Jump to content
The Lotus Eaters: Share Bug Reports and Feedback Here! ×

How does the relic random work?


mcsushi89

Recommended Posts

This is not a rant post, and I'm not questioning how does RNG/PRNG work, I'm only questioning how does the RELIC random work. It's kind of hard for me to believe that a company which made lobbies use peer to peer connection and not actual, normal (for MMO games) servers, would make the random mechanism actually unpredictable and untraceable, like for example, based on an external factor, such as noise or electricity(if you don't know what I'm talking about, https://en.wikipedia.org/wiki/Hardware_random_number_generator ). Again, since I know there has to be a way to predict relic drops, I only want to know what is it. Please don't write me in the replies something like "it's random, it's supposed to be unpredictable", or "do you even know what random means?". Yes, I do know, I'm not questioning how I got once (from specifically my relic) 4 rare drops in a row(0.01% or 10%^4), then a bit later 3 rare drops in a row(0.1% or 10%^3), and there were cases where I did 8 rad shares in a row with 0 gold drops. (32 relics, 10% each, 0 drops). That's how random works, I know. But I still can't believe it's impossible to predict the drops from the relics, there has to be a way. If you don't have anything helpful or positive to answer, just don't waste time on this post please.

 

Link to comment
Share on other sites

As a mathematician i can tell you that even though real randomness can't be achieved digitally, the pseudo random algoritms that are used are impredictible enough that you can't realistically predict future numbers.

By realistically i mean that if you had all the data from all the drops and they always used the same pseudo rng and you had a huge computational power for like, months or years, you could maybe, maybe predict a couple numbers (starting from the last number imputted when you started calculating of course) in some cases. Also, you would need to know if the same rng gives numbers to all relics, for each relic individually or what. Also also, knowing future drops wouldn't mean that you can manipulate it to get different result, so it would be a not so useful information.

You can think about it like this: you can take the actual pseudo rng used and another one similar enough (or maybe the same with a sligtly different seed) so that they both give the same results up until the last number that you saw, but that give different results on the next number, so it would be actually impossible to discern which one the game is using. This is, of course, assuming you know all the numbers from the rng and that the same rng with the same seed is always used; also they probably use a 100 number rng that is translated into the 6 possible result that we see ingame so the information we get as player is even more reduced, making it 100% impossible to predict future numbers.

This is all assuming they use at least a decent pseudo rng of course, but they really have no reason to do otherwise.

tl;dr: it's impossible.

Link to comment
Share on other sites

14 minutes ago, Am_Sha3gar said:

You can think about it like this: you can take the actual pseudo rng used and another one similar enough (or maybe the same with a sligtly different seed) so that they both give the same results up until the last number that you saw, but that give different results on the next number, so it would be actually impossible to discern which one the game is using. This is, of course, assuming you know all the numbers from the rng and that the same rng with the same seed is always used; also they probably use a 100 number rng that is translated into the 6 possible result that we see ingame so the information we get as player is even more reduced, making it 100% impossible to predict future numbers.

ok, that helped. So you think they are using pseudo rng as well?

Link to comment
Share on other sites

It's probably possible (with probably certain dose of uncertainty) but it would require lot of relics opened before. For "single user"(1) PRNGs some patterns may repeat. There might be some errors (e.g. after certain number of numbers it outputs only 1).

That's assuming that they use 1 PRNG for 1 player. For "multiple user" (2) PRNGs it's more complicated because multiple user pick numbers from one PRNG. It makes PRNG more (real) random. In this case it would be much harder. I don't have information on this subject so... yeah, it's closer to "impossible" in this case.

(1) & (2):

I don't think anyone uses those terms. Single user means only one user picks numbers from PRNG. Multi user means that many (at random) user pick numbers from 1 PRNG. I haven't found too much information on 2nd case (multi user).

4 minutes ago, Am_Sha3gar said:

This is all assuming they use at least a decent pseudo rng of course, but they really have no reason to do otherwise.

They probably don't use some bad one (I don't remember the name but there is one where, for some arguments, it outputs "bad" numbers).

However after that... I don't have good feelings about it. Some "rare" stuffs are easily to get and some are BAD (solo 26 radiants => 0 uncommon stuffs - not counting formas).

Link to comment
Share on other sites

46 minutes ago, mcsushi89 said:

Again, since I know there has to be a way to predict relic drops, I only want to know what is it. 

 

What I meant by this is, for example, if I know the next number of the algorithm that generates random numbers from 1 to 10000 is 9100, for example, if I use an intact/exc/flawless relic, I won't get the gold drop, but instead the second silver drop. but if I use radiant, I'll get the gold drop. and in some cases, where the number is 9999 for example, it doesn't matter what refinement it is, the drop will be gold(so I don't even have to refine the relic and waste void traces). Something like that. This would allow to make trace usage optimal, and not ever run out of them, because you don't know how the relic random works. If this gets solved, there literally won't ever be a problem with void traces or bad drop, ever. if you see you're gonna get a bad drop, just use a relic where all the drops are bad, until you see you're getting a good silver/gold/specific cases good bronze drop on the next relic.

Link to comment
Share on other sites

2 minutes ago, (PSN)Madurai-Prime said:

Even if you figured it out, what are you gonna do differently anyway? Try to not rush and obsess over a specific item and just enjoy playing for it. 

All you can do is radshare.

What I meant by this is, for example, if I know the next number of the algorithm that generates random numbers from 1 to 10000 is 9100, for example, if I use an intact/exc/flawless relic, I won't get the gold drop, but instead the second silver drop. but if I use radiant, I'll get the gold drop. and in some cases, where the number is 9999 for example, it doesn't matter what refinement it is, the drop will be gold(so I don't even have to refine the relic and waste void traces). Something like that. This would allow to make trace usage optimal, and not ever run out of them, because you don't know how the relic random works. If this gets solved, there literally won't ever be a problem with void traces or bad drop, ever. if you see you're gonna get a bad drop, just use a relic where all the drops are bad, until you see you're getting a good silver/gold/specific cases good bronze drop on the next relic.

Link to comment
Share on other sites

12 minutes ago, mcsushi89 said:

ok, that helped. So you think they are using pseudo rng as well?

Well, yeah, it's what is used in these cases.

1 minute ago, mcsushi89 said:

What I meant by this is, for example, if I know the next number of the algorithm that generates random numbers from 1 to 10000 is 9100, for example, if I use an intact/exc/flawless relic, I won't get the gold drop, but instead the second silver drop. but if I use radiant, I'll get the gold drop. and in some cases, where the number is 9999 for example, it doesn't matter what refinement it is, the drop will be gold(so I don't even have to refine the relic and waste void traces). Something like that. This would allow to make trace usage optimal, and not ever run out of them, because you don't know how the relic random works. If this gets solved, there literally won't ever be a problem with void traces or bad drop, ever. if you see you're gonna get a bad drop, just use a relic where all the drops are bad, until you see you're getting a good silver/gold/specific cases good bronze drop on the next relic.

You are assuming that the same number is used in both cases. What if a different seed is used for each type of relic? In that case you would get 9100 using an intact while, for example, 2172 using a radiant. We know nothing of how they use the rng, we can't just assume things: it could depend on the squadmembers and their rng, it could depend on the state of the relic, on the name of the relic, on the number of rounds, on the type of the mission for what we know, there's no way for us to check.

Link to comment
Share on other sites

@mcsushi89 You will not be able to design a program that could predict with any sort of accuracy what Relic drop you will get next. All item drops in the game are governed by the same RNG, and while it is pseudorandom (no such thing as true random with a digital system), it is well enough designed that you will not be able to predict the next number in the sequence (there was actually a serious issue involving CPU security when the pseudorandom number generator in the CPU could be predicted, modern CPUs have since been improved dramatically in that area).

 

If you had enough money, resources and time, sure you probably could, but you'd have to devote a literal supercomputer to the task. It'd be akin to brute-forcing AES-256 bit encryption. And, chances are, the RNG system gets little tweaks every now and then (updated drop weights, inclusion or removal of items from the table, etc, would radically change outcomes of any predictive model) would invalidate any existing model and you'd have to start over..

 

Long story short, it's a fools errand. It would also require reverse engineering the game, which violates the EULA, and would most likely result in being permanently banned from the game.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

Guest
This topic is now closed to further replies.
×
×
  • Create New...