Flight Rising Discussion

Keeping this thread as a counter of how long it'll take me to get multi-eyes on my Fifth Anniversary dragon. So far we're up to 462+ [center]* * *[/center] I have a[url=http://www1.flightrising.com/forums/bug/2461757] bug report[/url] up with some fairly solid evidence that there may be dragons that [i]cannot[/i] get the higher rarity eyes. I know I've heard of others who dropped more than 300 vials on one dragon without seeing Primal or Multi-gaze, and would love to have them weigh in and see if we can't pin down what may be at the root of the bug. (For those who'd rather not follow the link and the wall'o'text it contains, yes, I was aware of the odds going in, and while there remains and will always remain a chance that this is simply bad luck, the odds of it being anything other than a bug are vanishingly small. From my numbers alone, it is 99-99.999% certain that something is wrong, and more people with more experience will assist with pinning down not if, but what.) Edit: 460 attempts brings me to 99% surety even using the absolute lowest calculated probabilities. Edit redux: We've had a response from one of the developers. [quote name="Kaepora" date="2018-06-16 13:53:54" ] Hi @sraben, I can confirm that neither dragon nor user-specific data are involved in generating a random eye type. Sly really does have the same probability of getting multi-gaze as all other dragons, but this is a case of particularly bad luck - on the order of a 1/100 chance after this many tries. While it doesn’t change the outcome, I do want to assure you that our testing before the launch, and the real world results after it, have yielded expected numbers. I do hope that RNG will be much kinder to you. [/quote] So that seems like it's the end of the story, and perhaps a fitting one: I was shooting for a perfect "Fifth Anniversary" dragon, and I've got one. One with 69,000g rare eyes. Thank you all for your support and I'm sorry I didn't end up with better news.

Haven't got any but I only spent like 30 scatters or so, so it doesn't count as "too many"

[quote name="chouki" date="2018-06-16 10:24:41" ] Haven't got any but I only spent like 30 scatters or so, so it doesn't count as "too many" [/quote] Yeah, that's part of the struggle. When some dragons simply can't get the higher rarity eyes it makes pinning down probabilities tough. So 'too many' right now needs to be 'overkill' to have much weight to your claim. 30 scatters is still painful if you walked away without feeling like you got anywhere.

i wasted 80 vials to get multi and i don't think i even got rare from that <:I idk how many that is compared to 300 tho holy hell

Even if there is nothing wrong and RNG hates you that much, they should reimburse you and restore at least half of the scatters you used. This is insanity.

I'm inclined to think it's RNG, simply because probability can never be pinned down, but good lord. I sincerely hope somebody addresses your problem.

I only used about 50 scatters. I did get one faceted eye, but nothing on the dragons I wanted.

Oof, and you just want to make a pretty eyewing dragon...

It may just be RNG being very mean (given the chance each time you roll is ~1%) but that many vials is absurd!

@Laluzi

Aye, It will always be possible that RNG is having a field day. If everyone on right now (~3,000) spent 450 vials and our lowest assumptions of probability are correct we’d expect about a half-dozen people to still be out of luck.

The only way I can address that is by finding those other players and seeing if there is some common thread, (all skydancers, or all hatched on the 8th etc.) or by continuing to throw vials.

We’ll see.

And I do appreciate the support. You all are good reasons to keep at it.

Somebody made a point about this potentially being a seed error - that if numbers aren't generated purely randomly, it might be that certain dragons are inherently advantaged or disadvantaged. This sounds pretty wild, but I know of one other equivalent example - in 3rd-gen pokemon games, specifically Emerald, two of the four numbers used for shiny chance calculation were set at character generation, one invisible and one not, and if you could figure out the invisible number, you were able to guarantee the game would generate a shiny by initiating a battle a certain (varying but tied to those two fixed values) number of milliseconds after the game booted. And theoretically, if one of those numbers had been very low (less than 8 in a 0-65355 range), your game would have been constantly generating shiny pokemon. I think the game actually made sure not to use those values for ID and invisible ID.

The reason this worked was because the game's generated numbers, which were refreshed many times per second, were not actually random. From an approximation they appeared so, but you would always get the exact same sequential results from the same starting value at load.

Anyway, because that digressed. While it seems unlikely to me that FR would use such a convoluted means of number generation, or involve dragon ID in the calculations at all (it's easier to just use rand(0,100), or whatever the equivalent is that accounts for floats, than to generate a constant seed for the site and then have it interact with dragon ID for results), I wouldn't use any more vials at the moment, until FR has responded.

A bit of a personal question, but would getting Multi-Gaze even feel like a victory anymore? It's a better outcome than continuing to fail, or stopping and letting all that cash and effort go to waste, but even getting it seems like a bitter triumph at this point. Or maybe it would be greater. I don't know.

@Laluzi

Aye, the problem is computers can't do 'random' very well (or at all if you want to get technical).

Sites like random.org actually will use things like radioactive decay or solar events to generate true random distributions. Most things don't need that level of randomness though, so the idea of seeds have been used to get pseudo-random 'good enough' calculations for computers pretty much since there have been computers.

As I understand it, most programs will tie in micro-second clock values tied to user-generated actions to produce random events. In the pokemon example you cited, to get past the random and into consistent results the user would not only have to know the variables, but also be able to act on the order of miliseconds, so it is a close enough approach.

The problem comes in when those numbers get screwy. Say the FR algorithm uses the hatch-time or the ID number, as well as the milisecond time-stamp of the user accepting the vial change. So say 'String of numbers A' tied to dragon plus 'String of numbers B' tied to timestamp, if within 0-x return Uncommon, if between x-y return Unusual, if within y-z return Rare, if between z-n return Faceted, if between n-m return Multi, if between m-p return Primal. By changing the ranges between variables you can tweak odds, and the system is complex and sensitive enough that users won't be likely to be able to abuse it.

The problem comes in if the hex value input from the dragon based 'String of numbers A' gets within a certain range it could be such that regardless of timestamp value 'string of numbers B' the sum can't ever add up to n-p.

It's a common enough bug, (you'll notice that many bug apologies on the site include something like "we apologize that players in the ID #####-##### range didn't get their gathering...." on the surface ID # shouldn't seem to change your gathering results, but it does, to help keep things 'random') but its a bear to pin down and even harder to prove when looking at things from the client side and when there's a 600$ price tag for checking!

Ah, well, yes. Computers have gotten better at approximations of random, but you can't actually tell a computer to pick a random value without already having some means of randomly picking a value. And so on. I actually didn't know that about drawing from simulated half-lives, but that's neat. The more you know!

Humans aren't great at picking truly random values either, but we at least have an idea of what random is.

Pokémon Emerald was an example of a poorly done seed - design allowed it to be manipulated by the user in a way that a website can't, and methods have gotten more sophisticated over time anyway. But for a website... I never actually did web programming, just assignments and modding, but for one-off events it was more useful to...

...oh. You know, it's funny. I went to school for this for four years and yet only now did it occur to me that the function called doesn't necessarily matter, because the function operates on the seed. I like to think I'm cleverer than this. Better late than never, I suppose.

In this case, it does matter whether or not the function uses dragon ID, but you're right that bugs demonstrably happen to ID ranges, and does raise the... if not probability, then seeming likelihood of this being related. The coding precedent exists. And RNG makes it difficult to prove there's an issue, as you said. Impossible, actually, just varying degrees of likely. But at this point, it definitely bears being looked at, and I wish you luck! I'll be checking in on this for sure.