How I Made BILLIONS Investigating Hypixel Skyblock RNG

Added: 2026-06-04

[00:00:00]You should always gamble when the others are stuck in your favor, which inside a casino is never.

[00:00:11]So, you should never gamble unless you're in a massive 200,000 player virtual economy run half by kids, half by lazy people. Go to the wiki, calculate the expected value of the loot box to be 190,000. go to the game and see it being listed for 130k, which by the way is the price you can buy it instantly for, or create a buy offer for only 100k. I mean, with odds like that, even just stack of boxes has already over 90% chance at making profit. And the odds are only getting better the longer you play, unlike mm- actual gambling. Not only that, but since you can open a box in less than a second, this should work out to somewhere between quarter to half a billion coins an hour, or about $123 an hour. if you wanted to buy that amount legally. But you see, there is a problem. Expected value only works when RNG behaves as expected. And inside Hypixel Sky Block, RNG is rigged.

[00:01:06]A words that have been spoken many times throughout the history of the game. Is it after someone lost all their money rolling dice? Okay, now how is that even one in six? Wait, you know it's not right. It has six eyes. Are you stupid or something? Didn't get a single handle after two dungeon runs. Ah, come on, man. I understand not getting handled after a single run, but two times in a row now. That's ridiculous. Or even by me in the last video after getting through entire progression for a thing without getting none of the RGs designed for the thing. Okay. How does this always happen, bro? It is literally first time. Okay. Well, that that's still always. But is it actually rigged?

[00:01:41]And if so, once understood, could it be rigged in your favor instead, thus making you infinite money? Well, the problem is that every time I was trolled by RNG, my sample size just simply wasn't big enough to have any meaningful conclusion. However, with Jerry boxes being fast to open, not restricted to a time or area, plus being literally just free XP and money, felt like a perfect opportunity. And so I have spent past few weeks meticulously writing down every drop in every condition, testing every even the most crazy theories I could think of, as well as falling into the rabbit hole of randomness deeper than ever before to once for all answer all these questions in the biggest Minecraft gambling experiment probably. Okay, but first I must show you something because depending when you are watching this, a long time coming collaboration with Epic Desk might be actually live. And if you know me as a guy who never sold anything, I don't really like selling stuff for the sake of it. That's why it took checking out what Epic Desk is for me to realize an opportunity of having a custom SLT desk Made by them is genuinely different.

[00:02:47]Plus, this guy who once beat me in proceeded to brag about it got 31 reviews on his design. And I thought we must get more. And so after weeks of designing process with an actual artist trying its hardest to bring my constantly evolving ideas into life, you get to now enjoy this majestic yet absurd scene of characters from SLT universe just chilling together on a nice day looking towards their next adventure. All that's put on a premium perfectly thick highest of quality material I have attached for you to use as a desk mat or I don't know just to have it while joining exclusive club of people that will ever get to say that because once it's gone it is actually gone forever with this being it one in a century opportunity for you to support me while getting something cool in return. I really hope you like it as much as I did. Link in the pinned comment. No, seriously. Back to the video. And before we get straight to my attempts at breaking the RNG, I must say I kind of lied to you. Okay. Well, I didn't lie lie. It's just that quickly Judgy just leave some important details out.

[00:03:50]>> That's called lying by omission.

[00:03:54]>> Oh well, I guess I did lie to you because while it is true that you can make half a billion coins an hour opening boxes, first you actually need the boxes. And there is a simple concept in economics known as supply and demand. And without getting into boring intricacies that also kind of allowed this method to exist in the first place. Basically, there was only enough of the boxes I could buy at the time without raising the price, which didn't add any effort, but for the amount of profit I made at the end. It took over a week waiting for all the boxes, which honestly I'm just saying for transparency because for me profit was the least of my concerns. I was here to uncover the truth.

[00:04:32]And to do that properly, first I needed a way to track the results, which I probably could have done by literally just writing them all down, but I'm lazy. So I just asked AI to create me a tracking app, which was the first time I tried doing something like this. So it kind of took me by surprise how it just did it perfectly first try and at the same time threw him in a tiny existential crisis realizing the more we will be able to get a finished product without the need to understand the process behind the LS need we will have to try other process behind it but time is short and that's probably a topic for my other channel especially that with everything set up it was time to start the test which I'm not going to lie I had no idea how so I just did and after first 400 boxes well Nothing happened.

[00:05:14]Everything seemed as expected, which is kind of expected when you do things as expected. But then I continued with another batch of 256 boxes. And this time I got a bit less Jerry Talisman than I should have, which doesn't mean anything, but I had a theory that contrary to the popular belief, probability does have memory.

[00:05:34]The one used in a lot of games at least.

[00:05:36]Because if it uses some sort of linear congreial generator, the probability over its full period should be fixed.

[00:05:42]Meaning if you get unlucky at the beginning, you are literally guaranteed by design to get more lucky later. And for anyone that knows one tiny problem with that reasoning, listen, I was young and stupid. Plus, technically what I said is not a lie.

[00:05:58]>> That's called lying biome.

[00:06:00]>> Okay, shut up. Well, so staying true to my theory as well as relying on a very limited knowledge of PRNG I had at the time, I just kept opening boxes, kind of curious if my luck will eventually flip and flip. It did, but not in the way I expected. In fact, that is the type of stuff that makes you start believing in higher power. And I know it's going to be terrible for attention, but I just want to play you the full unedited clip.

[00:06:27]I mean, we are somewhat up. We are up from the last time, but we are still down in total. So, I would say let's just open some more.

[00:06:43]I don't know, two more stacks on the server.

[00:06:56]Okay. Okay. Why does this happening every time.

[00:07:03]Okay, if I'm going to leave and if I'm going to pick something from the chest.

[00:07:21]Oh my god.

[00:07:26]Imagine.

[00:07:28]Okay. Oh. Uh. Um.

[00:07:38]Hello.

[00:07:39]Um.

[00:07:42]Uhhuh.

[00:07:45]Okay.

[00:08:03]Now, what are the odds of getting 6.6% drop three times in a row? Well, that's easy, you may say. You just need to multiply the number three times, giving us 0.02% odds. That's pretty rare. In fact, for the rest of the whole test, that was the only instance, it happened. Now, what are the odds of making a weird choice?

[00:08:22]like suddenly deciding to repeat the same sequence of actions three times in a row even though you know they will not serve any purpose. Well, that's a bit more complicated. You may continue because to answer that question precisely, we would have to look at brain and stuff. But I mean just purely statistically speaking looking at all the times I got green jar talisman and could have thought of doing it. My mind decided to suspend all this belief and do it for I don't know shits and giggles this one singular time. Well, pretty freaking low. So now what are the odds of those two deciding to synchronize together in the single compatible window they had?

[00:08:59]Well, not zero is the only logical answer I can give you right now because I'm sorry, but I'm kind of skeptical that me opening a chest was somehow the clue to predicting your drops all this time. So anyway, ignoring that whole bit for a second and looking at the overall session, well, my luck actually did somewhat flip. Not quite getting to the expected results, but pretty close with the overall conclusion for the first session being that if you're going to do things as expected, well, you're going to get close to the expected results.

[00:09:27]But now was the time for the fun part of the video, which is doing anything but what is expected in hopes of getting some unexpected results.

[00:09:37]At the first test I had in mind since the very beginning was changing the lobby after each opened box. The idea being that in the single lobby you'd be under a single seed and thus allowing whatever randomness system there is to do whatever it is supposed to do and so never deviate too much from the expected results. However, if you were to change the seed each time, thus kind of randomizing the randomness. My theory was that now it should be at least possible for me to start seeing some unexpected results.

[00:10:11]Okay, this is crazy that we still didn't get another talisman. I know that I'm losing money, but I want to see how far is this going to go. Another Jerry Sheen gun. And still no talisman. Another Jerry Sheen gun. Another and well, if you agree that getting two times more jarish guns than green jar talismans when you are supposed to get them five time less in the very first session on top of that is a bit of an anomaly. Then you could say my theory was kind of working two times more jarish guns even though there is a six time higher chance to get talisman. Hello. So was that it? Was that the proof that RNG is rigged? Well, almost because if from that point the pattern continued just a little longer would become practically impossible to happen by itself. So did it? Well, correct question is would it? Because immediately after getting that result, Sky Block conveniently crashed and I was put inside the lobby with an admin. Why?

[00:11:15]I don't know. probably unrelated or is it because after coming back as if someone wanted to eliminate my suspicions now I was getting more green Jun and le guns than I should have when even if the first anomaly happened just by chance there should be nothing that would make the game suddenly start compensating. What even is this? Why am I getting so many all of a sudden?

[00:11:37]Unless my first theory about the lag balancing out was kind of correct, but it still wouldn't make sense since now I was changing the lobby after each attempt. Unless I only thought I was, while in reality, I was just coming back to the same lobbies with the same C. Or maybe I didn't even open enough boxes for the first anomaly to be even considered anomaly. Well, realizing I only created more questions and answers and attempting to answer at least some of them, I tried switching between different kind of lobbies, not just hub.

[00:12:02]And after noticing that this time I'm definitely coming back to the same lobbies, I realize I'm going to need two things. One, a way to track the servers I'm on. And two, which is something I probably should have done from the start, is one big sample size test. So there would be no questions that I did not open enough boxes. And that could also serve as a reference point for any other test. However, that created its own sets of problems because even though marking down drops was way more efficient than writing them down for about 5,000 boxes that I was about to do, that would still take ages. Luckily, it was around the time when I realized that I'm stupid and was marking down drops when they are being literally sent out in the chat. So, all I needed to do was to download the chat logs.

[00:12:46]Genius, I know. Okay, I mean, it wasn't that easy because now I needed a way to analyze those massive logs that were now way too long to be pasted into any AI prompt box. But instead of just paying OpenAI to let me upload more files, I just asked it to create a Python script that would compress those logs into a string of letters representing each drop, which worked perfectly. Although I'm not sure what exactly I was expecting AI to do, but I guess a little bit more than just look at how many items I got at how many I should have get, say everything seems fine and call it a day. Which I mean, it is one of the criterias you should be looking at. But if for example you flipped a coin 100 times and the first 50 outcomes were heads, the next 50 were tails.

[00:13:28]Technically you also got a perfect amount of each outcome, but you wouldn't exactly call it random and definitely not non-exloitable. All I'm trying to say that there is a lot more stuff you should be looking at, which I had a bit of a hard time trying to explain that to AI. So in the end I just decided to create an app that basically in its core would be automatically scanning the logs based on a defined load table and then the recovered stream of drops would be a subject to more and more tests that I thought would be helpful and a few hardcore VIP coding sessions later. Here I present you the RNG Analyzer 3000, which is just so fun because all I need to do is to drop in the log files and then I can see everything. Starting with the visual drop stream where I can just individually select each item and try to look for any obvious patterns. Oh, there's a kind of pumpkin sideways.

[00:14:19]Look, there is a dinosaur. Yeah, those are not the patterns you should be looking for. But anyway, moving on onto the global statistics section containing the global statistics as well as just expected versus actual amount of items.

[00:14:32]Like for 20,000 coins, we got only 42% which there was only 3% chance of getting that result or less, which I mean that's pretty low. Where for mining XP, we just got bang average. Then there are drops split by server, which doesn't really apply here since there's only one server. Then there are statistics of just the first drop on each server, which again doesn't really apply here.

[00:14:52]Then we transition to the transition matrix, which is essentially just looking at the chance of getting each item after getting each item. For example, if you flipped heads, that shouldn't matter for your next flip, and you should again have an equal chance to get either heads again or tails. Whereas with my earlier example, there was more like 98% chance of getting heads after getting heads, and which is essentially what this section is trying to detect.

[00:15:16]Then there is a probability chart cuz why the hell not where you can sort of see the results converging to their expected value yields. Although for 20,000 gold there seem to be converging below the expected value which I'm not sure what is that about but maybe it's fine. Then there is extreme section showing you the most extreme examples of RNG during this session which I'm pretty sure normally this is not being looked at since it's literally just try picking the most extreme examples unless you're going to have examples so extreme they shouldn't happen like ever. So that is the reason I included this, which in our case, the most extreme example is getting 10 farming XP in 17 attempts, which on its own is insanely rare with one in 77,000 attempts. But if we look at the chance of that happening somewhere during the 5 12,000 box session, the probability jumps to 7%, which I mean is still pretty rare considering that now you have only a single attempt to get that 7% result.

[00:16:09]And for another attempt, you have to open another 5,000 boxes, but calling it impossible would be kind of a stretch.

[00:16:16]And at last, we have just a simplified dopamine version of results of each session just for you. What does all of it mean? Well, I'm glad you asked because there we have the first section from 0 to 1 sigma representing the pink area and is where most of the results should stop. But if they don't, we cross to the unusual section 1 to two sigma representing the blue area and meaning that the results were well pretty unusual. but still happening fairly often. But then things get more interesting because we cross into the rare section representing this tiny area. And while some of the sessions will end here, I mean they have to. Are those going to be among the few sessions that you did? Well, depends on how many sessions you did, but probably not. But hold on, because now we are no longer on the scale in this shouldn't happen section. And unless we are talking on a scale of the whole server with thousands of people opening thousands of boxes and someone getting a result above that line for you alone to simply put this should not happen. But then finally we get to the RNGs broken line 4 and a half sigma one in hundreds of thousands or even millions of sessions. So if one of the few sessions that you did did cross that line, well it is literally not impossible in any practical sense. Yeah, it is impossible. So our goal is to basically cross that line because if we do that could only mean one thing. The RNG has been cracked.

[00:17:38]So what was the result of our big sample size reference test? Well, nothing special or that's what I thought before later checking the transition matrix. But for now, I didn't see that plus. My mind was just too excited to now having the tool to just go ahead and test as many things as possible. and test. I did first coming back to the switching lobby strategy that last time gave us the most promising results. Well, this time a lot less fireworks. In fact, it was almost weird how flat the results were. Then I tried switching only between the main lobbies, which even though resulted in two of the drops crossing into the rare section, that still is not that rare.

[00:18:20]Although, it's fine cuz it wasn't even the goal of this test. What I wanted to test if some lobbies are for some reason luckier than others. So after switching only between the lobbies I could have easily come back to and determining in which one I got significantly better drop rate during a short test, I would then come back to it for a long test hoping that I don't know the drop rate stays the same. Spoiler alert, it didn't. In fact, turned out to be just a huge scam because even though you can see the green cherry talisman bar being well into the rare section, that only represents the deviation from the expected results this time into the red.

[00:18:56]So, not only I lost like 50 million coins of expected profit because I opened like 1,500 boxes there, the deviation wasn't even big enough for me to at least be able to actually call it rigged. In a need for a distraction from this unsuccessful gambling, I felt like doing something fun. So, I did some more gambling, which wasn't even part of any test. I just randomly run across sky block opening boxes. But well, I guess here are the results of this session.

[00:19:22]But more so, I used that time to think of more tests I thought would be helpful. Like the next test, which was conditional switching, aka switching lobby every time I got a specific drop, like 20,000 coins, hoping that that will somehow mess up with randomness. Again, it didn't. I tried doing the same except resetting the seed every time I got a rare drop like a green jar talisman. And even though this time we got more promising results, they once again did not cross the line where we could start to suspect something foul. The same could not have been said about the next test though, which was opening boxes in as equal intervals as possible. As in, if for a last test, you could say at least it was somewhat suspicious. For this one, it wasn't even close. I mean, this was almost suspicious how nonsuspicious it was compared to every other test. I mean, look at this. Is flatline, perfect RNG. Who are you trying to impress?

[00:20:11]Anyway, I did try a couple more things from the previous videos, like setting myself to perfect coordinates, fully relogging after each attempt, or even changing my name after each attempt in case that was somehow tied to anything.

[00:20:23]But once none of those gave me an as well, and having already spent over a week with nothing concrete to show for, I was honestly ready to call it because even though a lot is not shown here, I did go through all the results. And yes, even the first test, the one with two times more jerosin guns than green jar talismans, that at the time felt impossible to just happen by itself, apparently didn't even cross this shouldn't happen line. I mean, to be fair, it got to the verge of crossing it. And that still doesn't capture how it felt getting that as your first attempt cuz apparently mathematically it doesn't matter what it does psychologically. Okay, I really thought I had something. Out of curiosity, I also cropped out the most extreme segment from that session and leave alone that it wouldn't even be statistically valid. Even that didn't cross RNG's broken line. Absolutely losing it at this point. I tried the same with the most extreme segments out of all the sessions, which at this point I don't even know what I'm doing.

[00:21:16]Definitely not science. But even then, none of those crossed RNG is broken line. And so absolutely defeated at this point. I just sat in the main lobby with the goal to open the remaining boxes just thinking to myself, "What a stupid idea and a giant waste of time was all this."

[00:21:35]But then, I don't know how, maybe it was the sheer luck I was chasing this whole time. Maybe fate. In the very same lobby, I was ready to waste my remaining chances, a mega party has formed with me, and the man himself, Future 77, somehow both ending up in it. Which to be fair, neither of us had any business being that party. Just bunch of people trying to be funny. Like a guy asking Future if he's going to do a fanboy cosplay at 77K. Like what do you even expect him to say? I mean, he's probably going to get mad, laugh it off, or just ignore you.

[00:22:06]Well, I must say I did not know this man as I thought I did because the response we got instead was a resounding yes, sir. which I'm not going to lie shook my entire understanding of true commitment to the cause because if he is willing to go to such lengths not to disappoint his single fan who am I to give up now I think he was then trying to explain that he was actually responding to something I did not care that was the version of future I wanted to believe in and that gave me strength not to stop not before there is nothing more to be tested and so with the newfound inspiration I proceeded to test quite literally everything. I gambled high. I gambled low. I gambled high while being high. I gambled while constantly falling to the void. I recreated the greatest jump of all time and gambled in the middle of it. I gambled inside the balloon. I gambled inside the goon. I continued inside the prison. And as a free man, I tried worshiping Dante, Aura, and even that one guy in the basement. Then it pains me to say so. But yeah, objectively, Aura was best. I even began observing cases of undisputed correlation such as getting better drop rate while wearing pants compared to without them or clear as day influence of competition achieving significant improvement in results while actively trying to outgamble other guy in my lobby. Bro, he's getting better jobs than me once for all. Proving that if you're not getting results you want, you clearly just need to try harder. I tried to reverse psychology by not gambling at all. Instead, going outside having life.

[00:23:33]Worst strategy by far. I didn't even get a single drop. Realizing I'm still about one everything away from testing everything, the next logical step was to test the next thing I didn't test yet.

[00:23:43]Like completing a chicken rice on thousand ping gambling the entire room.

[00:23:48]And if you think that's a joke, it legit took me like half an hour to complete the single. The current record is 52 boxes. If anyone is wondering, nobody's wonder what am I. And at this point, I can literally hear you asking, okay, well, I sort of understand all the other stuff. There is like no physical way that would have to do with anything.

[00:24:10]Yeah. Well, what if I tell you that one of the attempts included one in 10,000 streak of me getting six talismans in just 13 attempts? I I mean that has to mean something.

[00:24:20]What is that? Did it cross the line?

[00:24:23]Okay. Well, no.

[00:24:29]So, probably not.

[00:24:32]A similar fate, though quite predictably, met every other one of those weird attempts I tried. And listen, even though it may seem it was all just for a fun bit, which to be fair, most of it was, there was also kind of a method to the madness here.

[00:24:46]Because sometimes, even though something was clearly unplanned by anyone, just by sheer coincidence of parts interacting with each other, relations emerge. And all it takes is to stumble upon one of those unplanned relations to follow the thread back to its source and expose the whole system. Which, to be fair, trying to do it on purpose is a bit of a hell marry. But now, I can say that I did not try even that. However, now with no more boxes left to open, I had no other choice but to concede. Defeat and accept the fact that I did not find anything, which now to come clean with you, I cannot say I was surprised. I mean, Hypixel is a multi-million dollar studio running for how many years. It would be silly of me to expect that first of all, Hypixel randomness does have a hole, and second of all, that I would be the one finding it, which doesn't mean I wasn't going to try. I mean, my mind would let me to just keep seeing those weird results and not at least confirm that they are in fact in order. And now seeing with my own eyes that the RNG, despite trying its absolute hardest mathematically, did not break even once, it would be disingenuous of me to try to spin it any other way than conclude that the RNG is fine.

[00:25:58]Or that's what I thought until I went through results again and saw a for sigma result in the transition matrix of the biggest session in which I didn't even do anything. So it was supposed to be a staple of perfect RNG. And yet for some reason 3D jar glasses transition to foraging XP literally three times more often than they should have. Not only that, but upon a closer inspection, in almost half of those cases, it transitioned to foraging XP again, resulting in this pattern that mathematically during 5 and a half thousand bucks session, I shouldn't expect to see even once appear five times instead with the deviation blowing past the RNG's broken line straight to I am not even joking, 6.9 sigma.

[00:26:40]Nice, but also impossible under normal energy. Although after a few seconds of excitement thinking I actually found something, the results started to feel a bit fishy because I started to think what is actually separating getting this pattern from getting I don't know 3D glasses followed by two farming XP instead or two foraging XP being followed by 3D glasses or farming XP follow. Yeah, you get the point and the answer is nothing. And because all of them have the same odds to appear, while it would be very difficult to get this one specific pattern to appear five times, getting one of the patterns like that to appear five times is a lot less rare. Which including that in the calculations didn't change the result as much as realizing that I was using a wrong formula in the first place. And for such tiny probabilities, the formula I should actually be using was this one, giving a result of 3.3 sigma if I predicted this specific result, or only 1.7 for a result of this nature, which is still rare considering that there is only 4% for any pattern like that to appear five times, but definitely not impossible. And so, still looking at the 6.9 sigma result, in my mind, I was like, okay, well, you almost got me. For a second, I thought there actually is going to be a twist, but at last, as expected, no twist today. Conclusion remains, RNG is fine. Time to pack up, except I decided to check one more thing. And like I said, I got glasses plus foraging combo 12 times, making for 12 spots for different items that could follow it. And it would be weird enough to see five of those spots be taken by foraging XP listed at only 13% drop rate. What if I tell you that almost the rest of them were taken by mining XP listed at the same rate, giving us two patterns that in total appeared nine times when even after adding their probabilities, on average, they still shouldn't appear even once. And when you think I'm done, I am not even done because pretty much every single instance of both of those patterns happen only during that one specific session. Well, for the remaining 15,000 boxes, I saw only the second pattern appear one more time, making for 27 times higher concentration in that one session. And this time, I try to be very careful to account for everything and not overestimate the exact chances. And as long as I'm not missing anything, even according to the most conservative calculations, there should be about 1 in 50,000 sessions. Not even to get the exact result, but just a result of this nature. Or in other words, I would have to be opening one box per second non-stop for eight and a half years spending net worth of top 100 richest players combined just for something similar to happen, which at this point it just cannot be random. And I wish I could give you some kind of neat summary of the absolute chaos that ensued from that. But I don't even know if that's possible because one second I'm exploring each exact step from user input to generating a specific outcome that explodes into bunch of different topics. From the exact algorithms behind different possible random number generators, the architecture behind a network as big as Hypixel to the minute things like the things influencing the variance in the delay of transfer of information, which only created more questions than answers. So the next second you can see me abusing a random cow inside the hub and discovering a weird pattern of the same exact 12 drop sequences repeating about every minute.

[00:30:01]Thinking it could just be a pre-written order, not actually using the same randomness code, I go back to the Jerry box data and what do I find? Two exact same 11 drop long sequence appearing not that far from each other. This time with 0% drop in the middle surrounded by the same exact drops with the chance of sequence like that repeating being apparently one in billions which feels like even more damning evidence. So, I message admins, they don't respond. So, I do a little bit more digging and realize what I have discovered is basically a birthday paradox, but on a much larger scale. What is a birthday paradox, you may ask? Well, it basically goes, how many people do you need for a 50% chance of someone sharing a birthday? Some may say, I don't know, like 180, no 23. Why? Basically, because of our main character syndrome, instinctively looking from our own perspective, which there is a pretty low chance of someone sharing the same birthday as you. But we forget that.

[00:30:50]Then the next person gets to check against everyone else. Then the next person, then the next person. And in the end, you are checking so many times that it is almost bound for someone to step on the mine. And in our case, there is just so many different possible 11 drop sequence per checks that it is not that weird to find a match. I feel stupid. I delete the message. But then I start to consider if I should consider the fact that they happen so close to each other and that had 0.4% drop in the middle. I am not sure. So, I decide to compile all of the information I have and give it to the smartest AI to look through. It has 10 minutes anime level internal dialogue about considering every single possible option before just giving up. Oh, I I just gave up. Left to myself, I'm trying to re-evaluate if the foraging pattern could also be the subject for the same problem. It does not seem so. And even if it still wouldn't explain the higher concentration in that one section.

[00:31:42]Getting desperate at this point. I'm trying to look through every single possible explanation. And the short answer is I don't know.

[00:31:51]But I have a plausible theory which involves two parts. First, randomness.

[00:31:55]And we are talking about Java most likely optimizing for speed rather than security. So probably achieved through Java util random using this equation.

[00:32:04]And if not, it doesn't change all that much. And I used to be confused how this equation by just keep feeding the results to itself somehow achieves the perfect desired distribution. Well, the answer is it's not. And to oversimplify it into oblivion, what is happening not literally in that order but effectively is that you have a circle with consecutive numbers. You can then split that circle to the desired distribution.

[00:32:24]And what this equation is doing is effectively shuffling the order in which the numbers appear. Then the program just picks the starting point the seed then just goes down the list and bam randomness. And because each state appears exactly one time before the cycle repeats. Okay, I mean there is a caveat but the point still stands.

[00:32:41]That's why it is technically true that getting unlucky at the beginning guarantees you getting more lucky later.

[00:32:46]But in the reality, the cycle is just so massive that you could spend your entire life and not get through even a fraction of the fraction of it. Thus never getting to the point when the lack would guarantee to flip. And the actual problem someone implementing randomness would be concerned with is that because the order is literally fixed. Just by observing enough consecutive drops, eventually there becomes only one possible place on the circle you could be knowing which would then allow you to predict every drop that follows as long as you are moving one RNG state at a time. And are you? Well, to answer that question, we move to part two. Hypixel network which consists of probably a data center and bunch of machines.

[00:33:28]Inside each physical machine there are some virtual machines. Each virtual machine is running a single server instance like hub one. Each server instance has a single main thread which is like a guy executing the orders like open the menu for this guy or show damage number for this guy etc. And that single main thread has a single RNG state. And every time you need a random result, the guy just looks at its current state, maps it onto the corresponding drop, and advances it to the next one.

[00:33:58]But now the problem with predicting your drops is that so is probably every other person or thing in that lobby. And even in a single second it took you to get from one box to another, the state probably has already advanced hundreds or even thousands of times. So, if what I said was even just conceptually true, in order to predict your drops, you would have to not only find the RNG state that gave you the drop, you would have to then predict the action of every player on that server, which even if that was just theoretically possible at this point, you might as well be taking a job from a god. But wait, what if there are no other players? Well, the RNG still doesn't stop because you still have mobs, particles, NPCs. But what if you remove everything? Well, that's where things become more interesting and could get us closer to the answer. I mean, the RNG still probably wouldn't stop because you still would have random block and environment checks, but those would update the RNG by the same amount each time. And if we go back for a second to the linear congreial generator formula, the thing is that you can actually pick the parameters to be anything except if you're not going to do it well enough. Then funnily enough, if you look at the generated numbers not individually but as pairs or triplets, then the results could start to fall on the visible parallel planes of the inherent latis structure. Or in other words, you could start to see patterns.

[00:35:21]Well, except it shouldn't even apply here since parameters. So something like Java to random were chosen specifically to pass the spectral test. Meaning the pattern should not be visible either way. Although that on the other hand applies only when you actually using the formula. Meanwhile by skipping the same amount of results each time you accidentally started using entirely new formula that wasn't even written anywhere in the code and that could start to produce patterns again.

[00:35:51]But now the question you may be asking is would it be exploitable if that's indeed what happened?

[00:35:57]Well yeah not really. I mean first of all you don't know the exact formula could be from Java random could be not. Second of all you don't know the size of a jump and even if I think the precision it would require to pull it off would be virtually impossible due to random delays you'd be exposed to in form of ping. But if you're not trying to exploit anything, just opening boxes as fast as possible, then I guess from time to time, it is plausible that the frequency of your opening boxes could get randomly aligned with the frequency of the game updates making some patterns appear more often than they should have, which would be kind of cool, if true, because I'm not even sure that this indeed what happened, but I already spent way too much time looking at math looking for something that may not even exist. So, I don't care. But what I am pretty confident about though is the theory of players sharing the same RNG, which I'm also realizing it could just be a common knowledge and I might be just stupid. Either way, it would make a lot of sense both logistically and how you can take the weakest RNG and make it the most unpredictable there is.

[00:37:04]Although, I guess except quantum randomness. But I mean, even quantum randomness have like probabilities and stuff, right? And you can never predict the next action of a sky block player. I mean, just factually true, which would make it for a bit of a bittersweet ending to my whole randomness saga. I mean, on one hand, it is kind of sad that we were never able to predict drops in the first place, which even though it was kind of expected, it was the slight possibility that was making things interesting. On the other hand, if that's the case, it is better to know that so I don't have to ever wonder if I could have positioned myself differently or done something that would prevent my terrible luck when all this time it was just Jamie who decided to waste the single RNG state that would give me a warden and heart on punching a zombie.

[00:37:49]You know what? you, Jamie. And also, you know what else is not so bad about the fact that RNG was not in fact broken? Yeah, as expected, money, which may not came from me breaking the Okay, I mean, technically it did came from the process of me breaking the RNG. It's just that it wasn't because the RNG was broken, but people don't do math.

[00:38:14]Listen, I have to make the title work somehow.

[00:38:20]So, you thought it was the end, right?

[00:38:22]Well, that's what I thought, too. But before anything, do you know how much 1,000x increase is? Well, to visualize it, basically takes you from $1,000 to a million. From a million to a billion, from a daily occurrence to one in 3 years. And do you remember that 11 drop sequence that just happened to miraculously repeat itself with like one and a half a billion chance? Well, that chance was calculated on the fact that I opened 5,000 boxes in that session. But I also opened something else about 1,000 less times, making observing any statistical event that much more unlikely. And why am I telling you all this? Well, because after all this, I remember that I do have a recording of this session. So, I can just go to the exact moments I got both sequences and just check if there is anything that connects them. The first pattern happened at 241.

[00:39:13]What we are looking for exactly, I'm not sure. I guess for something to stick out be the same for both instances. Okay, we are almost there at 24D.

[00:39:25]Wait, wait a second. I just run out of boxes at 240. Oh, no, no, no, no, no.

[00:39:31]There is no freaking way I open this chest again. There is no way, right?

[00:39:37]There's no way. Don't you dare.

[00:39:41]I've opened. Yep. Is literally the pattern. I mean, it still doesn't mean anything cuz the second pattern happened at 307.

[00:39:50]We have to go like No, no, no, no, no.

[00:39:54]Why are we at 306 and I'm about to run out of boxes suggesting that I'm going to refill the boxes again? I mean, there is still no guarantee that I'm going to open that. No, don't don't you dare.

[00:40:06]Don't.

[00:40:09]This This is some kind of witchcraft. I made my piece. No, no, no, no, no. I'm leaving. I made my piece. Goodbye.

[00:40:27]Also by my desk mat. It is very cool in a limited time and I tried and okay bye.