How the 1.10 Pindleskin "nerf" failed and Snapchip suffered

[Luhkoh]

A while back I heard on dbrunski125's stream a claim that that Snapchip was bugged and dropped poorly. Since I had recently seen the tournament on here that included snapchip and @ffs 's theory crafting for fire sorc snapchip runs, I thought I would investigate this further.

It turns out reddit/discord/youtube user ubeogesh is the one who made this claim, and he helped me come up with a lot of the information that follows. He also tested these theories through manipulating the treasureclassex.txt file, and I will include a link to his youtube video summarizing this later in the post. As that could be construed as somewhat spoilery, I wanted to add the disclaimer near the beginning to NOT READ FURTHER IF you don't want to see theories that were confirmed by manipulating the treasureclass file. But I assume this type of testing is how we figured out drop mechanics in the first place.

TLDR: Pindle and Thresh Socket are fine, the nerf failed. But don't bother running Snapchip or Frozenstein.

The Pindle Nerf

So at release of 1.10, blizzard tried, and failed, to nerf thresh socket and pindleskin. They did so by adding a new treasureclass for the two of them: Act 1 Super Ax through Act 5 (H) Super Cx. Their treasure class assignment can be seen in the superuniques.txt file. The one of interest is the TC assigned to hell Thresh and Pindle: Act 5 (H) Super Cx.

The attempted nerf was changing the "Unique" factor in the fifth column to 512 rather than the old value of 800. As @ffs details here, this WOULD greatly decrease pindle's efficiency. However if you look in the picks for TC Act 5 (H) Super Cx, the relevant pick is Act 5 (H) Uitem C. If we then look at that TC, the "Unique" factor is 800. As the ATMA creator notes here, the game uses the maximum value found for this factor when going through the TC tree. So we can conclude from this that Pindle and Thresh are the same as they've always been.

Based on new info from Jarulf and Ruvanal, it has been discovered that Pindleskin and Threshsocket are not as "nerfed' as one would expect. The basic idea that Blizzard had to "nerf" Pindle and Threshsocket was to drastically reduce the odds of them dropping uniques and sets. In order to implement this, a new TC was created for them. Every TC has an associated bonus to the chances of unique/set/rare/magic items dropping and this is where Pindle and Thresh got hit hard.

Every TC is defined recursively in terms of "lower" TCs such that you need to traverse the drop tree to determine what item gets dropped. However, the game keeps a running check of the maximum bonuses. Because the TC that Pindleskin and Threshsocket drop from intersect paths with the TC that Snapchip and Frozenstein drop from, the Pindleskin TC gets it's bonuses upgraded to what Snapchip/Frozenstein would have. This means that Pindleskin drops exactly like those two while Threshsocket has a slightly worse chance than Pindle because his mlvl in Hell is 84 (this affects the chance for a given quality) as opposed to Pindleskin who is mlvl 86 in Hell.

However, this failed nerf had an unexpected consequence. Enter the...

Treasure Class Upgrade

As described in the item generation tutorial this is SUPPOSED to happen when a monster's mlvl is greater than the level of its TC (column 3 in treasureclassex.txt). The game would then see if there is a TC of a greater level within the same "group" (column 2 in the text file) and upgrade the monster's treasure class to that one. This is why mlvl 88's in Ancient Tunnels or The Pit can drop any item in the game despite naturally having a TC specific to their act.

Now I've heard rumored from several places, and I think among them some of the drop calc creators, that superuniques do not upgrade TC's. This is incorrect and I'm not sure where the rumor started. I will describe how this was proven incorrect in the testing section near the end. For now, suffice to say that superuniques upgrade their TC's in Nightmare and Hell just like anything else. However, the algorithm described in the item generation tutorial is slightly wrong.

All monsters have treasure classes, written in monstats.txt file. When you kill a monster in normal difficulty the game simply uses its TC to drop items. However in Nightmare and Hell difficulty the given TC may be upgraded. When determining the TC for a monster, the game selects the highest level TC in the group that is equal to or less than the Mlvl. If there is no group for TC of this monster, than it doesn't upgrade TC. Treasure classes that are included in the selected TC aren't upgraded.

What the game really does is look at the the "next" TC down in the list from the monster's native one. If that "next" TC is in the same group, and is less than or equal to the mlvl, then the game checks the next one. On until the TC being checked is in a different group or the TC lvl is higher than the mlvl.

This seems pedantic, because the TC lvl's in a given group always ascend down the file EXCEPT in group 18 (the ones relevant to superuniques). When blizzard added in the "Super x" classes in the attempt to nerf Pindle and Thresh, they pasted them directly below the old Super classes. So when a monster drops from the regular Super classes (Act 1 Super A through Act 5 (H) Super C), the game does this "next" check for TC upgrades. As Act 5 (H) Super C has a lvl of 96, and no relevant monster has a mlvl of >96, the upgrade check will always stop at or before Act 5 (H) Super C, before proceeding into the the Super x classes, EXCEPT if the monster's native TC is Act 5 (H) Super C. The "next" TC in the list here, has a lvl of 0 (Act 1 Super Ax). Therefore, when a monster's native TC is Act 5 (H) Super C, it will get unfavorably "upgraded" to whatever Super x class comes in just under the mlvl.

What monsters drop from Act 5 (H) Super C? There are a list of 5 in the SuperUniques.txt file, but only two actually made it into the game: Snapchip Shatter and Frozenstein. Since they are both mlvl 86, their TC gets "upgraded" to Act 4 (H) Super Bx, which in effect causes them to drop the same as Infector or Vizier in the CS. Not too good. So that is why Shapchip is bugged, and I wouldn't think we would want to run him in more tourneys. Lastly I'll describe the txt file manipulation that went into proving the statements above.

Testing/Proof

Reddit/discord/youtube user ubeogesh did all of this treasureclassex.txt manipulation testing, and I'm grateful. He made a youtube video summarizing his method here.

Basically he changed it so that the relevant TC's has 100% chance to drop a specific rune, and saw which rune popped out, proving they indeed upgrade as described. Not included in the video, he also tested Dark Elder to show that superuniques who are not Snapchip or Frozenstein do upgrade as intended.

He was also unaware that the pindle nerf had failed and tested this. He set the unique factor to zero in all but one of pindle's TC's in the recursive tree for each one, and he was still always able to drop a unique, confirming that the maximum value is always used. That's it. Hope its useful!

 

[sir_cyclops]

Very interesting! Diablo has been through some crazy patches/changes...

 

[ffs]

@Luhkoh thanks! As mentioned I am not that savvy when it comes to the ins and outs of item generation, so it certainly helps to have others do some testing. I’ll take a closer look once I’m back from my extended vacation but the results of the recent Snapchip-themed tourney are in line with your findings, when Snapchip runners (granted, there weren’t many though) didn’t have amazing results either. I will address that in the 101 as well, as it seems he just isn’t a prime target in the end. It will be some before I get to it but it’s on my list of things to do once I’m back.

 

[Luhkoh]

@Luhkoh thanks! As mentioned I am not that savvy when it comes to the ins and outs of item generation, so it certainly helps to have others do some testing. I’ll take a closer look once I’m back from my extended vacation but the results of the recent Snapchip-themed tourney are in line with your findings, when Snapchip runners (granted, there weren’t many though) didn’t have amazing results either. I will address that in the 101 as well, as it seems he just isn’t a prime target in the end. It will be some before I get to it but it’s on my list of things to do once I’m back.

Sounds great! Hope you enjoy your time off, and we'll look forward to having you back. Big fan of the 101 post. As far as I know (which may not be very far) the Superunique TC upgrades have never been understood very well, let alone that little "bug" from the pindle nerf. So hope it's helpful. I know I found it fascinating!

 

[Carny]

Very interesting finding. Not only does it confirm that Snapshit and Frozenstein are bugged but the big thing it demonstrates is that Superunqies upgrade TCs just like normal Uniques. Unfortunately it doesn't have an impact on the areas where it would matter like CS, so Infector, Vizier and de Seis will still never be able to surpass TC78.
Good Find.

 

[T72on1]

Very interesting !!!

 

[darkstarhub]

While I have to read things like that twice to make sure I have a firm handle on them, they are very interesting. Pindle has never really dropped very good for me, but that is more or less just my luck and RNG. I have never ran Threshsocket much so if he indeed is not nerfed it could be a nice change of pace.

 

[Luhkoh]

While I have to read things like that twice to make sure I have a firm handle on them, they are very interesting. Pindle has never really dropped very good for me, but that is more or less just my luck and RNG. I have never ran Threshsocket much so if he indeed is not nerfed it could be a nice change of pace.

Thresh Socket is indeed not nerfed, however he's only mlvl 84, so he can't drop a few things that pindle can. See here, but as far as i know, he is a good target for the the big 3 (griffons, dweb, fathom), and I'm sure is easier than pindle for some characters. If you do it let us know how it goes!

 

[peti45]

maybe this solve the problem. In the SuperUniques.txt you need to type for snapchip and frozenstein
TC Act 5 Super Cx
TC Act 5 (N) Super Cx
TC Act 5 (H) Super Cx
From
TC Act 5 Super C
TC Act 5 (N) Super C
TC Act 5 (H) Super C
i think this works

 

[Fruit]

Cool research, thanks for posting!

He was also unaware that the pindle nerf had failed and tested this. He set the unique factor to zero in all but one of pindle's TC's in the recursive tree for each one, and he was still always able to drop a unique, confirming that the maximum value is always used.

Setting the modifier value for a certain quality to 0 does not mean that the TC can not drop an item of that quality. I don't think there's a bug here, and Pindleskin successfully uses the "appropriate" modifier values from "Act 5 (H) Super Cx", the TC he drops from. Or if I'm misunderstanding you, then please explain like I'm a child. Edit: Sorry, didn't read properly. You meant that the value is taken from the "Act 5 (H) Uitem C" TC, which sounds plausible.

The rumor that Superunique TCs are not upgraded might have come from the fact that any monster that's flagged as being a boss will not have its TC upgraded, and among these monsters are the Superuniques: Griswold, Radament, Summoner, and Nihlathak. Not that it would matter, because the TCs of these Superuniques are not in any group, so even if they weren't bosses, their TCs still wouldn't be upgraded. This is also the reason why the TC of those Defilers (in act 5) aren't upgraded; these monsters are also bosses.

 

[Luhkoh]

@Fruit Interesting stuff on the background for the incorrect assumption on superuniques not being upgraded. I didn't know that about the defilers, very cool.

And yeah the pindle section was not implying a bug. Only that the effort to nerf pindleskin was so poorly informed that they didn't realize the nerf did nothing because of a regular game mechanic (taking the highest value for unique modifier in the recursive TC tree). I see you figured it out but I thought I'd respond anyways! And this had been corroborated with the ATMA creator's quote in the article, but I was not aware that the way ubeogesh tested this recursive nature was not quite foolproof (meaning I didn't know pindle could drop a unique if the unique modifier was zero). However what he did was set Super CX unique modifier to zero and Uitem C Unique modifier to 1024 (which i believe makes a unique drop every time? but I didn't do this testing so I'm not sure), and did a few pindle runs, and then tested vice versa on the unique modifiers. Since its unlikely pindle would drop a unique several times in a row with the unique modifier of zero, I would imagine it still proves the point convincingly. But tell me if I'm wrong and I'll edit that part. Thanks!

The "bug" would solely be that snapchip and frozenstein get unfavorably upgraded. The failed pindle nerf is just the amusing background to how that happened.

@peti45 Modifying the tc file is not allowed on this forum. You are correct that what you said would make them drop like pindleskin.

 

[Fruit]

@Luhkoh
Appreciate the response!
Not that I don't trust ATMA's creator, but I went ahead and took a look myself, and can confirm (after debugging it) that the highest values are used for quality determination. So Pindleskin starts with Act 5 (H) Super Cx: 512, 654, 972, 1024. Then the next TC is Act 5 (H) Uitem C: 800, 800, 800, 1024. So the first two values are upgraded, while the other two remain the same, and it ends up as: 800, 800, 972, 1024. I didn't know about this. Still learning new stuff about this game regularly.

Setting a modifier value to 1024 will indeed force that quality, so it's fair to say that proves the point after a bunch of runs.

 

[Luhkoh]

@Fruit Fair enough, thank you for confirming it for sure. Positive is always better than nearly positive. And "confirmed" was perhaps the wrong word for ATMA creators post, and corroborated would have been better . I always appreciate your work on stuff like this!

 

[helvete1]

I think there are more monsters who count as bosses. Countess, hephasto, smith comes to mind. I'm pretty sure the countess doesn't upgrade, based on only being capable of Lo rune, while other bosses in the same level can Zod.

Corpsefire too, I don't think he can drop very high despite mlvl 86 iirc

 

[Luhkoh]

@helvete that's just a matter of looking at the TC files. Countess (and the others you mentioned I believe) has her own TC which is in its own group. Upgrades can only happen within the same group. The superuniques whose native TC is "Act N Super..." or somesuch are all potentially available for upgrade.

It's fun this thread is getting some discussion a long time after I posted it. I didn't expect it but am enjoying it

 

[Fruit]

It's fun this thread is getting some discussion a long time after I posted it. I didn't expect it but am enjoying it

I had somehow missed the thread, but luckily it got bumped. Learned multiple new things because of it!

@helvete
True, there are more that don't upgrade, but on top of what Luhkoh mentioned, not all of them are considered Superunique (for example Blood Raven isn't). Corpsefire does gets its TC upgraded though, it goes from Act 1 (H) Super A (TC63) to Act 4 (H) Super A (TC78).