I had not forged a great deal recently, and when I did it was only 1 or 2 items at a time (rift and gleambow totems).
Today I realised I needed some diamond hammers, so went back to my usual, forging 5 hammers at a time (always went with 4 or 5 and the same technique).
I noticed a significant difference in favourable rolls going from single item forges to 5 at a time.
I went with 5 diamond hammers, 2 effect gum then corrupt 2, it took 12 attempts to get devastating damage, then it was attempt 15 before getting all boons that I tried for (before this would take 4/5 at most).
I thought maybe it was just a bad run, so I tried again with 5 diamond slingbows. 17th attempts just to hit devastating damage, using 2 effect gum, then corrupted boon, after 20th attempt I gave up!
After I was suspicious it took so many attempts, so I tried a single diamond hammer, on the second attempt I got the results I was looking for, and just to make sure it was not a fluke, I tried a single shovel and same again, second attempt, devastating damage (8)+all around+heavy duty(7).
I don’t know if anyone else has noticed more favourable rolls on smaller batch forges.
Unfortunately, it’s far too expensive for me to test in larger numbers.
If youd like, i can run a pile of these via the test server and post the results…
Were there particular compounds you wanted looked at jic that makes a difference?
This is the issue with rng based systems. Negativity bias makes it seem like a given process is unfair etc. I bet if you run 100 forges of this on test it would show that it is just random.
It was mostly rolling devastating damage, as that always the first boon I aim for.
But I did notice a difference with the return of items after using deconstruction resin as well, I was much more likely to get mats returned with single forges.
It seemed!
Are we absolutely positive that the test forge runs exactly the same as live?
Quite possibly, it just stuck out like a sore thumb!
And I know it was a small sample, but 2 multi forges in a row needing 15+ attempts followed by 2 single forges only needing 2 attempts, with exactly the same criteria does suggest a difference.
.>> Insert random stuff of how Trundamere hates RNG in the forge <<
and as such it is more likely you are just experiencing a bad bit of RNG, due to the way the law of large numbers can work, when you look at small sample sizes.
And due to the way RNG works, any speculation and testing to prove anything is extremely hard to do.
and since this is a re-run I will link below my prior response
That is the definition of random. You could get 100 fails as you always start at the same place. 100 iterations is actually to small a sample set but would give you a rough idea. To really start pinning numbers you would need more like 1000 iterations. Also if you had gotten great rolls on the first iteration of mass forge would you be posting about it? This is where negativity bias comes into play =)
Nope, we do not know that they are the same. But ive not noticed the difference and I’ve forged quite a lot in both. And its probably safe to assume that a code change which impacts rng in general or the forge itself will cause differences in results. Ive probably forged hundreds and hundreds of items between the two systems. Perhaps there is a relatively cheap method to compare them. A few hundred gums and compounds perhaps… Hmmmm
I forge a good amount on average. (Usually anywhere between 10-40 forges every other night or so) but I usually do smart stacks of 9. Even with the smart stacks I’ll have some that are perfect fist try, and some that take a large 10 tries or so. I see that much difference even within smart stacks on the regular basis, so I’m about 80% confident that it’s just rng. Would be interesting to see though if someone can run a test.
I do know that if you stack a gum up to about 12 it acts pretty reliably as a trait reduction as well. Even with other additive gums (at much less than 12). Its kind of weird to watch a trait or two scaling up to 10/10 when you have other gums applied at the same time.
