About RNG and drop chances. Lets try to understand how that works

xAarionx

So, I've seen a lot of complains about RNG and chances of Drop, and even suspecting ZoS for messing up the drop chances so they can control what the players are getting, like the peoplçe complaining that got 7 of a single equip in a row, whille the odds of the happening is statisticly almost impossible
I will oppose that idea and try to explain why thats is actually possible, an not that unlikely to happen

Before we start I'd like to address that english is not my native language, so forgive-me for any mistakes and feel free to correct anything that I write here alright? this is an informative topic, and I'd like to get my ideas as clear as possible for everyone.
I'm also a physicist with a masters degree on particle physics, so I studied a lot of quantum mechanics. Therefore, although I don't have much more than the basic knowledge of programing, I understand a quite a bit of statistics and how that works.

So starting the topic, the first thing that I'd like to address is something that may shock some of you, but it's true: "there's absolutely nothing trully random at the universe!"
Note convinced? think thats an absurd? let-me explain it with a simple example:

Think about the most basic random thing that you know: A coin. Throw it on the air as many times you want and you can get "random" results of heads or tails with 50% chance for each correct?
But... what if i could control all, the initial conditions of the problem? If i could control the face that starts upwards, the force that is applied to the coin, the distance that the forçe is applied in relation to the center of mass of the coin. it's diameter etc.

If you could do that you actually could manipulate the results to give the face you want upwards (Lets say for example: tails) 100% of the time! thats awesome!

Lets think another example: a D6 Dice. You have six faces each one with 1/6 of chance of ending up upwards correct? but, again if you could control all the variables relevant to the problem in the initial state of the dice, You could control the results with a 100% of accuracy

So, why we don't do that? Well, it's simple, not only some of this variables are incredibly dificult to control requiring a extremelly delicate apparatus do do that (like in the coin example), but in most of the cases, that's so many variables involved on the problem that is virtually impossible to control (like in the dice example)
That's why e analyse these problems using statitics! any problem thats is studied using probability is done so because the number of variables is so great thet is inpossible to control the system as a whole.

Even in quantum mechanics, the reason why we study it using a probability model is because we are dealing with things so small and so delicate that the very action of measuring it causes changes in the system, thats why it's impossible to analyse it with a more objective and precise model!

So, "you're tripping? why the Hell did you make all that dissertation talking about real life fenomena? what this thing has to do with a game?"

let-me answer that question, with another question:

"If in real life, in the very nature, there isn't a single event that's trully random, why do you think a program is trully random?"}

It Isn't!

Here I leave my field of expertize, and will start to make some assumptions based on basic programming knowledge, anyone who wants to add somenthing or correct something that i say, feel free do do it!

On programming, to make something random, what you really do is create a pseudo-random variable. Something like, that time on the clock on the moment, the number of transistors charged in your processor, the number of seconds that you are online, basically anything that you can't control, and throw that value into an equation. the equaton will give a number between a limited set of results. and each value correspond to a diferent event on the algorythm.



Too dificult? Let-me think of the simplest possible example

lets supose that a certain monster has 50-50 chance to drop between a dagger or a staff
When you kill him, a pseudo-random number is chosen, and dropped to the equation.
This equation is made so all the possible results value that it can give are between 1 - 100

If the result is between 1 - 50, The monster will drop the dagger
If the result is beteween 51 - 100, the monster will drop the staff

That's basicaly, a simplified model of how the game can choose to give you a certain loot. Of course this is just a simplified model, and in thruth the generation of probability can be much more complex than that. The thing is, the core idea remains the same.

And that little thing is where the devils hide, because that equation can involve not only one pseudo-random variable, but as many as the programmer wants it to take. The more variables it uses, the more dificult is to control the outcome of the equation, the more "random" the event is.

That's why certain tricks that people do to control RNG - like reloading your map, reload the stations etc, may actually work! because if the variables are set, for example, the moment you enter the region, reloading your map may give you better numbers with better chances to get the item!

An thats why you can get several of the same item in a row, sorry, but you just got stuck in a single lucky (or unlucky) number.

And also, thats a way for the game developers actually control an item droppage
You can simply set the current numer of the same item on the server as a variable on the equation, on a way that the more of the same item aready exist in the game, the lower the chance of getting it. (gatchas do That A LOT)

So that's it folks, I hope i have been informative, and this topic will prevent certain complains in the future
Anyone that know more about programming feel free to add to the discussion, an even correct-me if i'm wrong.
Thank you.

SirAndy

You are completely overthinking this. I've worked in the gaming industry for many years.

There is not a single game out there that does NOT use weighted loot tables. Not one.
Repeat after me: There is no pure RNG in video games!

And no, i'm not talking about the mostly esoteric distinction between mathematical true RNG and computer generated "pseudo" RNG.
For the purpose of a computer game, your computer is perfectly capable of generating random numbers that are for all practical purposes random.
Especially if you use a more complex algorithm than the default for most OS implementations.
I have used variations of the Mersenne (Twister) in the past with very good results.

But all that aside, again, to repeat my point: There is not a single game out there that does NOT use weighted loot tables

Just let that sink in for a minute and then ask yourself, who is going to benefit from those weights?
You, the player? Or the game, trying to keep you playing?

xAarionx

SirAndy wrote: »

You are completely overthinking this. I've worked in the gaming industry for many years.


And no, i'm not talking about the mostly esoteric distinction between mathematical true RNG and computer generated "pseudo" RNG.
For the purpose of a computer game, your computer is perfectly capable of generating random numbers that are for all practical purposes random.

And how the computer generate this Variable?
He has to take that from somewhere, I assure you.

And I'm not denying that the game don't use weighted values, in fact, my whole explanation is one of the ways of how can they do this

And overthinking about things is actually my job lol

Amottica

Does anyone have a sample size large enough to be meaningful. Not sure sample size would be needed. If there are 10 possible drops I would suggest at least a thousand. If 20 or more drops then significantly more than that.

ThorianB

SirAndy wrote: »

You are completely overthinking this. I've worked in the gaming industry for many years.

There is not a single game out there that does NOT use weighted loot tables. Not one.
Repeat after me: There is no pure RNG in video games!

And no, i'm not talking about the mostly esoteric distinction between mathematical true RNG and computer generated "pseudo" RNG.
For the purpose of a computer game, your computer is perfectly capable of generating random numbers that are for all practical purposes random.
Especially if you use a more complex algorithm than the default for most OS implementations.
I have used variations of the Mersenne (Twister) in the past with very good results.

But all that aside, again, to repeat my point: There is not a single game out there that does NOT use weighted loot tables

Just let that sink in for a minute and then ask yourself, who is going to benefit from those weights?
You, the player? Or the game, trying to keep you playing?

Its pretty obvious when you are on a new character. Whatever weapon you are using almost never drops. I will get 100 great swords before i get a single lightning staff for example. When i made my necro i didn't get a RNG drop lightning staff from level 23 up to level 50 and it was a SLOW leveling experience as i was doing a mix of daily quests and storylines. I got every other weapon 10 times over during the same period. I couldn't even get a white lightening staff to drop. About 10% of my weapon drops were resto staffs though. Got a lot of ice staffs. I was so frustrated as soon as hit 50 i just bought my CP 160 staff. I was going to farm it myself but i wanted to get it before TES 10 came out.

SirAndy

Again, all these discussions are meaningless once you realize that the game is only interested in giving you a stick with a carrot dangling from the end of it.

Playing a stamina character? Your RNG is weighted towards magicka drops.
Playing a magicka character? Your RNG is weighted towards stamina drops.


Code like that is common place in games.
I literally wrote thousands of lines of code dedicated to giving you the impression of having a fair chance to get something you want.
But it's never "fair". It's never "pure" RNG.
So any discussion about how computer RNG works is always doomed from the very start because it never reflects how games actually use RNG.

Again, i have worked in the gaming industry for many years as a programmer.
Just get me drunk and i'll chew your ear off for hours with all sorts of anecdotes.

Elsonso

xAarionx wrote: »

but... what if i could control all, the initial conditions of the problem? If i could control the face that starts upwards, the force that is applyied to the coin, the distance that the forçe is applyied in relation to the center of mass of the coin. it's diameter etc.

If you could do that you actually could manipulate the results to give the face you want upwards (Lets say for example: tails) 100% of the time! thats awesome!

Yeah, but then some @^#% butterfly in Africa flaps it's wings and your coin lands balanced on the edge.

Saucy_Jack

To the OP: quantum mechanics, not to mention the Heisenberg Uncertainty Principle, would like a word with you.

xAarionx

Saucy_Jack wrote: »

To the OP: quantum mechanics, not to mention the Heisenberg Uncertainty Principle, would like a word with you.

The Heisemberg uncertainty principle comes EXACTLY from the fact that measurements change the state of the particle
Just like that
That doens't mean that nature itself is like that, that just means that when we intervene we lose information about the system. Quantum mechanics actually adds this loss of information to the mathematical model as the Uncertainty Principle.

Diminish

SirAndy wrote: »

You are completely overthinking this. I've worked in the gaming industry for many years.

There is not a single game out there that does NOT use weighted loot tables. Not one.
Repeat after me: There is no pure RNG in video games!

And no, i'm not talking about the mostly esoteric distinction between mathematical true RNG and computer generated "pseudo" RNG.
For the purpose of a computer game, your computer is perfectly capable of generating random numbers that are for all practical purposes random.
Especially if you use a more complex algorithm than the default for most OS implementations.
I have used variations of the Mersenne (Twister) in the past with very good results.

But all that aside, again, to repeat my point: There is not a single game out there that does NOT use weighted loot tables

Just let that sink in for a minute and then ask yourself, who is going to benefit from those weights?
You, the player? Or the game, trying to keep you playing?

Couldn't agree more. In direct relation to the OP's post, and using their coin example...

Loot Table | Weight
Heads 90
Tails 10

SUM of the item weight
100

Provide a range for each outcome
Heads 10,100
Tails 0,9

Generate a pseudo-random number within the max bounds of your range (i.e. the sum of the overall weights in the loot table)
* if the random number is 0-9, then the result is tails, if the number is 10-100, then the result is heads

Pretty basic example of a weighted loot table that can be quite easily manipulated to control outcomes. As you can see, it is not as simple as flip a coin 10 times, and you always have a 50% chance of either outcome... in the above example you can see how this has been manipulated to produce a 10% chance of landing on tails, and a 90% chance of landing on heads... even though there are only 2 possible results. Sure, you as the coin flipper see this for what it is, a fair chance at either outcome. Which is exactly what a game developer wants you to think when you are chasing their pixels for hours on end; it pads their numbers. If anyone thinks ESO does not have a system in place for this then you are bat **** crazy. If anything, theirs are much more complex loot tables, and likely use a seed state (or even multiple seed-states) which could be calculated from literally ANYTHING... could be player based, could be character class, could be multiple factors per character or even per account, RAM state, frames, the list goes on and on.

xAarionx

I'v played Gatchas people, I have no illusions that the drop rate is weighted, I just wanted to throw a discussion about that matter (and in fact learning a little bit more about programming) end also help people to avoid empty complains

Nevertheless ther is some possible experiments that we can actually make to test wich variables are invoved in the weighting. Although, they'll require some patience

Let's think about one possible weighting value, lets just say, subscription
We need o take a group of, lets say, 40 people, all of the same class and same character type (magicka, stamina health)
20 of them must be ESO+ members (it's the experimental group), the other 20 must not (Control group)

We dvide each of them if 5 party members of 4 people each, and all of them try to do a single dungeon a many times as possible (like, lets say, 200 times? i think it's enough to substancial evidence) Then compare the drops of each group.

How many staffs the Experiment group has, how many the control group got

That way not only we can discover if Being a subscriber is a weighting variable, but we can also estimate by how much is that weigh and also have empirical evidence of the RNG manipulation

We may repeat this experiment to anything that we think it's a variable, just changing the experiment group and the control
For exemple, set one group for stamina, and other for magicka. Do the runs and compare the results

Not sure if anyone would have the patiance to do it but i think it would be fun to see these values - and who knows, knowing this can actually give us a hint about how to manipulate it in our favor =P

Ackwalan

Would you like to test the RNG in ESO for yourself? Enter the PvP zone, go to the kill player quest board, get a kill player quest delete and get a new quest, repeat. The kill player quest is random for which class you get (- any class you have already done that day). Keep on getting a quest and deleting it, you quickly see that not all classes get an equal pull. Furthermore, you will get groups of patterns that are just to repeatable to be by chance. You will also see it bug out and get same class 10+ times in a row. As a bonus, you will get a different result when you are the only person there compared to when others are also using the board.

There are some serious flaws for whichever RNG program ZOS uses.

zaria

xAarionx wrote: »

SirAndy wrote: »

You are completely overthinking this. I've worked in the gaming industry for many years.


And no, i'm not talking about the mostly esoteric distinction between mathematical true RNG and computer generated "pseudo" RNG.
For the purpose of a computer game, your computer is perfectly capable of generating random numbers that are for all practical purposes random.

And how the computer generate this Variable?
He has to take that from somewhere, I assure you.

And I'm not denying that the game don't use weighted values, in fact, my whole explanation is one of the ways of how can they do this

And overthinking about things is actually my job lol

Pseudo random number generators has some problems in special cases like cryptography where you encrypts gigabytes of data and you might find an pattern.

For games no, ESO and other online games has the extra random layer of lots of people playing and need random numbers, did my attack crit, that loot did I get from the mudcrab. Probably used in the background for spawns and AI behavior.

The issue is the loot tables, it might also be how gear is given out but back then I farmed for SPC I never got any jewelry outside of the dungeon quest named necklace.

Diminish

Ackwalan wrote: »

Would you like to test the RNG in ESO for yourself? Enter the PvP zone, go to the kill player quest board, get a kill player quest delete and get a new quest, repeat. The kill player quest is random for which class you get (- any class you have already done that day). Keep on getting a quest and deleting it, you quickly see that not all classes get an equal pull. Furthermore, you will get groups of patterns that are just to repeatable to be by chance. You will also see it bug out and get same class 10+ times in a row. As a bonus, you will get a different result when you are the only person there compared to when others are also using the board.

There are some serious flaws for whichever RNG program ZOS uses.

FYI, you can do this when farming the Abah's Watch motifs as well. You can essentially manipulate the type of quest you are given from the quest board, as some are much faster than others to complete. They may have patched this by now for all I know, I haven't done one of these quests since shortly after they released. Made a small fortune from selling the motifs while they were new though, lol.

DarcyMardin

So when one is farming for magicka sets, one should use a stamina alt and vice versa?

deleted221106-002999

DarcyMardin wrote: »

So when one is farming for magicka sets, one should use a stamina alt and vice versa?

Not necessarily...

It depends on how zos have weighted the drops.

Having said that, there seems to be an abundance of anecdotal 'evidence' that suggests that the staves drop for non-light armour wearers etc but I expect that's just observational bias.

And having said that, the times I have seen burning spell weave staves drop, it's been from heavy armour wearing tanks.

Best optimisation strategy, in the absence of defintive information regarding the weighting of game rng, is simply to increase number of repetitions, either/or in more runs/tries and/or with more in group. Which is obviously by design.

And four-leaf clovers, rabbit feet, stirring tea widdershins......

twev

DarcyMardin wrote: »

So when one is farming for magicka sets, one should use a stamina alt and vice versa?

That will slant the table, slightly.
Another way to think it through is to play a stam character AND a mag character, and swap gear when it suits you.
Actually using the gear after you swapped would just put you back in the normal weighted RNG table, tho.

deleted221106-002999

xAarionx wrote: »

Saucy_Jack wrote: »

To the OP: quantum mechanics, not to mention the Heisenberg Uncertainty Principle, would like a word with you.

The Heisemberg uncertainty principle comes EXACTLY from the fact that measurements change the state of the particle
Just like that
That doens't mean that nature itself is like that, that just means that when we intervene we lose information about the system. Quantum mechanics actually adds this loss of information to the mathematical model as the Uncertainty Principle.

Actually, it[edit](heisenberg uncertainty)[/edit] results mathematically - the more precision you can apply in calculation of any given property of, for example, a particle, blurs the precision for other properties of that system/particle.

So, you could know with pretty amazing accuracy the position of a particle but, as a consequence of the maths, you lose detail on energy etc.

What you're referring to is more about superposition (the property of any system to exist in all possible states until 'locked' by observation), best illustrated by Schrodinger's eponymous moggy.

edit: spelling/removing 1 multiverse variant of my post
edit: further disambiguation/universe excision

Zodiarkslayer

SirAndy wrote: »

Again, all these discussions are meaningless once you realize that the game is only interested in giving you a stick with a carrot dangling from the end of it.

Playing a stamina character? Your RNG is weighted towards magicka drops.
Playing a magicka character? Your RNG is weighted towards stamina drops.


Code like that is common place in games.
I literally wrote thousands of lines of code dedicated to giving you the impression of having a fair chance to get something you want.
But it's never "fair". It's never "pure" RNG.
So any discussion about how computer RNG works is always doomed from the very start because it never reflects how games actually use RNG.

Again, i have worked in the gaming industry for many years as a programmer.
Just get me drunk and i'll chew your ear off for hours with all sorts of anecdotes.

Here. Have a beer on me. 🍺
And another one. 🍺
😄

zaria

Souterain wrote: »

DarcyMardin wrote: »

So when one is farming for magicka sets, one should use a stamina alt and vice versa?

Not necessarily...

It depends on how zos have weighted the drops.

Having said that, there seems to be an abundance of anecdotal 'evidence' that suggests that the staves drop for non-light armour wearers etc but I expect that's just observational bias.

And having said that, the times I have seen burning spell weave staves drop, it's been from heavy armour wearing tanks.

Best optimisation strategy, in the absence of defintive information regarding the weighting of game rng, is simply to increase number of repetitions, either/or in more runs/tries and/or with more in group. Which is obviously by design.

And four-leaf clovers, rabbit feet, stirring tea widdershins......

Trying to figure out that gear player have and use this in drop calculations would complicate stuff a lot who increase chance of bugs and could easy be counter productive especially once players figure this out.

Now something like crown crate logger for other drops would be interesting.

Kwoung

Ackwalan wrote: »

Would you like to test the RNG in ESO for yourself? Enter the PvP zone, go to the kill player quest board, get a kill player quest delete and get a new quest, repeat. The kill player quest is random for which class you get (- any class you have already done that day). Keep on getting a quest and deleting it, you quickly see that not all classes get an equal pull. Furthermore, you will get groups of patterns that are just to repeatable to be by chance. You will also see it bug out and get same class 10+ times in a row. As a bonus, you will get a different result when you are the only person there compared to when others are also using the board.

There are some serious flaws for whichever RNG program ZOS uses.

We actually saw this yesterday. There were 5 of us in comms trying to pull Brindle, every one of us pulled the same quest every time, and when it changed, it changed for all of us and we all continued to pull the same quest until it changed again. We finally ended up pulling Brindle and didn't even need to share it, as all 5 of us got it at the same time. It looked a lot more like a cycle the board went through giving out quests than any sort of random generation of them.

The RNG for items could be as simple as that as well, it is 12:03, Ice Staffs are dropping now until 12:17, at which time flame staffs will drop for 1 minute before bows come into rotation for the next hour. The illusion of RNG can be done in many different ways.

remosito

if we'd really want to get to the bottom of it we could. with raw data from pc.

1. addon
a. logs all 4man and trial weapon drops.
b. optionally character info that might be relevant (tank, healer, mag, sta, armor weight, weapon types)
c. optionally uses group loot to log rest of party drops as well
(not sure if b and c can both be achieved (get char info about party members)

2. uploader client
a. Parses the logs created by addon and
b. uploads new data to server

3. Server
a. Listener that receives data from clients
b. analyzes data and updates statistics with new data
c. webserver that displays said statistics

Pretty much exactly like what TTC does with auction house data...just with weapon drop data.


(I'd be pretty useless for 2. and 3. but could chip in on 1)

Mayrael

There is not such thing as random number in programming, its always pseudo-random. Next thing is that you would have to use Gauss-Laplace distribution to check if numbers gathered during tests are really random or there is somekind of other algorithm used, and we would need a freaking lot of data to do this.

Zodiarkslayer

Tl;dr: don't jump to conclusions!

I am German and I do not have an extensive vocabulary when it comes to programming or physics. But I have participated in several projects to decode RnGs and loot tables. All were successful in the end. The first was for the original Final Fantasy 12 on PS2. Way back when.

From my experience you have to validate the exact loot process first (mulistaged?), then the effectual variables and only after that look what the impacts are when you change any of the isolated variables. I think many people jump to conclusions here.

I guess that, when an instance is created on the server side, it includes a fixed loot table for all players in the instance, that doesn't change until you leave that instance. This has several advantages over perpetually changing tables. Server/Client exchange rate, for example. That table will be choosen based on things that can NOT change until you leave said instance. Attribute Point dispensation for example. Or choosen Morphes even.

My other experience is that one part RnG is always a system generated number. Likely on the server side. The other is a table of numbers, that consists of lines and columns. And then, depending on player actions, the reported field (RF) is changed/shifted by a fixed factor. Like everytime a light attack happens tge RF shifts one line down. Or everytime an ability happens the RF shifts a column to the right. You get the idea.
Looting chests and drops will likely not change it, but still require a rng read out.

Now the thing with weighted lists is simply to change the loot dispensation via the loot table, instead of the whole RnG process.

Following that knowledge/insight/argumentation it would only be possible to influence drop chance for any specific item, by isolating what factors are influencing the choice of loot table, not the RnG itself. That will probably be a fixed algorithm for any given type of gameplay (overland/dungeon/trial).

I would personally only programm an algorithm, that reads out attribute points, adds them up and divides them by player number. Then look if I have more than 50 in any attribute and choose a loot table for that specific attribute. Now if a Stam Character plays in a group of only Mag Characters he would still get only Mag drops. Or vice versa. That would fit peoples anecdotes.

cyberjanet

To understand RNG, you need to read this book:

https://www.goodreads.com/book/show/32509131-dungeon-born

The dungeon decides, and the dungeon has a sense of humour.

JTD

Do the developers even know the drop rates/percentages of items in this game? (a genuine question) As I've never seen loot tables like in WoW or FF or other mmorpg's.

zaria

Mayrael wrote: »

There is not such thing as random number in programming, its always pseudo-random. Next thing is that you would have to use Gauss-Laplace distribution to check if numbers gathered during tests are really random or there is somekind of other algorithm used, and we would need a freaking lot of data to do this.

Again the random numbers are more than random enough for game use.

The issue is weighting. Say you buy an 5 key container from an undaunted quest giver.
You expect its 50% chance of getting one of the sets and 1/3 chance of say medium weight.
However some monster sets who are seen as good has an very low chance of dropping getting 6 of the other set in a row has an 1.6% chance.
I guess the real chance here is more like 80/2o.

For arena weapons we don't know how they do this: some say you fist get weapon type, then weapon withing that type:
DW, 2H, S&B, destruction staff, bow or restoration staff. Last has only one type, while you have 4 one handed weapons and 3 2H and destro staffs.sword and shield is another set

This makes some sense, for dungeons I guess jewelry is an additional type and the shield is an type without weapon.
And think many sets has tweaked values.

Mayrael

zaria wrote: »

Mayrael wrote: »

There is not such thing as random number in programming, its always pseudo-random. Next thing is that you would have to use Gauss-Laplace distribution to check if numbers gathered during tests are really random or there is somekind of other algorithm used, and we would need a freaking lot of data to do this.

Again the random numbers are more than random enough for game use.

The issue is weighting. Say you buy an 5 key container from an undaunted quest giver.
You expect its 50% chance of getting one of the sets and 1/3 chance of say medium weight.
However some monster sets who are seen as good has an very low chance of dropping getting 6 of the other set in a row has an 1.6% chance.
I guess the real chance here is more like 80/2o.

For arena weapons we don't know how they do this: some say you fist get weapon type, then weapon withing that type:
DW, 2H, S&B, destruction staff, bow or restoration staff. Last has only one type, while you have 4 one handed weapons and 3 2H and destro staffs.sword and shield is another set

This makes some sense, for dungeons I guess jewelry is an additional type and the shield is an type without weapon.
And think many sets has tweaked values.

Yes but coders very often go easy with random number generators and use some stupid ways of initialising them (most common was using current time so we would have always different initilialisation number - but this has some issues to) thats just a one thing to check. Next thing is that despite being as close as possible to real random number I can't believe RNG is totaly random in ESO. The amount of some set parts drops to others simply don't cover with mentioned Gauss-Laplace distribution - deviations are to big to be random (but to prove it we would need to cover it with hard data). It's a bit of a reverse engeneering trying to find the right algorithim this way and is far from simple, sometimes things we wouldn't think of can have impact on drop chance.

Varana

But that's the point people were trying to make:
The larger part of non-randomness in ESO drops is most probably not due to how the RNG is initialised, but due to how the resulting outcome is applied.
I.e. the random number is for all intents and purposes truly random. But it isused to choose between non-equal outcomes. Determining how the RNG is set up, doesn't help a lot for determining drop chances. Yes, you might get a slightly different drop at a different time (for instance) but the inequality is not created at that level, it comes later. Like whatever you do to change your RNG outcome, you will never drop the BSW Inferno staff because that isn't even in the range of possible drops for you.

Elsonso

Varana wrote: »

But that's the point people were trying to make:
The larger part of non-randomness in ESO drops is most probably not due to how the RNG is initialised, but due to how the resulting outcome is applied.

Yes. All games bias the results to meet game design. Otherwise, opening a chest would result in the same chance of getting a lump of coal as it would the Prismatic Greatsword of Invulnerability.

What is really at hand is how much ZOS uses a bias that responds to the in-game situation. This can appear in quite a variety of different ways, and for different purposes.

Blade users getting staff weapons is one such case. Maybe they do this to promote grinding, as this is a key design point for MMO games. Maybe they do this to promote player-to-player trading. Maybe there are cases where it is both. Evidence that ZOS does this is anecdotal, at best.

Practical reward limits are another case. If the game design is such that certain items are limited across the server, once that limit is reached, the item is no longer available until the number drops. I can see where this can also help prevent "random" flooding of "rare" items over a short period of time, like with event loot that appears uneven over the course of the event. I have long suspected that ZOS does this for events, and possibly very rare crafting supplies.

There are others, but they increase in terms of "slimy business practices" from here.

zaria

Mayrael wrote: »

zaria wrote: »

Mayrael wrote: »

There is not such thing as random number in programming, its always pseudo-random. Next thing is that you would have to use Gauss-Laplace distribution to check if numbers gathered during tests are really random or there is somekind of other algorithm used, and we would need a freaking lot of data to do this.

Again the random numbers are more than random enough for game use.

The issue is weighting. Say you buy an 5 key container from an undaunted quest giver.
You expect its 50% chance of getting one of the sets and 1/3 chance of say medium weight.
However some monster sets who are seen as good has an very low chance of dropping getting 6 of the other set in a row has an 1.6% chance.
I guess the real chance here is more like 80/2o.

For arena weapons we don't know how they do this: some say you fist get weapon type, then weapon withing that type:
DW, 2H, S&B, destruction staff, bow or restoration staff. Last has only one type, while you have 4 one handed weapons and 3 2H and destro staffs.sword and shield is another set

This makes some sense, for dungeons I guess jewelry is an additional type and the shield is an type without weapon.
And think many sets has tweaked values.

Yes but coders very often go easy with random number generators and use some stupid ways of initialising them (most common was using current time so we would have always different initilialisation number - but this has some issues to) thats just a one thing to check. Next thing is that despite being as close as possible to real random number I can't believe RNG is totaly random in ESO. The amount of some set parts drops to others simply don't cover with mentioned Gauss-Laplace distribution - deviations are to big to be random (but to prove it we would need to cover it with hard data). It's a bit of a reverse engeneering trying to find the right algorithim this way and is far from simple, sometimes things we wouldn't think of can have impact on drop chance.

I thought all used the compilers random generator function. Know C# and vb has it.
However ESO has an added random scrambler, its lots of other players on the server who also ask for rng all the time, all the attacks need an crit check and all the loot even from plants need checks.
You will not get the next number in the list if you request the next number some seconds later as other players used it.

Its lots of myths here, member in elder scroll 4 Oblivion it was an common belief your luck stat affected the quality of loot.
So many cased an fortify luck spell just before opening the door to an delve and created it.
I tested it with an mini mod, simply added some chests with huge number of items inside an delve and no difference in quality if luck was 0 or 100.