except it isnt. a bad run of rng is how people use confirmation bias to cope
3 Likes
They won’t, not because they’re super well thought out and other studios might steal them, but they would cause a massive outrage if they released them, same for enchanting or any “RNG” system in this game.
“bad runs of RNG” don’t happen 98% of the time.
That’s the problem. And you’ll find thousands of people that agree.
After seeing the above results, how did you still come to the conclusion that there would be a massive outrage when it would say 25% chance in this case?
Where do you get 98% from?
I’m being facetious, but not far off.
I can come here and show numbers of my temper rolls. That would show there is a weighted algorithm.
Just as much as the poster showing there isn’t.
Draw a poll. Ask if people were able to get their 2 tempers more often or less often.
Just go look at how many posts are here about it.
Go look at the youtube videos from pro streamers that notice it who are tempering FARM more items than you and I.
Things are not “25%” as the would seem.
correct, but your imaginary 98% of the time is also due to your confirmation bias. like i have tempered hundred if not thousands of pieces of gear with no problem. but if i brick something because it is frustrating that moment sticks out and you remember it more vividly. so forget the 100 times prior that everything was fine you only focus on that 1 time. tempering doesnt fail 98% of the time
1 Like
What are the possible ranges of each affix that you rolled? I think that is where the problem is. If all of the affixes have the same ranges, i.e. 1-50, then you get a good chance of getting close to an even number of each possible affix.
The issue is on groups that are uneven. This is an exaggerated example.
Health: +1 - +100
Damage: +1 - +5
Now, pick a random number between 1 and 105. You are far more likely to get Health than Damage.
If this is how they programmed it, then it’s a problem. It should be 1) pick a random affix 1-5, then 2) pick a random number in the possible range of the affix that was selected. Then the affixes won’t be weighted.
That theory is fine on paper, but if I am BRICKING, more than I am HITTING, that sticks.
Let’s take a book that has 4 tempers in it. So in theory (and by what the poster concluded) is that your temper would HIT 25% of the time, right?
Confirmation bias then more heavily weighs on hitting something that you had LESS of a chance of doing.
I have a 75% chance to get the wrong one.
I am saying that I have hit wrong ones 5 times in a row (6 if you include the free one) FAR MORE than 75%. Which is why it’s frustrating. I’d be perfectly content with a 75% chance to hit a wrong one. From my data, it is FAR from 75, and more in the 90’s.
Congratulations, you figured out that 75% is 3 times as big as 25%. Of course you will find 3 people before you find 1 that got his desired stats when tempering 1 item. That does not mean the system is not working as intended or that there is some hidden shenanigans going on. Thats exactly whats saying on the tin.
Now compare that to the chance you had to find an item with the right stats in season 3. It has vastly improved.
No, not with a tiny sample size. It does not mean that within 4 rolls you should have the stat you want, not at all.
And thats just confirmation bias of confirmation bias.
1 Like
Nice work.
My impression has certainly also been that the rolls are not weighted. It is hard to tell for sure, but it just seems way more likely than the opposite.
Note: other aspects of item creation and rerolls do seem to be weighted. Only talking about tempering.
My view on Blizzards incompetence has no lower limit, but there is no way they would have designed it like that. Not even from the company that, for reasons, made player stashes shared across everyone in a server shard.
This is true, however +ranks to skills on items, especially amulets and gloves is heavily weighted. Gloves in the old system could also roll a base value of +4 and rank up to +5. Unique items could go even further in a few cases.
1 Like
except it is your confirmation bias, not data. you are wearing atleast 9 pieces of gear. you brick an upgrade to 1 slot a few times but neglect the fact you got the rolls you needed on other pieces because you only focus on your loss. if you actually tracked successes vs. failure you would see it does even out. your confirmation bias is what is blinding you. you believe it is weighted so when you fail you see this as confirmation that it is weighted and that it is designed for you to fail and feel frustration. this is why scientists follow very strict protocol and detail their process exactly when performing experiments, so they can exclude the possibility of the result being influenced by preconceived notions
3 Likes
Congratulations. You took the posters word for it and assume everyting is right, and not people that put THOUSANDS of hours and HUNDREDS of THOUSANDS of tempers.
So, using your logic (and what I recommended in my first reply), the posters tests is “too small of a sample size”.
I’d love to continue the debate with you, but I feel like I’m in a conversation with a living, breathing Dunning-Kruger graph.
you’re right, I’m wrong. That is the only way a debate ends with you.
I mean the data he provided is way bigger than the data you or any streamer provided. Feel free to show us your work.
1 Like
This is why I made my sorc blizzard + FO. I am able to make every temper except frost bolt work at the expense of more focused damage. It won’t clear as high or hit as hard and takes 12+ minutes to clear the same pit level my blight Necro with 5 uniques can do in 4 minutes, but it still works.
Following guides and getting stuck by not having options plays right into the system they have laid out for you. You might win some, but you will lose more. When you gamble the house always wins.
Running my particular blend of mashed up unique builds means I engage less with the annoying RNG systems and more with the game itself, but I have a much lower ceiling. That is the compromise I had to make in the name of fun. Now if only I can get an Uber or two to drop naturally so I don’t have to keep exploiting the iron wolves rep system…
Run this in a category that is not so specific. What I mean is you picked one where it doesn’t matter if one is weighted over the other, the player is clearly using cutthroat abilites. 3 out 4 of the rolls are fine for most. And Attack speed (the 4th) is not really useless if you are not close to cap1 or cap2.
Try it in the one where you have 4 skills to choose from, and select one of the skills in your skill tree and see how many times that does not come up.
While I appreciate OP’s testing, I would be more interested in seeing the success rate of getting the 2 tempers that you want on an item within 7 rolls over a large sample, which is really what a lot of people (including myself) seem to be having a hard time with, in practice.
This implies that there is any reason to believe they would function differently. There is no reason that I have seen to do so.
If someone dies that test and shows differently, I’m all ears. There is no reason to suspect that Blizzard actively wrote a different function to handle one category over another, rather than simply feeding different values into the same function. To do so without data is to seek to justify a confirmation bias - it is attributing malice and effort where efficiency and laziness both lead to the simpler and more likely outcome.
The idea that you offered this as a solution, without data to show there is any reason to believe this is the way it would function, then accuse other people of suffering from the dunning Krueger effect 3 posts later, is unassailably ironic.
Occam’s razor; there is ABSOLUTELY no way this is how the function is programmed, for many reasons. Here are the most prescient:
- Different stats use vastly different scales and would have to be normalized into a single scale first to do so. I.E. crit in a range from 6-8% vs vuln damage in a range from 50 to 100%. Do you think that crit is 1/25th as likely roll as vuln damage? The answer is no
- Some values aren’t even the same TYPE of number - how do you compare a %crit to a +skill?
- Given that you somehow account for the above, what is the scale used to normalize values that don’t end on a whole number? If you’re doing 6-8% crit, how do people roll 7.5% and not +1.5 levels of x?
- What values does the computer use for % values? Since these numbers are almost definitely calculated as decimals behind the scenes, are you saying they are converted to their % whole number, that is used to inform their roll range, and then converted back into 0.x again?
- Even we’re all this true… Why in the hell would they not simply do the very obvious and almost definitely real way the system works, which is very simply resolve the winning stat, then roll the range separately?
3 Likes
Beforehand kuddos by the effort.
And to use a famous motto:
While your intention was praiseworthy, I would be more cautious before taking down other theories, maybe you just oversimplified your report or maybe you have not considered all factors involved.
1 . Did you got notes about how much items you “broke” on trying to get a +Vulnerable weapon?
2 . You rolled all weapons til the end when as soon getting Vulnerable should switch to other that would be the real use case.
3 . Also considering the real use case, a better test would be trying to achieve not 1, but 2 desired affixes as example, Puncture twice cast (remember, also stoping when achieving at least a minimal roll to reproduce user behavior).
4 . Also considering the real use case, would be good just use as target 2~3 GAs weapons.
While your initial study is amazing (as I said) if you really decided to get this as a pet science project, my suggestion would be consider make from this an initial step before lighting the fireworks, also consider human factors as (you are trying to convince highly biased people as the tinfoil me 
)
- On evaluate your post, the first question is how the hell this guy stored 3.000 items on our 696/account item slots (considering the 300 shared stash + 33 * 12 allowed chars, by the way, you can add 24 equiped slots)?
- Is it recorded/registered or should we go just “trust me bro”?
I hope you get this as a constructive feedback, remember that until Newton Physics/Math break when not considering all factors, and you are studying the work from one literally assumed “SYSTEMs Designer”. 
1 Like