Yea I figured we’d make a new post eventually, just gathering data to how to format it first, even though we kind of hijacked this thread. I did find something else interesting though. Druid is the only class to have a Tempered Recipe with 6 affixes on it (aside from the one everyone can use). All the other classes only have 3 to 5 for each of their recipes. Now I’m even more curious to see what the ratio of tempered recipes is for each class alone.
For example is there one class that gets more 3-affix recipes or one that gets more 5-affix recipes, etc. Just gathering my thoughts, ignore this part lol.
I disagree, and I’m not even sure if the tempering manuals are currently intended to have weighted affixes or if it is a bug in some specific cases. If it is intentional, I think they should lean into by explicitly designating certain affixes in a manual as rarer ones. It would be a lot more fun to actually hit those and it would help them design manuals that are more interesting. Like a manual that usually gives you more FO damage or lower FO resource cost, but every once in a while gives you a chance to double cast FO.
why should that even be a thing at all? in my opinion in an arpg a fully perfectly geared character should be somethin to chase for, but nothing you realistically achieve. Hence why you dont “need” that to finish the contents. Its just something for people who like to chase for something they “want”.
at the end of the season you might have a few 3GA items that are rolled nicely.
yea i think itll lead down to a google docs or something like that, where hopefully a few people end up getting together to collect data.
Right that makes sense, but I was saying do it first with a random number generator that you know is weighting things equally, to get a sense of what that would look like. I bet most people’s sense of this is wildly inaccurate.
No, because it doesn’t mean it’s “best in slot” just because you get the affixes you want. Unless you get a perfect 3GA that rolls max values on both affixes you want, there is always a better piece.
Yes, because the nature of Diablo 4 as it was presented to players is a live service game where players are enticed to pay every season. Meaning within that season it should be reasonably quick to reach a high power level to keep even casual players engaged and able to complete seasonal content. It’s detrimental to Blizzard’s design of the game for something grinded for to only happen every hundred hours or so, because remember that their business model wants us playing limited duration seasons, not eternal.
Now personally I think live service Diablo can go to hell and it’s a terrible thing to design mechanics around, but it serves Blizzard’s own purpose to ease up on tempering, whether we agree with it or not.
no need to become agressive, because actually, they would, especially with open trading. Trade your way to get 3 GA items, or in some peoples cases, RMT for it, and temper to your hearts content. Bam, perfect gear in 2 weeks.
Firstly I doubt that such a thing exists, but I could be wrong.
Secondly and more importantly it isn’t required as you can carry out a chi-squared or find the P-value (or some other statistical test appropriate for the data) to determine how likely the given result would be on a fair RNG, i.e. one equally weighted. This isn’t hard, I did it for a dodgy D12 once for the boardgame Dead of Winter where my copy came with a misshapen die.
Comparing to another set of random data even if the RNG is ‘fair’ is just adding another random element and is also more difficult. We know the most likely outcome on a fair RNG and equal weighting - it would be equal outcomes for all affixes because that’s what has the greatest number of possible combinations.
I do agree there!
One thought I’m having, is that it’s a bit like when you finally find your car keys and always think “they are always in the last place I look!!”. Well of course they are because after that you stopped looking.
Similarly here, people will nearly always stop when they get the right affix - it’s always the last one your roll! And just like you looked in many wrong places before you found your keys, you can roll many wrong Tempers before getting the right one. And nobody remembers or gets annoyed when they find their keys in the first place they looked.
Not actually sure how that would affect stats - I don’t think it should. But it certainly does affect ones perception of the randomness involved.
No, this is missing the point. Of course you can calculate the probabilities, but the experience of tempering is a sequence of random outcomes, with different sequences having a different gut feel. You could go through and try to categorize how different sequences might feel the same in order to calculate the probability of each grouping. But at the end of the day the question is about how it feels, and simulating it with a fair random generator will give you a sense of what that would feel like and whether or not the live tempering actually feels different. The tiny discrepancies at the edges of computational pseudo-random numbers aren’t relevant, and just telling people the probabilities wouldn’t help them understand it.
I agree with the general sentiment. Tempering shouldn’t just be a free run at the stat you need, but getting your stat of choice is too hard right now. Bricking good items is painful.
So here’s my suggestion: As I find tempering recipes that I’ve already learned, let me allocate them in a point system that improves the probability of getting the stat that I’m after. Don’t let me push it to 100%, but let me earn my way through playing the game into a higher chance of rolling my preferred stat.
Example: as a thorns barb, I’m trying to stack bash cleave on all four of my weapons. That’s clearly my preferred stat from that tempering recipe. I’m trying to use it over and over as I upgrade gear on my way to top level. As I progress, the cost of missing a roll for it is going to get higher too, because the perfect piece of gear will be harder to find. So by the time I find that really good piece, if I’ve got a higher chance to roll the right stat, then the universe aligns, and I get that hard-earned reward with less chance to brick that piece I’ve worked so hard to find.
This points system would have to be reconfigurable too, since allocating points would be very build-specific.
I have doubts there is any weighting going on. I thought it might be at one point but the more I’ve thought about it, it doesn’t make sense. Unless they were actively scanning your character and making an educated guess at what you want and then weighting against you, they can’t know what is important to you. Are some affixes more desirable? Possibly, but again depends on what you are going for that is desirable.
Take resistances, I’ve rolled duplicates tons of times on them, but which resistance would they weight? They don’t know what you want presumably. Same with all the manuals, they can’t know which affix we are really after except in a very few build specific manuals, again assuming they are even checking what we are doing with our character. No affix is really more important than another. Sometimes I want %damage, other times I want damage to close or crowd controlled.
I just don’t see them trying to weight the affixes myself. That’s not to say there isn’t a problem in the algorithm causing the problem. If I’m honest I get what I’m rolling for far more often than I don’t so I don’t know. It is frustrating though. But I don’t want things changed just to make it artificially easier to get desired rolls. If there is a problem with the code fix it, but otherwise leave it alone.
Yeah, I am kind if with you on this one as I am not sure if the system should be changed. However, considering it changes to have a way to keep what you rolled, maybe a fair compromise is the following:
Equal chances on any roll (if this is not already the case)
When you are getting a new temper you have the option to get the new affix or preserving the previous one. However, if you choose to preserve you can’t reroll anymore that tempering slot.
About the can’t roll the same affix idea, I think it is tricky, for example when using 3 affixes manuals it might become to easy. I would say that if the weights are a thing, the removal of them can mitigate it a lot.
Hey, not bad it could be a Blessing system, maybe instead of points we could bless the item to influence the outcome of the tempering. As you said this should also be gated through playing (uber whispers or some special kind of Crusaders/church missions?).
But once again, I am still not sure that there is a problem in this aspect of tempering.
I see there are two opinions on changing and keeping the current tempering system.
IMO, the tempering system is not good and should change because there is no return point like Masterworking. That’s all!
But I would like to remind one thing that the tempering system is not an issue worth debating because whether or not there is a tempering reset feature, I am sure no one can surpass Pit 150 alone this season. This is the big problem and why don’t some of you give us a chance to have a perfect gear through tempering reset to deal with the Pit?
You make a good point, I certainly have my doubts and think it’s probably not weighted in the case of Tempering and Masterworking. Enchanting, there seems to be a consensus there is some weighting towards the ‘standard’ affixes like +Stat and +Health, and I have ideas why that might actually be correct. For example there might be ‘trees’ where it first selects a branch (+Health or +Skill) and then a twig if there are any (which skill?) and finally the amount.
It did occur to me though, that different affixes may have different numbers of possible outcomes, and depending how the affix is generated that could indirectly weigh for or against certain affixes.
For example, +Passive ranks I have seen are either +1 or +2. But +Health could be anything from, say, +500 to +600. My assumption is that the RNG first determines the affix type and then seperately determines the value. But if that’s not the case, then an affix with only 2 possible outcomes might be less likely to occur than an affix with 10 possible outcomes.
If we had the data it would be easy to carry out a p-value test and put this to bed once and for all. It wouldn’t address possible streakiness but it would almost certainly confirm/refute whether affixes are weighted. But for 1 person to collect it would take quite a while and be boring, plus the mats while not rare are finite so 1 person might not have enough to do thousands of Tempers. And Masterworking would be a whole other thing…
