Bug Impossible Drop

Discussion in 'General Archive' started by sebastian_fl, Jun 1, 2017.

Dear forum reader,

if you’d like to actively participate on the forum by joining discussions or starting your own threads or topics, please log into the game first. If you do not have a game account, you will need to register for one. We look forward to your next visit! CLICK HERE
Thread Status:
Not open for further replies.
  1. sebastian_fl

    sebastian_fl Count Count

    Summary
    It is somehow possible to drop two unique items with identical base stats and enchantments despite the probability of that being close to zero (less than 0.0000000001% which is mathematically impossible) taking into account 5 random variables with a wide range of possible values.

    Details and images

    Some time ago, I have dropped Q4 2-handed weapon with the base damage value of 420-660. It is a decent value, but I wanted to improve it further, as my q7 weapon has better.

    I have also dropped a Riddle with close to perfect stats, all thee base stats are yellow. And it was before the 'yellow' was nerfed.

    Yesterday, while farming for better weapon, I have got another Q4 Staff with identical base values (420-660).

    [​IMG]

    I would be surprised, but it is probable, as the enchantments were different. Probability of that was (guessing) about 1%. So guess you would also say - random, no big deal and no bug.

    But the next run I got Riddle, with all three base values and all enchantments identical to the one I owned.

    [​IMG]

    Im not sure if the PRNG is bugged, or is it a display error, or something else. Fortunately, I have both weapon and Riddle dyed, so know for sure which one was old and which is potentially bugged. But I keep both for now if you guys want to check some data on my account.

    Regards.
     
  2. cdeepal

    cdeepal Forum Baron

    It is not a bug.

    PRNG - The name is pseudo random number generator. It is not truly random. It is pseudo. That means it tries to simulate a random number. Most probably it uses the current time as the seed. When you do that, the resulting random number within milliseconds could be the same.

    That is the explanation for the same item being dropped at the same time. It is not only true for boss uniques, but for everything else. For example, I got two green weapons with 76% golden lines from the same mini-boss once. And so far, those are the only green weapons with golden lines I got for months of hunting.

    It is perfectly fine. This is the way random number generation works. Nothing to fix.
     
  3. sebastian_fl

    sebastian_fl Count Count

    I have masters in applied math and I'm in software business for 15 years. I have developed a bunch of randomizers myself, as well as tons of hashing algorithms, Huffman, Fano codings etc.

    So trust me, I perfectly know how the prng works. And there is no way that could happen with any modern method. Millis are never used as a seed, this would be junior dev if it is. The seed should not repeat. The queue of the random function calls should have not been stack with my client calls only, as there are thousands other calls happening at the same time.

    So if you are right, and I (system) indeed got the same seed with some miracle (and hit the same sequence # of that same seed=same sequence), and somehow the seed is client specific, this is a major development gap. If it is client specific, first of all it is insecure, and it is possible to break it with injection. It is also unfair, as some people may be never getting some seeds due to the way their networking works, or the clock.

    Even if millis are used as a seed, and they reset the seed for each client at restart, I should have hit the same millis to originate the seed, because (oldmillis+1) will create a whole another sequence, and there is no way to hit the same sequence 5 times with another seed, unless the algorithm has some serious issues.

    To go even further, imagine the programmatic start of an app at the same time the other day, previously learning the sequence by crafting 100 items, or doing jesters (assuming it is client specific).

    Not even saying, there might be a non prng explanation of what have happened, something related to this very specific item, for example. Something related to their recent tuning of what the yellow base stats are. Etc.

    So, all I'm saying, it is abnormal event worth investigating. If they are not careless, which I hope isn't the case.
     
    Last edited: Jun 3, 2017
    EmilyRose and _Baragain_ like this.
  4. trakilaki

    trakilaki Living Forum Legend

    You may have masters in applied math ... but you don't have masters in Drakensang Drop :D
    Those two never get along.
    I had this theory ... long time ago ... and proved it in action (not scientifically using math). Just search some of my posts. :p
    Here is another example of copy/pasted stats (green to blue item craft)

    [​IMG]
     
  5. _Baragain_

    _Baragain_ Living Forum Legend

    Interesting.
    If it were not for the Riddle case, I'd have said that it was unlikely, but not a bug since the damage only consists of +Damage and +Max Damage where Max damage is a fixed value and only Damage changes (test it with random T4 staffs if you'd like) meaning that the damage gap is always the same.

    The Riddle is much more interesting, especially since it has identical enchantments on top of the identical base stats.

    To me it is further proof of just how broken the PRNG is in this game that something like this can even happen.

    Come now, we already know that DSO only hires the best and brightest.
     
    Morinphen likes this.
  6. krawler2018

    krawler2018 Junior Expert

    i think u should be glad u even got unique drops i played q2 yesterday and killed arachna at least 10 times and i didnt get any uniques
     
  7. EmilyRose

    EmilyRose Forum Commissioner

    nothing is impossible for DSO system random . my question is always the same : is it really random or every characters have their own drop ?
     
  8. Grasschewer

    Grasschewer Forum Apprentice

    Hi, I play normal, try and see the system there.
    You are showing T4. Do you know what any Tier is compared to Normal.
    You complain about small random T4 then what is Tier supposed to do.
    Also Unique have set what and random what? They are likely to be the same, ref more than Tier system.

    Ofcourse I don't have real Tier experience as I play painful about 1/50 and tend to die early into it.
     
  9. cdeepal

    cdeepal Forum Baron

    And you assume I don't have any of those?
    Not only that I have a masters in computer engineering and applied mathematics. I have done everything you have mentioned and more.

    You assume that the PRNG will be developed the way you think it will be developed. It doesn't have to be. There are many ways a random number can be generated, so you cannot make an assumption of one way or the other. And why does it have to be perfect anyway?

    What is wrong with getting two similar items at the same time? I don't see a problem with that.
     
  10. sebastian_fl

    sebastian_fl Count Count

    If you had any of those, you would know that there lots of basic principles of good prng. Unless you have copy-pasted it from stackoverflow, where guys do indeed use dummy seeds.

    Those principles include uniform distribution, forward and backward unpredictability, noncompression using Huffman/Fano methods (Maurer test) etc. Randomized should be tested with a bunch of serious tests to make sure it meets all the requirements.

    Instead, what we likely have here is using default dummy Random() method like one from java which always returns the same value as it always starts with the same seed.

    Given that the randomized is not client specific, the drop I got could not have possible dropped unless there are major flaws in the code.

    If you want to argue further, dig in into details and possibility of above. If you don't want to do that, or you can't, please ignore this thread as it has nothing to do with you and it's goal is to report weird behavior of the game I must report per ToC.

    For example, tell me, what is the probabilities of the following scenarios:
    1. It was a different seed, which somehow ended up with the same sequence of X+5 numbers of from a set of a very wide range. X is a number of random calls to identify the Riddle being dropped and not something else. I could be 1 call in case of a large loot table, or, more likely 2 or more calls. So we talk about at least 7 sequence number.
    2. What is probability of hitting the same seed AND the same sequence number AND no other calls to the method so I can take the whole sequence to get this exact item again.

    Lastly, again, are you 100% sure that this isn't an issue with inventory, an item, a mix of two, or something else so it should not have been reported?

    ---

    Mods, please do some clean-up over here, some guys clearly missed the thread with their drop discussion.

    Also, decide what you wanna do with the thread, as there is nothing to discuss here. The issue is obvious to any software developer, besides one, I guess. Should you/they accept it or reject, there is no point of not having this thread closed.
     
    Last edited: Jun 3, 2017
  11. SillySword

    SillySword Regular

    Long long time ago, I tried to find Keen's items from normal Gorga. If there's any legendary item, I took a screen shot. It's so lucky or unlucky that I had three pairs of same loot, same item and number of coins. I thought the PRNG God of Dracania fell asleep.
     
  12. humunkulus

    humunkulus Forum Apprentice

    How did you get 0.0000000001% ? According my calculation probability should be 0.00000001% and i dont see reason why it shouldnt be possible.
     
Thread Status:
Not open for further replies.