
Re: Implementation of the IAS nerf
j9y7a7 wrote:
A quick glance at the IAS nerf reveals that the nerf was most likely implemented by lowering the affix ranges in a rather naive way. So if the affix microcode was "push; 20; push; 25; call; rand2; ret;" before it appears that they simply changed it to "push; 10; push; 11; call; rand1; ret;" (it would be nice if somebody could verify this).
I did confirm that this is how it was done. Whenever I process a new patch it gives me a list of deltas that includes any Attribute formula changes.
For example:
[Haste 10] has_row_changed = true | mod_algorithm_01 |
push 20.000000, push 25.000000, func, rand2, push 100.000000, div, return
push 10.000000, push 11.000000, func, rand2, push 100.000000, div, return
j9y7a7 wrote:
The result is that you could have had a 25% roll before and now end up with 10%, i.e., going from a perfect roll to the worst-case roll. This is quite baffling because you would expect that they use the power given by their system! Instead, they could have changed the microcode to "push; 20; push; 25; call; rand2; push; 0.5; call; mul; ret;". My only explanation for this is that they opted for the simple way given that the new ranges are rather small. Still interesting though.
Agreed. Have you confirmed that this actually happened? Although from looking at the formulas, I don't see how it could NOT have happened. After all, all the attribute values are based on the item seed and there's no way that they changed those.