it tbh

[lost laurel]

From what I understand is the same digit of seeds is controlling too many things at once

[shell steppe]

I never had jumbo be in the first companion choices, and I get a shop at the end frequently, and lumin parts are almost always IN the shop for me, and the second companion is ALWAYS generated after the first battle, that's I think the one hard-coded thing.

[surreal barn]

It’s not selection bias, it’s in the data. I thought it was selection bias until I tracked it over about 125 runs. It’s a hard thing to articulate, so I apologize if I didn’t make much sense. I’m simply trying to express that many aspects of the previous run release themselves 3-5 in the same way before they start to change. I was also just giving examples. I wasn’t saying those exact things were too common for everyone. They were just examples. I’m saying when something occurs in the RNG it has a tendency to repeat multiple times before it stops. If it were one aspect it wouldn’t be weird. When it’s several, it’s frequently because of how the code is using it. I just brought it up because it’s similar to things I’ve fixed before and wanted to help.

[surreal barn]

I just had six runs in a row without a companion after the second battle, but everything you said is more often my experience as well. All I’m saying is that whatever the scenario is, it will repeat several times before it flips. It’s because they’re trying to force specific outcomes that there is a higher chance of frequent repeats. (At least, from what I’ve seen in other projects, that’s the generally the cause.) I’m not judging whether they’re good or bad, it that alternative scenarios do not occur, but just saying that they cluster regardless of what they are.

[shell steppe]

for me usually if one of those happens it doesn't happen again until quite a few runs after

dunno what to tell you, you probably just have abnormal luck

that's heavily unlikely though

[surreal barn]

Do you play on Windows, Switch, or Steamdeck?

[shell steppe]

Windows

[surreal barn]

Obviously I can't say this for certain, but that might actually be the difference. I play on Steamdeck and the game doesn't seem to be able to work in the same way it does on Windows. For example, it writes states less frequently. There may be an issue that results from the Proton translation that makes the game playable. If that's the case, then this is actually a bug. Thanks! I might be able to figure it out because of this. 🙂

[lost laurel]

What he means is when the game chooses values to randomize, too many things are determined by the same value

[shell steppe]

no but they say that it's that when some specific event out of these happens it happens many times in a row but doesn't necessarily cause the other ones to happen

[lost laurel]

No

Not really no

[shell steppe]

@lost laurel that's literaly what they say though

[lost laurel]

Ok I misread what you said

How do I do this so much…

[nimble relic]

words hard

[bitter apex]

Could you share the data that you tracked?

[sleek drum]

Yea do that

Also cause of how rn works depending on the initial seed ur seeds start to repeat after X rolls, so maybe u got very unlucky with initial seed

Trying to set run seed to smth new might help?

[surreal barn]

I'm not familiar with a way to set the seed. Do you mean with a mod? I'm not using mods. I'm on a Steamdeck so I can't, as far as I know. I do want to make and play with mods eventually, but I've got a crazy win streak going and I don't want to play with mods until I lose. :-p

[surreal barn]

I can, but you might regret asking. 😛 I typically don't track everything—just the choices I make—and then keep notes when I need to get my thoughts out or am trying to figure something out like this situation (but usually just a strategy). I'm almost done with the tracking so, just warning you, this is very long (and contains spoilers since these are all vanquish runs, but I'm assuming if you're in this thread you probably don't need a spoiler warning): https://ghostme.at/wildfrost-journal/

[bitter apex]

From this journal alone, I don't think you can make any conclusion about the RNG being sticky. Especially not about your example of Jumbo being tied to getting a shop at the end.

Unrelated, are you sure that Steamdecks don't let you use steam workshop? I don't have a steamdeck so I don't know, but that sounds surprising to me.

[shell steppe]

Steam deck uses Linux, it 100% can mod thru the steam workshop, I don't know for sure but it'd be wild if it couldn't

[surreal barn]

I'm not saying the things I mentioned necessarily tied together—Jumbo and no shop at the end are definitely not tied together—but that specific things repeat frequently before changing. And I understand I could track every element of the game and not make a definitive conclusion because I'm always missing half of what's there, but there is definitely something weird happening on the Steamdeck. I've been comparing this to Windows (and today, iOS) and the Steamdeck shows repetition. It's gotten to a point where I can guess more than half of what's going to happen based on the first battle rewards—but only on the Steamdeck. I still don't know exactly what is happening, and it's probably impossible for me to figure it out from just playing, but there is intentional weighting in the game and it seems to repeat in the same way on the Steamdeck. For example, if you see a Spice Sparklers you'll have a Shellbo in the run as well (and vice-versa). That's been the most reliable pair I've seen so far. I don't think I'm going to get anything too reliable with cards unless I track how and where they appear very thoroughly (which I think is even too much for me). It's going to take awhile to figure out what I can from what I have. I'm starting to put things together to looks for patterns. The charms are the most reliable, so I can at least show their approximate probabilities per tribe: https://airtable.com/appsT2Zc1nX2yunmS/shrEkc5xf1Hdf3RFD/tblb6IYp9xEz9RONm

It was probably too early to bring this Steamdeck situation up in here, but I thought someone else might have noticed it. After 1200-some runs it's just impossible not to see it happening. It's weird and I just really want to understand what's going on. I get a little obsessed with patterns.

[surreal barn]

In the past I tried running mods. I installed everything through the Linux desktop. The game wouldn't start for me. This had to have been over six months ago, though, so I imagine things have changed. I was able to get it running on Windows at the time but only some mods worked (like controlling the seed).

[shell steppe]

Six months ago there wasn't official mod support, only the bepinex stuff