RNG Taken Too Far!

[Skeezy]

1st off thanks I was scared this was becoming less civil.

1. Disbelief is entirely understandable. Also dbz dokkan was #1 game in app store for multiple years until pokemon go and then it went back to #1 (now I dont even think its top 5). Its made by bandai/namco japan a well known developer. The game has been put for almost 5 years. Updates and rate increases have made it so I cannot provide evidence as ssr (super super rare) is common now. I honestly dont fault anyone for doubting this as seeing is believing.

2. I know that all player bases mostly requore data for obvious reasons.

The summon.
The three things invloving the summon of dokkan are
A. streak success/fail for extended period of tme
B. Singles over multis for $money conservation
C.the trick (tap manipulation)

A. Lows / highs streaks lasted way too long in some instances and it was very noticible. Why? i have no clue. When your hot streak is lasting for multiple hours without any designated rate change you can tell. Much like on here if you did 5 multi’s and all had 1 or 2, 5* and then you cashed and got near same result repeated to your hearts content. If you really could just keep repeating would you really need to go back and average everything to see if you were successful now that your sitting on a mountain of 5*'s? Especially if you repeated this process whenever it occured. Would really need to average it in order to know your above the usual rate when almost every attempt is succesful?

B. Is more of my observation as an ex whale. Singles were mostly just as rewarding as multi’s however if you pull what you want early. For example if I failed to get desired card with multi’s or couldn’t do multi’s, spread out singles would still get me the desired character in dokkan. Again purely observation but flawless record inpractice.

C… The character rarity for dokkan was normal, rare, super rare (SR), and super super rare(SSR). The games early years anything above rare was scarce even just sr (this was well before rate displays were mandatory). Meaning you could easily do multiple multi’s and not even get a SR. People would easily empty their wallets chasing the SSR (up to $1000) and wind up with a handful of sr to show for it.

The trick (tap manipulation) allowed me to pull them back to back to back. It did not work like a cheat code however. I would have to try it which invloved doing a multi. Once it was successful a 2nd multi must be done as 1st try could have been dumb luck. Once 2nd multi was succesful I kind of knew and would continue with success.
Once they added summon animations and zi would get the same one 3-5 times in a row was the confirmation I needed to know that it was some type of flaw. The fact I could do it repeatedly and actively predict my success was all the confirmation I needed.

Also I observed things other players paid no attention to that displayed how un randowm dokkans summon was. I also paid attention to my failed characters which were often from the same category.

For example players complained about getting 4 of the same SR character in a multi, I noticed that often the other SR, and Rs would be from that same category. Helps to know dragon ball but two categories were ressurction F (movie) and gt (series). I noticed doing back to back multis that a lot of my failed cards were gt characters in some multi’s almost all the R and SR would be from the same category. Later on in another sitting almost all my R and SR would be ressurection f characters even though I didnt get any ressurection f characters in the previous sitting. Things like that helped ensure me that it wasnt random. After a while as the player base summoned more and more many more players began to question the games randomness even those who disagreed with me as much as Brogg. Though I only had 2 players actually apologize to me after my 3 year debate, not necessarily because I was right but, because they saw whatever it is they saw which let them know that no, its not true random.

Farming wasnt even worth proving to others. Some players saw this on their own others refused believe anything other than true random.

Again no one has to believe me but Im stating my piece for others to read and decide for themselves about rng in general.

[StealthNinja]

These posts have been taken too far! You all are spinning in circles about something that is theoretical… I’m glad you all find reading these paragraphs and then copy/pasting them to comment about some issue you have with it… I did a 10 pull once and 6 of them were 5* including a HOTM… did I land on this magical epic pull because I clicked on a box three times then scratched my ■■■ exactly at 3:46 pm? Who knows and I don’t really care. Good luck spinning in circles!

[Brobb]

@Skeezy

If you really could just keep repeating would you really need to go back and average everything to see if you were successful now that your sitting on a mountain of 5*'s? Especially if you repeated this process whenever it occured. Would really need to average it in order to know your above the usual rate when almost every attempt is succesful?

You would not need to “average” anything. You would just need to record your results and compare them to expectations - whether calculated from published odds, or derived from the experience of the player base. Was this not the very first thing you did?

@Skeezy

For example if I failed to get desired card with multi’s or couldn’t do multi’s, spread out singles would still get me the desired character in dokkan. Again purely observation but flawless record inpractice.

So I think you’re saying here that multiple singles were just as effective as multiple pulls. Is this significant?

@Skeezy

I would have to try it which invloved doing a multi. Once it was successful a 2nd multi must be done as 1st try could have been dumb luck. Once 2nd multi was succesful I kind of knew and would continue with success.

This is the bit that makes it sound like your ‘system’ didn’t work. If you try a multi and it’s not successful, then you assume the algorithm doesn’t favour you and stop. If it is successful, you try again, but if the second isn’t successful too then you stop. If both succeed, then you assume the entrails favour you, and continue pulling… until you fail for a bit, whereupon you assume that something has changed, and stop.

The effect of this ‘system’, when outcomes are random, will be that the only times you will make a lot of of pulls will be during streaks of success. Your failures will be just as plentiful as anyone else’s, but you’ll spread them over more days.

How could you escape being deluded into thinking you had uncovered something significant? The same way you could overcome the scepticism of those who just want to see the numbers to back up your claims: by recording your results, measuring them, and comparing them to expectations.

@Skeezy

The fact I could do it repeatedly and actively predict my success was all the confirmation I needed.

@Skeezy

For example players complained about getting 4 of the same SR character in a multi, I noticed that […blah blah blah…] to know dragon ball but two categories were ressurction F (movie) and gt (series). I noticed doing back to back multis that […blah blah blah…] a lot of my failed cards were gt characters in some multi’s almost all the R and SR would be from the same category. Later on in another sitting […blah blah blah…] Things like that helped ensure me [I think you mean ‘assure’ me] that it wasnt random. After a while as the player base summoned more and more many more players began to question the games randomness even those who disagreed with me as much as Brogg [heh heh heh] Though I only had 2 players actually apologize to me after my 3 year debate, not necessarily because I was right [shocking] but, because they saw whatever it is they saw which let them know that no, its not true random.

I stopped getting headaches several years ago (don’t know why). After reading all that I almost miss them.

All I’m asking for here is a little bit of data analysis to back up the assertions (“noticed” this, “noticed” that). It seems like not only is there none, there has not even been any attempt at any, which is shameful.

It’s a total mischaracterisation to say that I “disagree” with you. You have asserted a whole bunch of stuff, I have asked for the evidence you collected to support your assertions, and you have replied that there is no evidence. I don’t “disagree” with you at all - there’s nothing to disagree with.

[Kikyo]

@Brobb

It’s a total mischaracterisation to say that I “disagree” with you. You have asserted a whole bunch of stuff, I have asked for the evidence you collected to support your assertions, and you have replied that there is no evidence. I don’t “disagree” with you at all - there’s nothing to disagree with.

All hail @Brogg , the most logical moustachioed cat of the land! @JonahTheBard would be proud, Gato Baffi.

[Garanwyn]

I’m going to let you in on a secret: don’t tell anybody, but they really don’t kick you off the forum if you choose not to read every thread or every post.

I’m not quite sure why the idea of this debate is so upsetting to you, but I promise if you just skip it, nothing bad will happen to you.

[TheSchmoo]

PRNG engines do typically use seed values, but repeating seed values does not necessarily generate a repeated pattern. (Maybe it did in the olden days.)

Incorrect, at least for the languages I typically program with (C/C++/Java/Perl). Using the same seed value generates the exact same pattern. That was the intent, and was necessary in order to validate programs including student code that I graded.

Even if a repeated pattern were generated, the worst PRNG from 1985 would still be perfectly fine for all E&P purposes

Disagree. If you mean EPOCH time, it is known to be a bad pattern seed generator and has been shown to be so in numerous papers. Your other examples are not good seeds either. Visit RANDOM.ORG.

And of course, we don’t actually know what E&P uses.

This reminds of the stink some years back when someone decompiled the source code to Magic the Gathering Online. It had a basic random() call with a timestamp. Anyone versed in the subject would be alarmed at such an ignorant and bad coding practice for such an integral part of the game. As it turns out (according to a senior person at WotC, who first tried to be all huffy about violating the ToS by decompiling the code blah blah middle finger), the client side code was only for the test deck shuffler. The real RNG function was done on the server, as it should be to avoid exploitation.

A better question is truly random a good idea? To cause frustration for players who then stop spending money and/or quit the game? Solutions to guarantee a top reward after time is a good idea, but that’s just my vote. Dilution of some extra 4*/5* here at there is not a concern to me, the game already has anti-inflation measures built-in.

[Garanwyn]

@Skeezy

Especially if you repeated this process whenever it occured. Would really need to average it in order to know your above the usual rate when almost every attempt is succesful?

Yup. That’s how science works. Because people often believe things are different than they really are. How big is a “mountain?” How many draws did you really do to acquire that mountain? How often did you do 3-pulls and get nothing, then decide the game was cold and stop?

Again, I’m not saying you’re wrong. But there are a lot of people who will tell you they’re great gamblers, and talk about all their winnings. They have a strategy that “works practically every time.” Their bank balance disagrees though. As would actual data.

Also, just like Kerridoc said:

@Kerridoc

why do you think that the PNRG quirks you were able to exploit in your previous game exist in E&P?

You say:

@Skeezy

My main point to the board and fellow players is when it comes to rng don’t be close minded. Never think you know for sure, because people have found exploits in this similar rng.

This is a really weird claim. How do you know the PRNG are “similar?” Do you mean, you think every last game with a PRNG is likely to have an exploitable flaw?

I mean, even supposing, for the sake of argument, that I accept your Dokkan claims as 100% accurate. Claiming that there’s any likeliness of a similar flaw here is like saying:

“Big National Bank made an error every week and deposited an extra $500 in the account of everyone who opened their account in February. So you should be open-minded about the possibility that opening an account at Other Mega Bank will also have some optimal month.”

Sure, I’m willing to admit it’s possible. But banks work really hard to NOT give away free money. So the default hypothesis should be that there isn’t a good month to open a bank account.

[Brobb]

@TheSchmoo

Using the same seed value generates the exact same pattern. That was the intent, and was necessary in order to validate programs including student code that I graded.

I’m a few years away from knowing anything serious about coding. Back when I used to pretend to understand anything about it, anyone seriously seeking random outcomes using a PRNG combined a seed (usually time-based) that was generated client-side, with a seed (usually time-based, but sometimes sequentially based) that was generated server side. This ensured that single seed manipulation could achieve nothing. Maybe things have relaxed since?

@TheSchmoo

If you mean EPOCH time, it is known to be a bad pattern seed generator and has been shown to be so in numerous papers. Your other examples are not good seeds either.

Okaaaay… I’d be happy to discuss the problems you think there might be. Do you feel there is some aspect that would b open to manipulation, given two million daily players? How, exactly?

@TheSchmoo

Solutions to guarantee a top reward after time is a good idea, but that’s just my vote.

That’s a perfectly valid opinion, but (of course) there are 560 million reasons to suggest you are wrong.

[Garanwyn]

I guess what I’m trying to say is: if you’re bored, go ahead and skip this. No one is making you read this thread. Were you seriously expecting a groundbreaking, novel debate in a thread entitled: “RNG Taken Too Far!”?

It’s literally fewer clicks to go back to the main forum page than to post here and tell everybody how much the debate bores you.

I mean, if it were called “New Info on Kunchen!” and then you saw this morass, I’d totally get your reaction

[Brobb]

Heh, heh, heh.

I’m always flummoxed by the posts that complain about people posting.

[Garanwyn]

That was definitely a weird Excel-ism. Not sure why it happened, but I’ve fixed it for you

[TheSchmoo]

PRNG combined a seed (usually time-based) that was generated client-side, with a seed (usually time-based, but sometimes sequentially based) that was generated server side. This ensured that single seed manipulation could achieve nothing. Maybe things have relaxed since?

I don’t see a good reason why you would want or need to accept any value from the client used to generate a seed. Don’t trust it, you can always use multiple values on the server. Time based was the ‘best’ method originally conceived, until the whole idea became better studied. Now it is widely mocked.

Okaaaay… I’d be happy to discuss the problems you think there might be. Do you feel there is some aspect that would b open to manipulation, given two million daily players? How, exactly?

Manipulation: No, not necessarily. My point was it doesn’t generate random results. The original code calls for EPOCH time returned a value in seconds. I can’t fathom how many E&P summons are done each and every second, but it’s far beyond 1. Every player would have the same loot seed who did a summon in that second! You would have 5*s and HotM’s appearing in bunches. That is not random. Time value generators were eventually upgraded to include the option of pulling microseconds, and there might be even finer values available now that processor clocks run as fast as they do. Those would be better options. So it all depends on how educated the development team is regarding the matter and/or how much they care about PRNG vs much ‘truer’ random results.

[Skeezy]

I concede. As usual someone will twist my words or read them how they want them to be read. Specifically the part where I could actively predict the summon using tap influence. Do i know exactly why it only worked? Hell no. Doubt is fine with me.

I dont think this board understands rarity levels of the games early life and its increase of summon rates and drop rates till now. Talking about a different game yes but this is how most games like this work. Rates increase over time (through updates over years) and its visible to everyone. So why would an obvious rate increase for a period of time not be visible otherwise? I know i know everything I’ll say is gamblers fallacy my warped little mind is just overworkimg itself to make things appear etc. Etc. Etc. Etc. Riiiight.

Or rather you would pretend rarity levels come and go while just farmin normally, to suit your arguement as though anyone could just farm in everything in 2 hours everytime on any game. and thats ok… its ok. No need to continue, I see where this is going.

I’ve said what I had to say, proceed.

[Garanwyn]

I get that you don’t like people doubting your claims or asking for proof, but that doesn’t mean anybody is twisting your words or is closed-minded.

Here’s my summary of the situation. You tell me where I’m wrong:

1. You claim a game you used to play, Dokkan, had an exploitable PRNG.

2. You didn’t collect detailed data to verify how good or bad your results actually were. Instead, you relied on your own sense of how very, very obviously good and repeatable your results were as confirmation of your beliefs

3. You don’t know, or claim to know, why or how the PRNG on Dokkan was actually flawed

4. You recently started playing E&P, and after a fairly small sample of summons (well less than 1000) you believe you have detected an exploitable pattern in the PRNG here too.

5. Again, you also haven’t kept track of any of your own summoning results in E&P, despite having no evidence or reason to suspect initially that there was something wrong with its PRNG. You are likewise relying on your own sense of correlations to figure out the claimed exploitable pattern.

6. Despite doubting your claims, several of us have volunteered to test them

[Skeezy]

Ok. thumbs up.

This debate isnt new to me either, and as I said its reached its platue. To continue would be trivial and pointless. As you said I dont have proof, only my word and my b***z. I merely wanted to add another point of view to the discussion.

…But…if someone says psuedo random is true random again, no lie, it might come back up lol.

[Garanwyn]

I promise it will come up again and again, from you or someone else Random data is super mysterious, and we’re pattern-seeking machines. It’s what nature has made us to be.

I also promise that I’ll do my draws in the middle of your window, spaced out over the days of Atlantis (or Wonderland) to maximize the ability to hit any hot spots. If I get a good pull sequence, I’ll follow up (money permitting). You can expect order of 120 samples from me.

[Skeezy]

I wish you guys had played dokkan lol.

You guys are awesome in that you will actually go and check for yourselves. In the beginning with my 1st findings on another board way back when, all I was asking for was help and volunteers. Turned into a monsterly unproductive argument.

It makes me think very highly of the community. Wish I actually had something more solid and skeezy proofed (lol) to give you guys for E&P but, alas I dont.

[Brobb]

@TheSchmoo

Manipulation: No, not necessarily. My point was it doesn’t generate random results. The original code calls for EPOCH time returned a value in seconds. I can’t fathom how many E&P summons are done each and every second, but it’s far beyond 1. Every player would have the same loot seed who did a summon in that second! You would have 5*s and HotM’s appearing in bunches. That is not random.

@Garanwyn has already explored this possibility:

@Garanwyn

That said, the question of consequences is an interesting one. The real downside is it massively inflates the effective correlation in the aggregate results of the PRNG. Suppose you generate a seed that initiates at a 1-in-10M result. If you have 100,000 people logged in and active during that second, and mix on ms, then we might randomly expect 100 of them to get that result. Bad, but not awful. If we mix on seconds, 100,000 people simultaneously get a 1-in-10M result, and that skews the aggregate statistics badly. On average, during testing, you’d never notice it, because any single client follows the correlation statistics of a single instance of the PRNG.

In the unlikely that E&P used a single, seconds-based seed for each summons, players’ results would probably be quite highly correlated (relatively, anyway). For individual players, however, results would remain unpredictable, unexploitable, and distributed consistently with E&P’s published probabilities.

We could debate whether philosphically those results would actually be random (I’m not certain either way, but it bears observing that tossing a coin produces a random outcome no matter how many people simultaneously bet on heads). In a practical sense, though, would that methodology offer any downside compared to pure randomness, for individual players? I still can’t really think of any.

[Garanwyn]

@Skeezy You’re welcome. I think the forumers here are a good crew. Hopefully, you’ll be with us a while, and you’ll get to see more of that first hand.

@Skeezy

Wish I actually had something more solid and skeezy proofed (lol) to give you guys for E&P but, alas I dont.

Lol. No worries, man. The worst that happens is I do the pulls I was going to do anyway, and get what I get. And the best that happens is I get more than my fair share of 5*. There’s literally no downside.

[Garanwyn]

@Brobb

We could debate whether philosphically those results would actually be random (I’m not certain either way, but it bears observing that tossing a coin produces a random outcome no matter how many people simultaneously bet on heads)

Interestingly enough, for my undergraduate degree, my minors were math and philosophy. I feel uniquely qualified for the upcoming debate. Bring it on!