The named item drop rate in this game is broken (with proof)

[Nokowi]

Here's a fun conspiracy theory....the game has a loot balance mechanism, it records how many times a named loot has been acquired. Once named loot has been acquired X number of times in Y time frame then the game may stop providing named loot or significantly reduce the chances. So, we've been looking at this thinking we each get a 33% chance at named loot on elite, when in reality the chest is adjusting itself to arrive at 33%. This is still fairly random, however it would make for a wide bell curve creating outliers on both sides.

What makes this theory interesting for me, is it's purpose in the game. This would make loot "feel special" b/c it would limit how many folks get some type of loot...making if feel rarer & more special in small social circles, like say a Guild

A second purpose for this theory is loot throttling for safety. Imagine SSG accidently released a bit of loot that was way more powerful than intended & folks were rushing to acquire that item...many folks would be over-farming it, but the chest would throttle the items availability, giving SSG time to change the item before to many folks acquire it. Plus if the chest re-rollers all focused on an item, it would drive up SSG profits by making it temporarily harder to acquire and netting SSG more shards.

I believe this was actually a real thing in the distant past for some content. While it's possible something could unintentionally persist, I do not think this is a part of their current design. It is much more likely that players can not comprehend randomness and/or that 'extra things' sometimes dilute the chances from the intended percentage.

 

[5 Other People]

well, I just think the forums needs more spice. you all are quite rational and calm.

carry on!

 

[Vua]

well, I just think the forums needs more spice. you all are quite rational and calm.

carry on!

Only took like 10 months to get there. And let's be honest, I can go from calm and rational one minute to full on douche the next. Though it's mainly for entertainment purposes. So check back in every once in a while.

 

[J1NG]

Meaning, if in ~300 pulls @Cowzrul has a drop rate of ~24%, then it's possible someone has a drop rate of ~42%...unlikely, but possible as we don't know the width of the bell curve.

I haven't done more testing yet as some of the test assistant toons are currently still on the old 32bit servers awaiting Ghost World status to bring them over. But in one quest that was being tracked, this is exactly what has happened, there was a spate of excessive drops (got "lucky" basically) but otherwise the drop numbers have persisted at the same level throughout (went to ~42%). But it is only in this one quest. Other quests remain at the 33% level or thereabouts.

Here's a fun conspiracy theory....the game has a loot balance mechanism, it records how many times a named loot has been acquired. Once named loot has been acquired X number of times in Y time frame then the game may stop providing named loot or significantly reduce the chances.

This doesn't exist in the game. (I can only say "I tested it", do not ask how, but I have)

J1NG

 

[norriskwondo]

I believe this was actually a real thing in the distant past for some content. While it's possible something could unintentionally persist, I do not think this is a part of their current design. It is much more likely that players can not comprehend randomness and/or that 'extra things' sometimes dilute the chances from the intended percentage.

This is a computer programed to operate with specific parameters, designed by people to do so. There is nothing truly random in this game. If it's working with a true odds algorithm as they claim, people should be getting loot they're working for in so many attempts. But their not getting their loot at all in the vast majority of scenarios. Thus, it's an intentional '**** yall' to those that are paying their bills.

 

[Cowzrul]

This is a computer programed to operate with specific parameters, designed by people to do so. There is nothing truly random in this game. If it's working with a true odds algorithm as they claim, people should be getting loot they're working for in so many attempts. But their not getting their loot at all in the vast majority of scenarios. Thus, it's an intentional '**** yall' to those that are paying their bills.

This assumes that every bit of code in the game functions as intended. I'm fairly sure this game is full of code that nobody understands and that has odd and unintended behavior.

 

[Magesty/Hawc/Scaredycat]

I am starting a new thread on this topic because the RNG poll includes a lot of discussion about pseudorandom number generators and particular items that are not the focus of this discussion.

I have suspected there to be an issue with named item drop rates for some time, and decided to collect data on my named item drops.

This data is collected on the four characters I regularly play, all on the same account. (Cowzrul, Cowzrule, Cowzrul-1 and Sneakycow)

All runs were completed on at least Elite, and many of them on R1 or higher. None of these runs include drop rate boosts of any kind.

I only logged drops in quests from Ravenloft, Sharn, Feywild, Isle of Dread, Vecna and Myth Drannor, which the developers have stated should have a drop rate of 33% on elite, with an additional 1% per reaper skull. I did not record specifically which runs were conducted at which difficulties, but my theoretical named item drop rate should be at least 34%, and likely somewhere in the neighborhood of 36% or higher.

I have created this Google Sheet to share the data I've been collecting recently.

https://docs.google.com/spreadsheets/d/e/2PACX-1vRcOSqdHfGgKFIAhm4MsrMm_IA_PlbSTyGh8Sa8d3SApFg2SYun-en9O9gdXdS8pwpXS6rWFL6I6DVJ/pubhtml

As shown, the rate at which I actually receive named items is 24% across 222 chest pulls. This, in itself, is not really compelling evidence of the drop rate being broken. I expect many of you would simply say "well, you're just unlucky". The likelihood of getting this outcome is incredibly low - 0.0464%, or 1 in 2153. There's more than a few thousand players of this game, I could simply be that super unlucky guy who stumbled across that 1 in 2153 chance, right?

Well, when you examine the outcomes in a little more detail, particularly the streaks in which I received no items, it becomes pretty obvious that the theoretical drop rate simply can't be the developer's intended 33%+. The likelihood of these outcomes occurring is so rare as to be unbelievable.

During these 222 chest pulls, I had notably long streaks without a named item of 18, 16, 15, 14, 10 and 11. There were also shorter streaks with no drops, but those are not particularly noteworthy.

The equation for the expected value of quest runs before encountering a particular streak of "no drops" of named items is given by ((p^-n)-1)/(1-p), where p is the probability of not getting a named item, and n is the number of chests in a row with no named item. I used a value of 67% chance of not getting a named item, although my rate *should* be lower than that when you factor in reaper bonuses.

For a streak of 18, this is 4090 chest pulls.
For a streak of 16, this is 1834 chest pulls.
For a streak of 15, this is 1228 chest pulls.
For a streak of 14, this is 822 chest pulls.
For a streak of 11, this is 245 chest pulls.
For a streak of 10, this is 163 chest pulls.

If you add all of these together to get a number of chest pulls, on average, that one would have to pull to observe these streaks within, you get 8384 chest pulls. I experienced all of these streaks in 222 chest pulls. This is a factor of over 37x more likely than expected to experience these streaks without receiving a named item. If this was occurring at a rate of 2x, or 5x, that would be within the realm of reasonable for a very unlucky person's experience. A factor of 37x means that something is fundamentally broken in the game. It does not require many thousands of pulls to demonstrate, as I've just shown. I'm essentially experiencing something with a probability of 1 in thousands, most times that I play the game, that I should be experiencing very rarely. The entire reason that I started logging this data is because I would regularly run entire sagas of expansion content, and get zero named items that dropped to me, so the logged data here is not cherry picked or some unusual streak of bad luck, it is my normal experience playing this game.

If anyone responds to this with "The probability of getting 10 heads flipped in a row is the same probability as any other sequence of coin flips", I will cry, tell you that's stupid and irrelevant, and point you to post #20 in this thread where I explain why. https://forums.ddo.com/index.php?threads/what-is-the-named-item-drop-rate-really.15872/

Eat more beef!

 

[Bobbryan2]

The infinite monkey theorem states that a monkey hitting keys independently and at random on a typewriter keyboard for an infinite amount of time will almost surely type any given text, including the complete works of William Shakespeare.

Some of the arguments on this thread center around the concept of luck, but I'd point out that we are all monkeys randomly getting loot and there should be some folks above the center of the bell curve and some folks below the center of the bell curve.

Meaning, if in ~300 pulls @Cowzrul has a drop rate of ~24%, then it's possible someone has a drop rate of ~42%...unlikely, but possible as we don't know the width of the bell curve.

So, what we could do is get a web-page that let's everyone record their observations. I know something like that is prone to a few bad monkeys intentionally tainting the data, but if you required the user to create a user name (and put a 48 hour window on new users) and only let a user add one observation every 5 mins, then you could begin to collect a very large data pool....the larger the data pool, the more significant and accurate it becomes.

I believe most folks are being honest here (my bias) and we're dealing with a loot system that SSG may or may not fully understand....looking at the state of the game, it's possible there is an issue with loot generation & it's also possible @Cowzrul is an outlier. Only more data can answer.

Yeah... if you guys are citing the infinite monkey theorem, then you're just trying to argue (in bad faith) that you can never know anything.
If one person got a drop rate of 24% and another got a drop rate of 42% with n>300 tries each... that's statistically significant.

No one is looking for certainty. People are looking to know things within a given confidence interval.

Even if the drop rate with 33%, and both people were rolling in the less than 1% possibility in alternative directions, that would still be very valuable data. But it would actually probably imply a problem with the RNG or data accumulation method.

 

[Magesty/Hawc/Scaredycat]

There's no real point to farming loot post-Vecna.

I totally agree with you. I’m not a big farmer anyway I’ve been playing the game since 2007 with a couple of breaks, the most farming I did was during the shroud era when we were running it every three days on clockwork with 8 characters to get your large ingredients. Since that I have rarely farmed, but it’s really futile with the new content that’s come out post dread. I still like the game. I just equip my characters with what I find and it works fine. Happy hunting everyone.

 

[Nokowi]

This is a computer programed to operate with specific parameters, designed by people to do so. There is nothing truly random in this game. If it's working with a true odds algorithm as they claim, people should be getting loot they're working for in so many attempts. But their not getting their loot at all in the vast majority of scenarios. Thus, it's an intentional '**** yall' to those that are paying their bills.

I think most people do not understand randomness. I have heard of professors that ask students for random sequences and they can always spot the student answers from random answers. Random is not 1-0-1-0-0-1-0-1-1. Random is 0-0-0-0-0-1-0-0-0-1. People tend to think random means that they get the average value, which is very far from the truth.

A couple percent drop rate can take 1000+ runs. A 15% drop rate can take 50+ runs.

Whether random design is good is a separate question.

 

[Fizban]

Do you have anything other than data that could have obviously been created?

 

[Zarrenn]

This assumes that every bit of code in the game functions as intended. I'm fairly sure this game is full of code that nobody understands and that has odd and unintended behavior.

The game is built on the guts of Asheron's Call, which was and still is one of the most impressive game engines of all time. The game allowed for a 500 square mile(actual size) world area but still had such low latency it supported "twitch" gameplay(this is not to be confused with the twitch streaming service). Development on AC began back in the mid 1990s.

For those of you keeping score at home that's a 30 year code legacy. There's bound to be some broken bits.

 

[MrBill]

Fallen age and Golden age are on the same loot table, independent from the other loot.

So what your saying is if I manage to roll 100 times on that loot table my odds of getting either a fallen age or golden age are the same?

 

[MrBill]

I believe this was actually a real thing in the distant past for some content. While it's possible something could unintentionally persist, I do not think this is a part of their current design. It is much more likely that players can not comprehend randomness and/or that 'extra things' sometimes dilute the chances from the intended percentage.

I always felt that way about Jibbers and I agree that they probably changed it in favor of diluting the loot tables. I still wonder about the loot tables. Does every item on said table have an equal chance to drop or are they weighted, as for example say on fallen/golden age weapons. Is a golden age a 100 and fallen age 1-35? They're both on the same loot table but the odds are stacked in the fallen age's favor.

 

[Rusty_helmet]

well, I just think the forums needs more spice. you all are quite rational and calm.

carry on!

I was banned for a whole week. I think that explains it.

 

[Buddha5440]

I love how people are trying to explain that a, supposedly, "random" thing is not working the way it should. While it is impossible to create a true RNG, the fact that it is random enough to cause so many 10/15/20/50+ page forum threads about how it is biasedly random or not random enough is just...

.

 

[Vladimir]

Part of the issue is there are so many potential variables that we are trying to navigate without being able to see under the hood.

random anecdote:
This last week I ransacked The Safehold on 30 characters (spread over 6 accounts, yes, I multibox). My goal was to get 2 x Legendary Medium Armor of the Artblade, so I was only paying attention to when I got something form the "rare table". The results: 1 L. Medium Armor of the Artblade (acquired fairly early on), 2 L. Solar Gems of Swift Charging (both acquired with 30 minutes of each other near the end of the week), and 9 L. Docents of the Warblade. NINE!

I know, a sample size of 240 chest pulls is rather small for statistical analysis... but what can I assume from this?
Well, 5% for rare table access, but this was WITH the djinni loot buff on all of them the whole time, so blargh. Probably still within error margins for the low sample size.
As for distribution, I can see 3 possibilities:
1) It is truly random, as true random will sometimes result in clumps of "results".
2) The items have uneven weights. Instead of 1/3rd 1/3rd 1/3rd it's something like 1/4th, 1/6th, 1/12th.
3) There is a complicated issue with how the game generates "randomness", and some factor or variety of factors can cause unintended clumping.

Option 3 is very possible imo, as not only have people seen similar behavior when farming loot in other quests (the one item they want never shows up, but they get the same different item over and over again... then at a later time, or a different person, has the opposite, getting repeats of what the first person wanted, but not the thing they were constantly getting)... but this is something that is known to happen in other games, and speedrunners and TAS creators exploit this knowledge to their advantage by manipulating RNG to get results they want.

I overheard a conversation between two NPC's in the Rusty Nail tavern that the RNG emulation scans your account and compares all possible loot drops with your account information and if something in a particular chest is missing from your account then the RNG will add a bias to the result and prevent that item(s) from dropping. I believe this is the main cause for lag also.

 

[Buddha5440]

I overheard a conversation between two NPC's in the Rusty Nail tavern that the RNG emulation scans your account and compares all possible loot drops with your account information and if something in a particular chest is missing from your account then the RNG will add a bias to the result and prevent that item(s) from dropping. I believe this is the main cause for lag also.

The lag is caused by the faulty logic in the conversation between the two NPC's...don't you know?

 

[GrizzlyOso]

Random number generation in the physical world, lets say using dice is different to a random number generation or emulation with computer coding. Has to do with Quantum mechanics. Interesting observations. When two boxing I have observed an irregularity with named loot drops, not the percentage drop rate but when named loot does drop (first open no re-rolls) I receive the same item on two different accounts. It appears there may be some bias with the event of the chest opening and the item appearing and that being replicated for other players. Notable however, this bias didn't extend to other variables such as mythic / reaper stats only the item itself. Not a big enough sample size for this to be of any use but just an observation I have made.

What is even happening around here.

Google: “what does quantum mechanics have to do with pseudo random number generator”
Google: “quantum mechanics definition”

Then rephrase the whole first part of your post.

 

[Cowzrul]

Do you have anything other than data that could have obviously been created?

This data was created. It was created by playing the game and logging the results of what named items dropped from which quests.

If I was going to just fabricate data with no observations, why would I have only generated a few hundred samples? I could have created tens of thousands of them to satisfy the people that don't understand sampling and statistics. Why would I create a couple hundred and slowly add another hundred or so over time? I don't understand this mentality of "I don't like this conclusion, therefore the data must be fake". Other than that it's a fairly convenient way to disagree with the data without any substance in the absence of a valid critique of the analysis. I don't find it a very compelling argument.