test content
What is the Arc Client?
Install Arc

Fudged Item Upgrade Odds

Dear Cryptic,

First, although I've been a member since 9/18, it still took several days to authorize me to use the forums. That's beyond frustrating.

Second, I saved RP items for weeks in prep for the X2 RP event. When I went to upgrade items, I burnt through 10 preservation stones in a row. Not on a 50% item. But on a 90% item. And it wasn't only me. Lots of other people reported this problem. I've never lost a single pres stone on a 90% item. 10 in a row suggests a design. I strongly suspect you fudged the odds so that people will spend Zen on pres stones.

Third, I don't mind spending money on a game. But I do mind it when you fudge mechanics to try and motivate me to spend money.
«1

Comments

  • sockmunkeysockmunkey Member Posts: 4,622 Arc User
    Oh boy, here we go again.
  • urabaskurabask Member Posts: 2,923 Arc User

    Oh boy, here we go again.

    Could just as easily argue that you have unfounded confidence in Cryptic on this.
  • litaaerslitaaers Member, NW M9 Playtest Posts: 871 Arc User
    Just for the record, yesterday I upgraded several R11's to R12's. On two of the tries, the first took only 5 PWards, and the other took 11. On a 3% chance.

    The odds are not just for you. They get spread around to the entire playerbase. I would wager that if they were able to generate a stat list of ALL of the RNG hits in the game (crits and other stuff included, not just RP), it would be a perfect Bell Curve.
  • nameexpirednameexpired Member Posts: 1,282 Arc User

    [...] When I went to upgrade items, I burnt through 10 preservation stones in a row [...] on a 90% item. [...] I've never lost a single pres stone on a 90% item. 10 in a row suggests a design.[..]

    Let me ask how often - before this event - you have "never lost a single pres stone on a 90% item"?
    statistics suggests 90 times. If you have so far only made 20 90% upgrades you should have the 70 90% upgrades remaining.
    On the other hand, I stole at least 30 of your 90% upgrades, which leaves you with only 40 left.

    Imaginary Friends are the best friends you can have!
  • plasticbatplasticbat Member, NW M9 Playtest Posts: 9,667 Arc User
    edited November 2016
    I have my mystic and unscientific procedure to do my RNG stuff. I went to an instance to make sure no other player in it. I hurry when the luck is on me. I slow down when the luck is against me.

    Yesterday, I did something simple and did not cost me an arm and leg if fails. No P-ward involves.
    I upgraded a bunch of enchantment from r5 to r7. That involves 40% and 30% I believe.

    Based on the experience yesterday, I found my luck was on a streak.
    It started on keep on failing. Fail a lot. Like 6 to 7 in a row.
    I took a break. Then, good luck came. It success 10 times in a row. I was also hurrying when it went successful.
    Then, the bad luck came back. Failed 6 in a row again. Then, I stopped. Take a break.
    Then, the luck went 'normal' afterward. One or two fails between success.

    Based on my experience in the past (not just yesterday) I also found that I had the worst luck in 20%. I had better luck to do 10% and 5% than 20%. No, it is not scientific founding (just happens to be like that for me).
  • minotaur2857minotaur2857 Member, NW M9 Playtest Posts: 1,140 Arc User
    I think rare streaks happen too often, there are times when the random number is not regenerated.

    You should take 100+ attempts on a 5% about 1 in 168 upgrades. Having not done 168 upgrades at 5% but had 6 go over 100, I call shenanigans on that.

    I've kept a log of every upgrade I've done in 2xRP for a long time.

    Over many thousands of RNG chances at various odds, I'm consistently taking 25-30% more attempts than I should.

    This 2xRP I got incredibly lucky at 25% took 17 such upgrades before one took more than 4 attempts (18 upgrades, 47 total attempts), but (5% 4 upgrades, 137 attempts) was less friendly. 20%s and 10%s were about 20% worse than average which is insignificant on the number of attempts I had.

    The total 32 successes, expected 227 attempts, actual 273 attempts is fairly consistent with what I would expect from experience, if anything a bit better.
  • oldbaldyoneoldbaldyone Member Posts: 1,840 Arc User
    People will say the odds aren't showing correctly, and people will say they are and that you just got unlucky. People will come around with astronomical odds that you could possibly fail a X% upgrade after X number of tries.

    In the end - nobody knows. Since there is no realistic way to test the system, and the data is not available to be parsed, all we have is their word that the numbers are accurate.

    My opinion is, the RNG gets stuck. best example is, go on a skill node hunt. Note down how many times you get it with the first try, how many times you fail and how many times you failed repeatedly on the same node. Your numbers will look something like this (number of kits used till success):

    1 Kit Used
    1 Kit Used
    1 Kit Used
    1 Kit Used
    1 Kit Used
    4 Kits Used
    1 Kit Used
    1 Kit Used
    1 Kit Used
    3 Kits Used
    1 Kit Used
    1 Kit Used

    Most skill nodes have a 75% success rate, but the numbers won't add up to that.

    Anyways, this comes up a lot and they aren't going to provide any acceptable proof that it is working. Just have to suck it up and understand that in general, the RNG gods will make it up to you eventually.
  • plasticbatplasticbat Member, NW M9 Playtest Posts: 9,667 Arc User
    edited November 2016
    Yes, I have the same feeling about random number is not regenerated too.
    For example, opening those RP bags.

    Using my isolated instance and hurry up when success method, I can get like 6 to 10 green in a row. No, it is not just once or twice. I got it very often. Yes, I slow down when I see a bunch of white.

    No, I don't know what the chance percentage for getting a green should be. However, I did not get anything like that if I opened the boxes in PE.
    Post edited by plasticbat on
  • litaaerslitaaers Member, NW M9 Playtest Posts: 871 Arc User
    A seperate RNG generator for each instance? Not sure that is how it works, but it's an interesting concept... might even have a grain of truth to it....
  • plasticbatplasticbat Member, NW M9 Playtest Posts: 9,667 Arc User
    edited November 2016
    litaaers said:

    A seperate RNG generator for each instance? Not sure that is how it works, but it's an interesting concept... might even have a grain of truth to it....

    Different seed. If different instance is a difference process in the server, it will have different seed because each process has it owns RNG initialized with a seed.
  • samaka#2511 samaka Member Posts: 568 Arc User
    In another one of these threads I do believe it was stated that Cryptic developed their own RNG library instead of using a standard library... that may have something to do with the strangeness of the RNG if true.

    Of course, the nature of RNG and our tendency to look for patterns in chaos could also generate "observer error" as we really have no idea what the RNG patterns are across every single RNG roll being made all day, every day. All we can see our our own rolls, and we have no idea how many numbers are being generated between each attempt at a skill node, for example.

    Also, just because the chance is 50% does NOT mean that if you perform the RNG check twice you are guaranteed to get a positive result. Each individual attempt has a 50% chance to succeed, that is all the percentage chance means. I have had a 50% chance fail 8 times in a row, while my partner has managed to land a 1% chance using preservation wards in less than 10 attempts 4 times, and 1 time where it took almost 20.

    All that said, when upgrading a 90% item, why are you using a ward? The actual upgrade components themselves are cheaper to buy than a preservation ward itself. You could sell the preservation ward on the AH, buy the green upgrade components and still have change left over! :)

  • btairbornebtairborne Member Posts: 352 Arc User
    I failed 52 times in an isolated instance on a 10% roll, so I consider it a myth and superstition.
  • namelesshero347namelesshero347 Member Posts: 2,091 Arc User
    edited November 2016

    I failed 52 times in an isolated instance on a 10% roll, so I consider it a myth and superstition.

    I tried this and I agree. The act of switching instances seems to work most of the time is because it slows you down which I think is more important due to the streaky nature of the RNG. You can lose a lot if you just mash the button very fast.

    I've have done four toons worth of upgrades to R11/R12/mythic. For the low percentage tries, I've had a few that hit early. I even once made two R12s with a single 10-pack of pres wards. Sadly what happens more often are upgrades that took four, five, or more times the expected number of wards to do. These outlier results happen way more often than I think it should. During the last double-RP, my last two upgrades were for R12s which I blew well over three 99 stacks of pres wards on. It was so aggravating. This double-RP, I didn't buy one pres ward; I bought 21 coal wards instead. Probably cost a bit more than the pres ward route, but I figure the premium saves some sanity and time.

  • litaaerslitaaers Member, NW M9 Playtest Posts: 871 Arc User
    No. Less than usual.
  • tripsofthrymrtripsofthrymr Member, Neverwinter Moderator, NW M9 Playtest Posts: 1,624 Community Moderator
    edited November 2016
    I used fewer preservation wards than statistics would predict this time around. If you have a large gaming population rolling dice (by upgrading) you'll have a fair number that get a pretty poor result. You will also get the same number that get results that are good by the same margin. The later group tend to be less vocal.

    I measured the results of 200-ish 10% upgrade attempts a while ago (not during this weekend) and got a result pretty close to statistical expectation.

    My gut tells me the RNG might be streaky (though fair in the mid-term ... streaks of good and bad rolls that do balance out) but I don't have any evidence to back up that theory. To understand how that could happen (not saying it does), check out https://www.random.org/analysis/
    Caritas Guild Founder (Greycloak Alliance)

    Sci-fi author: The Gods We Make, The Gods We Seek, and Ji-min
  • diogene0diogene0 Member Posts: 2,894 Arc User
    During the last double RP the numbers of preservation wards I needed was an anomaly. I even had a 160 wards rank 10-11 upgrade. This time I got really lucky and plenty of 2-3 tries 5% chance upgrades.

    There is plenty of evidence that RNG isn't flawed in this game. If anyone is interested, then I'll explain how and why.
  • minotaur2857minotaur2857 Member, NW M9 Playtest Posts: 1,140 Arc User
    edited November 2016
    I'm a statistics graduate, and the reason I keep the stats I keep is because once you get a large enough sample, you can approximate the binomial distribution with a normal and the confidence limits are reasonably easy to calculate.

    I have no confidence that the stated odds are correct, and I suspect it's because the trials are not independent (ie if the previous attempt failed, the chance of a failure on the next attempt is higher than it should be), hence my suggestion that the RNG is not being regenerated. This lowers the overall success rate because you probably do some mucking around after a success which may reset it but not after a failure.

    Interestingly, I used to play another originally Cryptic game (CoH) and the RNG there was dodgy as hell, I got streaks of 24 successes then 5 failures in a row at a mixture of 75/80/85% chances for example, I wonder if we're using the same one here.
  • mightyerikssonmightyeriksson Member Posts: 842 Arc User
    edited November 2016

    I'm a statistics graduate, and the reason I keep the stats I keep is because once you get a large enough sample, you can approximate the binomial distribution with a normal and the confidence limits are reasonably easy to calculate.

    I have no confidence that the stated odds are correct, and I suspect it's because the trials are not independent (ie if the previous attempt failed, the chance of a failure on the next attempt is higher than it should be), hence my suggestion that the RNG is not being regenerated. This lowers the overall success rate because you probably do some mucking around after a success which may reset it but not after a failure.

    Interestingly, I used to play another originally Cryptic game (CoH) and the RNG there was dodgy as hell, I got streaks of 24 successes then 5 failures in a row at a mixture of 75/80/85% chances for example, I wonder if we're using the same one here.

    ^ The bolded part is pretty much what I have deduced after countless attempts at upgrades.

    Also, something to consider is how the player experiences the upgrade process, I usually take a break after upgrading a large number of enchants, due to being annoyed by the wonky rng.

    By then I am in no mood what so ever to give any real money to Cryptic, but had the upgrade process felt fair and acceptable, I would be in a completely different spending mood...
  • samaka#2511 samaka Member Posts: 568 Arc User
    I really do wonder then if Cryptic rolled their own RNG engine instead of using an established system that would have the kinks already worked out. RNG is not a trivial task to implement correctly from what I've read online about the subject.
  • lldtlldt Member Posts: 197 Arc User


    Interestingly, I used to play another originally Cryptic game (CoH) and the RNG there was dodgy as hell, I got streaks of 24 successes then 5 failures in a row at a mixture of 75/80/85% chances for example, I wonder if we're using the same one here.

    In a controlled internal computer environment, there is no true randomness (you need to introduce an external physical phenomena to achieve that).

    An algorithm that generates a number or sequence of numbers, isn't random at all. The algorithm is always the same. Something is used as a seed to determine which permutation of results are generated from that algorithm (the simplest being a small time interval). In an MMO where an enormous number of results need to be generated, the algorithm can't be overly complicated and resource intensive.

  • santralafaxsantralafax Member, NW M9 Playtest Posts: 2,896 Arc User
    urabask said:

    Oh boy, here we go again.

    Could just as easily argue that you have unfounded confidence in Cryptic on this.
    I see we are in agreement.
  • urabaskurabask Member Posts: 2,923 Arc User

    In another one of these threads I do believe it was stated that Cryptic developed their own RNG library instead of using a standard library... that may have something to do with the strangeness of the RNG if true.

    Of course, the nature of RNG and our tendency to look for patterns in chaos could also generate "observer error" as we really have no idea what the RNG patterns are across every single RNG roll being made all day, every day. All we can see our our own rolls, and we have no idea how many numbers are being generated between each attempt at a skill node, for example.

    Also, just because the chance is 50% does NOT mean that if you perform the RNG check twice you are guaranteed to get a positive result. Each individual attempt has a 50% chance to succeed, that is all the percentage chance means. I have had a 50% chance fail 8 times in a row, while my partner has managed to land a 1% chance using preservation wards in less than 10 attempts 4 times, and 1 time where it took almost 20.

    All that said, when upgrading a 90% item, why are you using a ward? The actual upgrade components themselves are cheaper to buy than a preservation ward itself. You could sell the preservation ward on the AH, buy the green upgrade components and still have change left over! :)

    One of my guildmates was a bit drunk and did that with his relic weapons and had to farm 15 more keys lol.
  • diogene0diogene0 Member Posts: 2,894 Arc User
    hank41 said:

    litaaers said:

    Just for the record, yesterday I upgraded several R11's to R12's. On two of the tries, the first took only 5 PWards, and the other took 11. On a 3% chance.

    The odds are not just for you. They get spread around to the entire playerbase. I would wager that if they were able to generate a stat list of ALL of the RNG hits in the game (crits and other stuff included, not just RP), it would be a perfect Bell Curve.

    You can't trust what fanboy's ^^^ say, they always defend the game no matter what.

    = "I have nothing interesting or relevant to say so I make it personal".
  • lowjohnlowjohn Member, NW M9 Playtest Posts: 1,061 Arc User
    I failed a 5% upgrade 75 times in a row over the weekend. (Including multiple map swaps and delays, not because I thought the RNG was weird but just to avoid frustration: when I fail a lot, I take a break and do something else.)

    Anyway. Failing 75 times in a row on 5% seems like TOO MUCH, right? WRONG. 75 fails in a row on 5% happens about 2.1% of the time - which is to say, we would EXPECT 2% of people upgrading a R10->R11 to fail 75 times in a row. Put another way, "of every 50 people upgrading enchants, one of them is me."
  • litaaerslitaaers Member, NW M9 Playtest Posts: 871 Arc User
    diogene0 said:

    hank41 said:

    litaaers said:

    Just for the record, yesterday I upgraded several R11's to R12's. On two of the tries, the first took only 5 PWards, and the other took 11. On a 3% chance.

    The odds are not just for you. They get spread around to the entire playerbase. I would wager that if they were able to generate a stat list of ALL of the RNG hits in the game (crits and other stuff included, not just RP), it would be a perfect Bell Curve.

    You can't trust what fanboy's ^^^ say, they always defend the game no matter what.

    = "I have nothing interesting or relevant to say so I make it personal".
    Boom.
  • lowjohnlowjohn Member, NW M9 Playtest Posts: 1,061 Arc User

    II have had a 50% chance fail 8 times in a row,

    Failing 50% 8 times in a row is a 1/256 chance - it's actually WAY less likely than failing a 5% 75 times in a row. But for comparison, it's actually pretty close to the odds of rolling an 18 for stats while building a D&D character. And, again, all it REALLY means is that of every 250 players trying that, one of them is you.

    All that said, when upgrading a 90% item, why are you using a ward? The actual upgrade components themselves are cheaper to buy than a preservation ward itself. You could sell the preservation ward on the AH, buy the green upgrade components and still have change left over! :)

    Event currency upgrades are just about the only reason to use wards on high percentages.
  • minotaur2857minotaur2857 Member, NW M9 Playtest Posts: 1,140 Arc User
    lldt said:


    Interestingly, I used to play another originally Cryptic game (CoH) and the RNG there was dodgy as hell, I got streaks of 24 successes then 5 failures in a row at a mixture of 75/80/85% chances for example, I wonder if we're using the same one here.

    In a controlled internal computer environment, there is no true randomness (you need to introduce an external physical phenomena to achieve that).

    An algorithm that generates a number or sequence of numbers, isn't random at all. The algorithm is always the same. Something is used as a seed to determine which permutation of results are generated from that algorithm (the simplest being a small time interval). In an MMO where an enormous number of results need to be generated, the algorithm can't be overly complicated and resource intensive.

    I understand that, but I have plenty of reason to believe from CoH that the exact same random number is being generated several times in a row under certain circumstances (basically the same seed). If it's the same algorithm here it would explain a lot.
  • macaran5123macaran5123 Member Posts: 122 Arc User
    Not to go into how people exploit the Neverwinter RNG, but it's pretty well established what the RNG seed is. That saying, the Neverwinter RNG relies on human interaction to attempt to introduce a more random state to a otherwise computationally weak RNG.

    Consider how _fricking many_ RNG rolls are done server-side every second with people all over the place in combat. As far as I'm aware upgrades/lockboxes/everything uses the same computationally unintensive RNG. In theory it should be almost truly random given the chaotic interactions of human players with the generator. In reality... there at least _appear_ to be some server side deficiencies with the way the RNG is shared.

    Unfortunately to really get a grasp on how the server passes out numbers, one would have to use it in upgrades thousands of times each in different places and situations _on live_. Testing on preview would not be sufficient. The moral of the story is, yes, the RNG is deficient, no, there isn't much you can constructively do about it, and likely, due to the codebase there is probably very little even Cryptic can do to introduce a second RNG just for upgrades without destabilizing the already fragile servers.

    My advice? If you want to approch true randomness to the RNG, do a cupple things:
    1) don't spam upgrade, wait a bit between upgrades to get a fully new RNG seed and calculation.
    2) MAYBE -- and I stress this is a maybe that could very well have no effect at all -- go to a zone with lots of combat, say WoD. Some people report doing so makes the RNG more stable, I for one have noticed no difference, though I don't use it much anyway.

    But at the end of the day, there isn't much you can really do. It's a computationally weak RNG by design. It has to be for the game to work without far more expensive servers. You can blame PWE for not paying for hardware, stripping Neverwinter profit for new game development, anything you want, but at the end of the day you're pretty much stuck with it.
Sign In or Register to comment.