|
|
| Author |
Thread Statistics | Show CCP posts - 0 post(s) |
Garth Mandra
The Southern Legion
The Umbra Combine
139
   |
Posted - 2013.10.07 04:27:00 -
[1] - Quote
This has been noticed before. It's more obvious with ferroscale plates.
The issue is to do with the way the stacking penalty is calculated and it is presumably a bug.
I do not know for sure whether this is just a UI issue or whether it affects in games stats too. |
Garth Mandra
The Southern Legion
The Umbra Combine
139
|
Posted - 2013.10.08 00:05:00 -
[2] - Quote
Grims Tooth wrote:It's not a bug, stacking penalties work the same way in EVE. For stat boosting modules, they are arranged from strongest to weakest with the strongest getting no penalty and the weakest getting the largest penalty. It's the opposite for stat reducing modules, they are arranged so that the largest negative effect is penalized the most.
Interesting. I'd never paid attention to those things in eve.
So let's say we're looking at resistance modules (which had this bug reported ages ago, I don't know if it is still current).
Three resistance mods 30%, 20% and a 10%. Equal to 0.7, 0.8 and 0.9 multipliers.
What I would expect is the 20% to get the first stacking penalty and the 10% to get the second.
However what appears to happen is the reverse. The 10% module get no stacking penalty, the 20% the first and the 30% the second.
What I would expect:
0.7 * (1 - 0.2 * 0.87) * (1 - 0.1*0.57) = 0.545 = 45.5% resists
What may be happening:
0.9 * (1 - 0.2 * 0.87) * (1 - 0.3*0.57) = 0.616 = 38.4% resists
Not to mention should a penalty be getting stacking penalties? And should it be ordered best to worst or worst to best (since it's a bad thing). |
Garth Mandra
The Southern Legion
The Umbra Combine
140
|
Posted - 2013.10.08 03:09:00 -
[3] - Quote
KalOfTheRathi wrote:It is an insignificant amount and is not worth fixing (IMHO). This is most likely tied to their floating point issues. Of which there are several. The display is in integers but the values are floats. Which means you can take a Merc down to 0/0 for Shields/Armor and they can still walk around ... and kill you. But their value is greater than zero as it could be 0.2 Armor.
This aforementioned stacking penalty/bonus calculations combined with their particular decision on handling floats is the reason. Not that their decision is wrong, it just has this side effect that is valid but can be confusing.
I think it is a different issue.
From what I recall (and I think I tested this with maths with the movement penalty a while ago) the stacking penalty for modules that reduce variables (move speed, resistance/damage taken, regulators too I guess) are ordered in the same way as modules that increase variables ie. highest to lowest, in the opposite order to what is expected.
I expect all modules to be ordered in terms of effectiveness and stacking penalties applied in that order.
So with one of each damage mod:
10% = 0.1 = 1.1 [no stacking penalty]
5% = 0.05 = 1.05 [1st stacking penalty]
3% = 0.03 = 1.03 [2nd stacking penalty]
The modules are ordered highest (0.1) to lowest (0.03).
For vehicle resistance mods:
10% = -0.1 = 0.9 [no stacking penalty]
11% = -0.11 = 0.89 [1st stacking penalty]
15% = -0.15 = 0.85 [2nd stacking penalty]
The modules are ordered highest (-0.1) to lowest (-0.15) where they should be ordered from lowest to highest.
This is what I suspect anyway. I'll see if I can do some definitive tests next time I log on.
|
|
|
| |
|