Infernal Pit SpecialChest runs: high probability of Zod and other HRs

[drmalawi]

I think this thread is fascinating, i love how you guys bump into each other. What i as a total science/math noob see is that the question that vmiguli asked himself and others in this thread was good enough to get people to join the discussion. And i think that is the essence of good science (asking a good question and bring it into a public discussion). You have to ask the a good question. And its equally good that people help/push the topic into a direction that helps get the question solved. Ive learned new stuff from reading it.

People have done superchests in A5 long before this thread was started.

Would you say that flat-earthers are doing good science too since they ask "good questions" that makes many scientists provide explanations of the roundness of the earth?

 

[Eriklop]

Well, all in all i think that the question is the earth really round is not a bad question. A negative result as a good result in science right. But i think there are certain rules to get to a good answer ( im no science buff ). And that's why i think its good that you try to be critical of the stuff that is said in this thread. I just looking and enjoying the conversation that's all.

 

[Fruit]

@Fruit Oh man, that is unfortunate indeed. That is the only logical evidence I could come out with having those screenshorts on hand. I thought that different "blocks of codes," as you say, would result in different patterns. In fact, we even don't know now the total amount of patterns since, apparently, some of them belong to multiple Cases. Anyway, thanks a lot for finding that out!
And for Zod, are there 4 different patterns or only 1 for all 6 Cases also?

Well, despite executing different blocks of code, all the cases end up using the same TC, as well as the same item drop function. The main differences between the cases are the evaluations and the decisions, though I'm not sure how to explain that properly without turning it into another big post.

As for Zod specificallly:

Case 1, 2, 3

Spoiler

---------------start_seed: 27589
Greater Healing Potion
Greater Healing Potion
Zod Rune
Greater Healing Potion
Gold
Arrows
Gold
Arrows
Bone Wand (Magic)
Gold
Scroll of Town Portal
Greater Healing Potion
Gold
Demonhide Sash (Magic)
Greater Healing Potion
Gold
Greater Healing Potion
Gold
Rancid Gas Potion
Partizan (Magic)
Scroll of Identify
Key
Gold
Dragon Shield (Magic)
Rancid Gas Potion
Super Mana Potion
Battle Axe (Magic)
Gold
Light Plated Boots (Magic)
Gold
Gold
Gold
Gold
Gold
Healing Potion
Healing Potion
Mana Potion
Mana Potion

Case 4

Spoiler

---------------start_seed: 27589
Greater Healing Potion
Greater Healing Potion
Zod Rune
Greater Healing Potion
Gold
Arrows
Gold
Arrows
Bone Wand (Magic)
Gold
Scroll of Town Portal
Greater Healing Potion
Gold
Demonhide Sash (Magic)
Greater Healing Potion
Gold
Greater Healing Potion
Gold
Rancid Gas Potion
Partizan (Magic)
Scroll of Identify
Key
Gold
Dragon Shield (Magic)
Rancid Gas Potion
Super Mana Potion
Battle Axe (Magic)
Gold
Light Plated Boots (Magic)
Short Spear (Magic)
Gold
Heraldic Shield (Magic)
Gold

Case 5

Spoiler

---------------start_seed: 27589
Greater Healing Potion
Greater Healing Potion
Zod Rune
Greater Healing Potion
Gold
Arrows
Gold
Arrows
Bone Wand (Magic)
Gold
Scroll of Town Portal
Greater Healing Potion
Gold
Demonhide Sash (Magic)
Greater Healing Potion
Gold
Greater Healing Potion
Gold
Rancid Gas Potion
Partizan (Magic)
Scroll of Identify
Key
Gold
Dragon Shield (Magic)
Rancid Gas Potion
Super Mana Potion
Battle Axe (Magic)
Gold
Light Plated Boots (Magic)
Short Spear (Magic)
Gold
Heraldic Shield (Magic)
Gold

Case 6

Spoiler

---------------start_seed: 27589
Greater Healing Potion
Greater Healing Potion
Zod Rune
Greater Healing Potion
Gold
Arrows
Gold
Arrows
Bone Wand (Magic)
Gold
Scroll of Town Portal
Greater Healing Potion
Gold
Demonhide Sash (Magic)
Greater Healing Potion
Gold
Greater Healing Potion
Gold
Rancid Gas Potion
Partizan (Magic)
Scroll of Identify
Key
Gold
Dragon Shield (Magic)
Rancid Gas Potion
Super Mana Potion
Battle Axe (Magic)
Gold
Light Plated Boots (Magic)
Gold
Gold
Gold
Gold
Gold
Healing Potion
Healing Potion
Mana Potion
Mana Potion

As you can see, it's always the same seed, with only minor differences between the patterns.

 

[vmiguli]

@Fruit Thank you again!

I got either Case 4 or Case 5 judging by the Short Spear appearance. But this does not shed any extra light to my hypothesis, unfortunately.

The master seed, usually dubbed "game seed", is the fuel for randomness, and even the map seed is created from (seeded by) this game seed. This game seed is what RTB would've been referring to; a number that is generated by some formula, and as input/seeds it takes a few timers...

...Now, besides generating the map seed, the game seed also generates an "object seed", which is used for a bunch of object-related stuff (mostly initializing certain object types), though in most cases it is the game seed that generates the initial/starting seed for a unit (player, monster, missile, item). Then, whenever a unit needs a random roll done, it is usually its own seed that fuels the random roll; think of things like the to_hit formula, damage rolls, AI checks for monsters, item drops, etc. (Missiles sometimes use the seed of their source rather than their own seed, for example for the to_hit roll.)...

...Now, the famous limited seed of chests is generated/seeded by this "object seed" that I mentioned. This means that on your way to your sparklychest, everything that uses this object seed will end up playing a role in what seed the sparklychest ultimately ends up with. The random roll of 10,000 also is seeded by this object seed...

... it is pseudo-randomness after all, and therefore it is possible that there is a certain bias towards (a) certain sparklychest case(s). This does not seem likely to me, but I wouldn't just trust my own gut feeling on this.

Now, If I understood everything correctly, the algorithm for Zod rune is the following:
Game seed generates the Object seed when I enter the game. This Object seed constantly updates while I am running towards the chest and stops at Object seed #27589 (a lucky one!) when I hit the chest. This Seed generates its own random roll that determines which of 10,000 is chosen to determine the Case for the forthcoming Zod (and this roll is independent from the chest: thus, there is no difference for Abaddon, PoA or IP for this procedure, right?). Then all items are being dropped according to the chosen Case.
So, the total pattern count for Gul+ HRs from the SparklyChests Act 5 is 45+68+54+48=215.
The total Object seed count for Gul+ HRs from the SparklyChests Act 5 is 68 (the maximum for Case 4: all other Cases are being included, is that a correct assumption?)
Therefore, there are 2 KEY processes: firstly, the choice of an Object Seed #27589 which results in Zod, and secondly, the Case scenario roll, which appears not to matter for Zod but might be crucial for other HRs that are present in Case 4 and are absent in some other Cases!
This means that we have 68 Gul+ outcomes out of 65,536 possibilities at the first step, and then part of these positive outcomes will be discarded IF NOT Case 4 is chosen. (This means that most of other units stay unchanged - since it is the same seed after all - but the position of the corresponding HR is being removed from the list in a particular Case; thus, "discarding" this pattern).
So, if the above understanding is correct, the math should look like this:
For Zod: 1 out of 65,536
For Cham: 3/65,536 - 12%*2/65,536 - 30%*1/65,536 - 38%*2/65,536 = 1.7/65,536 = 0.0026 %
This is an identical mathematical operation to what was done earlier (in the table) adapted to the "timing" and "logic" of the RNG procedure after Fruit's explanations.
Please, correct me, if I am wrong!

 

[Edmond Dantes1]

Been away from the game for a while and decided to do a handful of runs for fun.
Run #3 a couple nights ago (players 3, as I was running Meph, AT, Trav, Pindle, etc).

View attachment 14162

 

[Dispo96]

So I'm confused, is this better or worse than LK? xD

 

[sir_cyclops]

So I'm confused, is this better or worse than LK? xD

Worse, unless you want a zod. Even then, act 5 superchests might be better. For cham, trav is better.

 

[vmiguli]

@Edmond Dantes Nice find, and thank you for your screenshot at Players 3!
This actually raises another block of important technical questions, like:
1. Is it possible to have more HR-responsible Object seeds at lower players settings (1,3,5)?
2. Object seed #27589, does it bring Zod at p5 or other ps? Does the unit formation for this seed go similarly or it will be the whole bunch of other stuff? IIRC, this was discussed in the LK thread showing some similarities among different player settings, but only in the "observation" mode without solid conclusions.
3... 4... etc.

So, @Disposition I hope to answer your question when I get closer to better understanding of these random (pseudo-random) processes for the SparklyChest unit generation. Hope, Fruit would eventually comment on my 68 HR Object seed assumption from the previous post when he gets a chance...

 

[Fruit]

Now, If I understood everything correctly, the algorithm for Zod rune is the following:
Game seed generates the Object seed when I enter the game. This Object seed constantly updates while I am running towards the chest and stops at Object seed #27589 (a lucky one!) when I hit the chest. This Seed generates its own random roll that determines which of 10,000 is chosen to determine the Case for the forthcoming Zod (and this roll is independent from the chest: thus, there is no difference for Abaddon, PoA or IP for this procedure, right?). Then all items are being dropped according to the chosen Case.

• Game seed is created.
• Game seed creates object seed.
• Object seed is constantly updated as you move towards your target chest, since this seed is used for many objects.
• When your target chest is finally spawned after you moved close enough, the object seed creates a seed for the chest. It will do a random roll of 65534 and then adds 1 to the result, so the chest will end up with a seed of 1-65534.
• (The object seed is probably updated a few more times before you physically reach your target chest.)
• You click the chest -- now the random roll of 10,000 is done that will determine the "case", and this roll is seeded by the object seed again.
• Now, based on the case, the chest will execute the appropriate code in the chest function. Every time the "item drop function" is called by the chest function, the seed of the chest will be used to determine which TC to continue with or which item to drop, and also to determine the quality of the item (unless it is forced to a certain quality). Etherealness, sockets and item stats are not seeded by the chest.

1. Is it possible to have more HR-responsible Object seeds at lower players settings (1,3,5)?

Technically yes, but in single-player I'm not sure if that actually happens. But that's talking total high runes; if we were talking specific high runes; that's fairly common. For example you may bump up the player number from 1 to 3, and then the total may go up from 20 to 25, but perhaps that p1 has 2 Ber patterns while p3 only has 1 Ber pattern.

Here's a real example that shows that a higher player number does not necessarily mean more high rune patterns. This is case 4 on p5:

       Total count:   68

 Patterns with Zod:    1
               Jah:    4
               Ber:    6
               Sur:   11
                Lo:    6
               Ohm:   13
               Vex:   14
               Gul:   13

Notice how the runes are different (no Cham Rune patterns for example), while the total is in fact the same as on p7?

2. Object seed #27589, does it bring Zod at p5 or other ps? Does the unit formation for this seed go similarly or it will be the whole bunch of other stuff? IIRC, this was discussed in the LK thread showing some similarities among different player settings, but only in the "observation" mode without solid conclusions.
3... 4... etc.

Player number has a big impact on the patterns, and it's very unlikely that you will find duplicate patterns between different player numbers.

For example here's seed #27589 from case 4 on p5, as you can see, the pattern is nothing like the one on p7:

Spoiler

---------------start_seed: 27589
Greater Healing Potion
Gold
Gold
Greater Healing Potion
Greater Healing Potion
Gold
Super Mana Potion
Broad Sword (Magic)
Scroll of Town Portal
Scale Mail (Magic)
Gold
Battle Axe (Magic)
Bolts
Colossus Crossbow (Magic)
Chain Mail (Magic)
Bolts
Scale Mail (Magic)
Arrows
Super Healing Potion
Oil Potion
Gold
Falcon Mask (Magic)
Crystal Sword (Magic)
Flawless Skull
Mage Plate (Magic)

And for example the Zod pattern from case 4 on p5 has a different seed than the Zod pattern on p7, and is a completely different pattern:

Spoiler

---------------start_seed: 34628
Stamina Potion
Gold
Gold
Scroll of Town Portal
Arrows
Gold
Great Hauberk (Magic)
Super Healing Potion
Super Healing Potion
Edge Bow (Magic)
Greater Healing Potion
Broad Sword (Magic)
Super Healing Potion
Super Mana Potion
Key
Greater Healing Potion
Gold
Mace (Magic)
Greater Healing Potion
Demon Heart (Magic)
Hatchet (Magic)
Gold
Gold
Rancid Gas Potion
Zod Rune

 

[vmiguli]

• Game seed is created.
• Game seed creates object seed.
• Object seed is constantly updated as you move towards your target chest, since this seed is used for many objects.
• When your target chest is finally spawned after you moved close enough, the object seed creates a seed for the chest. It will do a random roll of 65534 and then adds 1 to the result, so the chest will end up with a seed of 1-65534.
• (The object seed is probably updated a few more times before you physically reach your target chest.)
• You click the chest -- now the random roll of 10,000 is done that will determine the "case", and this roll is seeded by the object seed again.
• Now, based on the case, the chest will execute the appropriate code in the chest function. Every time the "item drop function" is called by the chest function, the seed of the chest will be used to determine which TC to continue with or which item to drop, and also to determine the quality of the item (unless it is forced to a certain quality). Etherealness, sockets and item stats are not seeded by the chest.

@Fruit Thank you for the detailed explanation!

So, the target chest is being spawned with a seed, thus its future potential outcome has been already defined at the moment of spawning. Now, the full list of all items that is generated by this particular Object seed is, in fact, the POTENTIAL OUTCOME for the chest. Then, after a roll of 10,000 is performed (although, it is not clear if the responsible for this roll Object seed remains the SAME from the "spawning" or is totally different due to its constant updates on the way to the chest) the chosen Case runs the proper code protocol which sorts out the actually dropping final item list from the "potential outcome".
The presented examples of the top rune lists in different Cases make it reasonable to assume that the item list produced by Case 4 is actually the "potential outcome" itself with no edits to it. Apparently, other Cases' protocols lead to shorter versions of the full list, sometimes eliminating the HR, and sometimes - not. (For example, Cases 4 and 5 are identical for the top runes but differ for Ohm). The main conclusion is that, if the above is correct, we have a total of 68 (Gul+) scenarios out of 65,536 and it is all the same for all SparklyChests in Act 5.

Now let us imagine that we work for Blizzard as professional programmers. And for a new patch for D2 we are given a task to DIFFERENTIATE act 5 SCs making one of it exclusively better then the others! Yes, in previous releases all of them were supposedly identical, but now Blizzard wants to make a specific one more profitable. How to do it without inducing major changes to the program?

I am not a programmer myself but I could use logic. From what we know it is impossible to make any preferences for a certain Object seed creation. The chest is spawned, Object seed is created - nothing we can do about it. Abaddon, PoA, IP - all of them will be the same. So, the only opportunity to make an influence on how things work - is the 10,000 roll. If we could trig this roll for a certain chest, then we could influence (or even choose) the Case running code! And it is not that complicated as it seems, we just need to include a parameter, or a function, or whatever in the execution protocol, so it should always yield a "random" number from the certain interval. Like, 1200-3199 for Case 4. I don't think it is too intricate to perform. And by this operation we definitely solve the task of our employee forcing, for example, the Infernal Pit SuperChest to be more productive than the others by running exclusively Case 4 full list item drop.

@Fruit what do you think? Is it technically possible?

 

[NarfBG]

@Fruit
Now let us imagine that we work for Blizzard as professional programmers. And for a new patch for D2 we are given a task to DIFFERENTIATE act 5 SCs making one of it exclusively better then the others! Yes, in previous releases all of them were supposedly identical, but now Blizzard wants to make a specific one more profitable. How to do it without inducing major changes to the program?

I am not a programmer myself but I could use logic. From what we know it is impossible to make any preferences for a certain Object seed creation. The chest is spawned, Object seed is created - nothing we can do about it. Abaddon, PoA, IP - all of them will be the same. So, the only opportunity to make an influence on how things work - is the 10,000 roll. If we could trig this roll for a certain chest, then we could influence (or even choose) the Case running code! And it is not that complicated as it seems, we just need to include a parameter, or a function, or whatever in the execution protocol, so it should always yield a "random" number from the certain interval. Like, 1200-3199 for Case 4. I don't think it is too intricate to perform. And by this operation we definitely solve the task of our employee forcing, for example, the Infernal Pit SuperChest to be more productive than the others by running exclusively Case 4 full list item drop.

For a programmer modifying the source code, it is entirely possible to even force that specific chest to always drop the same item(s) every single time. It's very simple logic and would require almost zero effort.

Why would you imagine anyone at Blizzard would want to do that though? That's an absurd scenario ...

 

[vmiguli]

For a programmer modifying the source code, it is entirely possible to even force that specific chest to always drop the same item(s) every single time. It's very simple logic and would require almost zero effort.

Thank you for the reply!
And modification of a 10,000 roll would also require "almost zero effort"?

Why would you imagine anyone at Blizzard would want to do that though? That's an absurd scenario ...

Well, does the differentiation of bosses of the same level look like an "absurd scenario"?

 

[NarfBG]

Thank you for the reply!
And modification of a 10,000 roll would also require "almost zero effort"?

I don't know what you mean by that, but I don't see why it would be a hard task.

Well, does the differentiation of bosses of the same level look like an "absurd scenario"?

That's not comparable. One is a gameplay variety mechanic and the other is an arbitrary high reward for no apparent reason. And you didn't answer my question ... why would we imagine something like that would ever be done?

 

[vmiguli]

That's not comparable. One is a gameplay variety mechanic and the other is an arbitrary high reward for no apparent reason. And you didn't answer my question ... why would we imagine something like that would ever be done?

Sorry, I do not agree on this. Here is my answer:

1. Logical argumentation.
The sets of monsters that are being spawned in Abaddon, PoA and IP sometimes differ in their kind, making one mini-level more (or less) difficult to progress (depends on the character though). Like, Reanimated Horde in IP - the game-creators might have thought that it would be the most difficult group of monsters to advance, and therefore had settled on better chances (Case 4) for the IP SuperChest. Just an example of such reasoning.

2. Irrational argumentation.
People just have fun.
You remember how many hidden features there are in all sorts of games? Even in early versions of Excel, I remember, there was a hidden F15 flight simulator or something like that. The Cow level itself proves the sense of irony in the Blizzard corporation. And this all is TOTALLY ABSURD from you rational point of view! And here we are talking about making sma-a-all chances of one chest (to drop a HR, for instance) a little better than that of others (not that anybody would notice it anyway, since it could be "felt" only after thousands and thousands of runs in a comparison mode). What is a big deal about it? It could be done easily. Just because they decided to do it for fun. Why not? "Why so serious?" (c)

 

[Fruit]

Something like that is certainly not impossible (in Diablo II), but I'm convinced that chests in act 5 are handled no differently from chests in other acts. I believe I would've picked up on it by now if there was anything extraordinary going on with act 5 chests, as I've gone through the process from chest creation, to initialization, to the chest function, to the item drop function, and I can't think of any other aspect being involved. I think that if chests in act 5 were exceptions, it would've stuck out.

 

[vmiguli]

Something like that is certainly not impossible (in Diablo II), but I'm convinced that chests in act 5 are handled no differently from chests in other acts. I believe I would've picked up on it by now if there was anything extraordinary going on with act 5 chests, as I've gone through the process from chest creation, to initialization, to the chest function, to the item drop function, and I can't think of any other aspect being involved. I think that if chests in act 5 were exceptions, it would've stuck out.

What I was saying, was not the comparison of Act 5 chests with the chests of other Acts, but the difference in those three mini-level SuperChests (SparklyChests) itself. Namely, the roll of 10,000 that defines the Case. Might it be different for those chests?

 

[NarfBG]

Sorry, @vmiguili, but your line of thought just strikes me as wishful thinking.

 

[NorthDakota]

I'm like reading and re-reading your post #130 and not understanding.

I understand that programming wise it would be trivial to alter it so that only the desired case was chosen. But what I'm not understanding is what's the point of bringing that up? Are you asking if there's certain player actions that could influence what case would be chosen? Are you asking a hypothetical question about whether or not the creators of the game put in a special set of actions for players to influence that case choice as a sort of easter egg?

 

[Fruit]

What I was saying, was not the comparison of Act 5 chests with the chests of other Acts, but the difference in those three mini-level SuperChests (SparklyChests) itself. Namely, the roll of 10,000 that defines the Case. Might it be different for those chests?

Nein, this is handled in the chest function: Once the function has determined that the chest is a "sparklychest", then it will immediately start with the random roll of 10,000, and then jump to the appropriate piece of code to execute. There's nothing here that checks which area the chest is located, or what the area level is, etc.

 

[vmiguli]

I'm like reading and re-reading your post #130 and not understanding.

I understand that programming wise it would be trivial to alter it so that only the desired case was chosen. But what I'm not understanding is what's the point of bringing that up? Are you asking if there's certain player actions that could influence what case would be chosen? Are you asking a hypothetical question about whether or not the creators of the game put in a special set of actions for players to influence that case choice as a sort of easter egg?

I guess, a quick summary of this thread should be handy for further discussion.
The point of my inquiry is to find out whether the topic of this thread is sustainable.
Basing on my personal experience, I suggested that the IP SC is more rewarding that the others of the same Act. There are only 2 apparent ways to either prove my point or prove me wrong.
1. Collect enough data for statistical analysis. One person's experience with less than 10k runs is obviously not enough. The best approach, I believe, would be to collect data from a consecutive dual runs in at least two areas for comparison: IP and PoA / Abaddon. My bet is IP will always beat the other areas in a decent sample amount.
2. Get a help from a professional programmer to see if the IP SC behavior is of any difference to ones from PoA and Abaddon. @Fruit is of a GREAT HELP on this subject explaining the basics of programming and RNG functioning by these chests. On one hand, we know it is possible to trig the program, on another - Fruit has just explained the identical working protocol for these chests. I personally don't know how difficult it is or how much time it would take to go through the code of two chests to see if there is actually any unexpected "easter egg". They should work identical - that is what is said - but what is it out there in reality we could know only after a scrupulous analysis of two chests mechanics in a comparison mode. Again, I suggest, if there is any difference - it should be connected to a 10'000 roll run by the chest.

That's not comparable. One is a gameplay variety mechanic and the other is an arbitrary high reward for no apparent reason. And you didn't answer my question ... why would we imagine something like that would ever be done?

Using your logic, @NarfBG would you explain, please, WHY the working protocols for Act 5 SuperChests and Act 5 SparklyChests are totally different?
Why would we imagine something like that would ever be done? (c)