Jump to content

A QUIRK OF THE RNG?


Zorba the Geek
 Share

Recommended Posts

Here is a bit of trivia for you all to ponder...  With six accounts for comparision I've noticed what might be an odd quirk of the RNG.  Please note that I'm not passing comment on drop rate because I have no evidence that this is an issue.

This observation applies to World 10, Heroes University, and looks at the order of drops in relation to the specialty class of each respective account.  Only account #2 might yet buck the trend, but, with Venam having just dropped in the last 2 hours, this account is following the same pattern.  If Noomye drops next the result would be compelling.  Look closely at the table below, colour-coded to clearly show the specialty class of each respective account.

image.png.2fc7c40f8b19fd1cbb116cec06330999.png

Is this just another case of randomness coincidence or is it a definite pattern?  What have others noticed?  I realise most of you would have only one account (or at most two).  Does a common IP address have something to do with how this pattern is manifest?  Sure, a sample of just six is hopelessly inadequate to draw any strong conclusions, that's why I'm keen to find out what others have found in regard to these three girls in World 10.

  • Like 3
Link to comment
Share on other sites

Data point #9: KH, Venam, Noomie, Mavin.

Perhaps it happens with other bosses too, and it is easy to know the order of girls drops as the default order in the harem is for date obtained.

Zorba, as you have the most number of accounts, could you see if that happened with the rest of the girls, (or at least with finalmecia who is the second most recent)?

Link to comment
Share on other sites

1 minute ago, jelom said:

Data point #9: KH, Venam, Noomie, Mavin.

Perhaps it happens with other bosses too, and it is easy to know the order of girls drops as the default order in the harem is for date obtained.

Zorba, as you have the most number of accounts, could you see if that happened with the rest of the girls, (or at least with finalmecia who is the second most recent)?

I don't know yet, but I can check.  This is the first time I kept such complete records because the two oldest accounts quickly followed the same pattern and gained my attention.  Might take me a day or two though since I have a lot of other things to do.  All this is really interesting so far!  Finalmecia might be closest for me to analyse because the drops happened reasonably close together as with Roko.  However, two of the accounts are outliers because some of Finalmecia's girls were obtained through EP and this skewed the results.

It would be amazing if we could anticipate the order of drops since this might influence our game strategy.

Link to comment
Share on other sites

there are just six drop options possible...

i wouldnt call it a big surprise this happens to someone, even your result is statistically rare, it had to happen at some point...

no offense Zorba, but could it be you are somewhat obsessed by the RnG in this game?

i believe you would enjoy the game much more if you would just lean back and take RnG as it is... totally random! ;)

Link to comment
Share on other sites

3 hours ago, Skiron said:

there are just six drop options possible...

i wouldnt call it a big surprise this happens to someone, even your result is statistically rare, it had to happen at some point...

no offense Zorba, but could it be you are somewhat obsessed by the RnG in this game?

i believe you would enjoy the game much more if you would just lean back and take RnG as it is... totally random! ;)

That's just it - it might not be totally random. There might be a limited range of choices offered to us that Kinkoid didn't tell us. We don't know the exact recruit rate for event girls, for instance, or the value of the pity boost for normal girls. And given that Zorba saw something happen that has a 1 out of (3^5 * 2^4 = 243 * 16 = roughly 3900) chance of happening, that no one else apparently thought to look at, it's worth checking. We might have discovered a hidden rule that decides what school you get first; who knows. Or we might have uncovered a bug in the random algorithm (this happens occasionally)!

Either way, we learn something - either that there's something, or that there's nothing. Looking for quirks in the RNG can be fun, too! ^_^

 

Why do you even have six accounts?

I can think of two possible reasons: greater chance to get a rare drop on at least one, and epic pachinko played at 6x the normal rate (assuming you have the free time in RL to pour into it).

Link to comment
Share on other sites

22 minutes ago, Hollo said:

That's just it - it might not be totally random. There might be a limited range of choices offered to us that Kinkoid didn't tell us.

it is not kinkoid offering limited options, it is math doing that...

there are simply just six orders three girls could drop...

the chance the same order happens on four accounts is 1:1296...

yes rare, but with this small limit of options it simply has to happen at some point...

 

and since there are already reports in this post with different drop-assemblies on the same classes,

you simply cant count classes in anymore, so even Zorba has already "two misses" on his six accounts,

which lowers the rareness of this happening by alot!

as the big C said earlier:

7 hours ago, Chthugha said:

Here we see what happens if coincidence is mistaken for a pattern.

 

EDIT:

i have to correct myself...

since Zorba wasnt aiming for a specific result in advance and there are six options to hit the same result,

the chance of hitting the same order on four accs is just 1:216...

 

Edited by Skiron
Link to comment
Share on other sites

They didn't happen the same way on just four accounts. They happened a specific way on six, in a way that suggested that it was school dependent. The chance the first one is of a particular school is 1 (100%), since it has to be something. For the next five, though, it happens the same way - the three other same-school accounts get the same school girl, and the other two get the exact same other school: 3^3 * 3^2 = 243. And then the second girl's school is also apparently constant, for five of the accounts, with the sixth not yet known. Again, the first one is just whatever it is, but the rest are consistent with it: 2^2 * 2^2 = 16.

243 * 16 = 3888. (If it turned out that the first recruit was determined by account school, we'd toss in another factor of 3, but we aren't asking that yet; we're asking what the chance is of what we actually saw happen.) And with tens of thousands of accounts AFAIK, this is of course statistically near certain to happen among six of them. But it didn't happen to exactly six accounts we cherry picked out of tens of thousands; rather, it happened to the exact six accounts that a single person was able to observe.

And then Zorba made the right noises: could this be due to a hash from IP address? Is six too small a sample? Can others report their recruits? Turns out it was indeed too small a sample, assuming Chthugha didn't have a burp in his record keeping.

It's like if you go to a casino where they've invented a die with 3888 sides, and on the grand opening of the mega-craps table, on the inaugural roll, they both come up with the same side up. Now everyone's staring at the dice and wondering what's up. Yes, those dice are going to be rolled millions of times, so someone was bound to get a pair someday, but if it's the very first roll, the pit boss can be forgiven for wanting to reroll the dice and make sure they're not loaded. :D

 

He wasn't mistaking coincidence for a pattern (or rather, for a cause); he observed a definite coincidence and was checking for a cause.

Link to comment
Share on other sites

@Hollo dont get me wrong please, i admit your math is correct, we just have different approaches ;)

i was going for the chance to hit 4 of a kind on 4 tries, then adding the two accounts who "missed it",

you added how likely it is that they show up with 2 of a kind...

 

let´s try another one:

given the fact we have 84ooo accounts on HH, equally divided 28ooo of them are KnowHow,

adding there are just six drop orders possible there are 4650 KnowHow which will roll the same order on a specific boss (on average)...

plus there are 4650 Hardcore accounts which will roll another, but same order on the same boss...

yes it is a rare chance that he holds exactly 4 of the one group and 2 of the other, but it doesnt mean a thing to the overall result...

 

and since it bothers me a little bit, here a fun fact to close this topic for me:

recently i posted an ironic thread where i tried to parody all the theories about drop rates all through this forum...

i set up the wierdest stuff i could come up with, like time, player-id and other crap and even called all them "conspiracy theories"...

long story short: it had to be deleted to avoid the unlikely chance that someone could actually believe it and im totally fine with that decision!

but here someone just hit an unlikely chance and tries to make it up to something nothing better, by calling out ip-adresses...

 

a stupid theory like this based on pure coincidence shouldnt be further supported by us discussing about it...

Link to comment
Share on other sites

6 minutes ago, Skiron said:

i was going for the chance to hit 4 of a kind on 4 tries, then adding the two accounts who "missed it", you added how likely it is that they show up with 2 of a kind...

Aye; I know what you did. :-) It's interesting, but it's not what Zorba was pointing out. It wasn't that four KH accounts had a specific drop order; it was that each account had the same drop order for the same school. Again, his thing was over 3 times as rare as yours. Even your example would have been worth mentioning; his, even more so. And this is important, because we need to start on the same page about what's being brought to our attention, before we can start arguing about whether it's more than coincidence.

14 minutes ago, Skiron said:

given the fact we have 84ooo accounts on HH, equally divided 28ooo of them are KnowHow,

adding there are just six drop orders possible there are 4650 KnowHow which will roll the same order on a specific boss (on average)...

plus there are 4650 Hardcore accounts which will roll another, but same order on the same boss...

yes it is a rare chance that he holds exactly 4 of the one group and 2 of the other, but it doesnt mean a thing to the overall result...

I agree that there are roughly 4650 KH accounts that will have the same order of drops from a specific world. You seem to be driving at the fact that it's possible to have four accounts all from that same set of ~4650. If so, I certainly agree. But what's the chance of that happening? The first is 1/6 likely; the next, also 1/6th; and so on, until we have a 1/1296 chance of all four being from that set. Just as you said above. (But again, that's not what Zorba was pointing out.)

22 minutes ago, Skiron said:

recently i posted an ironic thread where i tried to parody all the theories about drop rates all through this forum...

i set up the wierdest stuff i could come up with, like time, player-id and other crap and even called all them "conspiracy theories"...

long story short: it had to be deleted to avoid the unlikely chance that someone could actually believe it and im totally fine with that decision!

but here someone just hit an unlikely chance and tries to make it up to something nothing better, by calling out ip-adresses...

I could make the same post myself, yes. :-) Most of the time, I'm with you: these things are like cockroaches. And it's tempting to see the OP as one example of this. In this one specific case, though, I think it's plausible. Why? Two reasons.

If an anomaly is to have a cause, it has to either be put there intentionally, or get there by accident. What's the chance that Kinkoid intentionally made an account's world drop order by school predetermined? Not high, admittedly - I might place it at 1% - but that's based on what I know of Kinkoid's policy (combined with other factors I think are relatively minor), and none of us know much about that. Which means our uncertainty on that is pretty high. So now it's 1% +/- 20%, say. It's a dartboard. But that's actually not drawing my attention; I'm pretty comfortable saying they're unlikely to do this on purpose.

So what about accidentally? Seems even less probable, right? Well, welcome to the crazy world of RNG. Or rather, pseudo-RNG (PRNG). (For all I know, you're aware of this, and you can skip this paragraph if so; I'm writing this for any reader.) All computerized RNGs are basically this algorithm that generates numbers which look random, but are actually the result of a preset algorithm taking a preset key as input. Here's the thing: it's possible to have an algorithm that gives you a skewed distribution of numbers, such that when you put it through several filters to set up a certain range of integers, you get anywhere from 10% to 100% more numbers from one end than the other. It can creep in unpredictably - you might see a completely normal distribution if your range is 1-117, for example; but if your range is 1-3, blam! Where are all these 2s coming from??

It can also creep in if your RNG algorithm isn't just a simple call directly to Math.random() or rand() or randint() or other standard language's RNG, but rather to some homebrewed thing your company cooked up because they wanted more reproducible tests or some programmer was too clever by half and added some code to it to detect hackers and you didn't know about it because that programmer was not you and they checked it in with a bunch of other changes and no one else noticed because your team didn't do a code review that day.

Which still means this is unlikely to happen by accident - a lot of stars have to align just right. But that's okay, because again, in this one specific case, it's easy to check with a quick query out to other users. (And hopefully other users aren't somehow pathologically inclined to only report confirmations. This method ain't perfect, but in this case it's probable enough that if someone has a counterexample, they'll say so. And look! They did.)

There's a strong temptation, I think, to err on the side of what you call "conspiracy theories". Or more accurately, anomalies with a definite cause. But there's also a strong temptation - in part because of a lot of people having an understandably bad reaction to conspiracy mongering - to err on the side of "no definite causes, ever!". That can be even more damaging in rare cases. (As in, millions of dollars in damage. Security crackers make a living from exploiting anomalies they discover, that everyone else considers too coincidental to be worth bothering with. Or biologists who discover a microbe doing something slightly weird, that leads to the next great cure that they'd never have found if they just considered it a silly theory.)


You and I aren't disagreeing on whether this specific anomaly had a cause; we're disagreeing on how to tell whether an anomaly is just a coincidence in general.

...I mean, yeah, it's just a bit piece in a game with sexy pictures, but let's not jump to the conclusion that all theories are stupid, please. :-)

 

  • Like 1
Link to comment
Share on other sites

Since there are just 6 drop orders, it would be more useful to check all bosses drops. If anybody's interested, here are mines (charm account, never changed class):

World #2: Sheheramazond, Kimie, Arcana
World #3: Solveig, Sung Tai KI, Rumiko
World #4: Miss Spook, The housemaid, Regine
World #5: Eugenie, Bonny, Jeanne
World #6: Lola, Pelagie, Kelina
World #7: Queen Titania, Leeditt, Sylvia
World #8: Stephanie, Cunegonde, Marina
World #9: Zoe, Edna, Salem,
World #10: Noomye, Venam, Mavin

  • Like 1
Link to comment
Share on other sites

ok... last reply xD

40 minutes ago, Hollo said:

...It can also creep in if your RNG algorithm...

true, i give you that... creepy things tend to happen ;)

42 minutes ago, Hollo said:

You and I aren't disagreeing on whether this specific anomaly had a cause; we're disagreeing on how to tell whether an anomaly is just a coincidence in general.

...I mean, yeah, it's just a bit piece in a game with sexy pictures, but let's not jump to the conclusion that all theories are stupid, please. :-)

you are right, shouldnt do that...

wait... let me quickly bury that axe...

*noise*

done :D

  • Haha 1
Link to comment
Share on other sites

1 hour ago, Daniele said:

Since there are just 6 drop orders, it would be more useful to check all bosses drops. If anybody's interested, here are mines (charm account, never changed class):

World #2: Sheheramazond, Kimie, Arcana
World #3: Solveig, Sung Tai KI, Rumiko
World #4: Miss Spook, The housemaid, Regine
World #5: Eugenie, Bonny, Jeanne
World #6: Lola, Pelagie, Kelina
World #7: Queen Titania, Leeditt, Sylvia
World #8: Stephanie, Cunegonde, Marina
World #9: Zoe, Edna, Salem,
World #10: Noomye, Venam, Mavin

For a comparison, I'm Know-How (never changed Class before) and here is the order in which I got the Girls: 

World #2: Kimie, Sheheramazond, Arcana

World #3: Rumiko, Solveig, Sung Tai Ki

World #4: The Housemaid, Regine, Miss Spook

World #5: Bonny, Eugenie, Jeanne

World #6: Lola, Pelagie, Kelina

World #7: Queen Titania, Sylvia, Leeditt

World #8: Stephanie, Marina, Cunegonde

World #9: Edna, Salem, Zoe

World #10: Mavin, Venam, Noomye

Edited by TheSonicSly
Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
 Share

×
×
  • Create New...